close

Вход

Забыли?

вход по аккаунту

?

From Microwave, Infrared to UV: Probing the Conformational Preferences for Biomolecules with Intramolecular Hydrogen Bonds

код для вставкиСкачать
FROM MICROWAVE, INFRARED TO UV——PROBING THE
CONFORMATIONAL PREFERENCES FOR BIOMOLECULES
WITH INTRAMOLECULAR HYDROGEN BONDS
by
Di Zhang
A Dissertation
Submitted to the Faculty of Purdue University
In Partial Fulfillment of the Requirements for the degree of
Doctor of Philosophy
Department of Chemistry
West Lafayette, Indiana
May 2017
ProQuest Number: 10265249
All rights reserved
INFORMATION TO ALL USERS
The quality of this reproduction is dependent upon the quality of the copy submitted.
In the unlikely event that the author did not send a complete manuscript
and there are missing pages, these will be noted. Also, if material had to be removed,
a note will indicate the deletion.
ProQuest 10265249
Published by ProQuest LLC (2017 ). Copyright of the Dissertation is held by the Author.
All rights reserved.
This work is protected against unauthorized copying under Title 17, United States Code
Microform Edition © ProQuest LLC.
ProQuest LLC.
789 East Eisenhower Parkway
P.O. Box 1346
Ann Arbor, MI 48106 - 1346
ii
THE PURDUE UNIVERSITY GRADUATE SCHOOL
STATEMENT OF DISSERTATION APPROVAL
Dr. Timothy S. Zwier, Chair
Department of Chemistry
Dr. Dor Ben-Amotz
Department of Chemistry
Dr. Lyudmila Slipchenko
Department of Chemistry
Dr. Paul G. Wenthold
Department of Chemistry
Approved by:
Dr. Timothy Zwier
Head of the Departmental Graduate Program
iii
To my family
iv
ACKNOWLEDGMENTS
First I would like to express my immense gratitude to my research advisor Prof.
Timothy S. Zwier. During all my Ph.D. years at Purdue University, Tim has always
provided strong supports to me, especially at my hardest time. His strong passion and
enthusiasm for science has set a perfect example for me to follow in my life. I feel
extremely fortunate to have such a great advisor. Thank you very much for your excellent
guidance on my graduate school studies, which makes my Ph.D. experience fruitful and
exciting.
I am grateful to all Zwier group members, past and present, you have helped me a
lot both academically and personally. I shall express my special thanks to Dr. Vanesa
Vaquero-Vara, she introduced me into the new world of microwave spectroscopy and
helped me step by step to do experiments independently. I would also like to thank Dr.
Ryoji Kusaka, Dr. Deepali Mehata-Hurt, Dr. Jacob Dean, Dr. Nathan Kidwell, Dr.
Zachary Davis, Dr. Joseph Korn, Dr. Nicole Burke, Dr. Patrick Walsh, Dr. Joseph Gord,
Dr. Brian Hays, Dr. Chamara Abeysekera, Dr. Andrew DeBlase, Khadija Jawad, Daniel
Hewett, Karl Blodgett, Alicia Hernandez Castillo, Sean Fritz, Piyush Mishra, Chris
Harrilal, John Lawler, Dewei Sun and Josh Fischer. I am extremely grateful to those
people I learned from and worked with, for the friendship we developed inside the group,
you have made this journey such a pleasure for me.
I must also thank the staff in chemistry department. I would like to thank the Amy
Instrumentation Facility. I owe many thanks to Tim Selby, Mark Carlsen and Dr. Ryan
Hilger for their strong and continuous supports on my instrumentation and data analysis. I
would also like to thank the clerical staff, in particular, Debbie Packer, Betty Hatfield,
v
Darci Decamp and Liz Hewitt. They helped me in various ways through my time in
Purdue university.
I am also grateful to my research collaborators inside and outside Purdue
University. Especially Dr. Xiao Zhu at Rosen center for advanced computing and
Professor Steve Shipman in New College of Florida. They provided key supports on my
research. I would also like to thank Dr. Yu-ting Huang from Prof. Chappell group in ECE
department of Purdue university for introducing the engineering back ground of
microwave instruments to me.
During my Ph.D career in Purdue I met lots of wonderful friends. Among them I
specially thank Dr. Hao Zhong, Dr. Yue Ren and Dr. Lei Tan for their support and
company.
Finally, this work could not be done without the support from my family, to whom I
dedicate this dissertation. To my parents, grandparents and girlfriend, your love are the
sources to keep me move ahead. Thank you very much for always being there for me.
You are the most important people in my life and I love you more than anything.
.
vi
TABLE OF CONTENTS
LIST OF TABLES ........................................................................................................... viii
LIST OF FIGURES .......................................................................................................... xii
ABSTRACT................................................................................................................... xviii
CHAPTER 1. INTRODUCTION .................................................................................... 1
1.1 Spectroscopy Study of Hydrogen Bonded Molecules in the Gas Phase ................... 1
1.2 Organization .............................................................................................................. 4
1.3 References ................................................................................................................. 7
CHAPTER 2. EXPERIMENTAL METHODS .............................................................. 10
2.1 Supersonic Jet Expansion and Molecular Beams .................................................... 10
2.2 Vacuum Chambers .................................................................................................. 11
2.2.1 Time-of-Flight Mass Spectrometer................................................................ 11
2.2.2 Chirped-Pulse Fourier Transform Microwave Spectrometer ........................ 15
2.3 Spectroscopic Methods ........................................................................................... 16
2.3.1 Resonant Two-Photon Ionization .................................................................. 16
2.3.2 Hole-Burning Spectroscopy........................................................................... 17
2.3.3 Resonant Ion-Dip Infrared Spectroscopy ...................................................... 18
2.3.4 Chirped-Pulse Fourier Transform Microwave Spectroscopy ........................ 20
2.3.4.1 Jet Cooled CP-FTMW ................................................................................. 20
2.3.4.2 Room Temperature CP-FTMW ................................................................. 27
2.4 Computational Methods .......................................................................................... 29
2.5 References ............................................................................................................... 31
CHAPTER 3.SINGLE CONFORMATION SECTROSCOPY OF
SUBEROYLANILIDE HYDROXAMIC ACID (SAHA): A MLECULE BITES ITS
TAIL ................................................................................................................................. 33
3.1 Introduction ............................................................................................................. 33
3.2 Methods ................................................................................................................... 36
3.2.1 Experimental Methods ...................................................................................... 36
3.2.2 Computational Methods.................................................................................... 38
3.2.3 Nomenclature .................................................................................................... 42
3.3 Results and Analysis ............................................................................................... 44
3.3.1 R2PI and IR-UV Holeburning Spectra .......................................................... 44
3.3.2 RIDIR Spectra of Conformers A-C .................................................................. 47
3.3.3 RIDIR Spectra of the SAHA-H2O Complex (Structure D) ........................... 56
3.4 Discussion ............................................................................................................... 60
3.4.1 Inherent Conformational Preferences of SAHA Monomer ............................. 60
vii
3.4.2 Disconnectivity Graphs, Isomerization Pathways, and Observed
Conformers ................................................................................................................ 63
3.4.3 Cis-amide Structures and Laser Desorption ..................................................... 73
3.4.4 Effect of Water on the Conformational Preferences of SAHA ....................... 76
3.5 Conclusions .............................................................................................................. 78
3.6 References ................................................................................................................ 80
CHAPTER 4. The Delicate Balance of Hydrogen Bonding Forces in D-Threoninol ... 84
4.1 Introduction .............................................................................................................. 84
4.2 Theoretical and Experimental Methods. .................................................................. 86
4.3 Results. ..................................................................................................................... 90
4.4 Discussion. .............................................................................................................. 98
4.5 References ............................................................................................................. 114
CHAPTER 5. BROADBAND MICROWAVE SPECTROSCOPY OF
PROTOTYPICAL AMINO ALCOHOLS AND POLYAMINES: COMPETITION
BETWEEN H-BONDED CYCLES AND CHAINS...................................................... 117
5.1 Introduction ........................................................................................................... 117
5.2 Experimental and Computational Methods ........................................................... 119
5.3 Results and Analysis ............................................................................................. 122
5.3.1 Nomenclature ............................................................................................... 122
5.3.2 Microwave Spectra ...................................................................................... 128
5.3.2.1 D-allo-threoninol .................................................................................... 128
5.3.2.2 2-amino 1,3 propanediol, 1,3-diamino-2-propanol and Propane-1,2,3triamine 134
5.4 Discussion ............................................................................................................. 145
5.4.1 Comparing Predicted Energies and Observed Populations ......................... 145
5.4.2 The Preference for Cycles Versus Chains as a Function of NH2/OH Content
158
5.4.3 Effect of the Methyl Group on Structural Preferences ................................ 166
5.5 Conclusions ........................................................................................................... 170
5.6 References ............................................................................................................. 171
CHAPTER 6. ROOM TEMPERATURE CHIRPED-PULSE FOURIER TRANSFORM
MICROWAVE SPECTROSCOPY OF ISOBUTANOL ............................................... 174
6.1
6.2
6.3
6.4
Introduction ........................................................................................................... 174
Experimental and Computational Methods ........................................................... 176
Results and Discussion .......................................................................................... 181
References ............................................................................................................. 198
VITA ............................................................................................................................... 200
PUBLICATIONS ............................................................................................................ 201
viii
LIST OF TABLES
Table 3.1. Summary of calculated (Calc) and observed (Obs) vibrational frequencies
and assignments of SAHA monomer and the SAHA-H2O complex,
calculated at the DFT B3LYP-D3BJ/6-31+G(d) level of theory. .................... 51
Table 3.2. Summary of calculated vibrational frequencies in the hydride stretch region
for 3 lowest energy structures in NH-π, H-TOH and H-T/OH-π H-bonded
architectures, calculated at the DFT B3LYP-D3BJ/6-31+G(d) level of theory.
.......................................................................................................................... 55
Table 3.3. Dihedral angles (degrees) along the C6 alkyl chain of the three observed
conformers of SAHA monomer from head-to-tail. .......................................... 63
Table 3.4. Calculated relative energies and free energies of 10 lowest energy conformers
of SAHA in the disconnectivity diagram at B3LYP-D3BJ/6-31+G(d) level of
theory. Assigned structures are marked in bold. The two lowest energy cisSAHA structures areadded at the end of the table below the dashed line........ 72
Table 4.1. Rotational parameters of the seven most stable conformations of
D-threoninol. .................................................................................................... 93
Table 4.2. Experimental rotational parameters of the seven conformers identified in the
spectrum. .......................................................................................................... 94
Table 4.3. Observed frequencies and errors for the quadrupole hyperfine components
assigned in MHz for conformer A of D-threoninol. ........................................ 99
Table 4.4. Observed frequencies and errors for the quadrupole hyperfine components
assigned in MHz for conformer B of D-threoninol........................................ 100
Table 4.5. Observed frequencies and errors for the quadrupole hyperfine components
assigned in MHz for conformer C of D-threoninol........................................ 101
Table 4.6. Observed frequencies and errors for the quadrupole hyperfine components
assigned in MHz for conformer D of D-threoninol. ...................................... 102
Table 4.7. Observed frequencies and errors for the quadrupole hyperfine components
assigned in MHz for conformer E of D-threoninol. ....................................... 103
Table 4.8. Observed frequencies and errors for the quadrupole hyperfine components
assigned in MHz for conformer F of D-threoninol. ....................................... 104
Table 4.9. Observed frequencies and errors for the quadrupole hyperfine components
assigned in MHz for conformer G of D-threoninol. ...................................... 105
ix
Table 4.10. Comparison of relative energies of the seven most stable conformations of
DTN at the indicated levels of theory (including zero point corrected
vibrational energies). ...................................................................................... 106
Table 5.1. Structures for all conformational minima (within 500 cm-1 of the global
minimum) of D-allo-threoninol. The structural types are given above,
along with their zero point corrected relative energies in cm-1 at the
B2PLYP aug-cc-pVTZ level of theory. ......................................................... 124
Table 5.2. Structures for all conformational minima (within 500 cm-1 of the global
minimum) of 2-Amino-1,3-propanediol. The structural types are given
above, along with their zero point corrected relative energies in cm-1 at the
B2PLYP aug-cc-pVTZ level of theory .......................................................... 125
Table 5.3. Structures for all conformational minima (within 500 cm-1 of the global
minimum) of 1,3-diamino-2-propanol. The structural types are given
above, along with their zero point corrected relative energies in cm-1 at the
B2PLYP aug-cc-pVTZ level of theory. ......................................................... 126
Table 5.4. Structures for all conformational minima (within 500 cm-1 of the global
minimum) of propane-1,2,3-triamine. The structural types are given
above, along with their zero point corrected relative energies in cm-1 at the
B2PLYP aug-cc-pVTZ level of theory .......................................................... 127
Table 5.5. Calculated rotational parameters and relative energies of the nine most
stable confirmations of D-allo-threoninol. ..................................................... 131
Table 5.6. Experimental rotational parameters of the three assigned conformers of
D-allo-threoninol. ........................................................................................... 132
Table 5.7. Calculated rotational parameters and relative energies of the eight most
stable confirmations of 2-amino 1,3-propanediol. ......................................... 136
Table 5.8. Experimental rotational parameters for the assigned conformers of 2-amino
1,3-propanediol. ............................................................................................. 137
Table 5.9. Calculated rotational parameters of the eight most stable confirmations of
1,3-diamino-2-propanol. ................................................................................ 139
Table 5.10. Experimental rotational parameters for the four assigned conformers of
1,3-diamino-2-propanol. ................................................................................ 140
Table 5.11. Calculated rotational parameters of the eight most stable confirmations of
propane-1,2,3-triamine. .................................................................................. 143
x
Table 5.12. Experimental rotational parameters of the four assigned conformers of
propane 1,2,3-triamine. .................................................................................. 144
Table 5.13. The percent populations of the observed conformers and comparison of their
relative energies calculated at the B2PLYP-D3BJ/aug-cc-pVTZ level of
theory.............................................................................................................. 147
Table 5.14. Energy differences and rotational constants for all conformers of D-allothreoninol within 500 cm-1 of the global minimum computed with various
levels of theory. Conformers were labeled with Roman numerals according
to their energetic ordering based on the first calculations. Additionally,
experimentally observed conformers are labeled with alphabetic letters. ..... 148
Table 5.15. Energy differences and rotational constants for all conformers of 2-amino1,3-propanediol within 500 cm-1 of the global minimum computed with
various levels of theory. Conformers were labeled with Roman numerals
according to their energetic ordering based on the first calculations.
Additionally, experimentally observed conformers are labeled with
alphabetic letters............................................................................................. 149
Table 5.16. Energy differences and rotational constants for all conformers of 1,3diamino-2-propanol within 500 cm-1 of the global minimum computed with
various levels of theory. Conformers were labeled with Roman numerals
according to their energetic ordering based on the first calculations.
Additionally, experimentally observed conformers are labeled with
alphabetic letters............................................................................................. 151
Table 5.17. Energy differences and rotational constants for all conformers of propane1,2,3-triamine within 500 cm-1 of the global minimum computed with
various levels of theory. Conformers were labeled with Roman numerals
according to their energetic ordering based on the first calculations.
Additionally, experimentally observed conformers are labeled with
alphabetic letters............................................................................................. 153
Table 5.18. Summary of the sets of dihedral angles associated with each of the
prototypical H-bonded structural types. (X, Y = OH or NH2). ...................... 162
Table 5.19. Summary of calculated XH…Y HB distances (in Å) and bond angles (in
degrees) of the lowest energy chain and cyclic structures for each molecule
in the series glycerol, 2-amino-1,3-propanediol, 1,3-diamino-2-propanol,
and propane-1,2,3-triamine at the B2PLYP-D3BJ/aug-cc-pVTZ level of
theory.............................................................................................................. 164
Table 6.1. Observed frequencies and errors for the assigned transitions in MHz for gA
conformer of isobutanol. ................................................................................ 182
xi
Table 6.2. Observed frequencies and errors for the assigned transitions in MHz for g’g
conformer of isobutanol. ................................................................................ 184
Table 6.3. Observed frequencies and errors for the assigned transitions in MHz for gg
conformer of isobutanol. ................................................................................ 186
Table 6.4. Observed frequencies and errors for the assigned transitions in MHz for AA
conformer of isobutanol. ................................................................................ 187
Table 6.5. Experimental rotational parameters of the four assigned conformers of
isobutanol. ...................................................................................................... 188
Table 6.6. Summary of all vibrational states calculated within 400 cm-1 of the ground
state of the global minimum calculated at B3PLYP-D3BJ/6-311++G(d,p)
level of theory.Vibrational states from gA, g’g, gg, AA and Ag conformers
are labeled in red, pink, blue , green and black respectively. ........................ 190
xii
LIST OF FIGURES
Figure 2.1. Schematic diagram of time-of-flight mass spectrometer chamber. ............... 14
Figure 2.2. Schematic diagram of chirped-pulse Fourier transform microwave (CPFTMW) chamber. .......................................................................................... 16
Figure 2.3. Schemes for (a) resonant two-photon ionization spectroscopy, (b) UV-UV
hole-burning spectroscopy, (c) IR-UV hole-burning spectroscopy and (d)
resonant ion-dip infrared spectroscopy employed in this work. .................... 20
Figure 2.4. Schematic of the jet cooled chirped pulse Fourier transform microwave (CPFTMW) spectrometer. ................................................................................... 22
Figure 2.5. Multiple segments with gaps timing diagram. (Modified from Guzik ADC
6000 series manual) ....................................................................................... 25
Figure 2.6. Width of the gas pulse measured by monitoring the signal intensity of OCS
transition at 12162.9 MHz against AWG delay time. ................................... 26
Figure 2.7. A typical example of the Guzik ADC 6131 configuration file. ..................... 26
Figure 2.8. Schematic for the room temperature chirped pulse Fourier transform
microwave (RT-CP-FTMW) spectrometer.................................................... 28
Figure 3.1. Chemical structure of suberoylanilide hydroxamic acid (SAHA). ................ 33
Figure 3.2. CNOH dihedral angle scan results for N-Hydroxypropanamide. (Prepared by
Karl Blodgett from Zwier group) .................................................................. 39
Figure 3.3. R2PI (top trace) and IR-UV HB spectra (lower traces) for SAHA. Asterisks
in the R2PI spectrum are tentatively ascribed to hot bands of conformer A.
Since the electronic chromophore of SAHA is closely related to that in
trans-formaniliide (tFA) and trans- acetanilide (tAA), their electronic
origins are shown in the figure for reference.29 ............................................. 44
Figure 3.4. RIDIR spectra for SAHA conformer A,B and C in the (a) hydride stretch
and (b) amide I/II regions. Calculated IR spectra at the
DFT B3LYP D3BJ/6-31+G(d) level of theory are shown below as stick
diagrams in black. These frequencies were scaled by 0.96 for free NH
stretch, 0.948 for hydrogen bonded NH and OH stretches and 0.985 for
amide I/II frequencies. Asterisks indicate infrared transitions used to
record IR-UV holeburn scans. ....................................................................... 47
xiii
Figure 3.5. Calculated optimized structures assigned for SAHA conformers A to C and
structure D, assigned to the SAHA-H2O complex, at the DFT B3LYPD3BJ/6-31+G(d) level of theory. The zero-point corrected relative
energies are included. .................................................................................... 49
Figure 3.6. (a) Lowest energy structures in NH-π, H-TOH and H-T/OH-π H-bonded
architectures. (b) Stick diagrams of the calculated frequencies and IR
intensities in the hydride stretch region. These frequencies were scaled by
0.96 for free NH stretch, 0.948 for hydrogen bonded NH and OH stretches.
The experimental RIDIR spectra for SAHA conformer A,B and C in the
hydride stretch region are plotted above. ....................................................... 54
Figure 3.7. RIDIR spectra in the (a) hydride and (b) mid-IR regions of SAHA
conformer D, the SAHA-H2O complex. Stick diagrams underneath were
simulated based on calculations using DFT B3LYP/6-31+G(d) (black) and
B3LYP D3BJ/6-31+G(d) (red) levels of theory. The asterisk in the hydride
stretch region indicates transition used to record the IR-UV HB spectrum.
These frequencies were scaled by 0.96 for free NH stretch and water OH
stretch, 0.948 for hydrogen bonded NH and OH stretches and 0.985 for
amide I/II frequencies. ................................................................................... 56
Figure 3.8. R2PI (top trace) spectrum for SAHA. Calculated TDDFT spectra at the
DFT B3LYP D3BJ/6-31+G(d) level of theory are shown below as stick
diagrams in black. The absolute frequencies of the S0-S1 origins are lined
up with experiment for conformer A and scaled by a factor of 0.46. ............ 60
Figure 3.9. (a) Disconnectivity graph for SAHA using the modified general Amber
force field (GAFF). Red asterisks indicate the locations of SAHA A and
SAHA B. (b) Close-up view of the dashed rectangle region of the SAHA
disconnectivity graph where the assigned structures for SAHA A and
SAHA B are located. The zero-point energy corrected relative energies
calculated at the DFT B3LYP-D3BJ level of theory are also indicated (X),
taken from Table 3.4. (Prepared by Dr. Xiao Zhu from Rosen center for
advanced computing and Karl Blodgett from Zwier group) ......................... 65
Figure 3.10. Stationary points along the lowest-energy isomerization pathway predicted
by GAFF between SAHA A, B, and C, calculated at the DFT B3LYPD3BJ/6-31+G(d) level of theory. ................................................................... 70
Figure 3.11. (a) Structures, labels, and zero-point energy corrected energies (relative to
SAHA A at the DFT B3LYP-D3BJ/6-31+G(d) level of theory) of the two
low-energy conformers of cis-amide SAHA. (b) Stick diagrams of the
calculated frequencies and IR intensities in the hydride stretch region,
using the same scale factors (0.948) for hydrogen bonded NH and OH
stretches. The experimental RIDIR spectra in the hydride stretch region are
presented in Figure 3.4(a). ............................................................................. 74
xiv
Figure 3.12. Calculated global minimum for SAHA water complex through B3LYPD3BJ/6-31+G(d) level of theory. ................................................................... 77
Figure 4.1. The seven most stable conformers of D-threoninol paired according to the
dihedral angles of their functional groups. The direction of H-bonds of the
functional groups (-OH and -NH2), α···β···γ (I) or γ···β···α (II), are given,
along with their zero point corrected relative energies in cm-1. .................... 87
Figure 4.2. The next seven most stable conformers for DTN (numbers 8-14 in energy)
paired according to the dihedral angles of their functional groups. The
orientation of the donor-acceptor interactions, αβγ(I) or γβα(II), are given
in parentheses, along with their zero point corrected relative energies
in cm-1. ........................................................................................................... 88
Figure 4.3. Pure rotational spectrum of D-threoninol from 7.5 GHz to 18.5 GHz. .......... 91
Figure 4.4. Quadrupole hyperfine structure in the 220-111 rotational transitions of the
two conformers exhibiting cyclic HB networks. ........................................... 97
Figure 4.5. Calculated (MP2/6-31++G(d,p)) interconversion barriers involving pairs
of conformers in DTN linked by internal rotation of the terminal acceptor
-OH groups. ................................................................................................. 108
Figure 4.6. Calculated hydrogen bond distances (in Å) in two comparable
conformations of glycerol and DTN, obtained using
MP2/6-311++G(d,p). ................................................................................... 110
Figure 4.7. HOMO’s of the two most stable predicted conformers of DTN. The
molecular orbitals are located mostly above the plane of the backbone. .... 112
Figure 5.1 (a) Experimental rotational spectrum of jet-cooled D-allo-threoninol from
7.5 to 18.5 GHz. (b) Close-up of the 15.85-15.95 GHz region with
calculated stick spectra due to conformer A (red), B (blue), and C (green)
below, showing the quality of the fit. (c) Further expansion of 5 MHz
regions around the 322-212 transitions of conformers A, C are shown in
part (b), in order to compare the experimental and calculated nuclear
quadrupolar splittings for the two assigned cyclic conformers. The F’-F”
labels are included in the figure. .................................................................. 129
Figure 5.2. Experimental rotational spectrum of jet-cooled 1,3-diamino-2-propanol
from 7.5 to 18.5 GHz. .................................................................................. 138
xv
Figure 5.3. (a) Experimental rotational spectrum of jet-cooled propane-1,2,3-triamine
from 7.5 to 18.5 GHz. Calculated stick spectra due to conformer A (blue),
B (green), C (yellow) and D(red) are shown below. (b) Further expansion
of 5 MHz regions around the 211-101 transitions of conformers A with
calculated nuclear quadrupolar splittings shown below. The F’-F” labels
are included in the figure. ............................................................................ 142
Figure 5.4. (a) Interconversion barrier between structure IV to structure I of D-allothreoninol at the B3PLYP-D3BJ/aug-cc-pVTZ level of theory. (b)
Interconversion barrier between structure IV to structure I for 2-amino-1,3propanediol at the B3PLYP-D3BJ/aug-cc-pVTZ level of theory. .............. 157
Figure 5.5. Calculated energy level diagrams for glycerol, 2-amino-1,3-propanediol,
1,3-diamino-2-propanol and propane-1,2,3-triamine through B2PLYPD3BJ/aug-cc-pVTZ level of theory. ............................................................ 158
Figure 5.6. Calculated structures for the full set of observed conformers of 2-amino-1,3propanediol with calculated HB distances (in Å) at the B2PLYP-D3BJ/augcc-pVTZ level of theory. Assigned structural types and their relative
energies calculated at the same level of theory are included. The global
minimum is shaded in red, while the lowest energy local minimum is
shaded in blue. ............................................................................................. 160
Figure 5.7. Calculated structures for the full set of observed conformers of 1,3-diamino2-propanol with calculated HB distances (in Å) using
B2PLYP-D3BJ/aug-cc-pVTZ level of theory. Assigned structural types
and their relative energies calculated at the same level of theory are
included. The global minimum is shaded in red, while the lowest energy
local minimum is shaded in blue. ................................................................ 160
Figure 5.8. Calculated structures for the full set of observed conformers of
propane-1,2,3-triamine with calculated HB distances (in Å) using
B2PLYP-D3BJ/aug-cc-pVTZ level of theory. Assigned structural types
and their relative energies calculated at the same level of theory are
included. The global minimum is shaded in red, while the lowest energy
local minimum, a bifurcated extended chain, is shaded in blue. Note that
the cyclic structure was not observed experimentally (see text for further
discussion). .................................................................................................. 161
Figure 5.9. The calculated structures for the full set of observed conformers of glycerol
with calculated HB distances (in Å) using B2PLYP-D3BJ/aug-cc-pVTZ level
of theory. Assigned structural types from ref. 11 and their relative energies
calculated at the same level of theory are included. The global minimum is
shaded in red, while the lowest energy local minimum is shaded in blue. .. 161
xvi
Figure 5.10. Calculated H-bond distance (Å) versus XH…Y bond angles (degrees),
at the B2PLYP-D3BJ/aug-cc-pVTZ level of theory, grouped by H-bond
types. ............................................................................................................ 165
Figure 5.11. Structural evolution from 2-amino-1,3-propanediol (center) to
D-threoninol (left) and D-allo-threoninol (right). Zero-point corrected
relative energies (in cm-1) calculated at the B2PLYP-D3/aug-cc-pVTZ
level of theory are included. Mirror image pairs in
2-amino-1,3-propanediol are included in the boxes. ................................... 167
Figure 5.12. Energy level diagrams for all conformational minima of D-threoninol and
D-allo-threoninol within 500 cm−1 of the global minimum, calculated at the
B2PLYP-D3/aug-cc-pVTZ level of theory. Conformers that are assigned in
the expansion are shown in red. Structures of the lowest energy chain and
cyclic conformers are also plotted with their Newman projections. ........... 169
Figure 6.1. The five most stable conformers of isobutanol calculated at
B3LYP-D3BJ/6-311++G(d,p) level of theory, along with their zero point
corrected relative energies in cm-1. .............................................................. 178
Figure 6.2. The experimental rotational spectrum of room temperature isobutanol
from 7.5 to 18.5 GHz. The upper part of the spectrum is experimental
data. The red trace is the Fourier transformed FID after up chirping (7.518.5GHz) pulse while the blue trace is from the down chirping pulse.
Black traces are collected background signals. Simulations of the fitted
rotational parameters are plotted in the lower trace. Different colors stand
for different conformers assigned. gA, g’g,gg and AA conformers are
labeled in red, pink, blue and green respectively. Close-up of the
16.75-17.00 GHz region is inserted, showing the quality of the fit............. 180
Figure 6.3. Calculated energy level diagrams for gA(red), g’g(purple), gg(blue),
AA(green) and Ag(black) conformers through B3PLYP-D3BJ/6311++G(d,p) level of theory. The relative energy for each vibrationally
excited state (to their respective ground state) is given under the solid line.
The ground state relative energy for each conformer to gA is given in the
parentheses. .................................................................................................. 189
Figure 6.4. Predicted room temperature spectrum for gA conformer. Vibrationally excited
states up to 300 cm-1 are included and their intensities are scaled according
to the Boltzmann distribution. (Prepared by Dr.Brian Hays) ...................... 193
xvii
Figure 6.5. Two-dimensional relaxed potential energy scan of isobutanol at the
B3LYP-D3BJ/6-31+G(d) level of theory. The surface was calculated at 10°
grid points, from -180° to +180° on ϕ and θ dihedral angles. ..................... 194
Figure 6.6. (a) Interconversion barriers among Ag, Ag’ and AA conformers of
isobutanol calculated at B3LYP-D3BJ/6-31+G(d) level of theory. (b)
Interconversion barriers among gg, gg’ and gA conformers of isobutanol
calculated at B3LYP-D3BJ/6-31+G(d) level of theory. .............................. 197
xviii
ABSTRACT
Author: Zhang, Di. Ph.D.
Institution: Purdue University
Degree Received: May 2017
Title: From Microwave, Infrared to UV——Probing the Conformational Preferences for
Biomolecules with Intramolecular Hydrogen Bonds.
Major Professor: Timothy Zwier
Hydrogen bonds play very important roles in the structural organizations of
biological molecules. The best way to study hydrogen bonded molecules in the gas phase
is through optical spectroscopic methods in the microwave, infrared and UV range
coupled with supersonic jet expansion. This dissertation characterizes the potential
energy surfaces of several biological molecules with intramolecular hydrogen bonds
through a combination of different spectroscopic and theoretical methods.
The single conformation spectroscopy of suberoylanilide hydroxamic acid (SAHA)
is presented first. SAHA is a histone deacetylase inhibitor and anti-cancer drug. Its
structure is very unique with a formanilide “head” and a hydroxamic acid “tail” separated
by a n-hexyl chain. As a result, the alkyl chain’s preference for extended structures is in
competition with tail to-head (T-H) or head-to-tail (H-T) hydrogen bonds between the
amide and hydroxamic acid groups. Three conformers of SAHA were distinguished and
spectroscopically characterized. A modified version of the generalized Amber force field
was used to generate a disconnectivity diagram for the low-energy portion of the
potential energy landscape of SAHA. This combination of force field and DFT
calculations provides insight into the potential energy landscape and how population was
funneled into the three observed conformers.
xix
For small molecules without a chromophore, their intramolecular hydrogen bond
patterns were studied through broadband microwave spectroscopy. The rotational spectra
of D-threoninol, D-allo-threoninol, 2-amino-1,3-propanediol, 1,3-diamino-2-propanol
and propane-1,2,3-triamine have been recorded under jet cooled conditions through
chirped-pulse Fourier transform microwave spectroscopy. In total, 22 conformers were
assigned in the expansion for those 5 small amino alcohol molecules. With three adjacent
H-bonding substitutes along the alkyl chain, both hydrogen-bonded cycles (3 H-bonds)
and hydrogen-bonded chains (2 H-bonds) were observed. Both families remain close in
energy. In D-threoninol, D-allo-threoninol, 2-amino-1,3-propanediol and 1,3-diamino-2propanol, H-bonded cycles are most highly populated while curved and extended chain
structures are favored by propane-1,2,3-triamine.
Finally, a room temperature detection method employing chirped-pulse Fourier
transform microwave spectroscopy coupled with a waveguide and an ultrafast digitizer is
described. The ultrafast speed of the digitizer enables up to 1 billion shots averages of the
sample molecule. In the collected room temperature spectra of isobutanol, 4 conformers
were observed and assigned. Aside from transitions from ground state, a large number of
transitions from vibrationally excited states were also observed and predicted through
theoretical methods.
1
CHAPTER 1.
1.1
INTRODUCTION
Spectroscopy Study of Hydrogen Bonded Molecules in the Gas Phase
Hydrogen bond is a weak electrostatic interaction between a covalently bonded
hydrogen atom (X-H) to another atom (A) with a lone pair of electrons. In general, X
and A are atoms with high electronegativity like nitrogen or oxygen atoms and X-H acts
as a proton donor to A. A hydrogen bond could be formed either between different
molecules or within the same molecule. Hydrogen bond is variable in length, usually
between 1.5-2.5 Å.1 The energies of hydrogen bonds could cover two orders of
magnitude, about 1kJ/mol to 200 kJ/mol with an average of 5-30 kJ/mol.2 The bonding
energies are generally less than covalent bonds, ionic bonds and metallic bonds, but
higher than van der Waals forces.
Hydrogen bond plays a very important role in the structural organization of
biological molecules. Watson, Crick and Franklin discovered that the two strands of
DNA double helix are held together by hydrogen bonds.3 Hydrogen bonding schemes
like alpha helix and beta sheet are crucial to stabilize the secondary structures of proteins
and nucleic acids and determine the three-dimensional foldings of these biological
macromolecules.4-7 In lots of cases, delicate balances between intermolecular and
intramolecular hydrogen bonds control the conformation of biomolecules, and also play a
key role in determining the functions of these biological molecules.
Different experimental methods have been adopted to study hydrogen bonds. The
hydrogen bond length in crystal structure could be determined by X-ray powder
diffraction (XRD) with a precision about 0.01 Å ~ 0.001 Å.8-10 In the aqueous phase,
infrared spectroscopy provides the most straightforward way in detecting hydrogen bond
2
since the νXH stretching mode is shifted to lower frequency range and shows an increased
intensity and broadening of half-band width.11-15 Raman spectroscopy could also provide
complementary information to the infrared results.16-18 Another powerful experimental
method for the study of hydrogen bond is nuclear magnetic resonance (NMR)
spectroscopy,19-21 in which the 1H downfield shifts are correlated with the strength of the
hydrogen bond. Electronic spectroscopy could also be adopted to study hydrogen bonds
if the chromophore portion is affected by the hydrogen bond, thus will affect the
frequency and intensity of electronic transitions.11
Gas phase study of hydrogen bonds provides the opportunity to explore the
inherent conformational preferences without the perturbation from environment, which
could also enable the comparison with results from quantum theoretical calculations.
However, in gas phase, the number density of sample molecules is much lower than
aqueous phase, so spectroscopic methods with high sensitivities are required. Laserinduced fluorescence (LIF) or multi-photon ionization coupled with mass spectrometric
detection are widely used in the ultraviolet region detections. To record conformer
specific vibrational spectra, IR-UV double resonance spectroscopic methods22-24 are
widely adopted, in which the removal of ground population by IR excitation laser is
probed as a dip in the collected UV induced ionization signal. Again, structural features
like intermolecular or intramolecular hydrogen bonds could be characterized by the shifts
of hydride stretch vibrations. For small molecules without an aromatic chromophore but
with large dipole moments, rotational spectroscopy is a powerful method to characterize
their three dimensional structures in the gas phase. The three rotational constants A,B and
C determined from the analysis of rotational spectra indicate the mass distribution of the
3
molecule. Besides, through isotopic substitution using Kraitchman method,25 the atom
positions could be independently determined.
Another challenge in the gas phase study is at room temperature, a large number of
quantum states are populated and the spectrum is composed of an enormous number of
individual lines. To cool the molecules into vibrational ground state, supersonic jet
expansion26-27 is adopted by seeding the sample molecules in buffer gas. Thus cools down
the translational temperature. The vibrational and rotational temperature could also be
decreased through the energy transfer in the gas bath. Thermal heating or laser desorption
are adopted to transfer molecules of moderate size into gas phase. To study large
biomolecules, optical spectroscopy is always coupled with tandem mass spectrometry.
Techniques like electrospray ionization 28 (ESI) and matrix-assisted laser desorption
ionization (MALDI)29 are adopted to bring large biomolecules into gas phase. Cryogenic
ion trapping30 is adopted to cool down the ion temperatures to the range of a few K
before interrogated with tunable UV/vis and IR laser. Through these cooling techniques
described above, the excited rotational and vibrational levels are significantly
depopulated and allows the characterization of individual conformers with double
resonance spectroscopic techniques.
In conclusion, among all the spectroscopic methods to probe hydrogen bond
conformation, optical spectroscopic methods in the microwave, infrared and UV range
are most suitable for the application to gas phase molecules or ions, especially when
combined with molecular beam techniques and other cryogenic methods, leading to
highly resolved spectroscopic signatures.
4
1.2
Organization
The overall goal of this thesis is to obtain spectral signatures of several
biomolecules with intramolecular hydrogen bonds in the gas phase using different optical
spectroscopic methods including infrared, UV and microwave spectroscopy. The
potential energy landscapes for these biomolecules are also widely explored with the aid
of different calculation methods.
Chapter 2 describes the investigation methods for biomolecules in the gas phase. It
starts from the mechanism of supersonic jet cooling and the vacuum chamber design.
Then introduces different spectroscopic methods including resonant 2 photon ionization,
double resonance methods and chirped-pulse Fourier transform microwave spectroscopy
with jet cooled condition or room temperature waveguide. A brief description for the
theoretical calculation methods that aid the structural assignment and discussion is
presented at the end of the chapter.
Chapter 3 focuses on the single conformation spectroscopy of suberoylanilide
hydroxamic acid (SAHA), an anti-cancer drug that causes growth arrest and
differentiation of many tumors. Single conformation UV spectra in the S0-S1 region and
infrared spectra in the hydride stretch and mid-IR regions are recorded using IR-UV holeburning and resonant ion-dip infrared spectroscopy, respectively. Three conformers of
SAHA are distinguished and spectroscopically characterized, all of which possess tightly
folded alkyl chains that enable formation of either a head N-H to tail C=O or tail N-H to
head C=O intramolecular hydrogen bond. A modified version of the generalized Amber
force field (GAFF) is developed to more accurately describe the hydroxamic acid OH
internal rotor potential and generate a disconnectivity graph for the low-energy portion of
the potential energy landscape of SAHA. This disconnectivity graph contains more than
5
one hundred minima and maps out the lowest-energy pathways between them, which
could then be characterized via DFT calculations.
Chapter 4 presents the broadband microwave spectroscopy study of intramolecular
hydrogen bonding patterns in D-Threoninol molecule (2(S)-amino-1,3(S)-butanediol), a
template used for the synthesis of artificial nucleic acids. Seven structures are observed in
the jet expansion, two are H-bonded cycles containing three H-bonds while the
other five are H-bonded chains containing OH···NH···OH H-bonds with different
directions along the carbon framework and different dihedral angles along the chain.
The two structural types (cycles and chains of H-bonds) are in surprisingly close
energetic proximity. Comparison of the rotational constants with the calculated
structures reveals systematic changes in the H-bond distances that reflect NH2 as a better
H-bond acceptor and poorer donor, shrinking the H-bond distances by ∼0.2 Å in the
former case and lengthening them by a corresponding amount in the latter.
Chapter 5 follows up the study of Chapter 4 and explores the rotational spectra of
other amino alcohols including D-allo-threoninol, 2-amino-1,3-propanediol, and 1,3diamino-2-propanol and the triamine analog, propane-1,2,3-triamine. Microwave
transitions due to three conformers of D-allothreoninol, four conformers of
2-amino-1,3-propanediol, four conformers of 1,3-diamino-2-propanol, and four
conformers of propane-1,2,3-triamine have been identified and assigned, aided by
comparison of the fitted experimental rotational constants with the predictions for
candidate structures based on an exhaustive conformational search using force field, ab
initio and DFT methods. With three adjacent H-bonding substituents along the alkyl
chain involving a combination of OH and NH2 groups, hydrogen-bonded cycles (3 H-
6
bonds) and chains (2 H-bonds) remain close in energy, no matter what the OH/NH2
composition. Two families of H-bonded chains are possible, with H-bonding substituents
forming curved chain or extended chain structures. Percent populations of the observed
conformers are extracted from the relative intensities of their microwave spectra, which
compare favorably with relative energies calculated at the B2PLYP-D3BJ/aug-cc-pVTZ
level of theory. In glycerol (3 OH), D-allothreoninol (2 OH, 1 NH2), 2-amino-1,3propanediol (2OH, 1 NH2), and 1,3-diamino-2-propanol (1 OH, 2 NH2), H-bonded cycles
are most highly populated, followed by curved chains (3 OH or 2 OH/1 NH2) or extended
chains (1 OH/2 NH2). In propane-1,2,3-triamine (3 NH2), H-bonded cycles are pushed
higher in energy than both curved and extended chains, which carry all the observed
population. The NH2 group serves as a better H-bond acceptor than donor, as is
evidenced by optimized structures in which H-bond lengths fall into the following order:
r(OH···N) ≈ r(OH···O) < r(NH···N) ≈ r(NH···O).
Finally, the last chapter (chapter 6) introduces a novel and powerful method by
employing chirped-pulse Fourier transform microwave spectroscopy coupled with a room
temperature waveguide and ultrafast digitizer. The room-temperature rotational spectrum
of isobutanol (2-methylpropan-1-ol) is collected with one billion shots of average. The
vibrational ground state spectrum is assigned with four conformers identified. Spectra
patterns for vibrationally excited states with energies up to 300 cm-1 above the ground
state are predicted through theoretical calculations.
7
1.3
References
1.
Jeffrey, George A. An introduction to hydrogen bonding; Oxford University Press:
Oxford, U.K., 1997.
2.
Steiner, T. The Hydrogen Bond in the Solid State. Angew. Chem. Int. Ed. 2002, 41,
48-76.
3.
Watson, J. D.; Crick, F. H. C. MOLECULAR STRUCTURE OF NUCLEIC ACIDS
- A STRUCTURE FOR DEOXYRIBOSE NUCLEIC ACID. Nature 1953, 171,
737-738.
4.
Kabsch, W.; Sander, C. DICTIONARY OF PROTEIN SECONDARY
STRUCTURE - PATTERN-RECOGNITION OF HYDROGEN-BONDED AND
GEOMETRICAL FEATURES. Biopolymers 1983, 22, 2577-2637.
5.
Pauling, L.; Corey, R. B.; Branson, H. R. THE STRUCTURE OF PROTEINS - 2
HYDROGEN-BONDED
HELICAL
CONFIGURATIONS
OF
THE
POLYPEPTIDE CHAIN. Proc. Natl. Acad. Sci. U.S.A. 1951, 37, 205-211.
6.
Chou, P. Y.; Fasman, G. D. BETA-TURNS IN PROTEINS. J. Mol. Biol. 1977, 115,
135-175.
7.
Frishman, D.; Argos, P. Knowledge-based protein secondary structure assignment.
PROTEINS 1995, 23, 566-579.
8.
Allan, D. R.; Clark, S. J.; Brugmans, M. J. P.; Ackland, G. J.; Vos, W. L. Structure
of crystalline methanol at high pressure. Phys. Rev. B 1998, 58, 11809-11812.
9.
Tauer, K. J.; Lipscomb, W. N. ON THE CRYSTAL STRUCTURES, RESIDUAL
ENTROPY AND DIELECTRIC ANOMALY OF METHANOL. Acta Cryst. 1952,
5, 606-&.
10.
Wong, P. T. T.; Whalley, E. FAR-INFRARED SPECTRUM AND NORMAL
COORDINATE ANALYSIS OF ALPHA-METHANOL. J. Chem. Phys 1971, 55,
1830-+.
11.
Murthy, A. S. N.; Rao, C. N. R. SPECTROSCOPIC STUDIES OF HYDROGEN
BOND. Appl. Spectrosc. Rev. 1968, 2, 69-&.
12.
Asbury, J. B.; Steinel, T.; Stromberg, C.; Corcelli, S. A.; Lawrence, C. P.; Skinner,
J. L.; Fayer, M. D. Water dynamics: Vibrational echo correlation spectroscopy and
comparison to molecular dynamics simulations. Journal of Physical Chemistry A
2004, 108, 1107-1119.
8
13.
Blume, A.; Hubner, W.; Messner, G. FOURIER-TRANSFORM INFRAREDSPECTROSCOPY OF C-13=O-LABELED PHOSPHOLIPIDS HYDROGENBONDING TO CARBONYL GROUPS. Biochemistry 1988, 27, 8239-8249.
14.
Garczarek, F.; Gerwert, K. Functional waters in intraprotein proton transfer
monitored by FTIR difference spectroscopy. Nature 2006, 439, 109-112.
15.
Keutsch, F. N.; Saykally, R. J. Water clusters: Untangling the mysteries of the liquid,
one molecule at a time. Proc. Natl. Acad. Sci. U.S.A. 2001, 98, 10533-10540.
16.
Ohmine, I.; Saito, S. Water dynamics: Fluctuation, relaxation, and chemical
reactions in hydrogen bond network rearrangement. Acc. Chem. Res. 1999, 32, 741749.
17.
Bondesson, L.; Mikkelsen, K. V.; Luo, Y.; Garberg, P.; Agren, H. Hydrogen bonding
effects on infrared and Raman spectra of drug molecules. Spectrochim. Acta Mol.
Biomol. Spectrosc. 2007, 66, 213-224.
18.
Triggs, N. E.; Valentini, J. J. AN INVESTIGATION OF HYDROGEN-BONDING
IN AMIDES USING RAMAN-SPECTROSCOPY. J. Phys. Chem. 1992, 96, 69226931.
19.
Huggins, M. L. 50 YEARS OF HYDROGEN BOND THEORY. Angew. Chem. Int.
Ed. 1971, 10, 147-&.
20.
Martin, T. W.; Derewenda, Z. S. The name is bond - H bond. Nat. Struct. Mol. Biol.
1999, 6, 403-406.
21.
Grzesiek, S.; Cordier, F.; Jaravine, V.; Barfield, M. Insights into biomolecular
hydrogen bonds from hydrogen bond scalar couplings. Prog. Nucl. Magn. Reson.
Spectrosc. 2004, 45, 275-300.
22.
Stanley, R. J.; Castleman, A. W. CLUSTER ION DIP SPECTROSCOPY OF
HYDROGEN-BONDED PHENOL(H2O)N CLUSTERS, N=0-4. J. Chem. Phys
1991, 94, 7744-7756.
23.
Tanabe, S.; Ebata, T.; Fujii, M.; Mikami, N. OH STRETCHING VIBRATIONS OF
PHENOL-(H2O)N(N=1-3) COMPLEXES OBSERVED BY IR-UV DOUBLERESONANCE SPECTROSCOPY. Chem. Phys. Lett. 1993, 215, 347-352.
24.
Pribble, R. N.; Zwier, T. S. SIZE-SPECIFIC INFRARED-SPECTRA OF
BENZENE-(H2O)(N) CLUSTERS (N=1 THROUGH 7) - EVIDENCE FOR
NONCYCLIC (H2O)(N) STRUCTURES. Science 1994, 265, 75-79.
9
25.
Costain, C. C. DETERMINATION OF MOLECULAR STRUCTURES FROM
GROUND STATE ROTATIONAL CONSTANTS. J. Chem. Phys 1958, 29, 864874.
26.
Levy, D. H. THE SPECTROSCOPY OF VERY COLD GASES. Science 1981, 214,
263-269.
27.
Smalley, R. E.; Wharton, L.; Levy, D. H. MOLECULAR OPTICAL
SPECTROSCOPY WITH SUPERSONIC BEAMS AND JETS. Acc. Chem. Res.
1977, 10, 139-145.
28.
Fenn, J. B.; Mann, M.; Meng, C. K.; Wong, S. F.; Whitehouse, C. M.
ELECTROSPRAY IONIZATION FOR MASS-SPECTROMETRY OF LARGE
BIOMOLECULES. Science 1989, 246, 64-71.
29.
Shevchenko, A.; Wilm, M.; Vorm, O.; Mann, M. Mass spectrometric sequencing of
proteins from silver stained polyacrylamide gels. Anal. Chem. 1996, 68, 850-858.
30.
Rizzo, T. R.; Stearns, J. A.; Boyarkin, O. V. Spectroscopic studies of cold, gas-phase
biomolecular ions. Int. Rev. Phys. Chem. 2009, 28, 481-515.
10
CHAPTER 2.
EXPERIMENTAL METHODS
Most of the experimental methods and instrumentations used in this thesis have
been described in detail previously.1-4 A general description will be given below. This
chapter will begin with supersonic jet expansion and vacuum chamber designs, followed
by the description of different spectroscopic methods including resonant two-photon
ionization, hole-burning spectroscopy, resonant ion-dip infrared spectroscopy, jet cooled
and room temperature chirped-pulse Fourier transform microwave spectroscopy.
Different computational methods adopted to interpret the experimental results are also
discussed at the end of the chapter. Experimental methods unique to specific projects will
be described in their respective chapters.
2.1
Supersonic Jet Expansion and Molecular Beams
Spectroscopy is a powerful tool in molecular structural determination and the
study of dynamics and kinetics of chemical processes.5-6 However, at room temperature,
rotational and vibrational energy levels higher than the ground state will have a huge
population and complicates the spectrum.7 To solve this problem, supersonic jet
expansion is adopted to achieve narrow distributed spectra corresponding to extremely
low internal temperatures.
To form a supersonic free jet, the molecule sample is vaporized and entrained in a
high pressure carrier gas (an inert gas like He, Ne, Ar). Then the gas mixture from a
reservoir is expanded into vacuum through an expansion orifice. When the diameter of the
orifice (D) is much greater than the mean free path of the gas molecule in the reservoir (λ0),
many collisions happen between molecules in the expansion. As a result, the internal
11
energy of the sample molecule is transferred into the translational energy of the buffer
gas.8-10 Cooling the thermalized molecule into ground state of different conformers. At
the same time, the translational temperature, defined by the peak width of the velocity
distribution, is also largely reduced as the enthalpy for random motion is transferred into
directed flow. Thus by using supersonic jet expansion it is possible to cool down the
vibrational, rotational and translational temperature to the order of 10-20K, 2-5K and 1K
respectively. The inhomogeneous broadening induced by higher temperatures is also
greatly decreased.
For all experiments described in this thesis, the samples were first vaporized
through either heating or laser desorption, then seeded in high pressure carrier gas and
introduced into the chamber through supersonic jet expansion before being interrogated
with tunable laser or microwave radiation. In experiments with time-of-flight mass
spectrometer, a skimmer is placed ~2 cm downstream to select the coldest part of the
expansion to form a molecular beam.
2.2
2.2.1
Vacuum Chambers
Time-of-Flight Mass Spectrometer
The experiments described in Chapter 3 were carried out in a molecular beam time-
of-flight mass spectrometer (TOF-MS) with mass selective measurements. A diagram of
TOF vacuum chamber is presented in Figure 2.1. The instrument is composed of 2
differentially pumped sections: the source section and the detection section. The source
section is pumped out to a base pressure around 10-5 mbar (working pressure about 10-4
12
mbar ) through a Pfeifer THM 1001 (1000 l/s, Ø = 10΄΄) turbomolecular pump backed by
a mechanical pump.
The carrier gas is introduced into the source section through jet nozzle sealed with
the source chamber. The jet expansion is then skimmed by a Beam Dynamics skimmer
(55° conical angle , 2mm diameter) to form a molecular beam and enters into a WileyMcLaren type11 ion source region inside the copper shroud. At the same time, the UV laser
is introduced into the same region orthogonal to the molecular beam and TOF tube. Upon
ionized by the UV laser, the cations generated are accelerated up a 1 m flight tube and
detected by the micro channel plates detector (MCP, R.M. Jordan, Co.) at the end of the
tube. The acceleration is induced by 3 electrodes with a voltage of +4225V (repeller),
+3950V (draw-out-grid) and 0V (ground) respectively. A set of steering plates are used
after the acceleration stage to steer the ion package towards the MCP in the detection
chamber. A second turbo molecular pump (Pfeiffer THM 261, 250 l/s, Ø = 6΄΄) is used to
pump down the detection section to a vacuum pressure of around 10-9 bar. The detection
chamber can be isolated from the source chamber through a pneumatic gate valve. The ion
signal collected by the MCP is amplified by 25 times through Stanford Research Systems
SR445 amplifier and monitored on a Tektronix 3032B oscilloscope. A home-written Lab
View data acquisition program is used to compile and plot the data.
For powder sample that is subject to thermal decomposition when heated to high
temperature through heating rope, laser desorption is used to bring the molecule into gas
phase. A graphite block with a dimension of approximately 50x3x2mm is used as the
desorption medium. The powder sample is rubbed into the surface of the graphite rod to
attain a smooth, visually uniform top surface layer. The block is then fixed in an aluminum
13
boat and attached to the end of a stainless steel rod. The graphite rod is inserted into the
chamber and placed directly underneath the nozzle orifice via a load-lock assembly to seal
the vacuum between the chamber and atmosphere. A linear actuator (NSC 200, Newport)
is applied to move the steel rod linearly to ensure exposure of new sample to the desorption
laser. A Nd:YAG laser (Continuum Minilite II) operating at 20 Hz (5mJ/pulse, 2mm beam
diameter) is used for desorption and was aligned through a window above the pulsed valve
directly onto the graphite rod. Ultra high purity argon is used as buffer gases (2-3 bar
backing pressure) in the supersonic jet expansion, pulsed at 20 Hz out of a pulsed valve
(General, Series 9) with a 500 μm diameter orifice.
14
Figure 2.1. Schematic diagram of time-of-flight mass spectrometer chamber.
15
2.2.2
Chirped-Pulse Fourier Transform Microwave Spectrometer
The experiments described in Chapter 4 and Chapter 5 were performed in the
chirped-pulse Fourier transform microwave (CP-FTMW) chamber. Figure 2.2 presents the
diagram of the CP-FTMW chamber. In general, the chamber is composed of 2 distinct
parts: a Balle-Flygare cavity12-14 with a laser entrance port and a broadband microwave
spectrometer. The latter was used to achieve high resolution broadband microwave spectra
described in this thesis.
The entire chamber is evacuated to a base pressure about 10-6 Torr (working
pressure about 10-5 Torr ) through a pumping system composed of 2 water cooled stainless
steel diffusion pumps (Varian VHS 10), backed by a roots blower (BOC Edwards EH 500)
and a roughing pump (Alcatel 2063). To reduce the spurious signal from reflections, the
broadband portion of the chamber is lined up with Eccosorb microwave absorber backed
with metal shielding (Emerson & Cuming HR-25/ML). The molecules presented in this
study are either solid or liquid at room temperature. Unlike in the TOF chamber, the sample
of interest is wrapped into either cotton or glass wool and placed into the sample holder
immediately behind the general valve. To bring the sample into gas phase, a heating rope
is used to heat the sample to its flash point. The sample is introduced into the chamber
through a pulsed Series 9 general valve (1mm orifice diameter) in a jet expansion, with
ultrahigh purity neon as carrier gas (around 0.7 bar). As a result, the rotational temperature
of the molecules is cooled to around 2-5 K in the expansion.
16
Figure 2.2. Schematic diagram of chirped-pulse Fourier transform microwave (CPFTMW) chamber.
2.3
2.3.1
Spectroscopic Methods
Resonant Two-Photon Ionization
Resonance enhanced two-photo ionization (R2PI) is used to achieve mass selective
detection of electronic spectra with vibrational resolution in the time-of-flight mass
spectrometer chamber described in section 2.2.1. In this technique, UV laser is scanned in
an appropriate region of interest for the specific molecule. When the UV photon is in
resonance with a S0 →S1 or S0 →Sn vibronic transition, a fraction of sample is excited to
the excited states. Then a second UV photon excites the molecule to the ionization
continuum immediately. The energy level diagram of this process is illustrated in Figure
2.3(a). When the excitation and ionization are from the same laser pulse, it is named one
17
color R2PI (1C-R2PI). In case the S0 →S1 transition energy is less than the half of the
energy needed to ionize the molecule, a second UV laser is introduced to ionize the
molecule, which is named two color R2PI (2C-R2PI). The two beams are counter
propagated with each other and overlapped in time and space. The laser and pulse timings
are controlled through a digital delay generator (Berkley Nucleonics BNC Model 555). By
collecting ion signals from light absorbed sample molecules in the probed mass channel,
we get the R2PI spectrum with information from all conformers presented.
2.3.2
Hole-Burning Spectroscopy
To determine the number of conformers presented in R2PI spectra and get their
individual spectra signatures. We employ a technique named hole-burning spectroscopy.
Two lasers are required in this technique, a hole-burn laser and a probe laser. If the holeburn laser is a UV laser, it is named UV-UV hole-burning spectroscopy (UV-UVHB). If
the hole-burn laser is a IR laser, it is named IR-UV hole-burning spectroscopy (IR-UVHB).
A schematic diagram of this method is presented in Figure 2.3(b) and 2.3(c). In UV-UVHB,
the UV hole-burn pump laser is fixed on a UV transition determined from the R2PI
spectrum (usually a S0 (v=0) →S1 (v=0) origin), then the UV probe laser is spatially
overlapped with the hole-burn laser in a counter propagated manner, but arrives ~200 ns
later than the hole-burn laser. Both lasers interact with the same part of the molecule in the
collision free region of jet expansion. The probe laser is operated at 20 Hz while the holeburn laser is operated at 10 Hz. When the hole-burn pump laser is resonant with a transition
that shares the same ground state as the probe laser, a depletion of signal is observed from
the probe laser. By collecting the difference in ion signals between hole-burn laser “on”
18
and “off” through a gated integrator (Stanford Research System, SRS) in active baseline
subtraction (ABS) mode, we are able to record a single conformation UV spectrum.
IR-UVHB is analogue to UV-UVHB but the UV hole-burn laser is replaced with an
IR hole-burn laser. The IR pump laser is fixed on an IR transition determined from RIDIR
spectrum (section 2.3.4) and the probe UV laser is scanned through the UV spectrum. Any
UV transition shares the same ground state with the IR transition will have a depletion in
signal. Like in the scheme of UV-UVHB, IR pump laser is pulsed at 10 Hz while the UV
probe laser is pulsed at 20 Hz. This method is especially helpful when the UV-UVHB
spectrum is hard to obtain. For example, the R2PI spectrum is over congested or the UV
S0 →S1 origin band of a minor conformer is weak and embedded with a strong band from
a major conformer. At the same time, this method also requires conformers to have
different IR signatures with their transitions separated spectrally.
2.3.3
Resonant Ion-Dip Infrared Spectroscopy
Resonant ion-dip infrared spectroscopy (RIDIRS) is used to record conformer
specific infrared spectra. The experimental scheme is presented in figure 2.3(d). The setup
is identical to IR-UVHB except the UV laser, running at 20 Hz, is fixed on a UV transition
determined from R2PI spectrum (usually a S0 (v=0) →S1 (v=0) origin), then the IR laser,
running at 10 Hz, is scanned through the region of interest ~200 ns prior to the UV laser.
Again, when the IR laser is in resonance with an IR transition from the same conformer
UV laser probes, a depletion in ion signal is observed by the probe laser and RIDIR
spectrum is obtained by collecting the difference between “IR on” and “IR off” using SRS
gated integrator in ABS mode.
A LaserVision optical parameter oscillator/optical
parameter amplifier (OPO/OPA) system is used to generate IR beam. In this nonlinear
19
optical process, the 1064nm injected light from a seeded Nd:YAG (Continuum Surelite
EX, Continuum 8000 series) is split by a 70/30 beam splitter. 70% of the beam is passed
into the OPA stage while the remaining 30% of the beam is reflected into the OPO stage
and frequency doubled to 532 nm through a KTP doubling crystal. In the OPO stage, the
532 nm (18794 cm-1) photon is used to pump two phase-matched KTP crystals and split
into signal and idler photons whose energies are dependent on the angle of the KTP
crystals, but are always summed to 18794 cm-1 due to conservation of energy. The idler
photon is then seeded into the OPA stage as the signal and is difference frequency mixed
with the 70% of the 1064nm light through a series of 4 KTA based OPA. The idler beam
from the OPA stage is selected using a “stack-of-plates” silicon polarizer and used as the
IR beam in hydride stretch regions (2800-3700 cm-1). Pulse energies for the idler beam is
around 12-15 mJ/pulse. To generate IR light in the amide I/II regions (1400-1800 cm-1),
difference frequency mixing of signal and idler beams from the OPO was carried out in a
AgGaSe2 crystal. IR laser powers are 0.5−1.0 mJ/pulse in the amide I/II region.
20
Figure 2.3. Schemes for (a) resonant two-photon ionization spectroscopy, (b) UV-UV
hole-burning spectroscopy, (c) IR-UV hole-burning spectroscopy and (d) resonant iondip infrared spectroscopy employed in this work.
2.3.4
Chirped-Pulse Fourier Transform Microwave Spectroscopy
For molecules without an aromatic chromophore but with substantial dipole
moments, broadband chirped-pulse Fourier transform microwave spectroscopy (CPFTMW) is a powerful way for the structural determination of different conformers in the
gas phase.15 These experiments can be conducted either in the CP-FTMW chamber
(section 2.2.2) with supersonic jet expansion or in a room temperature waveguide as will
be discussed in detail later in section 2.3.4.2 and chapter 6.
2.3.4.1 Jet Cooled CP-FTMW
The supersonic jet cooled CP-FTMW spectrometer is based on the design from
Prof. Brooks Pate in University of Virginia.16 A schematic of the CP-FTMW electronic is
presented in Figure 2.4. In the broadband (7.5-18.5 GHz) set up, an arbitrary waveform
21
generator (AWG, Tektronix 7101) is used to generate a linearly swept chirped pulse from
1.875 GHz to 4.625 GHz in 1 μs with a 1V peak to peak intensity (1.7 dBm). The pulse is
then sent into a 5 GHz low pass filter (Lorch 10LP-5000-S) to remove high frequency
anharmonics before being amplified by a pre-amplifier (Minicircuits ZX60-6013E-S) to
form more uniform power intensity (+10 dB). Then the bandwidth of the AWG output
(1.875-4.625 GHz ) is expanded by 4 times through a quadrupler (Phase One PS06-0161)
to 7.5-18.5 GHz. A step attenuator (Weinschel AF117A-69-11) is used to attenuate the
signal intensity before being fed into a 200W traveling wave tube amplifier (TWTA,
Amplifier Research 200T8G18A). A reflection isolator (Ditom Microwave Inc., DMI
6018) is placed between the attenuator and TWTA to protect the electronics at the
broadcasting side.
The amplified polarizing pulse from TWTA is sent into the chamber through a 24
inch jacketed flexible waveguide (Western Test Systems WRD-750). A microwave horn
antenna with gain enhancer (Amplifier Research model AT4004) is used to broadcast the
polarizing pulse into the chamber. The sample is introduced into the chamber through
supersonic jet expansion in a perpendicular way to the polarizing pulse to minimize
Doppler effect. The distance between the face plate of the general valve to the interaction
region is about 4 inches. Upon interaction with chirped pulse, a macroscopic polarization
is induced in the sample and the free induction decay (FID) is detected by another gainenhanced horn antenna placed opposite to the broadcasting horn. The distance between
the broadcasting horn and receiver horn is about 7 inches.
22
Figure 2.4. Schematic of the jet cooled chirped pulse Fourier transform microwave (CPFTMW) spectrometer.
The collected FID frequency range is also from 7.5-18.5GHz. To display it in a 12
GHz oscilloscope (Tektronix TDS6124C). It needs to be amplified and down converted
through a set of electronic components at the receiver end.
The FID signal collected will be first sent into a p-i-n diode limiter (Aeroflex
ACLM 4619F-C36-1K) to contain excess power from the polarizing pulse. The peak
input power threshold for the p-i-n diode is 1 kW. A solid state switch (Advanced
Technical Materials S1517D) is placed after the p-i-n diode to protect the electronics at
23
the receiver side. The switch will be closed for about 1.2 μs during the polarizing pulse
and initial noise decay after the pulse, then the molecular FID is passed through the
switch and amplified by a low noise amplifier (LNA, Miteq AMF-6F-06001800-15-10P
+45 dBm). In order to be displayed on a 12 GHz oscilloscope, the signal is down
converted by mixing with the output from a 18.9 GHz phase-locked dielectric resonator
oscillator (PLDRO, Microwave Dynamics PLO-2000-18.90 ) through a mixer (Miteq
TB0440LW1). A band pass filter (Lorch 7CF7-18900/100-S) is used to remove spurious
signals from the PLDRO sine wave. After the mixing stage, 2 side bands are created (0.411.4 GHz / 26.4-37.4 GHz) and a 12 GHz low pass filter (Lorch 7LA-12000-S) is
adopted to remove the upper sideband (26.4-37.4 GHz) before being recorded and
digitized on the oscilloscope (Tektronix TDS6124C).
To average FID signal in time domain, phase stability is crucial. A Rb-disciplined
crystal oscillator (Stanford Research Systems FS725) is used to provide 10 MHz
synchronized signal for 12 GHz oscilloscope and drives a phase-locked loop (Wenzel
Associates 501- 10137B), where the 10 MHz signal is up converted to 100 MHz and
connected to the AWG, 18.9GHz PLDRO and a 20 output Masterclock (Thales Laser).
The Masterclock is fixed at 10 Hz and is used to trigger the AWG and the pulsed
valve. The timing of the pulsed valve is usually set to be 1300 μs ahead of the chirped
pulse to make sure the polarizing pulse will interact with the molecule sample cooled in
the supersonic jet expansion. A digital delay/pulse generator box (DG535, Stanford
Research Systems) is used to send out TTL pulses to control the timing of the TWTA and
the solid state switch, which is usually determined through a oscilloscope to make sure
the “┴” TTL pulse for TWTA will cover the whole chirp while the “┬” TTL pulse for
24
solid state switch is ~100 ns wider and covers the whole “┴” TTL pulse. The DG535
box could be triggered by either the AWG or the Masterclock. To record the FID signal,
the 12GHz oscilloscope is triggered by the marker from AWG channel, usually 0.5-1 μs
after the high power chirped pulse to avoid the initial noise decay. The Masterclock
triggered AWG and the pulsed valve are running at 10 Hz, however, due to the
acquisition limitation of the 12 GHz oscilloscope (Tektronix TDS6124C), the maximum
repetition rate in this experimental set up is limited to around 5 Hz with a FID collection
time of 4 μs, and will be further decreased to around 3 Hz if we increase the FID
collection time to 20 μs.
To overcome this problem and increase the repetition rate of our experiments, an
ultrafast 13 GHz digitizer (Guzik ADC6131) was introduced into the lab in summer 2014
to replace the original Tektronix 12 GHz oscilloscope. A multi-FID detection scheme is
also developed to collect up to 20 FIDs within each gas pulse. The detection scheme is
presented in figure 2.5. In short, the Masterclock is still used to trigger the pulsed valve
and the AWG. However, instead of sending out only one chirped pulse, the AWG is
sending out 20 chirped pulses per trigger in a sequence mode., A 16 μs long FID is
collected after each chirped pulse and the total length is within the width of the gas pulse
(~ 1500 μs). Figure 2.6 presents the width of the gas pulse measured by monitoring the
signal intensity of OCS (12162.9 MHz) against different AWG delay time. Again, the
Guzik digitizer is triggered by the marker associated with each chirped pulse from the
AWG. However, it is noticed that all the external syncs after the first one are ignored by
the digitizer. So it is crucial to set the gap length between segments and accumulation
25
zones correct in the configuration file for the Guzik digitizer. An example of the
configuration file is presented in Figure 2.7 and explained in detail below.
Figure 2.5. Multiple segments with gaps timing diagram. (Modified from Guzik ADC
6000 series manual)
26
Figure 2.6. Width of the gas pulse measured by monitoring the signal intensity of OCS
transition at 12162.9 MHz against AWG delay time.
Figure 2.7. A typical example of the Guzik ADC 6131 configuration file.
27
In the configuration file presented above, "acc_len_ns" specifies the length of
accumulation zone in nanoseconds and the range is from 24 to 16384 ns with a 4 ns
increment. The gap between accumulation zone in nanoseconds is defined by "acc_gap_ns"
with a supported range from 0 to 1,000,000,000 ns (1s). In segment average mode,
"acc_segment_num" determines the number of segments to be processed while the gap
between segments is set through "acc_segment_gap_ns" ranging from 0 to 1,000,000,000
ns (1s). Within each segment, "acc_PRI_num" determines the number of Pulse Repetition
Intervals to be processed. In the set up presented above in Figure 2.7, 20 FIDs are collected
per segment / gas pulse and each FID lasts for 16 μs with a 34 μs interval gap to the next
one. In total, 20X10000=200K averages are performed, which will only takes ~ 16 min
(1000s). So, when compared with the previous set up with Tektronix 12 GHz oscilloscope,
our detection speed is increased by 40-60 times with the Guzik digitizer through this multiFID detection method.
2.3.4.2 Room Temperature CP-FTMW
Through a fruitful collaboration with Prof. William J. Chappell in ECE department
of Purdue University and Prof. Steve Shipman in the chemistry department at New College
of Florida, room-temperature chirped-pulse Fourier transform microwave (RT-CP-FTMW)
experiments are also explored in our lab with either a 10m long WRD 750 waveguide or a
home built large electrical volume coaxial (LEVC) transmission line. Most of the
experiments with the LEVC cable are already described in detail in the Ph.D. thesis of Dr.
Yu-ting Huang17 from Prof. Chappell group. So this section is only focused on the RTCP-FTMW set up with the WRD 750 waveguide, which is constructed by Dr. Ryan Hilger
28
from Jonathan Amy facility for chemical instrumentation and Alicia Hernandez Castillo
from Zwier group.
Contrary to the low temperature spectrum, room temperature spectra suffer from
largely increased variety of transitions and weak spectral line intensities. With the
advantage of fast coherent averaging of the new digitizer, much stronger signal to noise
ratio (SNR) has been achieved and in other way, the input power can be reduced to achieve
a desired SNR that meets the detection threshold.
Figure 2.8. Schematic for the room temperature chirped pulse Fourier transform
microwave (RT-CP-FTMW) spectrometer.
The schematic for this RT-CP-FTMW design is presented above in figure 2.8.
Instead of using supersonic jet expansion, the liquid sample is introduced into the room
29
temperature cell through repeated freeze-pump-thaw cycles using liquid nitrogen as
coolant. At the same time, since the repetition rate of the experiment is no longer limited
by the 10 Hz general valve, this experiment can be performed with an ultrafast frequency
up to 100 kHz. In the current set up, AWG is sending out a 10 μs frame in continuous mode.
Within each frame, there is a 1μs chirped pulse followed by a 4 μs FID collection time.
Again, the gap between them is usually set to be 0.5 μs to avoid the initial noise decay. It
is noticed that the damage threshold for the p-i-n diode limiter (Aeroflex ACLM 4619FC36-1K) is only 2 W in continuous mode. To protect electronics at the receiver end, the
gain of the 200 W TWTA (Amplifier Research 200T8G18A) is limited to 5%.
Alternatively, a 3W solid state amplifier (Microwave Power L0818-32-T358, +40 dBm)
could be used to replace the 200W TWTA in this RT-CP-FTMW set up.
2.4
Computational Methods
To interpret the spectra obtained from the experimental methods listed above,
theoretical calculations in different level of theories are performed to construct the whole
picture for molecular spectroscopy investigation.
For most of the molecules studied in this thesis, a conformational search is
performed first using Amber* force field through Maestro Macromodel18 suite of programs.
In general, 10,000 iterations are adopted with a convergence threshold of 0.0001 gradient.
The energy window for saving structures is always set to be 50 kJ/mol. The outputs from
the conformational search are subject to higher precision geometry optimizations and
energy / frequency calculations using ab inito or DFT methods through Gaussian 09.19 In
general, DFT provides a reasonable method in predicting harmonic vibrational frequencies.
30
However, to correct dispersion effects for modest-sized molecules, M05-2X hybrid
functional and B3LYP-D3BJ level of theory are adopted for energy predictions.
In addition to the ground state geometry optimizations and energy / frequency
calculations. Excited states (TD-DFT), transition states (TS, QST2, QST3) and intrinsic
reaction coordinate (IRC) calculations are also widely used in this thesis to explain the
physical background of the spectroscopic experimental results.
31
2.5
References
1.
Dean, J. C.; Buchanan, E. G.; James, W. H., III; Gutberlet, A.; Biswas, B.;
Ramachandran, P. V.; Zwier, T. S. Conformation-Specific Spectroscopy and
Populations of Diastereomers of a Model Monolignol Derivative: Chiral Effects in
a Triol Chain. J. Phys. Chem. A 2011, 115, 8464-8478.
2.
Gord, J. R.; Walsh, P. S.; Fisher, B. F.; Gellman, S. H.; Zwier, T. S. Mimicking the
First Turn of an alpha-Helix with an Unnatural Backbone: Conformation-Specific
IR and UV Spectroscopy of Cyclically Constrained beta/gamma-Peptides. J. Phys.
Chem. B 2014, 118, 8246-8256.
3.
Vaquero-Vara, V.; Zhang, D.; Dian, B. C.; Pratt, D. W.; Zwier, T. S. Delicate
Balance of Hydrogen Bonding Forces in D-Threoninol. J. Phys. Chem. A 2014, 118,
7267-7273.
4.
Walsh, P. S.; Kusaka, R.; Buchanan, E. G.; James, W. H., III; Fisher, B. F.; Gellman,
S. H.; Zwier, T. S. Cyclic Constraints on Conformational Flexibility in gammaPeptides: Conformation Specific IR and UV Spectroscopy. Journal of Physical
Chemistry A 2013, 117, 12350-12362.
5.
Zwier, T. S. Laser Probes of Conformational Isomerization in Flexible Molecules
and Complexes. J. Phys. Chem. A 2006, 110, 4133-4150.
6.
Dian, B. C.; Florio, G. M.; Clarkson, J. R.; Longarte, A.; Zwier, T. S. InfraredInduced Conformational Isomerization and Vibrational Relaxation Dynamics in
Melatonin and 5-Methoxy-N-Acetyl Tryptophan Methyl Amide. J. Chem. Phys.
2004, 120, 9033-9046.
7.
Reinhold, B.; Finneran, I. A.; Shipman, S. T. Room Temperature Chirped-Pulse
Fourier Transform Microwave Spectroscopy of Anisole. J. Mol. Spectrosc. 2011,
270, 89-97.
8.
Lubman, D. M.; Rettner, C. T.; Zare, R. N. How Isolated are Molecules in A
Molecular-Beam. J. Phys. Chem. A 1982, 86, 1129-1135.
9.
Levy, D. H. The Spectroscopy of Very Cold Gase. Science 1981, 214, 263-269.
10.
Smalley, R. E.; Wharton, L.; Levy, D. H. Molecular Optical Spectroscopy with
Supersonic Beams and Jets. Acc. Chem. Res. 1977, 10, 139-145.
11.
Lubman, D. M.; Jordan, R. M. Design for Improved Resolution in A Time-Of-Flight
Mass-Spectrometer Using A Supersonic Beam and Laser Ionization Source. Rev.
Sci. Instrum. 1985, 56, 373-376.
32
12.
Balle, T. J.; Campbell, E. J.; Keenan, M. R.; Flygare, W. H. New Method for
Observing the Rotational Spectra of Weak Molecular-Complexes - Krhcl. J. Chem.
Phys. 1979, 71, 2723-2724.
13.
Balle, T. J.; Campbell, E. J.; Keenan, M. R.; Flygare, W. H. New Method for
Observing the Rotational Spectra of Weak Molecular-Complexes - Krhcl. J. Chem.
Phys. 1980, 72, 922-932.
14.
Balle, T. J.; Flygare, W. H. Fabry-Perot Cavity Pulsed Fourier-Transform
Microwave Spectrometer with A Pulsed Nozzle Particle Source. Rev. Sci. Instrum.
1981, 52, 33-45.
15.
Zhang, D.; Bocklitz, S.; Zwier, T. S. Broadband Microwave Spectroscopy of
Prototypical Amino Alcohols and Polyamines: Competition between H-Bonded
Cycles and Chains. J. Phys. Chem. A 2016, 120, 55-67.
16.
Brown, G. G.; Dian, B. C.; Douglass, K. O.; Geyer, S. M.; Shipman, S. T.; Pate, B.
H. A Broadband Fourier Transform Microwave Spectrometer Based on Chirped
Pulse Excitation. Rev. Sci. Instrum. 2008, 79.
17.
Huang, Y. T. Microwave Chemical Sensing Using Overmoded T-line Designs and
Impact of Real-time Digitizer in the System. Ph.D. Dissertation, Purdue University,
West Lafayette, IN, 2014.
18.
Mohamadi, F.; Richards, N. G. J.; Guida, W. C.; Liskamp, R.; Lipton, M.; Caufield,
C.; Chang, G.; Hendrickson, T.; Still, W. C. Macromodel-an Integrated Software
System for Modeling Organic and Bioorganic Molecules Using Molecular
Mechanics. J. Comput. Chem. 1990, 11, 440-67.
19.
Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.;
Cheeseman, J. R.; Scalmani, G.; Barone, V.; Mennucci, B.; Petersson, G. A., ; et
al. Gaussian 09, revision C.01; Gaussian, Inc.: Wallingford, CT, 2010.
33
CHAPTER 3.
SINGLE CONFORMATION SECTROSCOPY OF
SUBEROYLANILIDE HYDROXAMIC ACID (SAHA): A
MLECULE BITES ITS TAIL
3.1
Introduction
Suberoylanilide hydroxamic acid (SAHA) is a histone deacetylase (HDAC) inhibitor
that binds directly to the catalytic site of the enzyme, thereby blocking substrate access.
SAHA is known to inhibit class I and class II HDACs and arrests cell growth of a wide
variety of transformed cells.1 SAHA has demonstrated significant anticancer activity in
both hematologic and solid tumors.2-3 Receiving approval by the U.S. Food and Drug
Administration (FDA) for the treatment of cutaneous T-cell lymphoma (CTCL)4-5 in 2006,
SAHA has become the lead compound in a promising new class of anticancer drugs.
Figure 3.1. Chemical structure of suberoylanilide hydroxamic acid (SAHA).
One of the most striking features of the structure of SAHA is its linear juxtaposition
of non-polar aromatic, polar amide, non-polar alkyl chain, and polar hydroxamic acid
groups. The two polar groups contain hydrogen bond donor and acceptor groups, enabling
34
hydrogen bonds to be formed, either internal to the molecule or with its surroundings,
involving both the amide group “head” and the chelating hydroxamic acid group “tail”.
The hexyl chain that links them gives great flexibility to the molecule in interacting with
its surroundings. For instance, the crystal structure of SAHA itself involves an array of
“linear” SAHA molecules in which the six-carbon chain is extended in an all-trans
structure that enables H-bonds “head-to-tail” and “tail-to-head” H-bonds between SAHA
molecules in adjacent layers.6
More importantly, in the crystal structure of the complex of SAHA with histone
deacetylase like protein (HDLP) complex,7 SAHA also adopts an extended conformation,
but this time with the aromatic “head” sitting at the entrance to a long, cylindrical HDLP
pocket. The 6-carbon alkyl chain extends down the length of the pocket, where its
hydroxamic acid “tail” can engage in bidentate chelation to a Zn+2 cation located at the
bottom of the polar HDLP pocket.
From a fundamental viewpoint, what is not yet established is the inherent
conformational preferences of the SAHA molecule in its isolated form.
This work
addresses that need. In the gas phase, the many intermolecular interactions with the
binding pocket or other SAHA molecules are removed, leaving only the molecule’s
intramolecular interactions to dictate its inherent conformational preferences.8-10 With one
hydrogen bonding group adjacent to ring and the other at the end of its long, flexible C6
hydrocarbon tail, SAHA is able to form “head-to-tail” and “tail-to-head” hydrogen bonds
involving the several donor and acceptor sites in the amide and hydroxamic acid groups
(Figure 3.1). With a flexible alkyl chain connecting them, one anticipates the potential
35
presence of several competing “head-to-tail” and “tail-to-head” conformational isomers,
with isomerization occurring on a potential energy landscape that is prototypical in form.
One might anticipate that “head-to-tail” and/or “tail-to-head” cyclic structures will
be low in energy due to the hydrogen bond(s) so formed. As Figure 3.1 shows, two
NH···O=C H-bonds are possible that constitute 11-membered rings (denoted as ‘C11’).
The OH group offers another H-bond donor site, to which it could bind to the head amide
group either alone, or in concert with the hydroxamic acid carbonyl group. These unique
bonding arrangements assume that the hexyl chain is able to loop back on itself with
minimal conformational strain. However, in pure alkanes, the extended ‘all-trans’ structure
is most stable for alkyl chains up to 18 in length, with each gauche defect destabilizing the
structure by about 2 kJ/mol.11 Thus, formation of H-bonded cycles in SAHA of necessity
occurs with some conformational strain in the alkyl chain. Furthermore, the sheer number
and increased floppiness of extended conformations argues for their dominance at higher
temperatures on entropic grounds.
Thus, it is fascinating to explore the inherent
conformational preferences of SAHA in the gas phase.
In this chapter the conformation-specific infrared (IR) and ultraviolet (UV) spectra
of the isolated SAHA molecule are presented, carried out under expansion-cooled
conditions in the gas phase. Transitions due to three different conformers of SAHA are
observed. Assignments are made for these conformers based on several pieces of
spectroscopic data. The conformational specific infrared spectra in hydride and amide I/II
regions serve as diagnostics for structural determination of individual conformers. The
single-conformation ultraviolet spectra also shed light on the conformations present, due
to large variations in the S0-S1 origin transition frequencies. The structures, relative
36
energies, and harmonic vibrational frequencies for many low lying conformational minima
of SAHA have been calculated using electronic structure methods, providing several points
of comparison between theory and experiment. Finally, as a means of assessing the overall form of the potential energy landscape for the molecule, we modify a common
molecular mechanics force field to more accurately account for the hydroxamic acid OH
internal rotation, and use it to calculate a disconnectivity graph for SAHA. This pictorial
summary of the potential energy landscape provides a useful means of understanding the
observed and ‘missing’ conformations, aided by predictions of the isomerization pathways
between them.
3.2 Methods
3.2.1 Experimental Methods
The experimental methods used in the present study have been described in detail in
Chapter 2. SAHA was purchased from Cayman Chemical at 98% purity, and used without
further purification. In the present case, laser desorption was used to vaporize the sample.
The powder sample was rubbed into the surface of a graphite rod to attain a smooth,
visually uniform top surface layer. The graphite rod was placed directly underneath the
nozzle orifice via a load-lock assembly. A linear actuator (NSC 200, Newport) was applied
to move the steel rod linearly to ensure exposure of new sample to the desorption laser. A
Nd:YAG laser (Continuum Minilite II) operating at 20 Hz (5mJ/pulse, 2mm beam diameter)
was used for desorption and was aligned through a window above the pulsed valve directly
onto the graphite rod. Ultra high purity helium and ultra high purity argon were used as
buffer gases (2-3 bar backing pressure) in the supersonic jet expansion, pulsed at 20 Hz out
37
of a pulsed valve (General, Series 9) with a 500 µm diameter orifice. The expansion was
skimmed by passing through a conical skimmer placed ~2 cm downstream to form a
molecular beam, which was subsequently photoionized in the ionization region of a timeof-flight (TOF) mass spectrometer. Trace water in the sample or gas handling lines led to
formation of the SAHA-water complex.
Monitoring the SAHA parent mass channel (m/z 264), one–color resonant twophoton ionization (R2PI) was used to record mass selected UV excitation spectra in the S0S1 region. The collimated, frequency doubled output of a Nd:YAG (532nm) pumped dye
laser was used as the ultraviolet light source. Fluorescein 548 and Rhodamine 6G were
used in the dye laser to cover the wavelength range from 275 to 281 nm at pulse energies
of 0.1-0.3 mJ/pulse at a 20 Hz repetition rate. Conformation-specific IR spectra were taken
using resonant ion dip infrared spectroscopy (RIDIRS) in the hydride stretch region (32003500 cm-1) and amide I/II (1450-1850 cm-1) regions. In this double resonance method, the
IR beam was generated by a Nd:YAG pumped optical parametric converter (LaserVision)
and was introduced into the chamber coaxially and counter propagating the UV laser beam.
To record a spectrum, the UV laser was fixed on a transition in the excitation spectrum
correlated with a single conformer while the infrared laser was turned through the region
of interest. The UV laser was pulsed at twice the frequency of the infrared laser, delayed
from the IR by 200 ns. When the IR frequency is resonant with a transition which shares
the same ground level as the UV laser, the IR pulse will remove a fraction of the ground
state population by absorption. The difference in ion signals between IR “on” and IR “off”
was monitored by scanning the IR laser and using a gated integrator (Stanford Research
Systems) in active baseline subtraction mode. To generate IR light in the amide I/II regions,
38
difference frequency mixing of signal and idler beams from the OPO was carried out in a
AgGaSe2 crystal. IR laser powers were 3-5 mJ/pulse in the amide NH stretch region, and
0.5-1.0 mJ/pulse in the amide I/II region. All RIDIR spectra were recorded by monitoring
the origin transition for each conformer in their UV excitation spectra respectively.
In order to record conformation-specific electronic spectra, IR-UV hole burning
spectroscopy was employed. The method uses an identical configuration to RIDIR
spectroscopy, except the wavelength of the IR hole-burn laser was fixed at a unique
ground-state vibrational transition of a particular conformer observed in the RIDIR spectra
while the UV probe laser was tuned through the wavelength region of interest. For all IRUV HB spectra, multiple infrared transitions in the RIDIR spectra were checked to make
sure the hole burn band was unique to a particular conformation.
3.2.2 Computational Methods
The long chain present in SAHA endows the molecule with a high degree of
flexibility, which makes it possible to adopt a large number of stable conformations and
increases the complexity of its potential energy surface (PES). Early attempts to use the
generalized Amber force field (GAFF) as a screening tool to identify low-energy minima
for optimization via density functional theory calculation showed that this force field was
inadequate to describe the hydroxamic acid functionality. This is not surprising, since the
hydroxamic acid moiety is not included among the test set of molecules (e.g., amino acids,
proteins) used in creating the force field.12 In particular, the weak stabilization between
the OH and C=O groups, forming a 5-membered H-bonded ring, is not correctly described.
39
Figure 3.2 represents the CNOH dihedral angle scan for one of the simplest
hydroxamic acids, N-hydroxypropanamide, based on the standard GAFF force field
(dashed line). The potential energy curve is nearly flat at dihedral angles within ±60° of
planar. As a consequence, the CNOH dihedral angle has no strong preference to maintain
planarity of the OH group (dihedral 0°), and this led to artificially low-energy structures in
the force field searches in which the tail OH group points out of the O(C=O)-C-N-O plane
to accommodate additional stabilizing interactions for the OH group, which in SAHA
includes the phenyl ring π cloud.
To correct this deficiency in the force field, we fit the CNOH hindered rotor potential
to one obtained from a relaxed dihedral scan of N-hydroxypropanamide carried out at the
B3LYP-D3BJ /6-31+G(d) level of theory, as illustrated by the blue line in Figure 3.2. The
modified GAFF force field dihedral angle scan result plotted in Figure 3.2 as the black line
better describes the dihedral angle preference close to zero degrees and generates structures
with energies in much closer agreement with both DFT predictions and experiment.
Figure 3.2. CNOH dihedral angle scan results for N-Hydroxypropanamide. (Prepared by
Karl Blodgett from Zwier group)
40
Armed with this modified force field, we pursued a different strategy than in
previous studies10 for screening structures for DFT structural optimization. Rather than
simply using the force field to locate minima, the generalized Amber force field (GAFF)
was used to generate a disconnectivity graph as a visualization tool to display the energies
of local minima on PES and their connectivity through transition states connecting them.
In such a disconnectivity graph, the end of each branch identifies the energy of a particular
conformational minimum. The nodal points represent collections of transition states in the
prescribed energy window that connect the minima below them. In this way, one can
trivially locate the highest energy barrier along the minimum-energy isomerization
pathway between any two minima on the graph.
The theoretical method for generating disconnectivity graphs has been described in
detail elsewhere.13 In brief, the PES was surveyed using the general AMBER force field
(GAFF), with atomic charges obtained from the semiempirical AM1 bond charge
correction approach. Local minima on the PES were located using a basin-hopping
algorithm14 within a canonical Monte Carlo simulation carried out by the GMIN 2.0
program of Wales and co-workers.15 We carried out up to 100 basin-hopping steps until
the global minimum was found and the step size was adjusted in each Monte Carlo step for
an acceptance ratio of 0.5.
For each minimum determined by the basin-hopping algorithm, transition states
were located by calculating the Hessian in GAFF and walking uphill in both directions
along the smallest eigenvalues using a hybrid BFGS/eigenvector-following transition state
search. All stationary points were converged to a root-mean-squared gradient of less than
4X10-6 kJ.mol-1 Ǻ-1. Then the minima connected to the transition states were identified
41
using the DNEB/L-BFGS method developed by Wales and coworkers.16-18 Previously
unknown minima were added to the growing database of minima, transition states, and
pathways, which were then used to generate the disconnectivity graph. Finally, we further
systematically expanded the tree by increasing the number of connections per minimum
through single-ended transition state searches and the overall connectivity of the
disconnectivity graph through parallel double-ended transition state searching.
The
disconnectivity graph so generated will be presented and discussed after the experimental
data has been considered.
Fifty unique conformational minima with lowest energies in the disconnectivity
graph were identified and served as starting geometries for further optimizations using
density function theory (DFT) calculations using the Gaussian 0919 suite of programs.
These calculations involved tight geometry optimizations followed by harmonic
vibrational frequency calculations using B3LYP20 or M05-2X21 hybrid functionals with the
6-31+G(d) basis set. The dispersion correction from Grimme and co-workers with
Becke−Johnson damping (D3BJ)22-23 was added to the B3LYP functional to account for
the London dispersion energy not correctly described by standard DFT calculations.
The order of the relative energies of the conformers was not changed significantly
between the M05-2X and B3LYP-D3BJ calculations. In this paper we use the B3LYPD3BJ calculations to report relative energies and calculated harmonic frequencies, as its
predictions were closer to experiment than the M05-2X calculations. The harmonic
frequency calculations aided in the assignment of conformational isomers observed in the
experiment. These frequencies were scaled by 0.96 for free NH stretch, 0.948 for hydrogen
42
bonded NH and OH stretches and 0.985 for amide I/II frequencies. These scale factors were
chosen by scaling the calculated IR frequencies of the assigned structure of SAHA
conformer A to the experimental frequencies, where the corresponding patterns provided
an unequivocal match between experiment and theory.
Finally, vertical excitation energies and excited state geometries were computed with
time-dependent density functional theory (TDDFT) with the same basis sets as mentioned
above.
3.2.3 Nomenclature
As discussed above, since the amide and hydroxamide groups of SAHA are
separated by a C6 alkyl chain, several different intramolecular H-bonding arrangements
are predicted to be possible in the gas phase by the theoretical calculations. Consequently,
the structures are grouped first into families by H-bonding pattern. The NH and C=O
groups adjacent to the phenyl ring are denoted as the head NH and head C=O, while the
corresponding groups in the hydroxamic acid are denoted as ‘tail’ groups. The size of a
hydrogen bonded ring formed by the NH or OH with the C=O groups is denoted as Cn,2425
where n stands for the number of atoms involved in the ring. In 38 out of the 50 structures
optimized through DFT calculations, the hydroxamic acid C=O and OH groups are cis to
one another, thereby forming a C5 ring. At the same time, in 19 of the calculated structures,
the head N-H group forms an additional intramolecular hydrogen bond with the tail C=O,
forming a C11 ring. This pattern is described as a Head to Tail pattern (labeled as H-T),
which are the preferred arrangement in most of the low energy conformers. Since the amide
groups at the head and tail are in reverse order to one another, C11 rings can also be formed
43
through intramolecular hydrogen bonds between tail NH and head C=O groups. In total, 9
of the calculated conformers adopt this Tail to Head pattern (labeled as T-H). In 7
calculated structures, the tail NH, instead of adopting a T-H hydrogen bond, points instead
to the π cloud of the phenyl ring, forming a weak NH-π intramolecular hydrogen bond
(labeled as NH-π ). 3 structures are predicted to form a head NH to tail OH intramolecular
hydrogen bond (Labeled as H-TOH). For the remaining 12 structures, the C5 rings at the
tail are broken and the tail OH group points to the phenyl ring, thereby forming an OH-π
intramolecular hydrogen bonds. Notably, these structures still retain the C11 H-T H-bond
as well, and are thereby labeled as H-T/OH-π. For all 50 of the calculated structures, the
amide group near the phenyl ring is in a trans-amide arrangement. The possibility for cisamide arrangement is also explored and will be discussed in detail in Section 3.4.3. It is
clear from the above description that several conformers of each H-bonding arrangement
(H-T or T-H) can be formed, which therefore must differ in the configuration of the alkyl
chain that links them. We therefore designate the full conformational structure by denoting
the alkyl chain dihedral angles along the hydrocarbon chain, numbering the C-atoms C1 to
C6 respectively from head to tail. Including the C=O carbons, there are five dihedral angles
along the alkyl chain, which are designated by α for C(=O/Head)-C1-C2-C3, β for C1-C2C3-C4, γ for C2-C3-C4-C5, δ for C3-C4-C5-C6 and ε for C4-C5-C6-C(=O/Tail). As is standard,
we will use gauche+ (g+), gauche- (g-), anti (a) and eclipsed (e) to describe dihedral angles
around +60°, -60°, ±180° and ±120°, respectively. Thus, the complete name of a structure
would contain both H-bonding pattern and C6 chain orientation, for example, T-H
(g+,g+,a,g+,g+).
44
3.3 Results and Analysis
3.3.1
R2PI and IR-UV Holeburning Spectra
Figure 3.3. R2PI (top trace) and IR-UV HB spectra (lower traces) for SAHA. Asterisks in
the R2PI spectrum are tentatively ascribed to hot bands of conformer A. Since the
electronic chromophore of SAHA is closely related to that in trans-formaniliide (tFA) and
trans- acetanilide (tAA), their electronic origins are shown in the figure for reference.29
The top trace of Figure 3.3 presents the one-color R2PI spectrum of SAHA in the S1
←S0 origin region, covering from 35590 to 36350 cm-1. Two distinct groups of transitions
are present in the spectrum, with a 450 cm-1 gap between them. The intensities of the
transitions in the higher wavenumber region are almost three times stronger than those in
the lower wavenumber region. Using infrared transitions determined from RIDIR
spectroscopy, a series of IR-UV hole burning spectra corresponding to each individual
45
conformation are presented below the R2PI scan in Figure 3.3. The R2PI spectrum divides
into transitions due to four unique structures (A-D).
The S0-S1 origin transition of conformer A is at 36095 cm-1 (A) and has short FrankCondon progressions involving vibrations of frequency 23 cm-1 and 67 cm-1 built off of it.
The IR-UV hole-burning spectrum of A (recorded with the IR at 3258 cm-1) accounts for
all the major transitions in the blue part of the R2PI spectrum, except the small transitions
marked by asterisks, which are tentatively ascribed to hot bands based on changes in their
intensity with desorption conditions. The background present in the R2PI spectrum but
missing in the hole-burning spectrum, is likely due to higher-frequency vibronic activity
arising from B-D.
The corresponding S0-S1 origins for B-D are shifted more than 400 cm-1 to the red
of A, appearing in close proximity of one another at 35648 cm-1(B), 35675 cm-1(C) and
35654 cm-1 (D), respectively. These spectra also exhibit some Frank-Condon activity in
low frequency vibrational modes, as summarized in Table 3.1. Based on the fact that the
S0-S1 origins of B-D differ from one another by no more than 27 cm-1, and the associated
hole-burning spectra display similar low frequency vibronic structures, it is clear that the
environments for the aromatic ring are similar in all three. This suggests that B-D are all
of one structural type while A is of a different type.
The frequency positions of the S0-S1 origins for A-D already provide some clue to
the structures involved. The electronic origins of B-D are well red-shifted from that of
trans-formanilide (tFA, 36004 cm-1) and trans-acetanilide (tAA, 35902 cm-1),26 while that
for conformer A is to the blue, as shown in Figure 3.3. Previous studies27 have identified
H2O complexes with tFA in which H2O acts either as H-bond donor to the C=O site or as
46
H-bond acceptor from the NH. Their electronic origins are shifted to the blue in the former
case (36,114 cm-1) and to the red in the latter (35,783 cm-1). Based on these simple
arguments, we anticipate that conformer A will be a T-H structure, while conformers B
and C are both H-T. We will see shortly based on their infrared spectra that this is indeed
the case.
It is worth noting that, in both tFA and tAA, previous studies have also observed a
minor cis-amide conformer (~6% of trans).28 The UV spectrum of the cis-amide conformer
of formanilide has its S1←S0 origin band ~1000 cm-1 red shifted from that of the trans
isomer.26 The corresponding transition in acetanilide has not been detected to date. In the
SAHA spectrum, there are several small peaks in the R2PI spectrum in Figure 3.3 around
36000 cm-1 (marked by asterisks) that do not burn out with conformer A. Their weak
intensity prevented recording RIDIR spectra of these bands; however, the balance of
evidence points to other these transitions as hot bands of A or water complexes rather than
other conformers of SAHA monomer. Scans taken over the 35590 – 36350 cm-1 region
revealed no further transitions not accounted for by A-D. As a result, experimental
evidence points towards all observed conformers arising from the trans-amide structure of
SAHA, the only isomeric form observed in solution or in the crystal.6 We will return to
this point in the discussion. Finally, in section 3.3.3, evidence is presented that structure
D is in fact due to a SAHA-H2O complex, since the IR spectrum contains two more hydride
stretch transitions. The ion signal appears in the monomer mass channel through its
efficient fragmentation following photoionization.
47
3.3.2 RIDIR Spectra of Conformers A-C
Figure 3.4. RIDIR spectra for SAHA conformer A,B and C in the (a) hydride stretch and
(b) amide I/II regions. Calculated IR spectra at the DFT B3LYP D3BJ/6-31+G(d) level of
theory are shown below as stick diagrams in black. These frequencies were scaled by 0.96
for free NH stretch, 0.948 for hydrogen bonded NH and OH stretches and 0.985 for amide
I/II frequencies. Asterisks indicate infrared transitions used to record IR-UV holeburn
scans.
Figure 3.4 presents a series of RIDIR spectra recorded in the hydride stretch (Figure
3.4a) and amide I/II regions (Figure 3.4b) for SAHA conformers A, B, and C. The S1←S0
origin transitions of the three conformers were used as monitor transitions for the RIDIR
spectra, resulting in the single-conformer IR spectra shown. The asterisks in the RIDIR
spectra denote the transitions used to record the IR-UV HB spectra in Figure 3.3. The
B3LYP-D3BJ /6-31+G(d) predictions for the best-fit vibrational frequencies and peak
intensities of the assigned structures are shown as stick diagrams immediately below the
experimental RIDIR spectra in Figure 3.4. Comparison between observed and calculated
vibrational frequencies in the hydride stretch and mid-IR are given in Table 3.1.
48
Since the SAHA molecule possesses one OH and two NH groups, three hydride
stretch fundamentals are anticipated in the RIDIR spectrum of each conformer. In all three
conformers, there is a single free NH stretch fundamental in the 3450-3500 cm-1 region.
While conformers B and C are at similar wavenumber positions (3497 cm-1), that for
conformer A is at 3457 cm-1, some 40 cm-1 lower. Thus, as with the UV spectra, conformer
A seems to be of one structural type, different from those of B and C. Based on calculations
of the conformations of SAHA, including extended conformers where both amide and
hydroxamic acid NH groups are free, it is clear that the frequency of the free NH stretch
fundamentals of head and tail NH groups are indeed different, with the free amide NH
about 40 cm-1 higher in frequency than the free hydroxamic acid tail. This provides a
second piece of evidence that conformer A is a T-H structure, while B and C are both from
the H-T family.
The RIDIR spectrum of conformer A also contains two hydride stretch fundamentals
shifted to much lower frequency (3250-3310 cm-1), which also show an increased intensity
and broadening, all of which are signatures of the formation of H-bonds.29-30 Similar
intense, broadened, low-frequency transitions are also present in conformers B and C,
indicating that all three conformers possess significant intramolecular H-bonds. However,
when compared with conformer A, all hydride stretches of B and C are shifted to higher
frequencies, consistent with weaker H-bonds in B and C than in A. For conformer B, scans
that extend down to 3200 cm-1 revealed no additional transitions, suggesting that the
broadened band at 3351 cm-1 may arise from an overlap of two H-bonded hydride
stretches31 that are more closely spaced than in conformer C.
49
In the amide I (1650-1750 cm-1) region, two transitions are resolved for all
conformers, corresponding to the two C=O stretches in the molecule. The 1450-1600 cm1
region is somewhat more complicated, containing transitions due to the NH bends of
amide and hydroxamic acid groups, and several weak benzene CH bend fundamentals.
Again, the patterns for conformer B and C resemble each other while the pattern for
conformer A is different. The sharp transitions in all three spectra around 1620 cm-1 are
due to aromatic C=C stretching modes.
Figure 3.5. Calculated optimized structures assigned for SAHA conformers A to C and
structure D, assigned to the SAHA-H2O complex, at the DFT B3LYP-D3BJ/6-31+G(d)
level of theory. The zero-point corrected relative energies are included.
50
Comparison of the experimental spectra with the scaled harmonic frequencies and
IR intensities of low-energy conformers of SAHA (stick diagrams in Figure 3.4) leads to
the structural assignments for conformers A-C presented in Figure 3.5. The match between
experiment and calculation, relative to alternatives (Figure 3.6 and Table 3.2), is
sufficiently good to make these assignments secure. As anticipated, conformers adopting
both H-T and T-H patterns are observed in the gas-phase.
Table 3.1 compares the observed and calculated vibrational frequencies for
conformers A-C in the hydride stretch and mid-IR regions. Low-frequency vibrations that
appear in the R2PI spectrum are also compared with the assigned structures, under the
assumption that electronic excitation will change these vibrational frequencies only
modestly from their ground state values. This latter comparison adds confirming evidence
to the assignments, but couldn’t be considered diagnostic on its own.
51
Table 3.1. Summary of calculated (Calc) and observed (Obs) vibrational frequencies and
assignments of SAHA monomer and the SAHA-H2O complex, calculated at the DFT
B3LYP-D3BJ/6-31+G(d) level of theory.
Torsional modes (cm-1)a
Hydride stretches (cm-1)b
ν1
ν2
Molecule
Obs Calc
Obs Calc
Obs
Calc
Obs
Calc
Obs
Calc
SAHA A
23
23
67
66
3256
3259
3306
3310
3457
3459
SAHA B
36
31
64
70
3497
3492
3351
3342
3351
3335
SAHA C
33
34
3497
3490
3373
3348
3390
3352
SAHA D
16
19
3496
3487
3339
3336
—c
3418
72
69
ν tail NH
ν OH
ν head NH
a
Unscaled harmonic frequencies and experimentally observed spacings of Franck-Condon
activity from the IR-UV holeburning scans. bHydride stretch fundamentals are scaled by
0.96 for free NH stretches, 0.948 for hydrogen bonded NH and OH stretches, respectively.
See text for further discussion. cNot found in the experiment.
Conformer A is assigned to a T-H structure (Figure 3.5A) labelled as T-H
(g+,g+,a,g+,g+). This structure is also the calculated global minimum at the B3LYPD3BJ/6-31+G(d) level of theory, consistent with the large intensity of its transitions in the
UV spectrum (Figure 3.3). The free head NH stretch of A appears at 3457 cm-1, similar to
the NH stretch fundamentals of trans-formanilide (3463 cm-1) and trans-acetanilide (3472
cm-1).32 The slight shift to lower wavenumber is potentially due to anti-cooperativity8 in
which the head NH bond is weakened by formation of an intramolecular H-bond to its
amide C=O group, which acts as an acceptor in the T-H intramolecular H-bond, forming
an 11-membered H-bonded ring (labeled ‘C11’, Figure 3.5).
The tail-to-head
intramolecular H-bond is short (1.90 Å, Figure 3.5A), producing a tail NH stretch
52
fundamental at 3256 cm-1. The tail OH group of conformer A is cis to the C=O group,
forming a C5 intramolecular H-bond with the C=O oxygen, which places the OH stretch
fundamental at 3306 cm-1.
In the amide I region, the head C=O is acceptor for the C11 T-H hydrogen bond. As
a result, its C=O stretch fundamental is shifted down in frequency to 1697 cm-1 when
compared to that of trans-fromanilide (1742 cm-1) and trans-acetanilide (1728 cm-1). The
tail C=O of the hydroxamic acid group appears at 1681 cm-1 due to the unique chemical
environment of this group and the C5 ring it forms with the tail OH. The amide II band
for the head NH group is at 1540 cm-1, slightly higher in frequency than in trans-formanilde
(1529 cm-1) and trans-acetanilide (1528 cm-1), reflecting the same indirect effect of the
strong H-bond to its C=O group. For the tail NH, the predictions of the calculations are
that its NH bend fundamental should be shifted to even higher frequency (~1570 cm-1).
However, this transition is predicted to have very weak intensity and wasn’t observed
experimentally.
The structure assigned to conformer B (Figure 3.5B) is calculated to be 5.1 kJ/mol
higher in energy than conformer A at the B3LYP-D3BJ/6-31+G(d) level of theory. Notably,
this structure is determined to be the lowest-energy member of the H-T family. Conformer
C, with its many spectral similarities to B, is assigned to a conformer in the same family,
shown in Figure 3.5C, with an energy only 1.2 kJ/mol higher than B. The difference
between these structures lies largely in the folding of their C6 alkyl chains, as reflected in
their dihedral angles, with B labelled as H-T (g+,g+,e,a,g-) and C labelled as H-T
(a,g-,g-,a,g-). In general, conformer B adopts a tighter loop for the C6 hydrocarbon chain
while a somewhat more extended chain is favored by conformer C. This leads to a slightly
53
larger distance between the head NH and tail C=O groups in C than B, and a different
approach angle for the NH…O=C H-bond.
As anticipated, for both conformer B and conformer C, the free tail NH stretch
fundamentals appear at around 3497 cm-1, about 40 cm-1 up from the position of the free
head NH stretch observed in conformer A. The tail OH retains the same structural
preference in forming a weak C5 H-bond with the tail C=O. However, in this H-T family,
the tail C=O also acts as an acceptor to the donor head NH group in the H-T H- bond,
forming a bifurcated double ring structure with both C5 and C11 H-bonds sharing the same
tail C=O group. As a result, both H-bonded tail OH and head NH stretches are shifted up
in frequency relative to those in A, revealing another anti-cooperative effect, in that one Hbond to the same acceptor group weakens the other. For conformer B, as the calculated
stick diagram suggests, those two transitions are overlapped with each other and thus forms
a broadened peak at 3351 cm-1. According to the calculation, the C5 tail OH is much weaker
than the H-bonded head NH stretch.
In conformer C, calculations predict that the relative wavenumber positions of the
H-bonded NH and OH stretch are reversed, appearing at 3390 cm-1 and 3373 cm-1
respectively. In the amide I region of both B and C, the tail C=O stretch is shifted to lower
wavenumber relative to A, reflecting its character as a double H-bond acceptor. The head
C=O stretch, which is now free, shifts up to around 1724 cm-1, similar to the value in transacetanilide (1728 cm-1). As expected, the head NH bending fundamentals are also shifted
slightly up in frequency to 1560 cm-1 for both conformers in the amide II region as a result
of engaging in a H-T intramolecular H-bond.
54
In arriving at the assignments, a large number of alternative low energy structures
and structures in other conformational families were compared with experiment, including
those engaged in NH-π, H-TOH and H-T/OH-π H-bonded architectures (Figure 3.6 and
Table 3.2). However, they were either completely inconsistent with experimental patterns,
or were significantly poorer matches. In addition, most of the structures belonging to
different families suffer from significantly higher energies than the assigned structures,
consistent with their absence in the expansion.
Figure 3.6. (a) Lowest energy structures in NH-π, H-TOH and H-T/OH-π H-bonded
architectures. (b) Stick diagrams of the calculated frequencies and IR intensities in the
hydride stretch region. These frequencies were scaled by 0.96 for free NH stretch, 0.948
for hydrogen bonded NH and OH stretches. The experimental RIDIR spectra for SAHA
conformer A,B and C in the hydride stretch region are plotted above.
55
Table 3.2. Summary of calculated vibrational frequencies in the hydride stretch region for
3 lowest energy structures in NH-π, H-TOH and H-T/OH-π H-bonded architectures,
calculated at the DFT B3LYP-D3BJ/6-31+G(d) level of theory.
a
ν tail NH
ν OH
ν head NH
NH-π
3429a
3302
3456
H-TOH
3446
3424
3286
H-T/OH-π
3471
3431
3333
Hydride stretch fundamentals are scaled by 0.96 for free NH stretches, 0.948 for hydrogen
bonded NH and OH stretches, respectively.
56
3.3.3
RIDIR Spectra of the SAHA-H2O Complex (Structure D)
Figure 3.7. RIDIR spectra in the (a) hydride and (b) mid-IR regions of SAHA conformer
D, the SAHA-H2O complex. Stick diagrams underneath were simulated based on
calculations using DFT B3LYP/6-31+G(d) (black) and B3LYP D3BJ/6-31+G(d) (red)
levels of theory. The asterisk in the hydride stretch region indicates transition used to
record the IR-UV HB spectrum. These frequencies were scaled by 0.96 for free NH stretch
and water OH stretch, 0.948 for hydrogen bonded NH and OH stretches and 0.985 for
amide I/II frequencies.
The RIDIR spectrum of structure D in the hydride stretch and mid-IR regions are
shown in Figures 3.7(a) and (b), respectively. The hydride stretch spectrum contains two
transitions above 3500 cm-1 that could not be ascribed to SAHA monomer. Instead, bands
at 3515 and 3661 cm-1 are due to a single H2O molecule in a SAHA-H2O complex. The
57
R2PI spectrum in the SAHA monomer mass channel (m/z=264) contains transitions due to
SAHA-H2O as a result of efficient fragmentation of the complex following
photoionization, by loss of H2O.33 Such efficient fragmentation has been observed on many
previous occasions,34 signaling a large geometry change accompanying photoionization as
the H2O molecule responds to the positive charge on SAHA+. Searches for these R2PI
transitions in the [SAHA-H2O]+ mass channel were unsuccessful, indicating that
fragmentation is near-complete. Interestingly, these transitions due to SAHA-H2O are
present despite the fact that we did not add any H2O to the expansion, and using laser
desorption to bring the SAHA sample into the gas phase. The likely source of the H2O is
either residual H2O in the gas lines or included H2O in the solid SAHA sample.
Based on the close proximity of the S1←S0 origin of the SAHA-H2O complex to the
origins of conformers B and C of SAHA monomer, we anticipate a similar H-T H-bonding
arrangement, either mediated by a H2O molecule as bridge, or supporting the
intramolecular H-T structures through stabilization elsewhere. The RIDIR spectra of
Figure 3.7 enable the deduction of several important structural elements of SAHA-H2O
even in the absence of calculations on the complex. For instance, the band at 3497 cm-1 is
at the same frequency as the free tail NH stretch fundamentals in SAHA(B and C).
Furthermore, the highest-frequency amide I band occurs at 1724 cm-1, within a few
wavenumbers of the free C=O stretch of the head amide group, and almost 50 cm-1 above
the C5 tail C=O stretch. Thus, the tail hydroxamic acid NH group of the observed SAHAH2O complex is free, as is the head amide C=O group.
58
There is no free H2O OH stretch (which would occur at ~3710 cm-1), indicating that
the H2O molecule acts as a double donor. The higher frequency transition at 3660 cm-1 is
at a characteristic frequency associated with formation of an OH…π H-bond with the
aromatic ring,33,
35
although alternative weak OH H-bonds must also be considered.
Finally, the water OH stretch transition at 3515 cm-1 is close to where an OH…O=C Hbond appears in other amide-H2O complexes.34, 36-37 This points to a SAHA-H2O H-T
structure in which H2O bridges as double-donor between the phenyl ring and the tail C=O
group.
Based on a thorough conformational search of SAHA-H2O, a structure for the
complex meeting the structural constraints outlined above was identified, whose predicted
vibrational frequencies and IR intensities match experiment reasonably well. The assigned
structure is shown in Figure 3.5D. Stick diagrams below the experimental spectra in Figure
3.7 provide the results calculated at the DFT B3LYP/6-31+G(d) (black) and B3LYP
D3BJ/6-31+G(d) (red) levels of theory. The match with experiment is somewhat better for
the B3LYP calculations without dispersion correction; however, a significant structural
rearrangement accompanies the inclusion of dispersion. While the energetics are better
accounted for with the dispersion correction, the head NH stretch gains significant intensity
and shifts to lower frequency, and could shift still further with larger basis sets. This
suggests that, in the experimental spectrum, the H-T head NH stretch may be overlapped
with the C5 tail OH stretch.
The assigned structure is shown in Figure 3.5D. As anticipated, the structure is a HT structure related to SAHA conformer B, but with the H-T NH…O=C H-bond largely
broken (RH…O= 3.60 Å) through insertion of the water as bridge between the hydroxamic
59
acid C=O and the phenyl  cloud. The energy of this structure is about 23 kJ/mol above
the global minimum structure identified by this search for SAHA-H2O, keeping the
assignment somewhat tentative. Many minima with different H-bonding patterns have
energies lower than the one we have tentatively assigned; yet, they are not observed, a point
to which we will return in the discussion section. Most of these structures are built off other
SAHA monomer minima, and the fits to the experiment are significantly poorer than the
assigned structure both in mid-IR and hydride stretch regions.
Finally, two additional pieces of evidence confirm and strengthen the assignment of
the observed SAHA-H2O complex to the structure shown in Figure 3.5D. First, the lowfrequency torsional modes (Table 3.1) predicted by theory for the assigned structure match
well with the Franck-Condon active progressions in modes of frequency 16 and 72 cm-1 in
the experimental spectrum. The 72 cm-1 mode is the intermolecular stretch of the water
molecule against the π cloud, which would be expected to have significant activity induced
by the ππ* electronic transition. Second, the relative wavenumber positions of the S1←S0
origins of SAHA (A-C) and SAHA-H2O(D) have the same relative ordering and, after
scaling, surprisingly close correspondence with experiment (Figure 3.8).
60
Figure 3.8. R2PI (top trace) spectrum for SAHA. Calculated TDDFT spectra at the DFT
B3LYP D3BJ/6-31+G(d) level of theory are shown below as stick diagrams in black.
The absolute frequencies of the S0-S1 origins are lined up with experiment for conformer
A and scaled by a factor of 0.46.
3.4 Discussion
3.4.1 Inherent Conformational Preferences of SAHA Monomer
A primary motivation for the present study was to understand the inherent
conformational preferences of SAHA monomer in the gas phase, where environmental
effects are removed. Whether in crystalline form or in its binding to the HDAC enzyme
pocket, the alkyl chain is extended in order to facilitate interactions with other SAHA
molecules or the binding pocket. Yet, in the absence of these environmental effects, the
C6 alkyl chain, which prefers an extended conformation, is countered by the stabilizing
effect of intramolecular H-bond(s) between the molecule’s “head” amide group and “tail”
hydroxamic acid.
Our study seeks to determine how many and which conformers are present in the
gas phase. Using laser desorption to bring SAHA into the gas phase, and cooling the
61
molecules in a supersonic expansion, we have recorded single-conformation UV and IR
spectra that led to the identification and assignment of three conformers of SAHA
monomer, shown in Figure 3.5. All three conformers are tightly-folded structures that
contain intramolecular H-bonds between “head” amide and “tail” hydroxamic acid groups.
The unique wavenumber positions of the free NH groups of head (3457 cm-1) and tail (3497
cm-1) provide characteristic spectroscopic signatures of H-T and T-H structures. In all
three conformers, the hydroxamic acid OH group is cis to the C=O group, forming a 5membered H-bonded ring (C5) that is opposite to its orientation when chelating the Zn+2
cation in the HDAC binding pocket, which uses the oxygen lone pairs of the C=O and OH
groups.7
In conformer A, the global minimum, the hydroxamic acid tail engages as H-bond
donor via its NH group to the C=O group of the amide head, forming an eleven-membered
H-bonded ring (C11) in a T-H structure. This H-bond is strong, with a H-bond length of
1.90 Å, leading to a broad and intense NH stretch fundamental at 3256 cm-1. In conformers
B and C, the direction of the H-bond is reversed, with the molecule’s head NH acting as
H-bond donor to the tail C=O group. These T-H hydrogen bonds also form C11 rings due
to the reversal in order of the NH and C=O groups in the amide and hydroxamic acid
moieties of SAHA. Conformers B and C have R2PI transitions about one-third the size of
those of A, consistent with their calculated relative energies about 5 kJ/mol higher than A.
They also have somewhat weaker head-to-tail H-bonds in the 3350-3400 cm-1 region, with
calculated hydrogen bond lengths of 2.11 and 2.14 Å, respectively.
In SAHA, the 6-carbon alkyl chain would by itself energetically prefer an extended
structure. However, as the labeling scheme in Figure 3.5 indicates, the presence of the
62
amide and hydroxamic acid groups at either end of the alkyl chain cause it to fold into a 4gauche (conformer A), 3-gauche (conformer C), or 3-gauche, 1-ecclipsed (conformer B)
turn that positions the tail and head groups where they can engage in a H-bond that
stabilizes the fold. The turn in A is just as in the pure alkyl chains, in this case (gα, gβ, a,
gδ, gε), and this ideal turn does indeed position the head and tail for a strong, near-linear TH hydrogen bond. In Table 3.3, we have listed not only the dihedral angles along the alkyl
chain (α−ε) but those on either side (NC12 and 56CN), which denote the orientation of the
first and last C-C bond relative to the amide or hydroxamic acid planes. It is noteworthy
that the alkyl chain prefers an out-of-plane orientation for these two ancillary dihedral
angles. The end result is that the amide and hydroxamic acid planes are nominally
perpendicular to one another in all three conformers (Figure 3.5), with different approach
angles for the H-bond so formed.
Each of these structures incorporates a turn in the alkyl chain. In previous studies
from Luttschwager et al.,11 Raman spectra provided spectral evidence that the straightchain n-alkanes prefer an extended structure up to n=17-18, with each gauche defect
providing a destabilization of ~2 kJ/mol. For pure alkyl chains longer than this threshold
length, the alkyl chain folds back on itself using a turn composed of four gauche ‘defects’,
configured as (gm-2, gm-1, a, gm+1, gm+2). By positioning this turn mid-way through the alkyl
chain (m=n/2), the two all-trans segments on either side are anti-parallel to one another,
where dispersive attractions can act in concert along these segments to stabilize the folded
structure.
63
Table 3.3. Dihedral angles (degrees) along the C6 alkyl chain of the three observed
conformers of SAHA monomer from head-to-tail.
Dihedral Angle (degrees)





Conformer
NC12
C123
1234
2345

456C
56CN
A
+86
+65
+75
-164
+72
+61
-118
B
-130
+62
+56
-111
+164
-65
+131
C
-95
+165
-58
-58
+164
-67
+133
cis-1
+130
-79
+62
+64
-179
+59
+63
cis-2
+144
-44
-46
+171
-63
-64
+126
3.4.2
Disconnectivity Graphs, Isomerization Pathways, and Observed Conformers
Single-conformation IR and UV spectroscopy provides a powerful tool for dissecting
a complicated spectrum into its constituent components due to individual conformational
isomers. Assignments are made by comparing the observed single-conformer spectra with
ab initio or DFT calculations. In order to make these assignments, however, classical force
field searches are typically used as a screening tool to locate conformers, and to order them
by force field energy as a means of prioritizing the quantum chemical calculations, which
are much more computationally intensive.
The field of single-conformation spectroscopy is now at a point where it is capable
of serving as the basis for refining these force fields, especially in their applications to
isolated, gas phase molecules, using the assigned structures as benchmarks for doing so.
Accurate force fields would then open up whole new classes of problems for exploration,
both in the size of the molecules which could be explored, and in going beyond
spectroscopy to understand the dynamics of conformational isomerization.38 In this latter
64
context, the disconnectivity graph serves as a powerful tool for summarizing the entire
potential energy landscape for the molecule, with all its conformational minima, transition
states, and linked pathways between individual minima.
In the present work on SAHA, we have taken steps in this direction, first in
modifying the hydroxamic acid CNOH dihedral potential within GAFF based on B3LYPD3BJ/6-31+G(d) calculations, as shown in Figure 3.2. Then, armed with these parameters,
we used the modified version of GAFF to create a disconnectivity graph for gas-phase
SAHA monomer, shown in Figure 3.9. We were motivated to construct the disconnectivity
graph for SAHA because of the prototypical nature of the conformational landscape, with
H-T, T-H, and extended conformers all possible, with their interconversion pathways and
their energetics difficult to intuit.
65
Figure 3.9. (a) Disconnectivity graph for SAHA using the modified general Amber force
field (GAFF). Red asterisks indicate the locations of SAHA A and SAHA B. (b) Close-up
view of the dashed rectangle region of the SAHA disconnectivity graph where the
assigned structures for SAHA A and SAHA B are located. The zero-point energy
corrected relative energies calculated at the DFT B3LYP-D3BJ level of theory are also
indicated (X), taken from Table 3.4. (Prepared by Dr. Xiao Zhu from Rosen center for
advanced computing and Karl Blodgett from Zwier group)
Figure 3.9(a) shows an overview of the GAFF disconnectivity graph for SAHA,
while Figure 3.9(b) focuses in on the section enclosed by a dotted red line, where
conformers A and B reside. The disconnectivity graph places stable conformational
minima at the end of vertical branches that denote their relative energies. These minima
are connected by branches that group together transition states of the same energy, grouped
into energy bins that are user-determined. In the present case, transition states are grouped
66
into bins separated by 1.00 kcal/mol (4.18 kJ/mol). The numbering of the conformational
minima denotes the order in which they were found in the search process.
Note, first, that conformers A, B, and C are all predicted by the modified GAFF
force field to be among the low-energy conformers. This gives confidence that the overall structure of the graph is meaningful, and that it can aid a deeper understanding of the
conformational landscape of SAHA. At the same time, the relative energies calculated by
the force field are not in perfect agreement with DFT B3LYP-D3BJ calculations, as can be
seen from the positions of the symbols (X) and Table 3.4. Therefore, in the discussion that
follows, we will use the disconnectivity graph to provide an overview of the potential
energy surface and isomerization pathways, but use the DFT B3LYP-D3BJ/6-31+G(d)
calculations to refine our arguments. Table 3.4 contains relative potential energies, both
with and without zero-point correction, and relative free energies calculated at 300 K.
In order to focus our analysis on the lowest-energy conformations and pathways,
the disconnectivity graph in Figure 3.9(b) is restricted to minima connected to transition
states within 33 kJ/mol of the global minimum, which captures the pathways connecting
A, B, and C. In total 159 minima were identified for SAHA, connected by 207 transition
states.
One of the insights to be gained from the disconnectivity graph is the way in which
it divides the potential energy surface into basins containing sets of low-lying minima
connected by transition states that are also comparatively low in energy. All the low-lying
minima on the potential energy surface have an intramolecular H-bond, whether NH…O=C
H-T or T-H H-bonds, and/or those involving a π H-bond (e.g., OH…π). The fully-extended
structure in which the alkyl chain is all-trans is calculated to be 44 kJ/mol higher than the
67
global minimum in the GAFF force field, 27 kJ/mol from DFT B3LYP-D3BJ/6-31+G(d).
In this sense the conformational minima that have no interactions between SAHA’s head
and tail are just off the top of the disconnectivity graph in Figure 3.9.
In SAHA, the low-energy isomerization pathways involve breaking/re-forming Hbonds, and reconfiguring the alkyl chain to bring the head and tail groups into contact with
one another. Our initial expectation was that the disconnectivity graph would have two
major basins involving H-T and T-H minima separated by a comparatively large barrier.
However, this is not the case. In actuality, the energetics of breaking H-bonds and
reconfiguring the alkyl chains are similar in size, leading to a disconnectivity graph with
appearance more like a banyan tree,39 with many low-lying minima separated by barriers
of approximately the same height. We use the disconnectivity In SAHA, the low-energy
isomerization pathways involve breaking/re-forming H-bonds, and reconfiguring the alkyl
chain to bring the head and tail groups into contact with one another. Our initial expectation
was that the disconnectivity graph would have two major basins involving H-T and T-H
minima separated by a comparatively large barrier. However, this is not the case. In
actuality, the energetics of breaking H-bonds and reconfiguring the alkyl chains are similar
in size, leading to a disconnectivity graph with appearance more like a banyan tree,39 with
many low-lying minima separated by barriers of approximately the same height. We use
the disconnectivity graph to understand why conformers A, B, and C appear in the
expansion-cooled gas phase sample, and why other low-lying minima are absent.
According to the GAFF force field, the SAHA minima are grouped more by having
common or similar dihedral angle patterns along the C6 hydrocarbon chain rather than via
their intramolecular H-bonding patterns. This is immediately evident from Figure 3.9(b),
68
where conformers A and B, which are T-H and H-T structures, respectively, are in the same
branch inside the red-dashed rectangle. This branch also contains the GAFF global
minimum (Min52), while SAHA C, also a H-T structure like B, is in a separate branch with
a different dihedral angle pattern.
According to GAFF, the global minimum for SAHA monomer is Min52, a
conformation that adopts the same dihedral angle pattern (g+,g+,a,g+,g+) as SAHA A, but
with a H-T rather than T-H H-bond. An obvious question, then, is how the isomerization
between A and Min52 occurs. Notably, after DFT re-optimization at B3LYP-D3BJ/631+G(d) level of theory, SAHA A becomes the global minimum on the disconnectivity
graph, with Min52 calculated to be 2.1 kJ/mol higher in energy.
The lowest-energy pathway between Min52 and SAHA A identified by the GAFF
disconnectivity graph was recomputed by DFT methods, and is shown in Figure 3.10. The
transition states between individual minima were confirmed by intrinsic reaction
coordinate (IRC)40 calculations. Note that the highest energy barrier between Min52 and
A is 13.1 kJ/mol, a barrier small enough that population can be funneled from Min52 to A
during the collisional cooling process in the early portions of the expansion. Note that the
isomerization from Min52 to Min56 involves breaking the H-T hydrogen bond by
reorienting the tail hydroxamic acid group, with the C5-C6-C(=O/Tail)-N(-H/Tail)
dihedral angle changed in so doing from +103° to -122° in the intermediate state (Min56).
In the second step, the entire formanilide ‘head’ rotates about its N(-H/Head)-C(=O/Head)C1-C2 dihedral angle from -141° to 85°, forming a short (1.83 Å) tail-to-head
intramolecular H-bond in SAHA A. The second barrier (TS297) is located only 6 kJ/mol
above the intermediate state. Thus, the pathway from Min52 to A involves breaking the H-
69
T and forming the T-H H-bond without changing the alkyl chain configuration, and does
so over a surprisingly small barrier. In fact, Rice-Ramsperger-Kassel-Marcus (RRKM)
rate constants for both steps are calculated to be around 1010-1011 s-1 at the average internal
energy of SAHA monomer at 300 K (<Evib> = 46 kJ/mol), much faster than the cooling
rate. So it is anticipated that the population of Min52 originally present in the laserdesorbed monomer is largely interconverted to SAHA A in the supersonic jet expansion
and is thus absent in the R2PI spectrum.
Similar arguments can be used to rationalize the absence of conformers Min56/139
and Min42/107, which are in the same sub-branches as SAHA A and SAHA B, respectively
(Figure 3.9b). As Table 3.4 shows, these four minima have very low isomerization barriers
(4-5 kJ/mol) to the lowest energy minima in their sub-basin (A or B). Thus, like Min52,
structural relaxation into A or B through interconversion over low energy barriers is
postulated to account for their not being observed experimentally.
70
Figure 3.10. Stationary points along the lowest-energy isomerization pathway predicted
by GAFF between SAHA A, B, and C, calculated at the DFT B3LYP-D3BJ/6-31+G(d)
level of theory.
For SAHA B, although it resides in the same general branch of the disconnectivity
graph as SAHA A, with the same dihedral angle pattern at the beginning of the C6
hydrocarbon chain, the differences in the last three dihedrals places the two conformers in
different sub-branches with a larger isomerization barrier that likely hinders their
interconversion. The barrier from B to A is predicted to be 26 kJ/mol by DFT (Figure
3.10), slightly higher (~3 kJ/mol) than the barrier between SAHA B and SAHA C
molecules. Both the A→B and B→C isomerization pathways are included in Figure 3.10,
and were verified by intrinsic reaction coordinate (IRC) calculations.
The other way in which the populations of the conformers could be modified is if
their free energy corrections are quite different from one another. Table 3.4 also includes
the relative free energies of the 10 lowest energy conformers of SAHA at the B3LYP-
71
D3BJ/6-31+G(d) level of theory. Free energy corrections were made at 300 K, not
knowing the internal energy of the laser desorbed molecules prior to cooling in the
expansion. Note first that the relative free energies for A, B, and C are three of the five
lowest, consistent with their large populations in the expansion. In free energy, they are
matched only by Min52 and Min139, which have already been argued to lose their
population during the collisional cooling process due to small barriers to A. Indeed, the
negative free energy correction for conformer C is consistent with its presence among those
observed.
Only two of the seven non-observed minima in Table 3.4 lack a low-energy cooling
pathway to A-C. Of these Min373 incorporates a H-T intra-molecular H-bond like that in
B and C; however, the tail OH, instead of engaging in a C5 ring with the tail C=O group,
points to the phenyl ring and forms an additional OH…π intramolecular H-bond. This
conformer has an energy only 4 kJ/mol above the global minimum and a free energy close
to SAHA B. With a high isomerization barrier to assigned conformers (>20 kJ/mol), it is
less likely that population initially in this structure would be lost by collisional energy
transfer. Min42 is somewhat higher in both ∆E (8.6 kJ/mol) and ∆G (5.0 kJ/mol), and thus
less of a concern. The present data cannot determine whether inaccuracies in the calculated
relative energies or barrier heights, or some anomaly of the laser desorption process led to
their not being detected.
72
Table 3.4. Calculated relative energies and free energies of 10 lowest energy conformers
of SAHA in the disconnectivity diagram at B3LYP-D3BJ/6-31+G(d) level of theory.
Assigned structures are marked in bold. The two lowest energy cis-SAHA structures
areadded at the end of the table below the dashed line.
ΔE (kJ/mol)
Free energy
correction
B3LYP-D3BJb
B3LYP-D3BJc
ΔG
(kJ/mol) e
Isomer
GAFF a
SAHA A
7.5
0.0
0.0
0.0
0.0
Min52
0
2.4
2.1
2.9
5.0
Min373
3.9
3.9
3.6
6.1
9.7
Min56
9.5
7.6
6.9
3.8
10.7
SAHA B
7.2
7.9
7.3
1.7
9.0
Min139
5.9
8.7
7.7
0.7
8.4
Min74
6.5
9.0
8.6
4.7
13.3
Min42
3.8
11.1
8.7
1.5
10.2
10.2
9.5
8.8
5.2
14.0
SAHA C
4.1
9.9
8.9
-2.7
6.2
cis-SAHA1
12.5
-8.3
-6.7
12.3
5.6
cis-SAHA2
16.8
-4.6
-2.9
12.7
9.8
(kJ/mol) d
(cool to A)
(cool to A)
(cool to A)
(cool to B)
Min107
(cool to B)
a
General Amber force field bCalculated relative energies (kJ/mol) at the B3LYP-D3BJ/6-31+G(d) level of
theory, without harmonic zero-point energy correction. cCalculated relative energies (kJ/mol) at the B3LYPD3BJ/6-31+G(d) level of theory, including harmonic zero-point energy correction. dGibbs corrections
calculated at pre-expansion sample temperatures (taken as T=300 K) at the B3LYP-D3BJ/6-31+G(d) level
of theory. eΔG=free energy correction+ΔE(B3LYP-D3BJ)
73
The significant free energy corrections of the low-lying conformers listed in Table
3.4 raises the question of whether these free energy corrections are sufficient to bring the
fully-extended conformers into contention with those in Table 3.4. To test this possibility,
free energy corrections were calculated for a set of extended-chain conformers. In every
case, their internal energies are so high (>25 kJ/mol at B3LYP-D3BJ/6-31+G(d) level of
theory) that the final ΔG value (>15 kJ/mol) is still much higher than the assigned
conformers. As a result, these extended chain conformations of SAHA are not observed in
our experiment, as would be obvious from their two free hydride stretch fundamentals.
3.4.3 Cis-amide Structures and Laser Desorption
The discussion in Section 3.4.2 neglected one possibility that must still be
considered. The head group of SAHA is an alkylated formanilide, Ph-NH-CO-R. The
amide group’s conjugation with the phenyl ring produces a smaller than typical energy
difference between trans-amide and cis-amide isomers, and in formanilide itself, both cis
and trans isomers were observed in the expansion.26 As a result, it was important also to
explore this possibility in SAHA, where the cis-amide structure could potentially engage
in hydrogen-bonding arrangements with the hydroxamic acid tail that are not available to
trans-SAHA. However, the standard force field parameters in GAFF places an artificial
barrier at the cis-amide configuration, since in peptides, trans-amide structures are
preferred over cis nearly exclusively.
As a result, we explored the cis-amide
conformational space by fixing the amide group dihedral angle in the cis configuration and
carrying out a force field search that identified over 400 cis-amide minima for SAHA.
Subsequent optimization of the low-energy structures at the DFT B3LYP-D3BJ/6-31+G(d)
74
level led to the identification of two conformers shown in Figure 3.11(a), which are
calculated to be lower in potential energy and free energy than any trans-amide SAHA
structure so far considered. All other cis-amide minima are significantly higher in energy
than these two structures.
Figure 3.11. (a) Structures, labels, and zero-point energy corrected energies (relative to
SAHA A at the DFT B3LYP-D3BJ/6-31+G(d) level of theory) of the two low-energy
conformers of cis-amide SAHA. (b) Stick diagrams of the calculated frequencies and IR
intensities in the hydride stretch region, using the same scale factors (0.948) for hydrogen
bonded NH and OH stretches. The experimental RIDIR spectra in the hydride stretch
region are presented in Figure 3.4(a).
Both these conformers engage in a pair of strong, bridging H-bonds, T-H NH…O=C
and H-T NH…OH, with different alkyl chain configurations. The cis-1 conformer has
75
shorter H-bonds (1.99 Å H-T, 1.89 T-H) than cis-2 (2.02 Å H-T, 1.95 Å T-H), consistent
with its lower energy. Both these conformers have calculated IR spectra (Figure 3.11b)
that are completely inconsistent with any of the observed conformers A-C, since no free
NH group is present in them, and both NH stretch fundamentals are shifted below 3350
cm-1.
We have searched over the 35590 to 36700 cm-1 region in R2PI for additional
transitions not yet accounted for. The TDDFT calculations at B3LYP D3BJ/6-31+G(d)
level of theory predict that cis-1 will have its S1←S0 origin slightly blue (+108 cm-1) of
SAHA A, but IR-UV hole-burning proved that all transitions blue of the SAHA A origin
are due to A. Thus, there is no experimental evidence for cis-SAHA structures in the
expansion. Furthermore, the barrier to interconversion from cis to trans is characteristically
high (>94 kJ/mol), so that interconversion during collisional cooling is out of the question.
With no experimental evidence for cis-SAHA in the expansion, we seek an
explanation for their absence. It seems most likely to us that the laser desorption process
produces only trans-amide structures. As a crystalline solid, SAHA exists exclusively in
the trans-amide form. Since laser desorption occurs out of a thin film of the solid directly
into the gas phase, and is done under gentle conditions designed not to decompose the
sample, we postulate that the initial internal energy of the desorbed SAHA molecules is
well below the barrier to trans→cis isomerization. As a result, no cis-amide conformers
are formed, and full thermal equilibrium of the gas phase molecules does not occur. As a
result, the SAHA monomer is a clear example in which the observed conformers are
affected by the means used to bring them into the gas phase.
76
3.4.4 Effect of Water on the Conformational Preferences of SAHA
The structural preference of SAHA molecule in the gas phase is widely discussed
in the main text. However, under physiological conditions, SAHA molecule may have
interaction with water molecule and unravel the hydrogen bonded cycles. As SAHA
molecule enters HDLP tube-like pocket, further stabilization effect will be received from
the interaction with the walls of the channel and finally leads to a fully extended
structure. In the present study, we have incorporated a single water molecule to SAHA as
a first step to shed some light on the water complexation effect to the monomer.
Water molecule could both donate and accept hydrogen atoms through
intermolecular H-bonds with the monomer. Thus, the conformational preferences
stabilized by intramolecular H-bonds could be largely affected with the addition of water.
If the water molecule could bridge to the monomer with minimum perturbation to the
existing intramolecular H-bonds, a similar conformational preference will be retained in
the water complex.34 However, in some cases,41 the water molecule inserts itself between
the previous intramolecular H-bonds donor and acceptor sites. Thus largely changes the
original conformational preference for the monomer molecule to which it is bound to and
may reshape the monomer molecule into a structure never found in its isolated form.
77
Figure 3.12. Calculated global minimum for SAHA water complex through B3LYPD3BJ/6-31+G(d) level of theory.
For SAHA water complex, as stated above, many low energy conformations with
SAHA molecule significantly reshaped due to the insertion of water are predicted in the
conformational search. For example, in the calculated global minimum (Figure 3.12), the
water molecule acts both as an acceptor to form an intermolecular H-bond with the tail NH
group and as a donor to form two intermolecular H-bonds with the head C=O group and
phenyl ring respectively. As a consequence, the H-T or T-H H-bonding architectures
assigned in the isolated forms are completely broken to form three new intermolecular Hbonds. At the same time, a head NH to tail OH intramolecular H-bond is formed while the
tail OH to tail C=O C5 member ring is still retained in the structure. However, the predicted
pattern for this structure does not match with the experimental result at all and has to be
ruled out as a possible candidate. In the assigned SAHA water complex, the SAHA
monomer retains its preference to form a H-T intramolecular H-bond pattern based on
conformer B, indicating strong stability of intramolecular H-bonding architecture with the
disturbance of water. The H2O molecule, at the same time, forms a double donor bridge
with the tail C=O group and the phenyl ring. As mentioned above, the C6 hydrocarbon
chain backbone dihedral angles stays similar with conformer B while γ dihedral angle
78
changes from eclipsed (e) to anti (a) configuration to accommodate the insertion of water
molecule and lengthens the head NH to tail C=O intramolecular H-bond length by about
1.5Å. The binding energy is determined to be 37.8 kJ/mol at the density function theory
(DFT) DFT/6-31+G(d) level of theory after a correction for basis set superposition error
(BSSE), similar to the binding energy of DPOE-(H2O)1 complex42 as we studied before.
Since SAHA conformer B is not the lowest energy structure in the assigned SAHA
monomers in the expansion, it is still an open question why water clusters based on SAHA
conformer A have not been observed in R2PI scans up to 36350 cm-1. Water complex
structures based on SAHA conformer A calculated at the B3LYP-D3BJ /6-31+G(d)//M052X/6-31+G(d) level of theory suggested the water molecule forms a double donor bridge
between the head C=O group and aromaticπcloud. However, the basis set superposition
error (BSSE) corrected binding energy for SAHA A calculated at B3LYP-D3BJ/6-31+G(d)
level of theory is only 31.4 kJ/mol, much smaller than the binding energy based on
conformer B (37.8 kJ/mol). Besides, in the calculated structure for SAHA A-H2O complex,
the C(=O/head)-N(head)-C1(phenyl)-C2(phenyl) dihedral angle is about 29°, and greatly
distorts theπmolecular orbitals in the phenyl amide group plane and raises the energy. Thus
the water molecule is preferable to bind with conformer B than conformer A to form a
water complex although conformer A is the global minimum in the expansion.
3.5 Conclusions
The present study of the conformational preferences of the gas-phase SAHA
molecule has revealed the presence of three conformers under jet-cooled conditions, all of
which are tightly-folded conformers involving tail-to-head or head-to-tail H-bonds. This
79
is in striking contrast to the extended structure SAHA takes up in crystalline form, or when
binding to histone deacetylase (HDAC) in its function as anti-cancer drug.
The
characteristic frequencies of the free NH groups of the amide head and hydroxamic acid
tail provide a clear diagnostic of structural type. The 6-carbon alkyl chain that serves as
chemical linkage is sufficiently long that several structural turns can be formed by it that
bring the hydroxamic acid functional group(s) into close spatial proximity with the head
amide group.
A disconnectivity graph created using GAFF provides insight to the over-all shape of
the potential energy landscape for SAHA. The basins on the potential energy surface have
similar alkyl chain conformations, and interconversion pathways from H-T to T-H
structures can occur by twisting the head amide and tail hydroxamic moieties successively
through hindered rotations that break and then reform new H-bonds between them without
reconfiguring the alkyl chain. The calculated barriers for doing so are somewhat smaller
than the energetic cost for breaking a H-bond in the absence of other compensating
attractions.
Alkyl chain reconfiguration is more energetically costly, preventing
isomerization between A, B, and C during the cooling in the expansion. Calculations
predict that two cis-amide structures have potential and free energies below any of the
trans-amide structures, but no experimental evidence exists for their presence in the
expansion. We postulate that the laser desorption of the solid, in which SAHA exists
exclusively in a trans-amide configuration, leads only to trans-amide structures in the gas
phase, which cannot interconvert to cis-amide on the timescale of the experiment.
80
3.6 References
1.
Johnstone, R. W.; Licht, J. D. Histone Deacetylase Inhibitors in Cancer Therapy: Is
Transcription the Primary Target? Cancer Cell 2003, 4, 13-18.
2.
Kelly, W. K.; Marks, P. A. Drug Insight: Histone Deacetylase Inhibitors Development of the New Targeted Anticancer Agent Suberoylanilide Hydroxamic
Acid. Nat. Clin. Pract. Oncol. 2005, 2, 150-157.
3.
Kelly, W. K.; Richon, V. M.; O'Connor, O.; Curley, T.; MacGregor-Curtelli, B.;
Tong, W.; Klang, M.; Schwartz, L.; Richardson, S.; Rosa, E.; et al. Phase I Clinical
Trial of Histone Deacetylase Inhibitor: Suberoylanilide Hydroxamic Acid
Administered Intravenously. Clin. Cancer Res. 2003, 9, 3578-3588.
4.
Salmi-Smail, C.; Fabre, A.; Dequiedt, F.; Restouin, A.; Castellano, R.; Garbit, S.;
Roche, P.; Morelli, X.; Brunel, J. M.; Collette, Y. Modified Cap Group
Suberoylanilide Hydroxamic Acid Histone Deacetylase Inhibitor Derivatives
Reveal Improved Selective Antileukemic Activity. J. Med. Chem. 2010, 53, 30383047.
5.
Grant, S.; Easley, C.; Kirkpatrick, P. Vorinostat. Nat. Rev. Drug Discov. 2007, 6, 2122.
6.
Griffith, D. M.; Szocs, B.; Keogh, T.; Suponitsky, K. Y.; Farkas, E.; Buglyo, P.;
Marmion, C. J. Suberoylanilide Hydroxamic Acid, a Potent Histone Deacetylase
Inhibitor; its X-ray Crystal Structure and Solid State and Solution Studies of its
Zn(II), Ni(II), Cu(II) and Fe(III) Complexes. J. Inorg. Biochem. 2011, 105, 763769.
7.
Finnin, M. S.; Donigian, J. R.; Cohen, A.; Richon, V. M.; Rifkind, R. A.; Marks, P.
A.; Breslow, R.; Pavletich, N. P. Structures of a Histone Deacetylase Homologue
Bound to the TSA and SAHA Inhibitors. Nature 1999, 401, 188-193.
8.
Kusaka, R.; Zhang, D.; Walsh, P. S.; Gord, J. R.; Fisher, B. F.; Gellman, S. H.; Zwier,
T. S. Role of Ring-Constrained gamma-Amino Acid Residues in alpha/gammaPeptide Folding: Single-Conformation UV and IR Spectroscopy. J. Phys. Chem. A
2013, 117, 10847-10862.
9.
Vaquero-Vara, V.; Zhang, D.; Dian, B. C.; Pratt, D. W.; Zwier, T. S. Delicate
Balance of Hydrogen Bonding Forces in D-Threoninol. J. Phys. Chem. A 2014, 118,
7267-7273.
10.
Zhang, D.; Bocklitz, S.; Zwier, T. S. Broadband Microwave Spectroscopy of
Prototypical Amino Alcohols and Polyamines: Competition between H-Bonded
Cycles and Chains. J. Phys. Chem. A 2016, 120, 55-67.
81
11.
Luttschwager, N. O. B.; Wassermann, T. N.; Mata, R. A.; Suhm, M. A. The Last
Globally Stable Extended Alkane. Angew. Chem. Int. Ed. 2013, 52, 463-466.
12.
Davis, Z.S. Exploring Conformational Preferences of Flexible Biomolecules
Utilizing the Instrument for Cold Ion Spectroscopy and Force Field Methods. Ph.D.
Dissertation, Purdue University, West Lafayette, IN, 2015.
Shubert, V. A.; Baquero, E. E.; Clarkson, J. R.; James, W. H., III; Turk, J. A.; Hare,
A. A.; Worrel, K.; Lipton, M. A.; Schofield, D. P.; Jordan, K. D.; et al. EntropyDriven Population Distributions in a Prototypical Molecule with Two Flexible Side
Chains: O-(2-acetamidoethyl)-N-acetyltyramine. J. Chem. Phys. 2007, 127,
234315.
13.
14.
Wales, D. J.; Doye, J. P. K. Global Optimization by Basin-Hopping and the Lowest
Energy Structures of Lennard-Jones Clusters Containing up to 110 Atoms. J. Phys.
Chem. A 1997, 101, 5111-5116.
15.
Wales, D. J. GMIN2.0; University of Cambridge: Cambridge CB2 1EW, UK.
http://www-wales.ch.cam.ac.uk/software.html.
16.
Wales, D. J.; Walsh, T. R. Theoretical Study of the Water Pentamer. J. Chem. Phys.
1996, 105, 6957-6971.
17.
Trygubenko, S. A.; Wales, D. J. A Doubly Nudged Elastic Band Method for Finding
Transition States. J. Chem. Phys. 2004, 120, 2082-2094.
18.
Liu, D. C.; Nocedal, J. On the Limited Memory BFGS Method for Large-Scale
Optimization. Math. Prog. 1989, 45, 503-528.
19.
Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.;
Cheeseman, J. R.; Scalmani, G.; Barone, V.; Mennucci, B.; Petersson, G. A., et
al. Gaussian 09, revision C.01; Gaussian, Inc.: Wallingford, CT, 2010.
20.
Becke, A. D. Density�Functional Thermochemistry. III. The Role of Exact
Exchange. J. Chem. Phys. 1993, 98, 5648–5652.
21.
Zhao, Y.; Truhlar, D. G. Density Functionals for Noncovalent Interaction Energies
of Biological Importance. J. Chem. Theory Comput. 2007, 3, 289-300.
22.
Grimme, S. Accurate Description of van der Waals Complexes by Density
Functional Theory Including Empirical Corrections. J. Comput. Chem. 2004, 25,
1463-1473.
23.
Grimme, S. Semiempirical Hybrid Density Functional with Perturbative Secondorder Correlation. J. Chem. Phys. 2006, 124, 034108.
82
24.
Gord, J. R.; Walsh, P. S.; Fisher, B. F.; Gellman, S. H.; Zwier, T. S. Mimicking the
First Turn of an alpha-Helix with an Unnatural Backbone: Conformation-Specific
IR and UV Spectroscopy of Cyclically Constrained beta/gamma-Peptides. J. Phys.
Chem. B 2014, 118, 8246-8256.
25.
Buchanan, E. G.; James, W. H., III; Choi, S. H.; Guo, L.; Gellman, S. H.; Mueller,
C. W.; Zwier, T. S. Single-Conformation Infrared Spectra of Model Peptides in the
Amide I and Amide II Regions: Experiment-Based Determination of Local Mode
Frequencies and Inter-Mode Coupling. J. Chem. Phys. 2012, 137.
26.
Manea, V. P.; Wilson, K. J.; Cable, J. R. Conformations and Relative Stabilities of
the cis and trans Isomers in a Series of Isolated N-phenylamides. J. Am. Chem. Soc.
1997, 119, 2033-2039.
27.
Dickinson, J. A.; Hockridge, M. R.; Robertson, E. G.; Simons, J. P. Molecular and
Supramolecular Structures of N-phenyl Formamide and its Hydrated Clusters. J.
Phys. Chem. A 1999, 103, 6938-6949.
28.
Cabezas, C.; Varela, M.; Caminati, W.; Mata, S.; Lopez, J. C.; Alonso, J. L. The
Two Conformers of Acetanilide Unraveled Using LA-MB-FTMW Spectroscopy. J.
Mol. Spectrosc. 2011, 268, 42-46.
29.
James, W. H., III; Buchanan, E. G.; Guo, L.; Geman, S. H.; Zwier, T. S. Competition
between Amide Stacking and Intramolecular H Bonds in gamma-Peptide
Derivatives: Controlling Nearest-Neighbor Preferences. J. Phys. Chem. A 2011, 115,
11960-11970.
30.
Walsh, P. S.; Buchanan, E. G.; Gord, J. R.; Zwier, T. S. Binding Water to a PEGLinked Flexible Bichromophore: IR Spectra of Diphenoxyethane-(H2O)(n)
Clusters, n=2-4. J. Chem. Phys. 2015, 142.
31.
Dean, J. C.; Buchanan, E. G.; James, W. H., III; Gutberlet, A.; Biswas, B.;
Ramachandran, P. V.; Zwier, T. S. Conformation-Specific Spectroscopy and
Populations of Diastereomers of a Model Monolignol Derivative: Chiral Effects in
a Triol Chain. J. Phys. Chem. A 2011, 115, 8464-8478.
32.
Miyazaki, M.; Saikawa, J.; Ishizuki, H.; Taira, T.; Fujii, M. Isomer Selective
Infrared Spectroscopy of Supersonically Cooled cis- and trans-N-phenylamides in
the Region from the Amide Band to NH Stretching Vibration. Phys. Chem. Chem.
Phys. 2009, 11, 6098-6106.
33.
Gotch, A. J.; Zwier, T. S. Multiphoton Ionization Studies of Clusters of Immiscible
Liquids .1. C6H6-(H2O)N, N=1,2. J. Chem. Phys. 1992, 96, 3388-3401.
83
34.
Walsh, P. S.; Kusaka, R.; Buchanan, E. G.; James, W. H., III; Fisher, B. F.; Gellman,
S. H.; Zwier, T. S. Cyclic Constraints on Conformational Flexibility in gammaPeptides: Conformation Specific IR and UV Spectroscopy. J. Phys. Chem. A 2013,
117, 12350-12362.
35.
Pribble, R. N.; Zwier, T. S. Size-Specific Infrared-Spectra of Benzenne-(H2O)(N)
Clusters (N=1 Through 7) - Evidence for Noncyclic (H2O)(N) Structures. Science
1994, 265, 75-79.
36.
Mitsuzuka, A.; Fujii, A.; Ebata, T.; Mikami, N. Infrared Spectroscopy of OH
Stretching Vibrations of Hydrogen-Bonded Tropolone-(H2O)(n) (n=1-3) and
Tropolone-(CH3OH)(n) (n=1 and 2) Clusters. J. Chem. Phys. 1996, 105, 26182627.
37.
Klyne, J.; Schmies, M.; Fujii, M.; Dopfer, O. Stepwise Microhydration of Aromatic
Amide Cations: Formation of Water Solvation Network Revealed by Infrared
Spectra of Formanilide(+)-(H2O)(n) Clusters (n <= 5). J. Phys. Chem. B 2015, 119,
1388-1406.
38.
Dian, B. C.; Longarte, A.; Winter, P. R.; Zwier, T. S. The Dynamics of
Conformational Isomerization in Flexible Biomolecules. I. Hole-Filling
Spectroscopy of N-acetyl tryptophan Methyl Amide and N-acetyl tryptophan
Amide. J. Chem. Phys. 2004, 120, 133-147.
39.
Despa, F.; Wales, D. J.; Berry, R. S. Archetypal Energy Landscapes: Dynamical
Diagnosis. J. Chem. Phys. 2005, 122.
40.
Gonzalez, C.; Schlegel, H. B. An Improved Algorithm for Reaction-Path Following.
J. Chem. Phys. 1989, 90, 2154-2161.
41.
Shubert, V. A.; Mueller, C. W.; Zwier, T. S. Water's Role in Reshaping a
Macrocycle's Binding Pocket: Infrared and Ultraviolet Spectroscopy of Benzo-15crown-5-(H2O)(n) and 4 '-aminobenzo-15-crown-5-(H2O)(n), n=1, 2. J. Phys.
Chem. A 2009, 113, 8067-8079.
42.
Buchanan, E. G.; Gord, J. R.; Zwier, T. S. Solvent Effects on Vibronic Coupling in
a Flexible Bichromophore: Electronic Localization and Energy Transfer induced by
a Single Water Molecule. J. Phys. Chem. Lett. 2013, 4, 1644-1648.
84
CHAPTER 4.
THE DELICATE BALANCE OF HYDROGEN
BONDING FORCES IN D-THREONINOL
4.1 Introduction
Nature uses ribose and deoxyribose as scaffolds for building nucleic acids to carry
genetic codes.1 But recently it has been discovered that simple acyclic diols like propylene
glycol2 and 2(S)-amino-1,3(S)-butanediol (D-threoninol)3 may be used to synthesize
artificial oligonucleotides that spontaneously fold with complementary strands into doublehelical structures, some more stable than natural DNA. This is an exciting result, since such
findings raise the prospect of designing new artificial duplexes (synthetic “foldamers”) that
do not rely on rigid preorganization of the single strand backbone, thereby creating a range
of artificial duplexes with binding properties that can be fine-tuned.
Intrigued by these results, we were curious to learn more about the common
structural feature(s) exhibited by these apparently dissimilar systems. Several important
chemical and biochemical scaffolds have hydrogen bonding groups spread across adjacent
carbons in an alkyl chain. Many of these incorporate hydroxyl (–OH) groups that can form
networks of intramolecular OH···OH···OH hydrogen bonds (HB’s). Indeed, earlier
spectroscopic studies of gas phase molecules have shown that glycerol, a prototypical
molecule with –OH groups attached to the three adjacent carbons, and ribose, with three
vicinal –OH groups and an additional “across-the-ring” –OH group, each exhibit
cooperative HB networks, resulting in either closed or open chain structures.4-7 These
structures, with exposed lone pairs of electrons on one or more faces, could play a key role
in catalyzing the formation of intermolecular assemblies.
85
In what follows, we use the technique of chirped-pulse Fourier transform microwave
(CP-FTMW) spectroscopy to examine the conformational properties of D-threoninol
(DTN) in the gas phase. Like glycerol (see below), DTN has three adjacent functional
groups, but an amino (–NH2) group replaces the central –OH group in glycerol.
Applications of the OH/NH2/OH scaffold in living systems include the molecules
serinol, a precursor in the synthesis of antibiotics,8 and sphingosine, which serves as an
anchor in the phospholipid bilayer.9 The –NH2 group is known principally as a HB acceptor
to its nitrogen lone pair, but –NH2 groups also can engage in HB donation. Indeed, our
results show that the stable DTN conformers also exhibit cooperative HB networks, as in
glycerol, though substitution of –OH by –NH2 reveals a different conformational landscape
arising from the stronger HB accepting character of the amino group.
86
4.2 Theoretical and Experimental Methods.
Force field calculations using Amber were first performed with the MacroModel10
commercial program suite (10,000 iterations, 0.0001 convergence threshold), yielding 85
stable conformations of DTN with energies less than 50 kJ mol-1. These were then
subjected to energy minimization using ab initio calculations at the MP2/6-311++G(d,p)11
level, with vibrational zero-point correction. Of these, 15 conformations with energies
below 500 cm-1 were predicted, the seven lowest in energy of which are shown in Figure
4.1. The next seven are described in Figure 4.2.
87
g+g-
(II) 0
g+g+
(I) 45
(I) 61
g-g-
(II) 134
(II) 70
g-g+
(I) 216
(II) 215
Figure 4.1. The seven most stable conformers of D-threoninol paired according to the
dihedral angles of their functional groups. The direction of H-bonds of the functional
groups (-OH and -NH2), α···β···γ (I) or γ···β···α (II), are given, along with their zero point
corrected relative energies in cm-1.
88
g+g-
402
406
g+g+
(I) 295
(II) 331
g-g-
(II) 285
(I) 345
g-g+
(I) 352
Figure 4.2. The next seven most stable conformers for DTN (numbers 8-14 in energy)
paired according to the dihedral angles of their functional groups. The orientation of the
donor-acceptor interactions, αβγ(I) or γβα(II), are given in parentheses, along with their
zero point corrected relative energies in cm-1.
89
Addition of a CH3 group to the backbone of glycerol breaks the molecular symmetry,
generating two different conformations for each original structure in glycerol. This new
methyl group also creates two chiral centers; thus, D-threoninol (2S, 3S) is one of a family
of four diastereomers with chemical formula C4H11NO2.
Its mirror-image form, L-
threoninol (2R, 3R) would share the same number of conformations with the same relative
stability, while the other two allo forms, (2S, 3R) and (2R, 3S), would exhibit different
conformational landscapes.
In the present work, we focus attention on the conformers of D-threoninol. These
conformers fall into two classes, cyclic and chain structures, and are labeled according to
the orientation of the functional groups, the number of intramolecular hydrogen bonds, and
the relative position of the heteroatoms. Two examples are shown below; I3(g+g-) and
II3(g+g-). Here, the Roman numerals I and II denote the direction of the H-bonds along the
carbon framework, α(OH)···β(NH)···γ(OH) (I) and γ(OH)···β(NH···α(OH) (II); the
subscripts 2 and 3 stand for the number of hydrogen bonding interactions within the
molecule; g, standing for gauche, describes the dihedral angle of N with respect to Oα and
the dihedral angle of Oγ with respect to N, being + or –, depending on the sign of the angle,
+60 or -60. The actual dihedral angles are typically within ±10° of this value.
Oα
Oδ
Nβ
I3(g+g-)
II3(g+g-)
90
In the CP-FTMW12 experiments, D-threoninol, a white solid (m.p. 57-61°), was
acquired from Sigma-Aldrich (97% purity) and used without further purification.
Preliminary studies were performed using the mirror-horn instrument at the University of
Pittsburgh;13 the majority of the experimental work was performed on the two-horn
spectrometer at Purdue University. The experimental methods used in the present study
have been described in detail in Chapter 2 and is only briefly described here. D-threoninol
was vaporized at 120° in a heating reservoir placed in front of a 1.8 mm General Valve
nozzle and seeded in Ne (99% purity) at a backing pressure of 1.8 bar. Pulsed at a rate of
10 Hz, the gas mixture was expanded into a vacuum chamber (10-5 Torr) and exposed to a
perpendicularly propagating microwave pulse that was chirped in 1 µs over the frequency
range 7.5-18.5 GHz. This 11 GHz bandwidth pulse was amplified by a 200 W travelling
wave tube amplifier and broadcast into the chamber through a microwave horn antenna. A
20 µs free induction decay (FID) was collected with a second horn antenna, amplified with
a 45 dB low noise amplifier, and down converted for display and processing on a 12 GHz
digital oscilloscope working at a sampling rate of 40 GS/s. The averaged signal from
10,000 gas pulses was background subtracted to remove electrical noise and any residual
harmonic frequencies.
4.3 Results.
Figure 4.3 shows the experimental microwave spectrum of DTN over the range 7.5
to 18.5 GHz. The spectrum contains hundreds of lines with widely varying intensities, and
likely includes separate contributions from several different conformers. Each conformer
is expected to exhibit a unique pattern of lines, but the different conformers have different
moments of inertia and rotational constants, so the overall spectrum is extremely complex.
91
Figure 4.3. Pure rotational spectrum of D-threoninol from 7.5 GHz to 18.5 GHz.
To deconvolute the experimental spectrum into the separate contributions of the
different conformers, we first used ab initio theory to predict the rotational parameters of
the predicted structures of DTN. Results for the seven lowest energy conformers are listed
in Table 4.1. Then, a semirigid-rotor model Hamiltonian14 was employed to predict the
rotational transitions of the two most stable conformers, and JB9515 was used as a
visualization tool to compare the predicted spectrum with the observed one. This led to
the identification of several strong lines in the spectrum. The experimental frequencies of
these lines were then input into Pickett’s SPFIT/SPCAT16 program, and refined values of
the rotational constants were obtained. These are listed in Table 4.2.
Examination of the data in Tables 4.1 and 4.2 reveals that the theoretical and
experimental rotational constants of a particular conformer are not identical; in most cases,
differences of more than 10 MHz were found between the calculated and observed
92
constants. (Typically, the calculated constants are larger than the observed ones.) At the
same time, the rotational constants of some of the conformers are nearly the same,
especially when they share a similar heavy atom configuration and differ primarily in the
orientation of their–OH and –NH2 groups. For example, the experimental values of A, B,
and C for the two more stable conformers are 3904, 1932, and 1701 MHz, and 3903, 1938,
and 1693 MHz, respectively, making it difficult to distinguish them. Therefore, other
methods must be used to assign the different sets of transitions to specific conformers.
93
Table 4.1. Rotational parameters of the seven most stable conformations of D-threoninol.
MP2/6-311+G(d,p)
II3(g+g-)
I3(g+g-)
I2(g+g+)
II2(g-g-)
II2(g+g+)
II2(g-g+)
I2(g-g-)
Aa (MHz)
3941
3924
3512
4618
3461
4194
4550
B (MHz)
1936
1945
2064
1579
2070
1732
1583
C (MHz)
1709
1704
1528
1480
1523
1308
1463
κb
-0.80
-0.78
-0.46
-0.94
-0.44
-0.71
-0.92
μac (D)
1.3
0.7
1.6
1.4
4.3
3.0
4.7
μb (D)
2.7
2.6
0.4
0.8
0.5
2.0
0.7
μc (D)
1.3
1.3
0.9
1.0
1.6
1.7
0.5
μT (D)
3.3
2.9
1.9
1.9
4.6
4.0
4.8
χaad (MHz)
-1.32
-4.91
-0.06
0.74
-3.60
-2.54
-4.85
χbb (MHz)
0.68
2.36
2.64
0.44
1.68
0.46
2.38
χcc (MHz)
0.64
2.55
-2.59
-1.19
1.92
2.08
2.46
0
45
61
70
134
215
216
ΔEMP2+ZPC (cm-1)e
A, B and C are the rotational constants. bRay’s asymmetry parameter, κ = (2*B-A-C)/(A-C). cµa, µb, µc and µT are the absolute values
for the electric dipole moment components and the total dipole moment. dχaa, χbb and χcc are the diagonal elements of the 14N nuclear
quadrupole coupling tensor. eRelative energies including zero point correction vibrational energies.
a
93
94
Table 4.2. Experimental rotational parameters of the seven conformers identified in the spectrum.
A
a
B
C
D
E
F
G
A (MHz) 3904.077(1)a
3902.915(1)
3513.696(2)
4584.786 (2)
3482.376(5)
4171.110(2)
4525.5(4)
B (MHz)
1931.925(1)
1938.430(1)
2037.084(1) 1568.3971(1) 2026.931(2)
1722.480(1)
1570.279(2)
C (MHz)
1701.108(1)
1693.264(1)
1515.886(1)
1472.179(1)
1507.379(2)
1301.685(1)
1460.121(2)
ΔJ(kHz)
0.16(3)
0.17(3)
0.22(2)
-
0.31(5)
-
0.18(3)
κ
-0.79
-0.78
-0.48
-0.94
-0.47
-0.71
-0.93
χaa (MHz)
-1.19(3)
-3.88(1)
0.08(2)
0.61(7)
-3.1(14)
-2.04(7)
-4.0(1)
χbb (MHz)
0.49(3)
1.65(1)
2.22(2)
0.33(5)
1.7(2)
0.59(7)
3.0(7)
χcc (MHz)
0.70(3)
2.23(1)
-2.30(2)
-0.94(5)
1.4(2)
1.45(7)
1.0(7)
Nb
31
33
14
22
20
23
15
σc / kHz
10.8
8.3
6.5
13.8
18.4
18.6
15.6
Errors in parentheses are expressed in units of the last digit. bNumber of fitted lines. cStandard deviation of the fit.
94
95
One parameter that proved useful in this regard is Ray’s asymmetry parameter, κ =
(2B-A-C)/(A-C).17 This constant gives information about the shape of the molecule; with
the limiting values of κ, -1 and +1, corresponding to prolate and oblate tops, respectively.
As can be seen from Table 4.1, the seven lowest energy conformers of DTN fall closer to
the prolate limit, but can still be divided into four different structural classes, based on the
dihedral angles of the functional groups; κ = -0.45 for g+g+, κ = -0.70 for g-g+, κ = -0.80
for g+g-, and κ = -0.93 for g-g-. Clearly, the two sets of rotational constants A and B,
associated to the most intense lines of the spectrum, belong to the same g+g- class, since
their experimental asymmetry parameters (-0.79 and -0.78) have values near -0.80. But the
remaining set of constants, C through G, must belong to one of the other classes, since their
κ values are very different.
We then proceeded to identify the other observed conformers of DTN. This was done
by removing the more intense assigned rotational transitions of the more stable conformers
from the experimental spectrum, and then comparing the remaining, less intense lines with
the predictions of theory. After many iterations, this process ultimately led to the
identification of five new structures with different sets of rotational constants, all of which
are summarized in Table 4.2. Altogether, we found two g+g+ forms, two g+g- forms, one
g-g+ form, and two g-g- forms. Of these, the g-g+ structure is unique since it is the only
one with a κ value near -0.70. Tentative assignments of the remaining spectra to specific
conformers within each class follow from comparing the observed rotational constants and
the relative intensities of a-, b-, and c-type lines in the spectra, with the predictions of theory.
The projections of the dipole moment along the a-, b-, and c-axes is particularly important
96
here, since reorientations of the -OH or -NH2 groups can substantially change the
orientation of the dipole.
Ultimately, it was discovered that conformers with similar κ values could be
distinguished by modeling the observed nuclear quadrupole splittings.18 These splittings
arise from a coupling between the nuclear spin angular momentum I of a quadrupolar
nucleus (I = 1 for 14N) and the angular momentum of overall rotation; different orientations
of I in the molecular frame have different energies owing to the interaction with a nonspherically symmetric electron distribution in the vicinity of the nucleus. The resulting
tensor is described by the formula χgg = eQqgg, where χgg are the nuclear quadrupole
coupling constant, eQ is the nuclear quadrupole moment (constant for every quadrupolar
nucleus), and qgg is the electric field gradient, with g = a, b, c. The different components in
a quadrupole-split rotational transition are designated by F, which can take on specific
values from (J-I) to (J+I), and obey the selection rule ΔF = 0, ±1. Ultimately, all transitions
were fit to a Watson non-rigid asymmetric rotor Hamiltonian corrected by the addition of
quadrupole coupling terms.
Figure 4.4 shows the quadrupole hyperfine patterns observed for the 22,0 ← 11,1
rotational transitions of the two g+g- conformers. Clearly evident is the fact that the two
patterns are different, owing to differences in the nuclear quadrupole coupling constants.
Comparisons with theory clearly show that the conformer with the smaller coupling is the
II3(g+g-) isomer that is predicted to be more stable, and that the conformer with the larger
coupling is the less stable I3(g+g-) isomer. The observed quadrupole patterns of the
remaining conformers also make possible an unambiguous assignation of the conformers
97
C, D, E, F and G to the I2(g+g+), II2(g-g-), II2(g+g+), II2(g-g+) and I2(g-g-) structures,
respectively.
A small number of very weak signals remain unassigned, and likely arise from
conformers predicted to have higher energies, and/or isotopomers of more stable structures.
Figure 4.4. Quadrupole hyperfine structure in the 220-111 rotational transitions of the two
conformers exhibiting cyclic HB networks.
98
4.4
Discussion.
A total of 172 lines (Table 4.3 to Table 4.9) have been observed and fit in the 7.5-
18.5 GHz region of the pure rotational spectrum of DTN. These have been assigned to
seven different conformers of the isolated molecule, whose relative energies (according to
MP2) span a range of only 216 cm-1, as shown in Table 4.10. Also included in this table
are the results from two DFT methods, using M05-2X and B3LYP functionals. The three
levels of theory all predict the two most stable conformers to be cyclic HB networks, in
which the direction of the cycle is reversed [(γβα) and (αβγ)], whereas the remaining five
structures are HB chains. The predicted relative energies of the seven observed conformers
differ among the different levels of theory. This is similar to the results obtained for
glycerol. In the case of glycerol,5 five conformers were assigned and identified with the aid
of theory; one of these is cyclic, whereas the remaining four are chains. The existence of
two low-energy cyclic structures in DTN may be traced to its lower symmetry.
99
Table 4.3. Observed frequencies and errors for the quadrupole hyperfine components
assigned in MHz for conformer A of D-threoninol.
J´
2
2
2
2
3
3
3
3
4
2
2
2
2
2
3
4
4
4
4
4
4
5
3
3
3
3
3
3
5
3
5
K´-1
1
1
1
1
1
0
1
1
0
2
2
2
2
2
1
1
0
2
2
1
1
0
2
2
2
2
2
2
1
2
0
K´+1
2
2
1
1
3
3
2
3
4
1
0
0
0
0
2
4
4
2
2
3
4
5
2
2
1
1
2
2
5
1
5
J´´
1
1
1
1
2
2
2
2
3
1
1
1
1
1
2
3
3
3
3
3
3
4
2
2
2
2
2
2
4
2
4
K´´-1
0
0
0
0
1
0
1
0
1
1
1
1
1
1
0
1
0
2
2
1
0
1
1
1
1
1
1
1
1
1
0
K´´+1
1
1
1
1
2
2
1
2
3
0
1
1
1
1
2
3
3
1
1
2
3
4
1
1
1
1
2
2
4
2
4
F´
3
2
3
2
4
4
4
4
5
3
1
3
2
2
4
5
5
4
5
5
5
6
4
3
4
3
4
3
6
4
6
F´´
2
1
2
1
3
3
3
3
4
2
1
2
1
2
3
4
4
3
4
4
4
5
3
2
3
2
3
2
5
3
5
νobs.
9007.385
9007.515
9699.835
9700.020
10541.405
10823.565
11232.860
12301.775
12870.275
13413.285
13662.815
13663.190
13663.425
13663.560
13685.715
14035.100
14348.530
14700.630
14700.780
14952.435
15513.355
16652.355
16815.500
16815.655
16910.135
16910.290
17507.945
17508.175
17514.235
17602.575
17817.155
νobs.-calc.
-0.001
-0.003
0.011
-0.003
-0.005
0.002
-0.009
-0.031
-0.005
0.001
-0.012
0.003
0.001
-0.011
0.013
0.019
0.007
-0.005
-0.013
0.008
0.031
-0.004
-0.002
-0.005
0.002
-0.005
0.005
0.010
-0.007
0.005
-0.005
100
Table 4.4. Observed frequencies and errors for the quadrupole hyperfine components
assigned in MHz for conformer B of D-threoninol.
J´
2
2
2
2
3
3
3
4
4
2
2
2
2
2
2
2
3
3
4
4
5
5
3
3
3
3
3
3
3
3
3
3
3
K´-1
1
1
1
1
1
1
1
0
0
2
2
2
2
2
2
2
1
1
1
1
0
0
2
2
2
2
2
2
2
2
2
2
2
K´+1
2
2
1
1
3
3
3
4
4
1
1
1
1
1
0
0
2
2
4
4
5
5
2
2
2
1
1
1
2
2
1
1
1
J´´
1
1
1
1
2
2
2
3
3
1
1
1
1
1
1
1
2
2
3
3
4
4
2
2
2
2
2
2
2
2
2
2
2
K´´-1
0
0
0
0
0
0
0
1
1
1
1
1
1
1
1
1
0
0
0
0
1
1
1
1
1
1
1
1
1
1
1
1
1
K´´+1
1
1
1
1
2
2
2
3
3
0
0
0
0
0
1
1
2
2
3
3
4
4
1
1
1
1
1
1
2
2
2
2
2
F´
3
2
3
2
2
4
3
4
5
1
3
1
2
2
3
2
4
3
3
5
5
6
2
4
3
2
4
3
4
3
2
4
3
F´´
2
1
2
1
1
3
2
3
4
1
2
0
1
2
2
1
3
2
2
4
4
5
1
3
2
1
3
2
3
2
1
3
2
νobs.
8982.665
8983.110
9718.120
9718.755
12254.920
12255.245
12255.430
12875.145
12875.425
13400.480
13401.835
13402.150
13402.415
13403.085
13668.515
13669.270
13725.020
13725.385
15441.805
15441.955
16648.625
16648.755
16788.105
16788.400
16788.930
16894.725
16895.035
16895.575
17523.825
17524.570
17630.100
17630.500
17631.225
νobs.-calc.
-0.001
-0.004
-0.003
-0.001
-0.004
-0.001
0.001
-0.009
-0.004
0.003
-0.002
0.003
-0.001
0.002
-0.002
-0.003
0.015
0.005
0.005
0.014
0.008
-0.011
0.000
0.000
-0.001
-0.009
0.002
0.000
-0.032
-0.002
0.012
0.010
0.008
101
Table 4.5. Observed frequencies and errors for the quadrupole hyperfine components
assigned in MHz for conformer C of D-threoninol.
K´+1
3
3
3
0
4
4
4
3
3
3
2
2
5
5
J´´
2
2
2
1
3
3
3
3
3
3
3
3
4
4
K´´-1
1
0
0
1
1
0
0
2
1
1
2
2
0
0
K´´+1
2
2
2
0
3
3
3
2
2
2
1
1
4
4
F´
4
4
3
3
5
5
4
5
5
4
4
5
6
5
F´´
3
3
2
2
4
4
3
4
4
3
3
4
5
4
νobs.
9811.439
10230.340
10230.620
12172.266
12979.704
13291.298
13291.538
14120.985
14969.257
14969.397
15039.412
15039.607
16276.978
16277.173
νobs.-calc.
0.000
-0.007
-0.004
0.001
0.000
0.016
0.003
-0.004
-0.008
-0.004
0.001
0.010
-0.008
0.003
102
Table 4.6. Observed frequencies and errors for the quadrupole hyperfine components
assigned in MHz for conformer D of D-threoninol.
J´
3
3
3
4
4
4
4
4
3
3
3
4
4
5
5
2
2
2
2
2
2
5
K´-1
0
1
1
1
0
2
2
1
1
1
1
1
1
1
0
2
2
2
2
2
2
1
K´+1
3
3
3
4
4
3
2
3
2
2
2
4
4
5
5
1
0
1
0
0
0
4
J´´
2
2
2
3
3
3
3
3
2
2
2
3
3
4
4
1
1
1
1
1
1
4
K´´-1
0
0
0
1
0
2
2
1
0
0
0
0
0
1
0
1
1
1
1
1
1
1
K´´+1
2
2
2
3
3
2
1
2
2
2
2
3
3
4
4
0
0
1
1
1
1
3
F´
4
4
3
5
5
5
5
5
3
4
2
5
4
6
6
3
3
3
2
3
1
6
F´´
3
3
2
4
4
4
4
4
2
3
1
4
3
5
5
2
2
2
1
2
0
5
νobs.
9112.672
11898.380
11898.530
11965.420
12139.730
12160.515
12183.075
12350.115
12475.525
12475.775
12475.920
14751.165
14751.350
14952.765
15157.950
15226.535
15228.795
15322.800
15324.790
15325.065
15325.355
15433.140
νobs.-calc.
0.004
-0.014
-0.028
-0.005
0.009
-0.008
-0.029
0.010
0.012
0.003
0.015
0.014
0.020
0.002
0.005
0.002
-0.005
-0.014
0.011
-0.016
0.013
-0.019
103
Table 4.7. Observed frequencies and errors for the quadrupole hyperfine components
assigned in MHz for conformer E of D-threoninol.
J´
3
3
3
3
3
3
2
2
4
4
4
4
4
4
4
5
5
5
5
5
K´-1
1
1
0
0
1
1
2
2
1
1
0
2
2
1
1
1
1
0
2
1
K´+1
3
3
3
3
2
2
1
1
4
4
4
3
3
3
3
5
5
5
4
4
J´´
2
2
2
2
2
2
1
1
3
3
3
3
3
3
3
4
4
4
4
4
K´´-1
1
1
0
0
1
1
1
1
1
1
0
2
2
1
1
1
1
0
2
1
K´´+1
2
2
2
2
1
1
1
1
3
3
3
2
2
2
2
4
4
4
3
3
F´
3
4
2
4
3
4
3
2
4
5
5
4
5
4
5
5
6
6
6
6
F´´
2
3
1
3
2
3
2
1
3
4
4
3
4
3
4
4
5
5
5
5
νobs.
9757.405
9757.670
10172.160
10172.360
11299.415
11299.650
12473.910
12474.410
12907.465
12907.590
13214.120
14045.490
14045.910
14888.975
14889.055
16005.220
16005.280
16182.445
17414.645
18290.010
νobs.-calc.
-0.021
-0.017
-0.016
-0.017
0.030
0.007
-0.003
0.009
-0.005
0.014
-0.002
0.025
0.036
-0.011
-0.020
0.006
0.014
0.010
-0.005
-0.038
104
Table 4.8. Observed frequencies and errors for the quadrupole hyperfine components
assigned in MHz for conformer F of D-threoninol.
J´
3
3
4
4
4
4
3
3
2
2
2
2
2
5
2
5
5
5
5
5
6
3
3
K´-1
1
1
1
2
2
1
1
1
2
2
2
2
2
0
2
2
2
2
1
1
1
2
2
K´+1
3
3
4
2
2
3
2
2
1
1
0
1
1
5
0
4
4
3
4
4
6
1
1
J´´
2
2
3
3
3
3
2
2
1
1
1
1
1
4
1
4
4
4
4
4
5
2
2
K´´-1
1
1
1
2
2
1
0
0
1
1
1
1
1
0
1
2
2
2
1
1
1
1
1
K´´+1
2
2
3
1
1
2
2
2
0
0
0
1
1
4
1
3
3
2
3
3
5
1
1
F´
3
4
5
4
5
5
4
3
3
2
3
3
2
6
3
5
6
6
5
6
7
4
3
F´´
2
3
4
3
4
4
3
2
2
1
2
2
1
5
2
4
5
5
4
5
6
3
2
νobs.
8411.750
8411.950
11165.860
12517.730
12517.985
12826.990
13010.454
13010.749
13814.961
13815.176
13864.646
14235.695
14236.155
14284.620
14285.395
15010.372
15010.542
15844.870
15914.845
15914.970
16575.310
16662.437
16662.662
νobs.-calc.
0.007
0.015
0.028
-0.001
0.001
-0.012
0.023
0.019
0.019
0.014
-0.003
0.002
-0.016
-0.038
-0.006
-0.008
0.010
-0.022
-0.035
0.035
0.013
-0.010
-0.014
105
Table 4.9. Observed frequencies and errors for the quadrupole hyperfine components
assigned in MHz for conformer G of D-threoninol.
J´
4
4
4
4
4
4
5
5
5
5
5
6
6
6
6
K´-1
1
1
0
2
1
1
1
1
0
1
1
0
2
2
2
K´+1
4
4
4
2
3
3
5
5
5
4
4
6
5
5
4
J´´
3
3
3
3
3
3
4
4
4
4
4
5
5
5
5
K´´-1
1
1
0
2
1
1
1
1
0
1
1
0
2
2
2
K´´+1
3
3
3
1
2
2
4
4
4
3
3
5
4
4
3
F´
4
5
5
4
4
5
5
6
6
5
6
7
6
7
7
F´´
3
4
4
3
3
4
4
5
5
4
5
6
5
6
6
νobs.
11895.270
11895.410
12091.445
12148.875
12335.565
12335.745
14863.760
14863.865
15092.160
15413.485
15413.585
18078.890
18168.545
18168.710
18272.045
νobs.-calc.
0.020
0.013
-0.044
0.001
-0.008
0.011
-0.021
0.009
0.010
-0.002
0.014
-0.001
0.002
0.003
-0.007
106
Table 4.10. Comparison of relative energies of the seven most stable conformations of DTN at the indicated levels of theory
(including zero point corrected vibrational energies).
a
II3(g+g-)
I3(g+g-)
I2(g+g+)
II2(g-g-)
II2(g+g+)
II2(g-g+)
I2(g-g-)
ΔEMP2+ZPC (cm-1)a
0
45
61
70
134
215
216
ΔEM052X+ZPC (cm-1)a
0
72
345
273
477
499
336
ΔEB3LYP+ZPC (cm-1)a
0
68
177
87
179
144
132
All calculations run using a 6-311++g(d,p) basis set.
106
107
We also estimated the relative populations of the different DTN conformers using
the relative transition intensities in the spectrum. Employing a method that is described
elsewhere,19 the intensities of each of the lines assigned to each specific conformer were
divided by the squares of the respective dipole moment components predicted by theory,
and then summed, to obtain an estimate of the total number of each conformer present in
the expansion.
After normalizing the results for all seven structures, the different
intensities were taken to be proportional to the relative populations of the different
conformers, yielding the results A > D > B > E > C > F > G. This “landscape” is somewhat
different from the predicted one, A > B > C, etc. Conformers A and B are the two Hbonded cycles with different directions of the H-bonds in the cycle, while D is the second
most stable of the chain forms, lying only 25 and 9 cm-1 higher in energy (according to
MP2) above conformers B and C, respectively.
The different relative abundances obtained from experiment may be influenced by
conformational interconversions during the cooling process. The next four higher energy
structures (Figure 4.2) are related to four of the observed conformations by a rotation of
the free –OH group in a chain structure. The charts in Figure 4.5 shows the interconversion
barriers between each pair of conformers calculated using MP2 methods. The barriers for
I2(g-g-) and for II2(g+g+) from the higher energy conformer to the lower one are each less
than 200 cm-1, while the corresponding barriers for the II2(g-g-) and I2(g+g+) conformers
are about 300 cm-1. In all four pairs, relaxation of the higher energy conformers to the lower
energy ones is likely,20 possibly decreasing the populations of the former and increasing
the populations of the latter in the experiment.
108
1200
800
600
400
800
600
400
340
189
200
200
0
0
10
60
1200
110
160
∡ HOCC / ͦ
10
210
ΔE / cm-1
600
313
200
100
150
200
250
∡ HOCC / ͦ
300
350
210
800
600
400
200
0
110
160
∡ HOCC / ͦ
I2(g-g-)
1000
800
400
60
1200
II2(g-g-)
1000
ΔE / cm-1
II2(g+g+)
1000
ΔE / cm-1
ΔE / cm-1
1200
I2(g+g+)
1000
178
0
100
150
200
250
∡ HOCC / ͦ
300
350
Figure 4.5. Calculated (MP2/6-31++G(d,p)) interconversion barriers involving pairs of
conformers in DTN linked by internal rotation of the terminal acceptor -OH groups.
In all of the conformers presented here, there is more than one HB stabilizing the
structure. Cooperative hydrogen bonding of this type is a consequence of the fact that both
oxygen and nitrogen have lone pairs of electrons, each of which can form a HB with a
hydrogen attached to an adjacent group. A similar behavior occurs on a larger scale in
liquid water, as well as in water clusters. There also is a redistribution of charge on initial
HB formation.21 In the model structure X–H···Y, the acceptor atom Y transfers some
electron density to the donor atom X, encouraging further HB acceptance by X, and
donation to Y. In water, cooperative hydrogen bonding increases the average O–H bond
length and decreases the average O···O distances.22-23
109
The co-existence of both cycles and chains in DTN is a characteristic property of its
energy landscape. Remarkably, the two different structural types have relatively similar
energies. While the cyclic forms have more HB’s, the smaller number of HB’s in the chain
forms is compensated for by their higher strength; chain HB’s are more nearly linear,
exhibit shorter HB distances and more relaxed structures. The predicted structures show
longer O–H distances when the hydrogen atom is linked to a nitrogen atom in the chain,
compared to the cyclic structures. Also, in DTN, the –NH2 group has more acceptor-like
character than the –OH group, making the OH···N HB’s shorter than the OH···O HB’s.
This is shown explicitly for two conformers of glycerol and DTN in Figure 4.6. In the chain
forms, the donor HB distance from the first (O)H to the central OH is larger in glycerol
than the corresponding distance in DTN, whereas the donor distance from the central OH
in glycerol to the third O(H) is smaller than it is in DTN. In the cyclic forms, there are
two shorter HB’s and one longer one in glycerol, and two longer HB’s and one shorter one
in DTN. Clearly, all of these effects can be traced to the amino group being a better Hbond acceptor but poorer donor than its OH counterpart. Interestingly, in both glycerol
and DTN, the cycles and chains have similar energies, since distortions along one
coordinate can be compensated for by complementary distortions along other coordinates.
The energy landscapes of DTN and glycerol are significantly different, when
examined in greater detail. The two molecules have equivalent conformations, as already
noted. But their predicted relative energies are different. Calculations at the MP2/6311++G(d,p) level give for glycerol5 a conformation similar to I2(g+g+) as the global
minimum, instead of the cyclic HB network, which is the second most stable form, 148
cm-1 to higher energy. All the lower energy conformations in DTN have their equivalent in
110
glycerol, spreading around a larger range of energies (750 cm-1). In this range, glycerol
also presents extra conformations, many of which feature only one HB. Conformations
similar to this for DTN are more unstable, and all lie above 700 cm-1. More surprising is
that glycerol has an additional conformation with respect to DTN, with an interaction (‒
OH)β···(‒OH)γ···(‒OH)α [(βγα) chain] lying at only 248 cm-1 above the global minimum,
in which the central –OH group only acts as a H-donor in an HB chain. The absence of this
kind of HB network in DTN provides further evidence for the stronger character of the
amino group as an H-bond acceptor.
Figure 4.6. Calculated hydrogen bond distances (in Å) in two comparable conformations
of glycerol and DTN, obtained using MP2/6-311++G(d,p).
111
DTN shares the same backbone structure with the amino acid L-threonine (THR),
apart from the acid group in the α position. Calculations of the energy landscape of THR
(at the MP2/6-311++G(d,p) level)24 revealed ten conformers with energies up to 1000 cm1
above the global minimum. Of these, the ones denoted as IIa, IIIαa and III’αb share a
similar structure with conformers I2(g+g+), II2(g+g+) and II2(g-g-), respectively, of DTN.
Only the first two were observed in the pure rotational spectrum of THR, the third one was
considered too high in energy (calculated at ca. 700 cm-1) to be observed in the free-jet
expansion. All three corresponding conformers of DTN have been detected in the present
work. It is interesting that the range of energies spanned by the conformers of THR is so
much larger than that in DTN. Likely, this is a consequence of the additional CO group of
the amino acid, a source of additional stabilization via the HB’s that may be formed with
it.
Finally, we return to the main theme in this work, a search for possible similarities
in the structural and/or dynamical properties of apparently dissimilar molecules like
glycerol, DTN and the riboses that might explain their common ability to act as scaffolds
for building nucleic acids. One factor that might play a role is that all such molecules
contain two or more adjacent hydroxy and/or amino groups that can form networks of
intramolecular OH···OH··· or OH···NH2··· hydrogen bonds. As a result, one side of the
carbon framework is electron-rich, and the other is electron-poor. This is illustrated for the
specific cases of the two DTN in the structures shown in Figure 4.7. Thus, establishment
of cooperative HB networks leaves “exposed” two sides of the molecule, an electron-rich
area to which electropositive groups might be attracted, and an electron-poor area to which
electronegative groups (such as phosphates and the nitrogenous bases) might be attracted.
112
A second factor is that such molecules exhibit many different cyclic and chain structures
with nearly equivalent energies. Despite the enhanced strength of the individual HB’s, they
are cooperative in nature, so that strengthening one weakens another. As noted by
Scheiner21 and others,25 perturbations induced by the formation of one HB are themselves
affected by the presence of other HB’s.
.
Figure 4.7. HOMO’s of the two most stable predicted conformers of DTN. The molecular
orbitals are located mostly above the plane of the backbone.
A third and final factor, possibly the deciding one, is that interconversions between
the different structures are relatively facile. Breaking a single HB in an isolated molecule
typically requires a large energy, 1000 cm-1 or more. But in systems where there is a
cooperative HB network, the interconversion energies may be significantly smaller, 300
cm-1 or less, since the breaking of one HB is accompanied by the formation of another. (kT
~ 200 cm-1 for a single molecule.) Indeed, in glycerol, a multi-step interconversion pathway
between an OH···OH···OH chain pointing in one direction and in the other along the chain
was calculated to have a barrier in this range.6 We have not explored these pathways in
detail in this work, but the calculations shown in Figure 4.5 provide some hint that a similar
113
facile pathways may be present in DTN, although those interconversions do not involve
breaking H-bonds, but only a reorientation of the acceptor OH group. It would be
interesting to calculate a more complete potential energy landscape including the various
interconversion pathways in DTN compared to glycerol. Thus, DTN and other related
molecules are anticipated to have relatively flat energy landscapes, which likely accounts
for their versatile behavior in biological systems.
114
4.5
References
1.
Stryer, L. Biochemistry, 4th ed.; W.H. Freeman and Co.: New York, 1995.
2.
Zhang, L. L.; Peritz, A.; Meggers, E. A simple glycol nucleic acid. J. Am. Chem.
Soc. 2005, 127, 4174-4175.
3.
Asanuma, H.; Toda, T.; Murayama, K.; Liang, X. G.; Kashida, H. Unexpectedly
Stable Artificial Duplex from Flexible Acyclic Threoninol. J. Am. Chem. Soc. 2010,
132, 14702-14703.
4.
Carcabal, P.; Jockusch, R. A.; Hunig, I.; Snoek, L. C.; Kroemer, R. T.; Davis, B. G.;
Gamblin, D. P.; Compagnon, I.; Oomens, J.; Simons, J. P. Hydrogen bonding and
cooperativity in isolated and hydrated sugars: mannose, galactose, glucose, and
lactose. J. Am. Chem. Soc. 2005, 127, 11414-11425.
5.
Ilyushin, V. V.; Motiyenko, R. A.; Lovas, F. J.; Plusquellic, D. F. Microwave
spectrum of glycerol: Observation of a tunneling chiral isomer. J. Mol. Spectrosc.
2008, 251, 129-137.
Maccaferri, G.; Caminati, W.; Favero, P. G. Free jet investigation of the rotational
spectrum of glycerol. J. Chem. Soc., Faraday Trans. 1997, 93, 4115-4117.
6.
7.
Cocinero, E. J.; Lesarri, A.; Ecija, P.; Basterretxea, F. J.; Grabow, J. U.; Fernandez,
J. A.; Castano, F. Ribose Found in the Gas Phase. Angew. Chem. Int. Ed. 2012, 51,
3119-3124.
8.
Andreessen, B.; Steinbuchel, A. Serinol: small molecule - big impact. Amb Express
2011, 1:12.
9.
Carter, H. E.; Glick, F. J.; Norris, W. P.; Phillips, G. E. BIOCHEMISTRY OF THE
SPHINGOLIPIDES .3. STRUCTURE OF SPHINGOSINE. J. Biol. Chem. 1947,
170, 285-294.
10.
Mohamadi, F.; Richards, N. G. J.; Guida, W. C.; Liskamp, R.; Lipton, M.; Caufield,
C.; Chang, G.; Hendrickson, T.; Still, W. C. MACROMODEL - AN INTEGRATED
SOFTWARE SYSTEM FOR MODELING ORGANIC AND BIOORGANIC
MOLECULES USING MOLECULAR MECHANICS. J. Comput. Chem. 1990, 11,
440-467.
11.
Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.;
Cheeseman, J. R.; Scalmani, G.; Barone, V.; Mennucci, B.; Petersson, G. A.; et
al. Gaussian 09, revision C.01; Gaussian, Inc.: Wallingford, CT, 2010.
12.
Brown, G. G.; Dian, B. C.; Douglass, K. O.; Geyer, S. M.; Shipman, S. T.; Pate, B.
H. A broadband Fourier transform microwave spectrometer based on chirped pulse
excitation. Rev. Sci. Instrum. 2008, 79, 053103.
115
13.
Bird, R. G.; Neill, J. L.; Alstadt, V. J.; Young, J. W.; Pate, B. H.; Pratt, D. W. Ground
State N-14 Quadrupole Couplings in the Microwave Spectra of N,N 'Dimethylaniline and 4,4 '-Dimethylaminobenzonitrile. J. Phys. Chem. A 2011, 115,
9392-9398.
14.
Watson, J. K. G. Vibrational Spectra and Structure; Durig, J.R., Ed.; Elsevier: New
York/Amsterdam, The Netherlands, 1977; Vol.6, pp 1-89.
15.
Plusquellic, D. F.; Suenram, R. D.; Mate, B.; Jensen, J. O.; Samuels, A. C. The
conformational structures and dipole moments of ethyl sulfide in the gas phase. J.
Chem. Phys. 2001, 115, 3057-3067.
16.
Pickett, H. M. SPFIT/SPCAT. http://spec/jpl.nasa.gov.
17.
Ray, B. S. Über die Eigenwerte des Asymmetrischen Kreisels. Z. Phys. 1932, 78,
74-91.
18.
Alonso, J. L.; Vaquero, V.; Pena, I.; Lopez, J. C.; Mata, S.; Caminati, W. All Five
Forms of Cytosine Revealed in the Gas Phase. Angew. Chem. Int. Ed. 2013, 52,
2331-2334.
19.
Neill, J. L.; Douglass, K. O.; Pate, B. H.; Pratt, D. W. Next generation techniques
in the high resolution spectroscopy of biologically relevant molecules. Phys. Chem.
Chem. Phys. 2011, 13, 7253-7262.
20.
Ruoff, R. S.; Klots, T. D.; Emilsson, T.; Gutowsky, H. S. RELAXATION OF
CONFORMERS AND ISOMERS IN SEEDED SUPERSONIC JETS OF INERTGASES. J. Chem. Phys. 1990, 93, 3142-3150.
21.
Scheiner, S. Cooperativity of multiple H-bonds in influencing structural and
spectroscopic features of the peptide unit of proteins. J. Mol. Struct. 2010, 976, 4955.
22.
Ludwig, R. The effect of hydrogen bonding on the thermodynamic and
spectroscopic properties of molecular clusters and liquids. Phys. Chem. Chem. Phys.
2002, 4, 5481-5487.
23.
Perez, C.; Lobsiger, S.; Seifert, N. A.; Zaleski, D. P.; Temelso, B.; Shields, G. C.;
Kisiel, Z.; Pate, B. H. Broadband Fourier transform rotational spectroscopy for
structure determination: The water heptamer. Chem. Phys. Lett. 2013, 571, 1-15.
116
24.
Alonso, J. L.; Perez, C.; Sanz, M. E.; Lopez, J. C.; Blanco, S. Seven conformers of
L-threonine in the gas phase: a LA-MB-FTMW study. Phys. Chem. Chem. Phys.
2009, 11, 617-627.
25.
Parra, R. D.; Streu, K. Hydrogen bond cooperativity in polyols: A DFT and AIM
study. Comp. Theor. Chem. 2011, 967, 12-18.
117
CHAPTER 5.
BROADBAND MICROWAVE SPECTROSCOPY
OF PROTOTYPICAL AMINO ALCOHOLS AND POLYAMINES:
COMPETITION BETWEEN H-BONDED CYCLES AND CHAINS
5.1
Introduction
Amino alcohols and their derivatives are widely used in organic synthesis and
medicinal chemistry. Reduced from a chiral pool such as the L-amino acids, β-amino
alcohols serve as chiral auxiliaries or ligands for asymmetric catalysis.1 Various amino
alcohol derivatives also exhibit antimicrobial2 and antifungal3 activity. As a result, the
amino alcohol group has been adopted in several antibiotics, including ethambutol4
prescribed for treatment of tuberculosis and other infections.
With a combination of amino and hydroxyl functional groups distributed along an
alkyl chain, the amino alcohols can engage in intramolecular H-bonds that dictate the
conformational preferences of the molecules.
One useful strategy for probing the
hydrogen-bonding architecture of conformationally flexible molecules is to study them in
isolated form in the gas phase, where supersonic expansion can be used to collisionally
cool the sample, thereby trapping the population in the low-lying conformational minima
where the conformations can be interrogated by a range of spectroscopic methods.5 6 7
When the molecule of interest incorporates a UV chromophore, IR/UV double resonance
methods can provide single conformation IR spectra that report directly on the hydrogen
bonding architecture via the hydride stretch fundamentals.8 However, they require the
presence of an aromatic chromophore, and provide less detailed structural characterization
than might be ideal.
118
Chirped-pulse Fourier transform microwave spectroscopy (CP-FTMW)9 is a
powerful alternative for the assignment and structural determination of different
conformational isomers in the gas phase. It allows for the acquisition of broadband spectra
at high resolution of molecules that possess permanent dipole moments. Moreover, the
intensities of the rotational transitions can be related in a straightforward way to the relative
populations. Thus, it is a well suited technique to examine the conformational properties
of the small amino alcohol molecules in the gas phase.
In a recent study from our group,10 CP-FTMW spectroscopy was used to determine
the conformational preferences of a first amino alcohol, D-threoninol. With functional
groups on three adjacent alkyl carbons (one NH2 and two OH groups), cooperative
hydrogen-bonded networks could be formed, with examples of both cyclic and chain
structures represented.
In what follows, we expand our study of the amino alcohols to include D-allothreoninol, 2-amino-1,3-propanediol and 1,3-diamino-2-propanol, and supplement these
studies with the triamine analog 1,2,3-triamino propane.
D-allo-threoninol is a
diastereomer of D-threoninol, differing in the chirality at a single site. In the other two,
the terminal methyl group is removed in order to simplify the potential energy surface. In
so doing, it is possible to concentrate more directly on the changes induced by the position
and number of NH2/OH groups along the alkyl chain.
119
This series has as its tri-alcohol analog glycerol, HOCH2-CH(OH)-CH2OH, whose
microwave spectrum was recently studied in detail by Ilyushin et al.11 The three adjacent
OH sites in glycerol are reminiscent of the sugars.12 As we shall see, the molecules in our
series have low-energy conformers that form both H-bonded chains and cycles, much as
occurs in glycerol. As a result, we can compare and contrast the ways in which the
chains and cycles compete with one another as a function of OH/NH2 make-up.
5.2
Experimental and Computational Methods
To identify the possible conformational minima associated with each molecule in the
series, an exhaustive conformational search was carried out using the Amber* force field
in the MACROMODEL13 suite of programs. Depending on the molecule, anywhere from
20-100 structures were found within the 50 kJ/mol energy window prescribed for the
search. These structures served as starting geometries for full optimizations using ab initio
and density functional theory (DFT) calculations via the Gaussian 09 suite of programs.14
For initial structure prediction, optimizations were carried out at the MP2/6-311++G(d,p)
level of theory. Rotational constants at this level of theory have been shown in previous
work to be in close agreement with experiment.15
While this level of theory is useful for structure determination, the calculated
energies are known to contain significant basis set superposition error.16 Therefore, further
calculations were carried out, using the redefined structures of the MP2/6-311++G(d,p)
method as starting point, for further optimization and energy determination. MP2/aug-ccpVTZ calculations explored the effects of increasing the basis set on the MP2 energies.
DFT calculations employing the B3LYP17 or M05-2X18 hybrid functionals, or B2PLYP19
double-hybrid functional, all with the aug-cc-pVTZ basis set, provided a range of methods
120
between which relative energies could be compared. Dispersion correction from Grimme
and co-workers with Becke-Johnson dampening was added to the B3LYP and B2PLYP
calculations.19,20 Both of these levels of theory were shown to give good results for the
relative energies in a recent study of monoglyme (CH3OCH2CH2OCH3), a molecule similar
in size to the aminoalcohols of interest here.21 Also, the M05-2X hybrid functional has
performed well in predicting relative energies in previous work.18 Tight convergence
criteria were employed, and zero-point energy corrections were included based on
harmonic vibrational frequency calculations.
The experimental methods for the CP-FTMW measurements have been described
in Chapter 2. Briefly, the solid sample was wrapped in cotton and inserted into a stainless
steel sample holder that was heated to ~ 130°C to obtain sufficient vapor pressure, and
entrained in neon carrier gas at a backing pressure of 0.7 bar. The sample holder was
located immediately behind a pulsed valve (Parker General Valve, Series 9) with a 1 mm
diameter nozzle orifice, operating at 10 Hz. 1 µs long frequency-chirped microwave pulses
spanning the 7.5-18.5 GHz range interrogated the jet-cooled molecules. Upon interaction
with the microwave field, a macroscopic polarization was induced in the sample, and a 20
μs free induction decay (FID) was collected and down converted for display and processing
on a 12 GHz digital oscilloscope operating at a sampling rate of 40 GS/s. 10,000 freeinduction decays (FIDs) were averaged in the time domain, both with (signal+background)
and without (background only) spectra. Limited sample sizes, especially for propane1,2,,3-triamine, prevented longer averages. The background contains resonances arising
from reflecting surfaces in the chamber, which were identified and removed from the
spectrum.
121
Two related methods were employed to extract relative populations of the observed
conformers from the microwave spectra. Intensities of transitions were tabulated without
an attempt to normalize for changes in microwave power with frequency.9 A simple,
approximate method employed in previous work,22 involved taking the sum of the
intensities of each type of microwave transition (a-, b-, or c-type) for a given conformer
(e.g., conformer X), dividing by the square of the associated component of the dipole
moment squared, and summing:
 
∑  ()
 2
+
∑  ()
 2
+
∑  ()
 2
.
Alternatively, a least-squares fit to the microwave spectra for each conformer was
obtained by first varying the temperature to obtain a best-fit temperature (Trot=1.5 K), and
then extracting the relative population of each conformer from the normalization constants
for each conformer obtained from the best-fit. The two methods yielded similar percentage
populations, consistent with estimated errors of about ±5% on these percentages.
122
5.3
5.3.1
Results and Analysis
Nomenclature
Despite the small size of these molecules, with three adjacent hydrogen bonding
substitutents involving a combination of OH and NH2 groups, several different
combinations of H-bonding architecture (e.g., cycle versus chain) and backbone dihedral
angles are possible. In each case, about 10 different conformations of each molecule are
predicted to have energies within 500 cm−1 of the global minimum (Table 5.1-5.4). D-allothreoninol (2S,3R), is a diastereomer of D-threoninol (2S, 3S). The other amino alcohols
and the triamine all lack the terminal methyl group present in threoninol, giving them
greater symmetry and eliminating one of the chiral centers. Nevertheless, a similar
nomenclature to the one used for D-threoninol10, which is adapted from one that is often
employed for small aliphatic amino acids, was used here.
Two examples are shown below for 2-amino-1,3-propanediol. After rotating the
molecules so that as many heavy atoms as possible are placed above the plane of an upsidedown V-shaped carbon framework, the substituents are listed as α-OH, β -NH2 and γ OH from left to right. Roman numerals I,II and III denote the direction of the H-bonds
along the carbon framework, α→β→γ (I), γ→β →α (II) and γ→α →β (III). A
subscript (2 or 3) is then used to denote the number of the intramolecular hydrogen bonds
within the molecule, 2 for H-bonded chains and 3 for H-bonded cycles. Finally, the
dihedral angles between adjacent heavy atom substituents are given in parentheses, with
angles within ±10° of +60° and -60° labeled as gauche (g+ and g-, respectively) while
123
those in the range between ± 90° and 180° are labeled anti (a). The dihedral angle between
Oα and Nβ is listed first, while that for Nβ with respect to Oγ is listed second.
Similar bonding patterns are also present in propane-1,2,3-triamine and 1,3-diamino2 propanol. In propane-1,2,3-triamine, since the β(NH2) has two hydrogen bond donors,
a unique H-bonding pattern (labeled IV) is possible in which the β amine donates both to
α and γ substituents, α(NH2)←β(NH2) →γ(NH2). Finally, in 1,3-diamino-2-propanol,
several structures with only a single H-bond were also predicted within 500 cm−1 of the
global minimum. However, as we shall see, these structures were higher in energy, and
not observed in the supersonic expansion.
Finally, it is worth noting that the three propyl derivatives that form the main
sequence studied here (2-amino-1,3-propanediol, 1,3-diamino-2-propanol, and propane1,2,3-triamine) are all symmetric, and thus have two identical minima on the potential
energy surface related by reflection of the standard configuration through a vertical plane.
This changes I3(g-g+) into II3(g+g-). In principle, tunneling can interconvert these
minima, but in practice no such tunneling splittings were observed. Since all minima of
these molecules come in such identical pairs, the relative populations are unaffected by
their presence.
124
Table 5.1. Structures for all conformational minima (within 500 cm-1 of the global minimum) of D-allo-threoninol. The structural
types are given above, along with their zero point corrected relative energies in cm-1 at the B2PLYP aug-cc-pVTZ level of theory.
II3 (g+ g-)
II2(g- g- )
I3 (g+ g- )
II2(g- g- )
I2(g- g-)
III2 (g+ a)
0 cm-1
58 cm-1
188 cm-1
203 cm-1
319 cm-1
401 cm-1
124
125
Table 5.2. Structures for all conformational minima (within 500 cm-1 of the global minimum) of 2-Amino-1,3-propanediol. The
structural types are given above, along with their zero point corrected relative energies in cm-1 at the B2PLYP aug-cc-pVTZ level of
theory
I2 (g- g-)
I3 (g- g+)
II2(g+ g+)
II2(g- g+)
II2 (g- g-)
II2(g- g-)
0 cm-1
55 cm-1
139 cm-1
201 cm-1
204 cm-1
264 cm-1
125
126
Table 5.3. Structures for all conformational minima (within 500 cm-1 of the global minimum) of 1,3-diamino-2-propanol. The
structural types are given above, along with their zero point corrected relative energies in cm-1 at the B2PLYP aug-cc-pVTZ level of
theory.
II3(g- g+)
II2(g+ g-)
I2(g+ g-)
I2 (g- g-)
0 cm-1
95 cm-1
112 cm-1
239 cm-1
126
127
Table 5.4. Structures for all conformational minima (within 500 cm-1 of the global minimum) of propane-1,2,3-triamine. The
structural types are given above, along with their zero point corrected relative energies in cm-1 at the B2PLYP aug-cc-pVTZ level of
theory
II2(g+ g+)
IV2(g- g+)
II3(g- g+)
II2(g- g-)
I2(g- g+)
II2(g- g+)
I2(g+ g+)
0 cm-1
151 cm-1
229 cm-1
293 cm-1
302 cm-1
360 cm-1
392 cm-1
127
128
5.3.2
Microwave Spectra
5.3.2.1 D-allo-threoninol
The experimental microwave spectrum of D-allo-threoninol over the range of 7.518.5 GHz is presented in Figure 5.1. With hundreds of lines observed with a range of
intensities, we anticipate that, as in the case of D-threoninol, separate contributions from
different conformers will contribute to the spectrum. Since each conformer exhibits a
unique pattern of lines, transitions from each are intermingled with one other.
129
Figure 5.1 (a) Experimental rotational spectrum of jet-cooled D-allo-threoninol from 7.5
to 18.5 GHz. (b) Close-up of the 15.85-15.95 GHz region with calculated stick spectra
due to conformer A (red), B (blue), and C (green) below, showing the quality of the fit.
(c) Further expansion of 5 MHz regions around the 322-212 transitions of conformers A,
C are shown in part (b), in order to compare the experimental and calculated nuclear
quadrupolar splittings for the two assigned cyclic conformers. The F’-F” labels are
included in the figure.
130
To analyze the spectrum, we focused attention on the nine predicted lowest energy
conformers within 500 cm−1 of the global minimum, as summarized in Table 5.5. The
calculated rotational constants for these conformers were used as a starting point in
providing predictions for the rotational spectrum, using a semirigid rotor Hamiltonian.23
The resulting spectrum was then plotted using the JB95 software developed by
Plusquellic24 as a visualization and analysis tool. Tentative assignments were made based
on the close match-up of several strong lines in the spectrum. These experimental
frequencies were then put into Pickett's SPFIT/SPCAT program25 to refine the rotational
constants, and expand the number of fitted transitions. Transitions due to three conformers
were identified in the spectrum, with calculated stick spectra for the three shown in red
(conformer A), green (conformer B), and blue (conformer C) below the experimental
spectrum. The number of fitted transitions N, the average standard deviation for each fit
(σ), and the resulting sets of fitted rotational constants (A, B, C) and inertial defect (∆) are
listed in Table 5.6. For the range of rotational transitions observed, centrifugal distortion
constants provided only marginally better fits with experiment, and so are not included in
the table. Figure 5.1(b) includes an expanded view of the 15.85-15.95 GHz region,
demonstrating the quality of the fit.
131
Table 5.5. Calculated rotational parameters and relative energies of the nine most stable confirmations of D-allo-threoninol.
Calculated rotational parameters and relative energies of the nine most stable confirmations of D-allo-threoninol.
I(B)
II (A)
III (C)
IV
V
VI
VII
VIII
IX
II2(g- g- )
3888/3883
II3 (g+ g-)
3141/3130
I3 (g+ g- )
3154/3180
II2(g- g- )
3851/3855
I2(g- g-)
3835/3836
III2 (g+ a)
3491/3491
II2(g- g+)
4376
I2 (g+ g+)
3861
I2 (g- g+ )
4412
1799/1795
2200/2208
2153/2139
1788/1785
1787/1782
2165/2153
1614
1946
1599
C (MHz)
1497/1492
1928/1934
1940/1932
1488/1486
1493/1494
1555/1546
1433
1655
1430
μa (D)
1.4
-0.2
0.1
-0.8
-4.4
0.2
-2.9
-1.7
3.8
μb (D)
-0.9
1.7
0.9
-0.7
1.2
3.2
2.5
-0.8
1.5
μc (D)
-1.1
-3.2
-2.9
3.5
-1.4
0.6
1.5
0.6
0.9
μT (D)
1.9
3.7
3.1
3.7
4.8
3.3
4.1
1.9
4.2
χaa (MHz)
0.73
1.08
-2.85
0.49
-5.01
-0.03
1.08
0.40
-4.49
χbb (MHz)
2.01
2.54
1.28
1.93
2.88
1.96
2.55
-0.45
1.67
χcc (MHz)
-2.74
-3.62
1.58
-2.42
2.13
-1.94
-3.63
0.05
2.82
κ
-0.75
-0.55
-0.61
-0.75
-0.75
-0.36
-0.88
-0.74
-0.88
58
0
188
203
319
401
A (MHz)
B (MHz)
a
ΔE(cm-1)
b
a
Calculated rotational parameters at the MP2/6-311++G(d,p) level of theory. Rotational constants at the B2PLYP-D3BJ/aug-cc-pVTZ
level of theory are also included after the slash (/). bCalculated relative energies (cm-1) at the B2PLYP-D3BJ/aug-cc-pVTZ level of
theory, including harmonic zero-point energy correction.
131
132
Table 5.6. Experimental rotational parameters of the three assigned conformers of D-allothreoninol.
Conformer A
Conformer B
Conformer C
A (MHz)
3132.501(1)
a
3859.899(1)
3167.001( 3)
B (MHz)
2180.140(2)
1785.658(1)
2129.223( 3)
C (MHz)
1919.830(2)
1489.387(1)
1920.516( 7)
Δ (kHz)
0.22(9)
0.06( 3)
0.55(1)
χaa (MHz)
0.41(2)
0.58(2)
-2.21(3)
χbb (MHz)
1.64(2)
1.86(3)
1.35(5)
χcc (MHz)
-2.05(2)
-2.45(3)
0.85(5)
N
45
43
20
σ / KHz
18.3
13.2
16.7
κ
-0.57
-0.75
-0.66
b
c
µa:µb:µc
c>b>a
a>c>b
c>b>a
a
Errors in parentheses are expressed in units of the last digit. bNumber of fitted lines,
including nuclear hyperfine components. cStandard deviation of the fit.
The three assigned conformers for D-allo-threoninol are all near-prolate asymmetric
tops with the Ray's asymmetry parameter κ=(2B-A-C)/(A-C)26 lying between -0.5 and 0.8. Two of them have similar κ values around -0.6 while the third one has κ= -0.75,
indicating two different structural classes based on the dihedral angles between adjacent
heavy atom substituents. Compared with calculation, structures II and III in Table 5.5 have
κ values around -0.6, a value consistent with a g+g- configuration . A κ near -0.75 is
consistent with any of the structures I, IV and V, which all share a g-g- configuration, or
structure VIII with its g+g+ configuration. Structures with similar κ values typically also
have similar rotational constants since they share the same heavy atom configuration. For
example, the experimental rotational constants for structures II and III in the g+g- family
are 3132, 2180 and 1920 MHz and 3167, 2129 and 1920 MHz, respectively. These
133
differences are not large enough to discriminate between them based on rotational
constants alone.
Conclusive evidence for structural assignments comes from the different
14
N
nuclear quadrupole splitting patterns, which are produced by the interaction between the
nuclear spin angular momentum I (I=1 for 14N) and the electric field gradient created by
the rest of the molecule at the site of each nucleus. Figure 5.1(c) shows 5 MHz regions
around the 322-212 rotational transitions of the two g+g- conformers, and compares the
experimental nuclear quadrupolar splittings with those calculated for the assigned
structures of the two cyclic isomers. The patterns are different enough to make a clear
assignment for conformer A as structure II, the II3 (g+, g-) conformer, while conformer C
is assigned to structure III, the I3 (g+, g- ) conformer. This fitting procedure determines a
set of nuclear quadrupole coupling constants χ, which are very sensitive to the orientation
of -NH2 group with respect to the principal inertial axis system.27 The fitted values for χ for
the assigned conformers are listed in Table 5.6 and match nicely with those predicted by
the ab initio calculations.
Structures I and IV have similar quadrupole coupling constants due to the similar
orientation of the amino group in these two conformations. To distinguish between them,
we make use of the direction of the dipole moment in the molecular frame, whose
projections on the inertial axes determine the relative intensities of the a-, b-, and c-type
microwave transitions in the spectrum. The rotational spectrum of conformer B shows
strong a-type, medium strength c-type, and relatively weak b-type transitions. These data
are consistent with the assignment of conformer B as structure I, which is predicted to have
134
μa > µc > μb. C-type transitions are predicted to be strongest in structure IV, counter to
experiment. Thus, we assign conformer B to structure I, labeled as II2(g- g- ).
5.3.2.2 2-amino 1,3 propanediol, 1,3-diamino-2-propanol and Propane-1,2,3-triamine
Similar procedures were followed in arriving at assignments for the rotational spectra
of
2-amino-1,3-propanediol,
1,3-diamino-2-propanol,
and
propane-1,2,3-triamine
molecules, respectively. The results are summarized in Table 5.7-5.12. These molecules
possess higher symmetry than D-threoninol and D-allo-threoninol molecules since the
methyl group at the end of the threoninol structure is removed from the carbon framework.
As a result, Ray's asymmetry parameter gains in importance over the dihedral angles of the
functional groups as a means of grouping them into different structural classes.
For 2-amino 1,3 propanediol, the predicted rotational constants and asymmetry
parameters for the eight lowest energy structures are summarized in Table 5.7, with κ value
ranges between -0.02 and -0.91. Conformers fall into sub-groups around four characteristic
κ values ( κ= -0.02, -0.3, -0.85, -0.91 ), with the κ= -0.91 family characteristic of the most
extended structures, while κ= -0.02 family is most compact.
When compared with the experimentally-derived κ values, a clear assignment for
conformer A to structure II, I3 (g-, g+), can be made, since it is the only one of the lowenergy structures with κ values around -0.3.
Conformer D is fit with rotational constants that give a κ=-0.91. Two κ= -0.91
conformers, structure V and structure VIII, are labeled as II2(g- g+), as they are extended
H-bonded chains, possessing similar rotational constants, dipole moment projections, and
nuclear quadrupole coupling constants. A close look at their structures reveals that they
135
are related to each other by a rotation of the free α(OH) group in the chain structure.
Comparing the measured percent populations with calculated relative energies (Table
5.13), structure V is favored as the structure observed for Conformer D. The fact that we
see only one of these structures suggests the possibility that conformational interconversion may be happening during the collisional cooling in the expansion. We will
consider this possibility further in the Discussion section.
There are two observed conformers (B and C) with κ=-0.85, and four calculated
structures in this category to choose from (structures I, III, IV and VI).
Conformer B is
unambiguously assigned to structure I through a comparison calculated and experimental
quadrupole coupling constants and electric dipole moments. Conformer C is then assigned
to structure III over structure VI, which differ once again only in the orientation of the free
α(OH) group in the chain structure. Exactly analogous arguments are used to assign
Conformer C to structure III over structure VI, the two unassigned conformers in the κ≈0.85 category.
136
Table 5.7. Calculated rotational parameters and relative energies of the eight most stable confirmations of 2-amino 1,3-propanediol.
a
A (MHz)
B (MHz)
C (MHz)
μa (D)
μb (D)
μc (D)
μT (D)
χaa (MHz)
χbb (MHz)
χcc (MHz)
κ
ΔE(cm-1)
b
I (B)
II (A)
III (C)
IV
V(D)
VI
VII
I2(g-g-)
6083/6103
2279/2268
1997/1988
-1.7
0.3
0.9
1.9
-0.29
2.64
-2.36
-0.86
0
I3 (g- g+)
4242/4224
3134/3146
2550/2549
0.3
2.7
1.7
3.2
-2.87
1.69
1.18
-0.31
55
II2 (g- g-)
5996/6017
2273/2263
1977/1973
4.4
0.9
1.4
4.7
-4.67
2.75
1.91
-0.85
204
II2(g+ g+)
6042/6058
2258/2251
1983/1979
1.2
1.4
3.2
3.7
-0.66
2.67
-2.02
-0.86
139
II2(g- g+)
7735/7727
1977/1975
1699/1695
3.5
-1.1
1.8
4.1
-3.94
1.79
2.16
-0.91
201
II2(g- g-)
6033/6009
2235/2239
1951/1954
2.0
0.1
0.9
2.2
-4.48
2.79
1.69
-0.86
264
III2 (g- a)
4173/4239
3167/3015
2196/2137
2.1
2.2
0.6
3.1
2.15
0.17
-2.32
-0.02
400
VIII
II2(g- g+)
7642/7662
1952/1950
1686/1681
1.9
-0.5
-0.1
2.1
-3.83
1.81
2.02
-0.91
304
a
Calculated rotational parameters at the MP2/6-311++G(d,p) level of theory. Rotational constants at the B2PLYP-D3BJ/aug-cc-pVTZ
level of theory are also included after the slash (/). bCalculated relative energies (cm-1) at the B2PLYP-D3BJ/aug-cc-pVTZ level of
theory, including harmonic zero-point energy correction.
136
137
Table 5.8. Experimental rotational parameters for the assigned conformers of 2-amino
1,3-propanediol.
Conformer
A
Conformer
Conformer
Conformer
D
B
C
A (MHz)
a
4208.577(8)
6049.922(1)
5981.55(7)
7679.43(7)
B (MHz)
3130.694( 1)
2265.006( 3)
2257.073(1)
1968.92(9)
C (MHz)
2527.345( 1)
1981.188( 4)
1965.833(9)
1689.01(1)
Δ (kHz)
0.59(7)
0.33(8)
0.32(3)
—
χaa (MHz)
-2.37(2)
-0.31(3)
-3.93(2)
-3.23(9)
χbb (MHz)
1.22(3)
2.35(4)
2.28(6)
1.64(4)
χcc (MHz)
1.14(3)
-2.03(4)
1.64(6)
1.59(4)
b
N
20
15
13
9
c
5.5
15.2
4.4
3.1
-0.28
-0.86
-0.85
-0.91
a>c>b
a>c>b
σ / KHz
κ
µa:µb:µc
b>c>a
a
a>c>b
b
Errors in parentheses are expressed in units of the last digit. Number of fitted lines,
including nuclear hyperfine components. cStandard deviation of the fit.
Based on the CP-FTMW spectrum (Figure 5.2), transitions due to four conformers
of 1,3-diamino-2-propanol have been assigned. A wider variety of κ value groups is
observed in this case than in 2-amino-1,3-propanediol. Five groups are identified (κ= 0.29, -0.42, -0.73, -0.88 and -0.92) in the eight conformations with energies within 500 cm1
of the global minimum. Among the higher energy structures are two (structures V and
VIII) that incorporate only a single hydrogen bond (κ= -0.29 and -0.73).
More complex hyperfine splitting patterns are observed with the presence of the
second nitrogen atom, leading to a new set of nuclear quadrupole coupling tensors χaa, χbb,
χcc for each structure as summarized in Table 5.9. In this way, within each sub-group
determined by different κ values, the predicted
14
N quadrupole coupling constants are
different enough to make an unambiguous assignment of conformers A,B,C and D to
structures I [II3(g-,g+)], III [II2(g+,g-)], II [I2(g+,g-)] and IV [I2(g-,g-)], respectively.
138
GHz
Figure 5.2. Experimental rotational spectrum of jet-cooled 1,3-diamino-2-propanol from
7.5 to 18.5 GHz.
139
Table 5.9. Calculated rotational parameters of the eight most stable confirmations of 1,3-diamino-2-propanol.
a
A (MHz)
B (MHz)
C (MHz)
μa (D)
μb (D)
μc (D)
μT (D)
N1χaa (MHz)
N1χbb (MHz)
N1χcc (MHz)
N5χaa (MHz)
N5χbb (MHz)
N5χcc (MHz)
κ
ΔE(cm-1)
b
I(A)
II(C)
III(B)
IV(D)
II3(g- g+)
4314/4323
3019/3002
2488/2480
0.3
1.9
1.7
2.6
-1.99
-0.35
2.34
2.17
-0.52
-1.65
-0.42
0
I2(g+ g-)
8073/8071
1969/1969
1711/1707
-2.1
-2.3
0.5
3.1
2.39
-4.55
2.16
2.92
2.24
-5.16
-0.92
112
II2(g+ g-)
8062/8053
1947/1948
1695/1692
3.3
-1.7
1.5
4.0
-1.93
0.72
1.21
2.39
-4.58
2.18
-0.92
95
I2 (g- g-)
6132/6119
2239/2211
1981/1960
-3.3
-2.4
0.7
4.1
2.79
-3.98
1.19
-4.95
2.28
2.67
-0.88
239
V
S (g+ anti)
5633/5640
2279/2269
1764/1758
-0.3
-2.2
0.5
2.3
-0.04
-2.19
2.23
2.42
-4.83
-2.41
-0.73
405
VI
VII
VIII
I2(g+ g+)
6209/6228
2219/2214
1968/1965
-2.4
-2.9
-0.9
3.8
1.61
-2.22
0.61
1.89
2.44
-4.33
-0.88
405
I2(g+ g+)
6103/6135
2206/2200
1962/1959
-2.8
-1.7
1.2
3.5
1.59
-2.14
0.55
0.69
-1.08
0.39
-0.88
377
S (anti g+)
4303/4484
2889/2729
2117/2061
1.4
-2.1
0.2
2.6
1.63
-1.71
0.08
0.97
-3.35
2.38
-0.29
490
a
Calculated rotational parameters at the MP2/6-311++G(d,p) level of theory. Rotational constants at the B2PLYP-D3BJ/aug-cc-pVTZ
level of theory are also included after the slash (/). bCalculated relative energies (cm-1) at the B2PLYP-D3BJ/aug-cc-pVTZ level of
theory, including harmonic zero-point energy correction.
139
140
Table 5.10. Experimental rotational parameters for the four assigned conformers of 1,3diamino-2-propanol.
A
B
C
D
A (MHz)
4304.640(5)
a
7983.566(2)
8000.795(8)
6056.698(1)
B (MHz)
2985.145(8)
1941.489(2)
1961.541(9)
2250.505(9)
C (MHz)
2465.185(1)
1687.348(2)
1701.152(5)
1996.815(1)
Δ (kHz)
0.71(5)
1.68(1)
0.25( 2)
1.77( 5)
N1χaa (MHz)
-1.66(7)
-1.55(6)
2.14(3)
2.16(5)
N1χbb (MHz)
-0.35(7)
0.73(7)
-3.78(3)
-3.19(5)
N1χcc (MHz)
2.01(7)
0.81(7)
1.64(3)
1.03(5)
N5χaa (MHz)
1.79(7)
2.09(3)
2.57(4)
-4.37(1)
N5χbb (MHz)
-0.45(8)
-3.73(4)
1.79(4)
2.19(1)
N5χcc (MHz)
-1.34(8)
1.64(4)
-4.37(4)
2.18(1)
b
N
11
21
17
13
c
σ / KHz
10.7
15.9
9.6
14.5
κ
-0.43
-0.92
-0.92
-0.88
a:b:c
b>c>a
a>b>c
b>a>c
a
a>b>c
b
Errors in parentheses are expressed in units of the last digit. Number of fitted lines,
including nuclear hyperfine components. cStandard deviation of the fit.
For propane-1,2,3-triamine, the asymmetry parameters of the calculated structures
divide into three groups with values near κ= -0.34, -0.85 and -0.91 (Table 5.11).
Unfortunately, due to limited sample size, only a small number of the most intense
rotational transitions could be definitively assigned at the present signal level, making our
assignments somewhat more tentative. Nevertheless, transitions due to two structures in
each of the last two families are tentatively assigned.
These are confirmed and
strengthened by nuclear quadrupole splittings associated with these transitions. Detailed
comparison of the values of the 14N quadrupole coupling constants lead to the identification
of conformer A and C as structure I and IV in the κ= -0.85 family, while conformers B and
141
D are assigned to structures II and VI in the κ= -0.91 family, respectively. Interestingly,
structure II has a central double-donor NH2 group (H2N←HNH→NH2) , and is formally a
IV2(g-g+) structure in our nomenclature, forming a bifurcated chain.
The lone calculated κ= -0.34 structure (structure III) adopts a cyclic conformation
with three connected NH…N intramolecular hydrogen bonds. The dipole moment of this
structure is rather small, but is directed almost exclusively along the out-of-plane c-axis.
Given the signal-to-noise ratio on the spectrum (Figure 5.3), we anticipated being able to
observe this conformer experimentally; however, no c-type transitions were observed near
their predicted frequencies. The present signal levels leads to an upper bound of 20% (for
S/N=3) on its percent population. It is unlikely that collisional removal of the cyclic
conformer population during the cooling process was occurring, since the calculated barrier
between the cycle and the global minimum chain, II2(g+ g+), is calculated to be around
1300 cm-1, large enough to trap population behind it during the cooling process.
142
Figure 5.3. (a) Experimental rotational spectrum of jet-cooled propane-1,2,3-triamine
from 7.5 to 18.5 GHz. Calculated stick spectra due to conformer A (blue), B (green), C
(yellow) and D(red) are shown below. (b) Further expansion of 5 MHz regions around
the 211-101 transitions of conformers A with calculated nuclear quadrupolar splittings
shown below. The F’-F” labels are included in the figure.
143
Table 5.11. Calculated rotational parameters of the eight most stable confirmations of propane-1,2,3-triamine.
a
A (MHz)
B (MHz)
C (MHz)
μa (D)
μb (D)
μc (D)
μT (D)
N1χaa (MHz)
N1χbb (MHz)
N1χcc (MHz)
N3χaa (MHz)
N3χbb (MHz)
N3χcc (MHz)
N5χaa (MHz)
N5χbb (MHz)
N5χcc (MHz)
κ
ΔE(cm-1)
b
I(A)
II(B)
III
IV(C)
V
II2(g+ g+)
5805/5810
2259/2240
1969/1950
0.4
0.1
1.8
1.9
2.51
-3.65
1.15
-0.43
2.32
-1.88
0.62
-2.47
1.86
-0.85
0
IV2(g- g+)
7575/7518
1929/1924
1686/1680
0
-2.6
1.6
3.0
2.31
-4.37
2.06
1.64
2.59
-4.24
2.31
-4.37
2.06
-0.92
64
II3(g- g+)
4143/4126
3007/2988
2452/2428
0.1
-0.1
1.4
1.4
1.71
-0.18
-1.52
-3.59
2.01
1.58
-1.82
-0.28
2.10
-0.34
201
II2(g- g-)
5837/5814
2227/2219
1930/1919
2.5
-1.1
0.1
2.8
1.69
-1.63
-0.06
-3.68
2.41
1.27
2.23
1.75
-3.98
-0.85
221
I2(g- g+)
7405/7356
1945/1939
1688/1682
-1.7
-1.6
1.6
2.8
2.60
1.42
-4.02
-2.64
1.19
1.44
2.35
-4.24
1.89
-0.91
171
VI(D)
II2(g- g+)
7393/7346
1920/1915
1669/1664
-2.9
-1.4
0.5
3.3
2.35
-4.27
1.92
-2.39
1.15
1.24
-1.42
0.84
0.59
-0.91
188
VII
I2(g+ g+)
5729/5716
2218/2208
1924/1912
-2.9
-0.0
2.0
3.6
0.99
-1.21
0.22
-3.78
2.35
1.43
1.65
-1.49
-0.15
-0.85
266
a
Calculated rotational parameters at the MP2/6-311++G(d,p) level of theory. Rotational constants at the B2PLYP-D3BJ/aug-cc-pVTZ
level of theory are also included after the slash (/). bCalculated relative energies (cm-1) at the B2PLYP-D3BJ/aug-cc-pVTZ level of
theory, including harmonic zero-point energy correction.
143
144
Table 5.12. Experimental rotational parameters of the four assigned conformers of
propane 1,2,3-triamine.
A
B
C
D
a
7630.95(5)
5777.45(2)
7309.08(8)
B (MHz)
2233.75(9)
1932.55(1)
2195.76(1)
1910.06(1)
C (MHz)
1943.57(1)
1663.83(8)
1907.41(1)
1657.83(7)
N1χaa (MHz)
3.56(1)
3.61(6)
1.65(1)
2.42(6)
N1χbb (MHz)
-4.46(4)
-4.26(8)
-1.70(3)
-4.87(1)
N1χcc (MHz)
0.90(4)
1.64(8)
0.04(3)
1.459(1)
N3χaa (MHz)
-0.14(2)
1.28(3)
-3.60(9)
-2.30(6)
N3χbb (MHz)
2.17(4)
3.49(1)
2.02(7)
1.27(2)
N3χcc (MHz)
-2.02(4)
-3.20(1)
1.58(7)
1.03(2)
N5χaa (MHz)
0.15(2)
3.64(6)
2.10(1)
-1.46(6)
N5χbb (MHz)
-1.94(2)
-3.33(5)
1.37(2)
0.49(1)
N5χcc (MHz)
1.78(2)
1.31(5)
-4.53(2)
0.97(1)
N
14
16
18
9
κ
-0.85
-0.91
-0.85
-0.91
A (MHz)
b
µa:µb:µc
a
5769.44(4)
c>a>b
b>c>a
a>b>c
a>b>c
b
Errors in parentheses are expressed in units of the last digit. Number of fitted lines,
including nuclear hyperfine components.
A small number of very weak transitions remain unassigned in each spectrum. They
might come from other conformers that lie at higher energies or 13C isotopes of more stable
structures. However, due to their weak intensities, no further assignments were possible.
145
5.4
Discussion
One of the primary motivations for studying the microwave spectroscopy of the
present series of aminoalcohols and triamines is the opportunity they afford for probing in
some detail the inherent conformational preferences of alkyl chains decorated with amino
and alcohol functional groups. Building on previous results for glycerol, with 3 OH
groups,11 prototypical propyl derivatives with two OH/one NH2, one OH/two NH2, and
three NH2 groups were studied. Firm assignments were made for a total of fifteen
conformers of four molecules were made, and their relative populations determined.
5.4.1
Comparing Predicted Energies and Observed Populations
Before considering the conformational preferences of the molecules in this series,
it is important first to test the levels of theory that are best used in making comparison with
experiment. The principal experimental results from the microwave spectra are the fitted
rotational constants, dipole moment directions, and nuclear hyperfine splitting parameters
used for conformational assignments. The experimental results (rotational constants,
dipole moment components and nuclear hyperfine splittings) are compared to the different
calculations that were used. The comparison revealed a good agreement among the
MP2and B2PLYP methods with the experiment, strengthening the conclusion10 that
MP2/6-311++G(d,p) is a good level of theory for matching experimental rotational
constants with theoretical predictions.
Having made conformational assignments, the other major experimental findings of
the present work are the percentage populations of the assigned conformations, extracted
from the relative intensities of the microwave transitions. The percent populations of the
146
observed conformers are summarized in Table 5.13 for the four molecules in this series.
The table also includes a comparison of their calculated relative energies at the B2PLYPD3BJ/aug-cc-pVTZ level of theory which has proven to be most consistent among the
methods across the series compared to experiment. In table 5.14-5.17, further comparisons
of the calculated relative energies at the MP2, B3LYP-D3BJ, and M05-2X levels with the
aug-cc-pVTZ basis set can be found. In the discussion that follows, we use the B2PLYPD3BJ calculations as the principal point of comparison with experiment. While the precise
energy differences vary from one method to the next, all the dispersion-corrected
functionals perform well with this large basis set, and in general appear to be converging
toward similar relative energies.
In most cases, the percent populations correlate
remarkably well with the calculated relative energies, given the small energy differences
involved.
147
Table 5.13. The percent populations of the observed conformers and comparison of their
relative energies calculated at the B2PLYP-D3BJ/aug-cc-pVTZ level of theory.
Molecule
Conformer
Population
Population
(%)a
D-allo-threoninol
2-amino-1,3-propanediol
1,3-diamino-2-propanol
propane-1,2,3-triamine
glycerol
a
d
(%)b
∆E(B2PLYP) c
(cm-1)
II (A)/ Cycle/ II3 (g+ g-)
58
48
0
I(B)/Curved chain/II2(g- g- )
37
44
58
III (C)/ Cycle/ I3 (g+ g- )
5
8
188
II (A)/ Cycle/I3 (g- g+)
52
56
55
I (B)/ Curved chain/I2 (g- g-)
37
33
0
III (C)/ Curved chain/II2 (g- g-)
8
6
204
V(D)/ Extended chain/II2(g- g+)
3
5
201
I(A) / Cycle/II3(g- g+)
34
40
0
III(B) / Extended chain/II2(g+ g-)
31
29
95
II(C)/ Extended chain/I2(g+ g-)
21
22
112
IV(D) / Curved chain/I2 (g- g-)
14
9
239
I(A)/ Curved chain/II2(g+ g+)
59
58
0
II(B) / Extended chain/IV2(g- g+)
22
20
64
IV(C)/ Curved chain/II2(g- g-)
17
18
221
VI(D)/ Extended chain/II2(g- g+)
2
4
188
1a/Cycle/II3(g-g+)
Large
0
2b/Curved chain/II2(g+g+)
Large
57
3b/Curved chain/II2(g+g+)
Small
205
5c/Extended chain/II2(g+g-)
Small
235
7/Curved chain/I2(g+g+)
Small
394
Population percentage determined by summing the total intensities of a,b and c type
transitions and dividing by the square of the theoretically-predicted electronic dipole
moments. bPopulation percentage determined by using the scale factors obtained from the
best fit to the intensities for each conformer in rotational fitting program (JB95).
c
Calculated relative energies (cm-1) at the B2PLYP-D3BJ/aug-cc-pVTZ level of theory,
including harmonic zero-point energy correction. dTaken from ref. 11 (Ilyushin et al.).
148
Table 5.14. Energy differences and rotational constants for all conformers of D-allothreoninol within 500 cm-1 of the global minimum computed with various levels of
theory. Conformers were labeled with Roman numerals according to their energetic
ordering based on the first calculations. Additionally, experimentally observed
conformers are labeled with alphabetic letters.
Conformer
∆E / cm−1
A / GHz
B / GHz
C / GHz
MP2-6-311++g(d,p)
I(B)
0.0
3.88836
1.79922
1.49748
II (A)
43.5
3.14055
2.20020
1.92887
III ( C)
263.0
3.15428
2.17373
1.93980
IV
262.5
3.85137
1.78760
1.48811
V
340.2
3.83549
1.78714
1.49292
VI
379.7
3.49065
2.16473
1.55481
MP2-aug-cc-pVTZ
II (A)
0.0
3.13048
2.24743
1.95733
I(B)
24.7
3.90319
1.81237
1.50523
III ( C)
202.9
3.19482
2.17161
1.95682
IV
212.7
3.87408
1.80020
1.49813
VI
351.7
3.50389
2.17626
1.56314
V
366.5
3.85468
1.79950
1.50517
M052X-aug-cc-pVTZ
II (A)
0.0
3.15771
2.22718
1.95984
I(B)
44.1
3.89927
1.81472
1.50943
III ( C)
97.4
3.17283
2.19118
1.95882
VI
202.3
3.50682
2.18793
1.56869
IV
218.1
3.87024
1.80389
1.50414
V
287.5
3.85439
1.80636
1.50618
B3LYP-D3BJ-aug-cc-pVTZ
II (A)
0.0
3.11414
2.19903
1.92282
I(B)
83.9
3.86866
1.78661
1.48332
III ( C)
170.0
3.18052
2.11721
1.91820
IV
199.0
3.84163
1.77634
1.47720
V
293.8
3.82633
1.77185
1.48532
VI
443.0
3.48390
2.13619
1.53585
B2PLYP-D3BJ-aug-cc-pVTZ
II (A)
0.0
3.13048
2.20817
1.93410
I(B)
57.7
3.88336
1.79520
1.49213
III ( C)
188.0
3.18025
2.13927
1.93200
IV
202.7
3.85507
1.78446
1.48577
V
318.7
3.83634
1.78253
1.49352
VI
400.8
3.49052
2.15282
1.54640
149
Table 5.15. Energy differences and rotational constants for all conformers of 2-amino1,3-propanediol within 500 cm-1 of the global minimum computed with various levels of
theory. Conformers were labeled with Roman numerals according to their energetic
ordering based on the first calculations. Additionally, experimentally observed
conformers are labeled with alphabetic letters.
Conformer
∆E / cm−1
A / GHz
B / GHz
C / GHz
MP2-6-311++g(d,p)
I (B)
0.0
6.08298
2.27903
1.99730
II (A)
145.9
4.24216
3.13372
2.55011
III( C)
215.1
5.99610
2.27260
1.97735
IV
247.2
6.04197
2.25823
1.98323
V(D)
297.2
7.73494
1.97747
1.69921
VI
443.5
6.03269
2.23523
1.95120
VIII
493.8
7.64171
1.95178
1.68639
VII
478.5
4.17251
3.16664
2.19577
MP2-aug-cc-pVTZ
I (B)
0.0
6.12347
2.28894
2.01000
II (A)
93.9
4.25602
3.17200
2.58433
IV
176.1
6.07615
2.27035
1.99982
III( C)
245.7
6.03677
2.28273
1.99288
V(D)
281.6
7.77637
1.98681
1.70569
VI
363.2
6.03638
2.25404
1.96997
VIII
421.6
7.71514
1.96023
1.69083
VII
428.8
4.21444
3.12904
2.18977
M052X-aug-cc-pVTZ
I (B)
0.0
6.10850
2.29667
2.01066
II (A)
25.8
4.25047
3.17705
2.57650
IV
174.3
6.06456
2.28200
2.00307
III( C)
235.0
6.04141
2.28658
1.99173
V(D)
311.4
7.79404
1.99173
1.70985
VI
336.5
6.01837
2.26713
1.97443
VIII
400.7
7.71090
1.96999
1.69948
VII
428.2
4.20142
3.18765
2.21251
150
Table 5.15, continued
Conformer
∆E / cm−1
A / GHz
B / GHz
C / GHz
B3LYP-D3BJ-aug-cc-pVTZ
I (B)
0.0
6.09473
2.25321
1.97461
II (A)
35.1
4.21242
3.12945
2.53139
IV
109.1
5.98926
2.22906
1.94415
V(D)
147.8
7.69746
1.96720
1.68735
III( C)
182.8
6.00522
2.24922
1.95965
VI
199.4
5.98926
2.22906
1.94415
VIII
226.6
7.63169
1.94303
1.67435
VII
379.7
4.26784
2.94439
2.10730
B2PLYP-D3BJ-aug-cc-pVTZ
I (B)
0.0
6.10311
2.26805
1.98813
II (A)
55.0
4.22355
3.14647
2.54895
IV
138.8
6.05825
2.25051
1.97877
V(D)
200.6
7.72703
1.97527
1.69499
III( C)
203.8
6.01743
2.26292
1.97262
VI
263.6
6.00909
2.23873
1.95369
VIII
303.8
7.66219
1.95025
1.68132
VII
400.4
4.23856
3.01511
2.13706
151
Table 5.16. Energy differences and rotational constants for all conformers of 1,3diamino-2-propanol within 500 cm-1 of the global minimum computed with various
levels of theory. Conformers were labeled with Roman numerals according to their
energetic ordering based on the first calculations. Additionally, experimentally observed
conformers are labeled with alphabetic letters.
Conformer
∆E / cm−1
A / GHz
B / GHz
C / GHz
MP2-6-311++g(d,p)
I(A)
0.0
4.31375
3.01907
2.48805
II( C)
114.6
8.07308
1.96924
1.71065
III(B)
146.4
8.06247
1.94711
1.69465
IV(D)
322.7
6.13189
2.20959
1.96084
V
408.6
5.63269
2.27915
1.76380
VI
455.3
6.20930
2.21917
1.96768
VII
466.1
6.10274
2.20599
1.96226
VIII
484.1
4.30274
2.88872
2.11715
MP2-aug-cc-pVTZ
I(A)
0.0
4.34555
3.03365
2.51266
II( C)
155.0
8.10768
1.98459
1.72075
III(B)
187.5
8.08790
1.96147
1.70451
IV(D)
340.8
6.14088
2.22841
1.97769
VI
447.0
6.23530
2.23673
1.98641
VII
461.4
6.13825
2.22189
1.97999
V
473.1
5.66482
2.28661
1.77219
VIII
519.7
4.43637
2.80611
2.09990
M052X-aug-cc-pVTZ
I(A)
0.0
4.36297
3.01952
2.50602
III(B)
4.9
8.08115
1.96594
1.70870
II( C)
28.4
8.10704
1.98583
1.72204
IV(D)
179.2
6.14088
2.23596
1.98055
VII
246.9
6.13667
2.22917
1.98057
VI
260.1
6.22538
2.24230
1.98570
V
323.0
5.62952
2.29108
1.77328
VIII
405.3
4.34098
2.88698
2.12130
152
Table 5.16, continued
Conformer
∆E / cm−1
A / GHz
B / GHz
C / GHz
B3LYP-D3BJ-aug-cc-pVTZ
I(A)
0.0
4.3107
2.9829
2.4617
III(B)
57.7
8.0302
1.9379
1.6839
II(C)
105.1
8.0457
1.9589
1.6981
IV(D)
195.3
6.0996
2.2003
1.9495
VII
341.3
6.13895
2.18377
1.94457
VI
391.8
6.22823
2.19910
1.95087
V
393.2
5.63341
2.25605
1.74985
VIII
497.3
4.51651
2.68321
2.03890
B2PLYP-D3BJ-aug-cc-pVTZ
I(A)
0.0
4.32274
3.00200
2.47960
III(B)
94.7
8.05287
1.94771
1.69246
II( C)
111.8
8.07061
1.96925
1.70731
IV(D)
239.0
6.11853
2.21106
1.96011
VII
376.9
6.13503
2.19986
1.95862
V
405.2
5.64037
2.26888
1.75875
VI
405.3
6.22804
2.21455
1.96472
VIII
490.7
4.48380
2.72868
2.06130
153
Table 5.17. Energy differences and rotational constants for all conformers of propane1,2,3-triamine within 500 cm-1 of the global minimum computed with various levels of
theory. Conformers were labeled with Roman numerals according to their energetic
ordering based on the first calculations. Additionally, experimentally observed
conformers are labeled with alphabetic letters.
Conformer
∆E / cm−1
A / GHz
B / GHz
C / GHz
MP2-6-311++g(d,p)
I(A)
0.0
5.79170
2.24902
1.95741
II(B)
150.6
7.52702
1.92394
1.68285
III
229.4
4.12808
2.99181
2.43798
IV( C)
292.5
5.79427
2.22545
1.92319
V
301.9
7.35188
1.93933
1.68479
VI(D)
359.9
7.35120
1.91382
1.66482
VII
392.0
5.68555
2.21583
1.91633
MP2-aug-cc-pVTZ
I(A)
0.0
5.82891
2.26248
1.97169
B
117.5
7.56353
1.93728
1.69298
III
228.0
4.15558
3.01160
2.46031
V
250.7
7.39826
1.95187
1.69437
IV( C)
259.6
5.82247
2.24294
1.94043
VI(D)
309.5
7.39243
1.92683
1.67494
VII
350.7
5.72527
2.23083
1.93285
M052X-aug-cc-pVTZ
I(A)
0.0
5.81937
2.26453
1.96699
B
121.3
7.56190
1.93842
1.69330
V
174.2
7.41891
1.95441
1.69633
VI(D)
182.3
7.39480
1.93225
1.68004
IV( C)
224.0
5.80711
2.24783
1.93687
VII
294.6
5.71706
2.23791
1.93153
III
296.2
4.16249
2.99959
2.44627
154
Table 5.17, continued
Conformer
∆E / cm−1
A / GHz
B / GHz
C / GHz
B3LYP-D3BJ-aug-cc-pVTZ
I(A)
0.0
5.79886
2.22612
1.93680
II(B)
40.8
7.48631
1.91513
1.67202
VI(D)
117.3
7.31951
1.90662
1.65609
V
123.1
7.33143
1.93015
1.67369
III
180.8
4.10953
2.97474
2.41065
IV( C)
195.6
5.80688
2.20450
1.90598
VII
216.8
5.70877
2.19327
1.89859
B2PLYP-D3BJ-aug-cc-pVTZ
I(A)
0.0
5.80999
2.24020
1.94956
II(B)
64.1
7.51778
1.92412
1.68034
V
171.0
7.35633
1.93881
1.68193
VI(D)
188.2
7.34598
1.91496
1.66385
III
201.4
4.12599
2.98800
2.42751
IV( C)
220.7
5.81368
2.21932
1.91899
VII
266.2
5.71562
2.20812
1.91168
To cite some examples, in D-allo-threoninol, 1,3-diamino-2-propanol, and propane1,2,3-triamine, the measured percent populations in the supersonic expansion follow the
calculated energy ordering and even the magnitudes of the energy differences in most cases
(Table 5.13). In 2-amino-1,3-propanediol, the lowest energy cyclic [I3(g-g+)] and curvedchain [I2(g-g-)] structures carry most of the population, also as predicted by theory. Even
in glycerol, where previous studies had not extracted relative populations quantitatively,
the qualitative population sizes match the calculated energies quite well. At the same time,
155
the correlation between experimental populations and conformer energies is not perfect.
For instance, in 2-amino-1,3-propanediol, B2PLYP calculations predict that the curved
chain conformer B [I2(g-g-)] is the global minimum, with the cyclic conformer A [I3(g-g+)]
slightly higher in energy, by 55 cm-1. However, the experimental percent population of A
(56%) is almost twice that of B (33%), pointing to the cycle as the global minimum.
More importantly, some structures predicted by calculation to have low energies are
"missing" in the expansion. For instance, in D-allo-threoninol, structure IV in Table 5.5
(II2(g-g-) configuration) was not observed experimentally although its energy is nearly
isoenergetic with structure III [I3(g+g-)], assigned to conformer C, which carries 8% of the
observed population. Structural relaxation28 into a lower energy conformation through
interconversion over a low energy barrier is suggested. This process can occur in the early
stages of supersonic expansion as a result of collisions with buffer gas. Figure 5.4(a) shows
the interconversion barrier between structure IV and structure I listed in Table 5.5 for Dallo-threoninol. Both these structures belong to the II2(g- g-) structural family and differ
from each other by a rotation of the free OH group in the curved-chain structure, effectively
switching which lone pair on the oxygen atom is used as H-bond acceptor site. Since a low
energy interconversion barrier of 275 cm−1 is predicted, the higher energy conformer is
able to relax into the lower energy minimum during the collisional cooling in the expansion.
This process will increase the population of structure I (conformer B), and decrease the
population of structure IV, which is not found in the expansion.
In 2-amino-1,3-propanediol, a similar low-barrier pathway exists between structure
IV and structure I (322 cm−1 barrier, Figure 5.4(b), or structure VIII and structure V (324
cm−1 barrier), explaining the inability to detect population in either structure IV or VIII
156
experimentally. Finally, in propane-1,2,3-triamine (Table 5.11), the cyclic structure III
[II3(g-,g+)] is not observed experimentally, despite being lower in energy than two others
which are observed (structures II and IV). The fact that the cycle is not observed is due in
part to its lower dipole moment, which points almost exclusively along the c-axis; however,
given the small energy difference between structures III and I, a low energy pathway may
be implicated as well.
Taken as a whole, these results provide further evidence that the relative populations
of conformers observed in the expansion are determined in part by their relative stabilities,
but also to some extent by the sizes of the isomerization barriers between them.29 Where
differences occur in the ordering of populations by energy, a low-energy isomerization
pathway is likely responsible for the observed discrepancy, draining population from a
higher-energy minimum into a lower-energy one during the collisional cooling process.
Indeed, Ruoff et al.29 have deduced based on small molecule cooling in expansions that
collisional removal of a higher-energy conformer is possible when the barrier to
isomerization is less than 400 cm-1, consistent with our deductions.
157
Figure 5.4. (a) Interconversion barrier between structure IV to structure I of D-allothreoninol at the B3PLYP-D3BJ/aug-cc-pVTZ level of theory. (b) Interconversion barrier
between structure IV to structure I for 2-amino-1,3-propanediol at the B3PLYPD3BJ/aug-cc-pVTZ level of theory.
158
Figure 5.5. Calculated energy level diagrams for glycerol, 2-amino-1,3-propanediol, 1,3diamino-2-propanol and propane-1,2,3-triamine through B2PLYP-D3BJ/aug-cc-pVTZ
level of theory.
5.4.2
The Preference for Cycles Versus Chains as a Function of NH2/OH Content
The small amino alcohol molecules studied in this work all possess three adjacent
H-bonding substituents as a combination of NH2 and OH groups along the carbon
framework. As a result, one similarity in their energy landscapes is that both cyclic and
chain HB patterns are present, and in close energetic proximity. The hydrogen-bonded
chains incorporate two H-bonds linking adjacent substituents (α→β→γ or γ→β→α) along
the carbon backbone. The cycles have three hydrogen bonds, closing the cycle by forming
a hydrogen bond between the α and γ carbons.
Figure 5.5 shows a set of energy level diagrams for 2-amino-1,3-propanediol (2
OH/1 NH2), 1,3-diamino-2-propanol (1 OH/2 NH2) and propane-1,2,3-triamine (3 NH2),
calculated at the B2PLYP-D3BJ/aug-cc-pVTZ level of theory that best matches with
experiment. As a point of comparison, the relative energies of different structures of
glycerol are also calculated at the same level of theory, and are included in the figure. In
all four members of the series, the lowest energy cyclic and chain structures are within
159
around 200 cm−1 of one another. 2-Amino-1,3-propanediol has the smallest gap, with chain
more stable than the cycle (∆E = Ecycle – Echain = 55 cm−1), while the largest gap is found in
propane-1,2,3-triamine, with chain preferred over cycle again (∆E = 201 cm−1).
The calculated structures for the full set of observed conformers in the series 2amino-1,3-propanediol, 1,3-diamino-2-propanol, propane-1,2,3-triamine and glycerol are
shown in Figures 5.6-5.9, respectively. The structures are divided into the three major Hbond structural types: cycles, curved chains, and extended chains. The observed structures
of 2-amino-1,3-propanediol are prototypical (Figure 5.6), with one cycle, two curved
chains and one extended chain detected in the expansion. As Table 5.18 summarizes, the
OCCC and CCCO dihedral angles (e.g., G, T) configure the terminal heavy atoms in
positions where they can form cyclic, curved chain or extended chain structures. These
dihedrals are both Gauche (but of opposite sign) in the cycle, the curved chain has one
Gauche and one Trans, and the extended chain has two Trans.
160
Figure 5.6. Calculated structures for the full set of observed conformers of 2-amino-1,3propanediol with calculated HB distances (in Å) at the B2PLYP-D3BJ/aug-cc-pVTZ
level of theory. Assigned structural types and their relative energies calculated at the
same level of theory are included. The global minimum is shaded in red, while the
lowest energy local minimum is shaded in blue.
Figure 5.7. Calculated structures for the full set of observed conformers of 1,3-diamino2-propanol with calculated HB distances (in Å) using B2PLYP-D3BJ/aug-cc-pVTZ level
of theory. Assigned structural types and their relative energies calculated at the same
level of theory are included. The global minimum is shaded in red, while the lowest
energy local minimum is shaded in blue.
161
Figure 5.8. Calculated structures for the full set of observed conformers of propane-1,2,3triamine with calculated HB distances (in Å) using B2PLYP-D3BJ/aug-cc-pVTZ level of
theory. Assigned structural types and their relative energies calculated at the same level
of theory are included. The global minimum is shaded in red, while the lowest energy
local minimum, a bifurcated extended chain, is shaded in blue. Note that the cyclic
structure was not observed experimentally (see text for further discussion).
Figure 5.9. The calculated structures for the full set of observed conformers of glycerol
with calculated HB distances (in Å) using B2PLYP-D3BJ/aug-cc-pVTZ level of theory.
Assigned structural types from ref. 11 and their relative energies calculated at the same
level of theory are included. The global minimum is shaded in red, while the lowest
energy local minimum is shaded in blue.
162
Table 5.18. Summary of the sets of dihedral angles associated with each of the
prototypical H-bonded structural types. (X, Y = OH or NH2).
Structure type
Cycle
Curved chain
Extended chain
XC(α)C(β)Y
gg+
gg+
g-
Dihedral angles
XCCC
XC(β)C(γ)Y
g+
G+
g+
T
gG
gT
g+
T
CCCX
GG
T
T
T
In order to form H-bonds, adjacent OH/NH2 functional groups must be gauche with
respect to one another, with XCCY dihedrals of approximately ±60o, as summarized in
Table 5.18. The H-bonded cycles have XCCY dihedral angles that change sign along the
carbon framework [(g+, g-) or (g-, g+)]. When combined with two Gauche XCCC and
CCCY dihedrals, the three OH/NH2 groups are positioned on the same side of the plane of
the carbon framework, where three H-bonds can be formed. Interestingly, this combination
of dihedrals built off the triangular carbon framework produces nearest-neighbor
(α−β, β−γ) and next-nearest-neighbor (γ−α) X…Y heavy-atom distances in the cyclic
structures that are nearly equal (2.8 ±0.1 Å, Table 5.19), so that an H-bond that closes the
cycle could be similar in strength to a nearest-neighbor H-bond in a chain.
The curved chains combine XCCY dihedrals of the same sign [(g+,g+) or (g-,g-)]
with one gauche XCCC and one trans CCCY dihedral, thereby directing the H-bonded
chain from above the plane of the carbon framework to in-plane, or vice versa. This trans
configuration relieves strain along the carbon backbone relative to the cycle. The extended
chains have both XCCC/CCCY dihedrals in the trans configuration, minimizing steric
strain in the carbon framework at both ends. XCCY dihedrals of opposite sign, (g-, g+) or
(g+,g-), produces two H-bonds of medium strength. In propane-1,2,3-triamine, the lowest
163
energy extended chain structure is a bifurcated chain, with the central NH2 group acting as
a double donor to the two terminal amino groups.
Table 5.19 compares the key structural parameters of the lowest energy chain and
cyclic structures for each molecule in the series of glycerol, 2-amino-1,3-propanediol, 1,3diamino-2-propanol, and propane-1,2,3-triamine. These structures are highlighted in red
(global minimum) and blue (second lowest) in Figures 5.6-5.9. Anticipating that a short,
near-linear H-bond is best, we looked for a correlation between the XH…Y H-bond distance
and bond angle (180o = linear). Indeed, this correlation is evident in Figure 5.10, which
plots the H-bond distance versus angle of the lowest-energy cyclic and chain structures of
each molecule, grouped by H-bond type.
The OH group is the better H-bond donor, forming all seven of the shortest H-bonds
(r(OH…Y) < 2.25 Å), while eight of the ten longest have the poorer NH as H-bond donor
(r(NH…Y) > 2.30 Å). The anticipation that NH2 groups would be better acceptors than OH
groups is not clearly borne out by the present data, in part because the range of H-bond
distances of a given type is substantial. Thus, along an alkyl chain, nearest-neighbor (α−β
or β−γ) and next-nearest-neighbor (α−γ) H-bonds follow the order: r(OH…N)≈r(OH…O)<
r(NH…N)≈ r(NH…O).
164
Table 5.19. Summary of calculated XH…Y HB distances (in Å) and bond angles (in degrees) of the lowest energy chain and cyclic
structures for each molecule in the series glycerol, 2-amino-1,3-propanediol, 1,3-diamino-2-propanol, and propane-1,2,3-triamine at
the B2PLYP-D3BJ/aug-cc-pVTZ level of theory.
Structural
Cyclic
Chain
type
Glycerol
OH…O
2-amino
Diamino Triamino Glycerol
2.11/134 2.10/136
2.33/108
2.19/117
2.21/112
2-amino
Diamino Triamino
2.62/97
OH…N
2.30/109 2.03/123
2.23/115 2.17/117
NH…O
2.40/106
2.33/107 2.53/102
NH…N
2.63/96
2.36/118 2.29/113
2.48/104
2.30/123
2.38/109
2.54/99
164
165
Figure 5.10. Calculated H-bond distance (Å) versus XH…Y bond angles (degrees), at the
B2PLYP-D3BJ/aug-cc-pVTZ level of theory, grouped by H-bond types.
While the cycle seems preferable over the chain by virtue of its extra H-bond, the
triangular structure leads to H-bonds that are quite strained, and of unequal strength. As
Figure 5.10 shows, the H-bonded chains (red symbols) are intermediate in length, pitting
two medium-strength H-bonds against what is often a combination of strong and weak Hbonds in the cycle.10
Interestingly, at the B2PLYP-D3BJ/aug-cc-pVTZ level of theory, H-bonded cycles
are preferred in glycerol and 1,3-diamino-2-propanol, while a curved-chain structure is
lowest in propane-1,2,3-triamine. In 2-amino-1,3-propanediol, the cycle and lowestenergy curved chain are close in energy, with calculation predicting the curved-chain lower,
but experimental populations pointing towards the reverse. As the preceding discussion
has amply illustrated, several counter-balancing factors contribute to the relative stability
166
of cycle and chain in any one molecule, leaving no simple explanation of these trends. It
is noteworthy, however, that the extended chains are quite high in energy in the molecules
with two terminal OH groups (glycerol and 2-amino-1,3-propanediol), but drop in energy
in the diamino and triamino analogs with one or more NH2 group on a terminal carbon.
5.4.3
Effect of the Methyl Group on Structural Preferences
Figure 5.11 illustrates the effect of lengthening the carbon framework by addition of
a methyl group at one end of 2-amino-1,3-propanediol. Depending on the chirality of the
newly-formed chiral center, either D-threoninol studied in previous work,10 or D-allothreoninol studied here, is formed. The preference for cycles over chains is retained in
both molecules.
This methyl group breaks the original symmetry of 2-amino-1,3-
propanediol, thus generating two unique conformers based on each parent conformation of
2-amino-1,3-propanediol, as illustrated in Figure 5.11.
167
Figure 5.11. Structural evolution from 2-amino-1,3-propanediol (center) to D-threoninol
(left) and D-allo-threoninol (right). Zero-point corrected relative energies (in cm-1)
calculated at the B2PLYP-D3/aug-cc-pVTZ level of theory are included. Mirror image
pairs in 2-amino-1,3-propanediol are included in the boxes.
The relative energies of the conformers in D-threoninol and D-allo-threoninol are
reasonably close to those in 2-amino-1,3-propanediol, indicating that the addition of the
methyl group does not change the energy landscape appreciably. More striking is the fact
that there are significantly fewer conformers of D-allo-threoninol detected in the expansion
compared to D-threoninol. This is in keeping with the energy level diagram calculated for
this pair of disastereomers, as presented in Figure 5.12. As described in previous work,10
fifteen conformations of D-threoninol were predicted by calculation to reside within 500
cm-1 of the global minimum at the MP2/6-311++G(d,p) level of theory, while only six
168
conformations were found for its allo form in the same range. As a result, seven lowest
energy conformers ( 2 cycles and 5 chains ) are assigned for D-threoninol in the densely
populated energy level diagram while only three ( 2 cycles and 1 chain ) are found for Dallo-threoninol. Calculations at the B2PLYP-D3J aug-cc-pVTZ confirm this conclusion
(Figure 5.12).
The lowest cyclic and chain structures of D-threoninol and D-allo-threoninol are
shown in the figure, together with Newman projections along the Cβ-Cγ bond. When
viewed along the Cβ-Cγchemical bond, the CH2OH group attached to the Cβ atom is in an
anti configuration relative to the CH3 group on the back carbon in D-threoninol for both
cyclic and chain structures. However, in D-allo-threoninol, with its opposite chirality on
the C γ atom, a gauche conformation is adopted in both cases. We infer that D-allothreoninol experiences a larger steric hindrance brought on by the different chiral center,
which raises the energy of the structures, and leads to a more sparsely populated energy
level diagram, with fewer conformers with measurable population, as observed.
169
Figure 5.12. Energy level diagrams for all conformational minima of D-threoninol and Dallo-threoninol within 500 cm−1 of the global minimum, calculated at the B2PLYPD3/aug-cc-pVTZ level of theory. Conformers that are assigned in the expansion are
shown in red. Structures of the lowest energy chain and cyclic conformers are also
plotted with their Newman projections.
170
5.5
Conclusions
The conformational preferences of a prototypical set of tri-substituted
aminoalcohols (2-amino-1,3-propanediol, 1,3-diamino-2-propanol, and D-allo-threoninol),
and the triamine analog 1,2,3-triaminopropane have been explored using a combination of
broadband microwave spectroscopy and theoretical calculations. Rotational constants,
dipole moment directions, and nuclear quadrupolar splittings are used to make firm
assignments of a total of 15 conformations of the four molecules. By placing NH2 and OH
functional groups on adjacent carbons along a propyl chain, hydrogen-bonded networks
can be formed. The low-energy structures show a remarkable variety of H-bonding
architectures, including cycles, curved chains, extended chains, and bifurcated chains.
These architectures are in close energetic proximity, with subtle differences between
molecules depending on the NH2/OH make-up and steric effects. Extending the sequence
of NH2/OH groups to four or more promises an even richer variety of possibilities worth
exploring.
171
5.6
References
1.
Soai, K.; Niwa, S. Enantioselective Addition of Organozinc Reagents to Aldehydes.
Chem. Rev. 1992, 92, 833–856.
2.
Chennakesava Rao, K.; Arun, Y.; Easwaramoorthi, K.; Balachandran, C.; Prakasam,
T.; Eswara Yuvaraj, T.; Perumal, P. T. Synthesis, Antimicrobial and Molecular
Docking Studies of Enantiomerically Pure N-alkylated beta-amino Alcohols from
Phenylpropanolamines. Bioorg. Med. Chem. Lett. 2014, 24, 3057-63.
3.
Hazra, B. G.; Pore, V. S.; Dey, S. K.; Datta, S.; Darokar, M. P.; Saikia, D.; Khanuja,
S. P. S.; Thakur, A. P. Bile Acid Amides Derived from Chiral Amino Alcohols:
Novel Antimicrobials and Antifungals. Bioorg. Med. Chem. Lett. 2004, 14, 773777.
4.
Mavrov, M. V.; Simirskaya, N. I. New Synthesis of Ethambutol and Related
alpha,beta-acetylenic Amino Alcohols. Pharm. Chem. J. 2013, 46, 730-735.
5.
Zwier, T. S. Laser Spectroscopy of Jet-cooled Biomolecules and Their Watercontaining Clusters: Water Bridges and Molecular Conformation. J. Phys. Chem. A
2001, 105, 8827-8839.
6.
Liu, K.; Brown, M. G.; Saykally, R. J. Terahertz Laser Vibration-rotation Tunneling
Spectroscopy and Dipole Moment of a Cage Form of the Water Hexamer. J. Phys.
Chem. A 1997, 101, 8995-9010.
7.
Chin, W.; Piuzzi, F.; Dimicoli, I.; Mons, M. Probing the Competition Between
Secondary Structures and Local Preferences in Gas Phase Isolated Peptide
Backbones. Phys. Chem. Chem. Phys. 2006, 8, 1033-1048.
8.
Kusaka, R.; Zhang, D.; Walsh, P. S.; Gord, J. R.; Fisher, B. F.; Gellman, S. H.; Zwier,
T. S. Role of Ring-Constrained gamma-Amino Acid Residues in alpha/gammaPeptide Folding: Single-Conformation UV and IR Spectroscopy. J. Phys. Chem. A
2013, 117, 10847-10862.
9.
Brown, G. G.; Dian, B. C.; Douglass, K. O.; Geyer, S. M.; Shipman, S. T.; Pate, B.
H. A Broadband Fourier Transform Microwave Spectrometer Based on Chirped
Pulse Excitation. Rev. Sci. Instrum. 2008, 79, 053103.
10.
Vaquero-Vara, V.; Zhang, D.; Dian, B. C.; Pratt, D. W.; Zwier, T. S.
Delicate Balance of Hydrogen Bonding Forces in d-Threoninol. J. Phys. Chem. A
2014, 118, 7267-7273.
11.
Ilyushin, V. V.; Motiyenko, R. A.; Lovas, F. J.; Plusquellic, D. F. Microwave
Spectrum of Glycerol: Observation of a Tunneling Chiral Isomer. J. Mol. Spectrosc.
2008, 251, 129-137.
172
12.
Carcabal, P.; Jockusch, R. A.; Hunig, I.; Snoek, L. C.; Kroemer, R. T.; Davis, B. G.;
Gamblin, D. P.; Compagnon, I.; Oomens, J.; Simons, J. P. Hydrogen Bonding and
Cooperativity in Isolated and Hydrated Sugars: Mannose, Galactose, Glucose, and
Lactose. J. Am. Chem. Soc. 2005, 127, 11414-11425.
13.
Mohamadi, F.; Richards, N. G. J.; Guida, W. C.; Liskamp, R.; Lipton, M.; Caufield,
C.; Chang, G.; Hendrickson, T.; Still, W. C. Macromodel-an Integrated Software
System for Modeling Organic and Bioorganic Molecules Using Molecular
Mechanics. J. Comput. Chem. 1990, 11, 440-67.
14.
Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.;
Cheeseman, J. R.; Scalmani, G.; Barone, V.; Mennucci, B.; Petersson, G. A., ; et
al. Gaussian 09, revision C.01; Gaussian, Inc.: Wallingford, CT, 2010.
15.
Alonso, J. L.; Perez, C.; Sanz, M. E.; Lopez, J. C.; Blanco, S. Seven Conformers of
l-Threonine in the Gas Phase: a LA-MB-FTMW Study. Phys. Chem. Chem. Phys.
2009, 11, 617-627.
16.
Grimme, S. Accurate Description of van der Waals Complexes by Density
Functional Theory Including Empirical Corrections. J. Comput. Chem. 2004, 25,
1463-1473.
17.
Becke, A. D. Density�functional Thermochemistry. III. The Role of Exact
Exchange. J. Chem. Phys. 1993, 98, 5648–5652.
18.
Zhao, Y.; Truhlar, D. G. Density Functionals for Noncovalent Interaction Energies
of Biological Importance. J. Chem. Theory Comput. 2007, 3, 289-300.
19.
Grimme, S. Semiempirical Hybrid Density Functional with Perturbative Secondorder Correlation. J. Chem. Phys. 2006, 124, 034108.
20.
Grimme, S.; Ehrlich, S.; Goerigk, L. Effect of the Damping Function in Dispersion
Corrected Density Functional Theory. J. Comput. Chem. 2011, 32, 1456-1465.
21.
Bocklitz, S.; Suhm, M. A. Constraining the Conformational Landscape of a
Polyether Building Block by Raman Jet Spectroscopy. Z. Phys. Chem. 2015, 229,
1625-1648.
22.
Neill, J. L.; Douglass, K. O.; Pate, B. H.; Pratt, D. W. Next Generation Techniques
in the High Resolution Spectroscopy of Biologically Relevant Molecules. Phys.
Chem. Chem. Phys. 2011, 13, 7253-7262.
23.
Watson, J. K. G., Vibrational Spectra and Structure; Durig, J.R., Ed.; Elsevier: New
York/Amsterdam, The Netherlands, 1977; Vol.6, pp 1-89.
173
24.
Plusquellic, D. F., JB95 Spectral Fitting Program, NIST, Gaithersburg, MD.
http://physics.nist.gov/jb95.
25.
Pickett, H. M., SPFIT/SPCAT. http://spec/jpl.nasa.gov.
26.
Ray, B. S., Über die Eigenwerte des Asymmetrischen Kreisels. Z. Phys. 1932, 78,
74-91.
27.
Blanco, S.; Sanz, M. E.; Lopez, J. C.; Alonso, J. L. Revealing the Multiple
Structures of Serine. Proc. Natl. Acad. Sci. U.S.A. 2007, 104, 20183-20188.
28.
Ruoff, R. S.; Klots, T. D.; Emilsson, T.; Gutowsky, H. S. Relaxation of Conformers
and Isomers in Seeded Supersonic Jets of Inert Gases. J. Chem. Phys. 1990, 93,
3142-50.
29.
Florio, G. M.; Christie, R. A.; Jordan, K. D.; Zwier, T. S. Conformational
Preferences of Jet-cooled Melatonin: Probing trans- and cis-amide Regions of the
Potential Energy Surface. J. Am. Chem. Soc. 2002, 124, 10236-10247.
174
CHAPTER 6.
ROOM TEMPERATURE CHIRPED-PULSE
FOURIER TRANSFORM MICROWAVE SPECTROSCOPY OF
ISOBUTANOL
6.1
Introduction
Rotational spectroscopy provides structural information for gas phase molecules and
is therefore a powerful tool for chemical sensing1 and structural determination of different
conformers.2-3 Since the development of the Balle-Flygare type spectrometer,4-6 it has been
routine for spectroscopists to perform rotational free induction decay (FID) detection for
molecules cooled in supersonic jet expansion. Thus, the sample molecules are rapidly
cooled to the vibrational ground state. However, certain kinds of studies are incompatible
with incorporation of the molecules in a supersonic expansion, and the spectra for
vibrationally excited states are also of research interest in their own right. For instance,
they provide information regarding local temperatures in the interstellar medium7-8 and
also play an integral part in kinetics studies under thermal conditions.9-10
Pioneered by Pate and co-workers11 from the University of Virginia in recent years,
the development of broadband chirped pulse Fourier transform microwave (CP-FTMW)
spectrometer has provided the ability to record a rotational spectrum measuring a
broadband region in single shot data acquisition. Compared to the Balle-Flygare type
cavity-FTMW spectrometer, the CP-FTMW design displays reasonable sensitivity while
greatly reducing the measurement time.
This chapter introduces the design and operation of a waveguide based room
temperature chirped pulse Fourier transform microwave spectrometer (RT-CP-FTMW).
175
To increase the signal to noise ratio of the spectrum, an ultrafast digitizer (Guzik ADC
6131) is adopted to perform up to 1 billion shot averages of the sample molecule. At room
temperature, the sample molecules are populated according to Boltzmann distribution in
the waveguide. Aside from the vibrational ground state, vibrationally excited states are also
present in the spectrum. We choose isobutanol (2-methylpropan-1-ol) as a test molecule,
which is a colorless, flammable liquid. With a high vapor pressure (9 mmHg at 20°C), it
did not require additional heating in the experiment. Its expansion-cooled spectrum has
been studied previously by Hirota and co-workers12 using Balle-Flygare type cavityFTMW spectrometer. Four conformers are assigned in the supersonic jet expansion. In the
current work, the molecular sample vapor is introduced into the waveguide without jet
cooling. As a result, many more transitions belonging to vibrational ground state are
observed and assigned. Most of these new transitions are Q type transitions from high
rotational quantum states, which enable us to make a more accurate assignment for the
centrifugal distortion constants.
Besides, transitions possibly corresponding to
vibrationally excited states of the lowest frequency torsional modes are also observed in
the spectrum.
176
6.2
Experimental and Computational Methods
The experimental method used in the present study has been described in detail in
Chapter 2 and is only briefly described here. The WRD 750 waveguide is chosen as a room
temperature cell based on the recent work by Shipman and co-workers.13-14 The isobutanol
sample was purchased from Sigma-Aldrich and subjected to 3 freeze-pump-thaw cycles
before being introduced into the 10 m coiled waveguide. The pressure inside the waveguide
was maintained at around 10 mTorr through a delicate balance between the sample inlet
valve (a needle valve and a stem valve) and a three stage pumping system (two diffusion
pumps backed by a roots blower and a roughing pump). To collect the microwave spectra
of isobutanol from 7.5-18.5 GHz, a 1 μs chirped pulse from 1.875-4.625 GHz was
generated by an arbitrary waveform generator (AWG 7101, Tektronix). Full bandwidth
was achieved by passing through a pre-amplifier (Minicircuits ZX60-6013E-S) and a
quadrupler (Phase One PS06-0161). Then the pulse was amplified by a 200 W traveling
wavetube amplifier (TWTA, Amplifier Research 200T8G18A). To protect the electronics
at the receiver end, the TWTA was limited to 5% gain. The collected free induction decay
(FID) signal was subsequently amplified by a low noise amplifier (LNA, Miteq AMF-6F06001800-15-10P) and down converted to 0.4-11.4 GHz by mixing with the output from a
18.9 GHz phase-locked dielectric resonator oscillator (PLDRO, Microwave Dynamics
PLO-2000-18.90). The FID was collected for 4 μs and averaged by an ultrafast 13 GHz
digitizer (Guzik ADC6131) for 1 billion shots, which took about 3 hours. Fourier
transform was applied to the time domain FID signal to yield frequency domain spectrum.
It is noteworthy to mention that the dephasing time in the room temperature waveguide is
much shorter than in the supersonic expansion chamber, probably due to high collision
177
rates with other molecules and waveguide walls. As a result, the intensities of transitions
at the edge of the chirp are affected to some degree by the short dephasing time. To
overcome this problem, FIDs from both up chirping (7.5-18.5GHz) and down chirping
(18.5-7.5GHz) pulses15 were collected to make sure all the information were obtained in
the broadband region.
To identify possible structures of isobutanol detected in the room temperature cell,
force field calculation using Amber* was first performed with Schrödinger MacroModel16
commercial program suite (10,000 iterations, 0.0001 convergence threshold). The output
structural minima were then subjected to full geometry optimizations using density
function theory (DFT) calculations through Gaussian 0917 suite of program. These
calculations involved tight geometry optimizations followed by harmonic and anharmonic
vibrational frequency calculations using B3LYP with 6-311++g(d,p) basis set. To account
for the London dispersion energy, correction from Grimme and coworkers18-19 with BeckeJohnson damping (D3BJ) was added to the B3LYP functional. To better predict the
information rich spectra, the calculated rotational constants for the vibrationally excited
states are scaled using the same factor determined from the ground state assignments.
178
Figure 6.1. The five most stable conformers of isobutanol calculated at B3LYP-D3BJ/6311++G(d,p) level of theory, along with their zero point corrected relative energies in cm1
.
Five structures with low energies are predicted in the conformational search, which
are summarized in Figure 6.1. To differentiate these conformers, a nomenclature based on
2 dihedral angles is used here. The H7-C3-C4-O dihedral angle is labeled as ϕ while the C3C4-O-H10 dihedral angle is labeled as θ. Dihedral angles within ±10° of +60° and -60° are
labeled as gauche (g or g’, respectively) while those in the range between ± 90° and 180°
179
are labeled as anti (A). In general, we choose heavy atoms to define dihedral angles for
nomenclature. However, for isobutanol, since C1 and C2 atoms are hard to be differentiated,
H7 atom is chosen in the definition of ϕ dihedral angle instead of carbon atoms. As a result,
when the ϕ dihedral angle is at anti configuration, the OH group is at gauche position with
methyl groups. It is noteworthy to mention that four isobutanol conformers (gA, g’g, gg,
Ag) predicted here have two identical minima on the potential energy surface related by
reflection of the standard configuration through a vertical plane. This will change g into
g’ configuration. No tunneling splitting has been observed in either the jet cooled cold
spectrum reported before or the room temperature spectrum presented in this work. This is
in accordance with the high barrier calculated associated with this motion.
180
Figure 6.2. The experimental rotational spectrum of room temperature isobutanol from 7.5 to 18.5 GHz. The upper part of the
spectrum is experimental data. The red trace is the Fourier transformed FID after up chirping (7.5-18.5GHz) pulse while the blue trace
is from the down chirping pulse. Black traces are collected background signals. Simulations of the fitted rotational parameters are
plotted in the lower trace. Different colors stand for different conformers assigned. gA, g’g,gg and AA conformers are labeled in red,
pink, blue and green respectively. Close-up of the 16.75-17.00 GHz region is inserted, showing the quality of the fit.
180
181
6.3
Results and Discussion
The room temperature chirped pulse Fourier transform microwave spectrum for
isobutanol over the range of 7.5-18.5 GHz is presented in Figure 6.2. The large number of
vibrationally excited states populated at room temperature lead to a high density of lines in
the spectrum. In the upper part of the spectrum, the red trace is the Fourier transformed
FID signal collected after up chirping (7.5-18.5GHz) pulse while the blue trace is from
down chirping pulse. The background signals are plotted in black. It is seen that in the
lower frequency region of the spectrum, the transitions from up chirping pulse (red trace)
suffered from weak intensities due to the short dephasing time. At the same time, the down
chirping pulse (blue trace) signals are also lost at the other edge of the spectrum due to the
same reason. It is estimated that there are more than 5000 peaks above the 3:1 signal to
noise threshold in the whole dynamics range. The typical (FWHM) for a transition is
around 700 kHz, limited by the total FID collection time as well as the collisional
broadening effect.
The assignment of the ground vibrational state of isobutanol was largely based on
the prior work by Hirota and co-workers.12 Transitions due to four conformers (gA, g’g,
gg AA) were identified in the spectrum. No transition from the Ag conformer was found
either in the previous or current work. At room temperature, a large number of rotational
states are populated, significantly increasing the number of transitions assigned for each
conformer. Tables 6.1-6.4 summarize all the rotational quantum levels for the transitions
assigned, with quantum numbers J ranging from 1 to 38. For each conformer, all the
transitions assigned were fit by a rigid rotor Hamiltonian including centrifugal distortion
182
through Pickett’s SPFIT/SPCAT program.20 The determined rotational constants A,B,C
and the quartic centrifugal distortion constants are summarized in Table 6.5.
Table 6.1. Observed frequencies and errors for the assigned transitions in MHz for gA
conformer of isobutanol.
J´
1
2
2
2
2
2
2
3
3
3
3
3
4
4
4
4
5
5
6
6
6
7
7
7
8
8
8
9
9
9
10
10
K´a
1
1
1
2
2
2
2
2
0
0
2
2
2
1
2
2
1
2
1
3
2
3
2
3
2
3
3
3
3
4
3
5
K´c
0
1
2
0
1
0
1
2
3
3
1
2
3
3
2
3
4
3
5
4
4
4
5
5
6
5
6
6
7
6
7
5
J´´
0
1
1
2
2
2
2
3
2
2
3
3
4
3
4
4
5
5
6
6
6
7
7
7
8
8
8
9
9
8
10
9
K´´a
0
0
0
1
1
1
1
1
1
1
1
1
1
2
1
1
0
1
0
2
1
2
1
2
1
2
2
2
2
5
2
6
K´´c
0
1
1
2
2
1
1
3
2
1
2
2
4
2
3
3
5
4
6
4
5
5
6
5
7
6
6
7
7
3
8
4
νobs.
11129.93750
18195.31250
15596.00000
14917.18750
14792.87500
12317.75000
12193.56250
16165.25000
14793.25000
12193.93750
11583.25000
10977.93750
18019.12500
15265.50000
11152.12500
9432.75000
13642.25000
11332.25000
17845.50000
17891.25000
12357.00000
17576.00000
14386.68750
15536.00000
17484.75000
16926.25000
12887.25000
17150.00000
10132.00000
16919.25000
18473.25000
14435.00000
νobs.-calc.
0.05995
0.08397
0.07952
0.07184
0.07949
-0.05762
0.07503
0.07411
0.00925
0.00479
-0.13523
0.06698
0.05625
-0.09292
0.05204
-0.12692
-0.04448
-0.08179
0.04910
0.06763
-0.05501
-0.06979
-0.06040
-0.14381
-0.00621
-0.10321
0.10133
-0.13354
-0.02770
0.32662
-0.06911
-0.34187
183
Table 6.1, continued
J´
12
13
14
16
17
18
19
22
25
26
29
30
33
34
K´a
4
4
4
5
5
5
11
6
7
7
8
8
9
9
K´c
9
10
11
12
13
14
8
17
19
20
22
23
25
26
J´´
12
13
14
16
17
18
18
22
25
26
29
30
33
34
K´´a
3
3
3
4
4
4
12
5
6
6
7
7
8
8
K´´c
9
10
11
12
13
14
6
17
19
20
22
23
25
26
νobs.
15385.00000
11847.25000
8607.25000
17146.00000
12979.50000
9294.75000
17142.75000
9647.00000
13937.50000
9737.25000
13928.00000
9625.50000
13680.75000
9360.50000
νobs.-calc.
-0.13339
-0.01516
0.04158
-0.07963
-0.05596
-0.06328
0.01683
-0.04776
-0.16212
-0.08996
-0.01089
0.15661
0.02809
0.11504
184
Table 6.2. Observed frequencies and errors for the assigned transitions in MHz for g’g
conformer of isobutanol.
J´
1
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
4
4
4
5
6
6
6
7
7
7
8
8
8
9
9
12
13
K´a
1
0
1
1
1
1
2
2
2
2
0
0
0
1
2
2
2
2
2
0
1
2
2
1
2
3
2
3
3
3
3
3
3
3
4
4
K´c
0
2
1
1
2
2
0
1
0
1
3
3
3
3
1
1
2
2
2
4
3
2
3
5
4
4
6
4
5
5
6
6
6
7
9
10
J´´
0
1
1
1
1
1
2
2
2
2
2
2
2
2
3
3
2
3
3
3
3
4
5
6
6
6
6
7
7
8
7
8
9
9
12
13
K´´a
0
0
0
1
0
1
1
1
1
1
0
1
1
1
1
1
2
1
1
1
2
1
1
0
1
2
3
2
2
2
4
2
2
2
3
3
K´´c
0
1
1
0
1
1
2
2
1
1
2
1
2
2
2
3
1
2
3
2
1
3
4
6
5
4
4
5
5
6
3
6
7
7
9
10
νobs.
11032.43750
12171.72000
18019.50000
13132.06250
15493.31250
11447.93750
14780.37500
14662.18750
12254.18750
12136.00000
17976.62500
12128.87500
14655.06250
17102.56250
11529.37500
16571.18750
18434.87500
10952.68750
15994.50000
16026.00000
14464.12500
11087.18750
11226.75000
17410.25000
12175.75000
17912.50000
17928.75000
17537.50000
15626.00000
16841.50000
18353.25000
13040.75000
16969.75000
10334.50000
15749.00000
12252.25000
νobs.-calc.
0.13237
0.02472
0.17253
0.05793
0.14137
0.03461
0.05535
0.14926
0.02419
0.11810
0.05314
-0.06467
-0.03351
0.03881
0.07937
0.11001
0.01720
0.09702
0.12766
-0.15524
-0.16414
0.12383
0.15049
-0.40578
0.03345
-0.18655
-0.00541
-0.14104
-0.05162
-0.04761
0.03597
0.08516
-0.05751
0.04529
-0.08151
0.09285
185
Table 6.2, continued
J´
16
21
27
29
33
37
38
K´a
5
6
8
8
9
10
10
K´c
12
16
19
22
25
28
29
J´´
16
21
27
29
33
37
38
K´´a
4
5
8
7
8
9
9
K´´c
12
16
20
22
25
28
29
νobs.
17773.75000
14548.50000
15132.00000
15359.50000
15361.00000
15159.25000
10488.75000
νobs.-calc.
-0.17389
0.05385
-0.01101
-0.24968
-0.56918
0.57754
-0.01571
186
Table 6.3. Observed frequencies and errors for the assigned transitions in MHz for gg
conformer of isobutanol.
J´
1
1
2
2
2
2
2
2
2
3
3
3
3
4
4
5
6
7
7
8
9
10
11
13
16
K´a
1
1
0
1
1
1
1
2
2
0
1
2
2
0
2
1
3
2
3
3
3
3
3
4
5
K´c
0
1
2
1
1
2
2
0
0
3
3
2
2
4
3
4
4
5
5
5
6
7
8
9
12
J´´
0
0
1
1
1
1
1
2
2
2
2
2
3
3
4
5
6
7
7
8
9
10
11
13
16
K´´a
0
0
0
0
1
0
1
1
1
0
1
2
1
1
1
0
2
2
2
2
2
3
3
4
4
K´´c
0
0
1
1
0
1
1
1
2
2
2
1
3
2
3
5
4
6
5
6
7
8
9
10
12
νobs.
11043.31250
10177.50000
12163.81250
18053.56250
13154.62500
15456.06250
11422.93750
12224.18750
14821.62500
17949.50000
17061.25000
18433.06250
16068.00000
15870.62500
9342.50000
13612.50000
17719.75000
10335.75000
15365.25000
16804.50000
17060.00000
11069.75000
15931.75000
11075.00000
16762.00000
νobs.-calc.
-0.01419
-0.01200
0.09096
0.02438
0.04327
-0.04959
-0.03288
0.14887
0.16034
0.07356
-0.01929
-0.12258
-0.29210
0.01971
-0.10567
-0.23456
0.18278
-0.26276
-0.18975
0.14949
-0.09041
0.42678
-0.11309
0.10678
-0.46293
187
Table 6.4. Observed frequencies and errors for the assigned transitions in MHz for AA
conformer of isobutanol.
J´
1
2
2
2
2
2
3
3
3
3
4
4
K´a
1
0
0
1
1
2
0
1
2
3
2
3
K´c
0
2
2
1
1
0
3
2
1
1
2
1
J´´
0
1
1
1
1
2
2
2
3
3
4
4
K´´a
0
0
1
0
1
1
1
2
1
2
1
2
K´´c
0
1
0
1
0
2
1
0
3
1
4
3
νobs.
10231.50000
14211.75000
11176.62500
18231.56250
15196.43750
9286.25000
16898.81250
15790.56250
11264.87500
12417.75000
14457.93750
13885.75000
νobs.-calc.
0.01149
-0.02416
-0.01760
0.01352
0.02009
-0.03159
0.01604
-0.01374
-0.00313
-0.00119
0.00572
0.00688
188
Table 6.5. Experimental rotational parameters of the four assigned conformers of
isobutanol.
gA
A (MHz)
B (MHz)
C (MHz)
DJ (kHz)
DJK (kHz)
DK(kHz)
dJ (kHz)
dK(kHz)
g’g
gg
AA
7597.203(22)a
7538.785(28)
7538.229(47)
6231.433(96)
3532.688(15)
3493.522(17)
3504.979(35)
4000.074(59)
2666.238(15)
2651.458(16)
2639.418(37)
3196.290(55)
5.85(25)
-0.348(78)
6.46(24)
1.7(44)
4.08(43)
-7.79(11)
0.01(22)
6.6(35)
0.174(118)
5.96(26)
-2.46(36)
-6.1(12)
0.132(24)
0.19(14)
0.580(86)
0.229(23)
3.61(42)
1.36(72)
-0.07(17)
1.4(41)
b
47
43
25
13
N
30
0
30
167
ΔEc
d
0
10
31
156
ΔE
a
Errors in parentheses are expressed in units of the last digit. bNumber of fitted lines.
c
Calculated relative energies (cm-1) at the B3LYP-D3BJ/6-311++G(d,p) level of theory,
without harmonic zero-point energy correction. dCalculated relative energies (cm-1) at
the B3LYP-D3BJ/6-311++G(d,p) level of theory, including harmonic zero-point energy
correction.
47 vibrationally excited states are predicted for 5 isobutanol conformers (including
Ag conformer) within 400 cm-1 of the ground state of the global minimum. Figure 6.3
presents the summarized energy level diagram. The Boltzmann’s factor for each vibrational
state is calculated and the related percentage is estimated by dividing the sum of all ground
states and the first and second (including combination bands) vibrationally excited states
(Table 6.6). In total, these 52 states (5 ground states + 47 vibrationally excited states)
account for about 60% of the total population and the ground state for gA conformer took
the largest portion (~4%) individually among all the states calculated, which is in
accordance with the large number of transitions assigned for gA conformer in the spectrum.
189
Figure 6.3. Calculated energy level diagrams for gA(red), g’g(purple), gg(blue),
AA(green) and Ag(black) conformers through B3PLYP-D3BJ/6-311++G(d,p) level of
theory. The relative energy for each vibrationally excited state (to their respective ground
state) is given under the solid line. The ground state relative energy for each conformer to
gA is given in the parentheses.
190
Table 6.6. Summary of all vibrational states calculated within 400 cm-1 of the ground
state of the global minimum calculated at B3PLYP-D3BJ/6-311++G(d,p) level of
theory.Vibrational states from gA, g’g, gg, AA and Ag conformers are labeled in red,
pink, blue , green and black respectively.
States
Ground State(ZPE)
Ground State(ZPE)
Ground State(ZPE)
39(1)
39(1)
Ground State(ZPE)
Ground State(ZPE)
39(1)
36(2)
38(1)
37(1)
38(1)
38(1)
36(1)
39(2)
36(1)
37(1)
39(2)
35(1)
39(1)
35(1)
39(1)
35(1)
36(1)
38(1)
36(1)
39(2)
38(1)
34(1)
37(1)
39(1) 38(1)
39(1) 38(1)
39(1) 36(1)
34(1)
39(1) 37(1)
37(1)
39(1) 38(1)
34(1)
39(2)
Energies(cm-1)a Boltzmann’s factor Percentageb (100%)
0
10.1
30.9
128.633
131.476
156.9
169.4
165.613
207.933
219.792
224.33
224.912
227.465
245.095
248.137
250.442
253.11
255.491
255.751
258.511
264.614
266.679
267.122
267.712
286.487
301.461
305.65
308.697
337.564
340.107
348.243
348.321
350.42
350.759
356.026
356.669
357.482
370.94
371.149
1
0.952715676
0.862263951
0.539607506
0.532300001
0.471196798
0.512450639
0.451912695
0.368900066
0.34850445
0.341001585
0.340051101
0.335912909
0.308678383
0.304207709
0.300863337
0.297038162
0.293665553
0.293299598
0.289442852
0.281093791
0.278323699
0.277733002
0.276948242
0.253100663
0.235561782
0.230876543
0.227527239
0.198110749
0.195709261
0.188219835
0.188149438
0.186264909
0.185962322
0.181323728
0.180765428
0.180061981
0.168807274
0.168638156
3.8393364
3.657796
3.3105214
2.0717347
2.0436788
1.809083
1.9674704
1.7350449
1.4163315
1.3380258
1.3092198
1.3055706
1.2896827
1.1851202
1.1679557
1.1551156
1.1404294
1.1274809
1.1260758
1.1112685
1.0792136
1.0685783
1.0663104
1.0632975
0.9717386
0.9044009
0.8864127
0.8735536
0.7606138
0.7513937
0.7226393
0.722369
0.7151336
0.7139719
0.6961628
0.6940193
0.6913185
0.6481079
0.6474586
191
Table 6.6, continued
States
Energies(cm-1)a
Boltzmann’s factor
Percentageb (100%)
39(2)
39(1) 37(1)
39(1) 36(1)
39(1) 35(1)
39(1) 35(1)
39(1) 36(1)
37(1)
35(1)
36(1)
38(1) 36(1)
38(2)
37(2)
39(1) 38(1)
377.205
377.497
381.482
384.099
386.001
386.342
387.642
389.247
390.993
391.597
393.241
395.325
398.744
0.163810661
0.16358142
0.160484781
0.158483137
0.157044054
0.156787433
0.155812952
0.154618194
0.153328878
0.152885368
0.151684684
0.150176194
0.147733797
0.6289242
0.6280441
0.6161551
0.6084701
0.602945
0.6019597
0.5982183
0.5936313
0.5886811
0.5869784
0.5823685
0.5765769
0.5671997
a
Calculated relative energies (cm-1) at the B3LYP-D3BJ/6-311++G(d,p) level of theory,
including harmonic zero-point energy correction. bPercentage is estimated through
dividing the Boltzmann’s factor by the sum of all ground states and the first and second
(including pair of states) vibrationally excited states.
From Figure 6.3 and Table 6.6, the ν39 vibrational state (CH2OH torsional mode) of
gA conformer is the lowest energy vibrationally excited state among all conformers, which
is predicted to have the largest population among all vibrationally excited states (~2%).
The predicted rotational constants for the fundamental of this torsional mode are slightly
smaller than the ground states, leading to red shifts on the main transitions. Attempted
assignment for the fundamental of this torsional state is based on the scaled calculated
rotational constants. However, due to the high line density in the spectrum, no firm
assignment has been achieved yet. It is noteworthy to mention that the ν39 vibrational
modes for g’g and gg conformers are of similar energies and below kT at room temperature,
raising the difficulties of the assignment. To illustrate the complexity of the spectrum, the
predicted room temperature spectrum for gA conformer is presented in Figure 6.4.
192
Vibrationally excited states up to 300 cm-1 are included and their intensities are scaled
according to the Boltzmann distribution.
193
Figure 6.4. Predicted room temperature spectrum for gA conformer. Vibrationally excited states up to 300 cm-1 are included and their
intensities are scaled according to the Boltzmann distribution. (Prepared by Dr.Brian Hays)
193
194
Figure 6.5. Two-dimensional relaxed potential energy scan of isobutanol at the B3LYPD3BJ/6-31+G(d) level of theory. The surface was calculated at 10° grid points, from 180° to +180° on ϕ and θ dihedral angles.
The potential energy surface (PES) of isobutanol is probed in a two-dimensional
relaxed potential energy scan along both H7-C3-C4-O (ϕ) dihedral angle and C3-C4-O-H10
(θ) dihedral angle at 10° intervals. The result is plotted in Figure 6.5. Different colors
stand for different energy levels. On the surface, it is clear to see there are 9 local minima
corresponding to AA, Ag, Ag’ g’A, gA, g’g, gg, g’g’, gg’ configurations respectively.
Due to the symmetry of the potential energy surface as discussed in the previous text,
those 9 structures will collapse into 5 independent structures as presented in Figure 6.1.
The absence of Ag (Ag’) conformer in the cold expansion can be explained by either the
high energy calculated or the interconversion between Ag (Ag’) and AA conformers
through a low barrier (~250 cm-1). This process often happens at the early stage of
195
supersonic expansion as a result of collision with buffer gas. Figure 6.6(a) shows the
minimum energy path among Ag (Ag’) and AA conformers as determined from the
relaxed PES scan result. The minimum energy pathway among gg,gg’ and gA conformer
is also added in Figure 6.6 (b) as a reference. It can be concluded from Figure 6.5 and
Figure 6.6 that the interconversion barriers along H7-C3-C4-O (ϕ) dihedral angle are much
higher than the interconversion barriers along C3-C4-O-H10 (θ) dihedral angle, probably
due to the large steric hindrance between OH and CH3 groups. The same steric hindrance
effect could also explain the reason for the high energy of AA conformer, since the OH
group is at gauche configuration with two CH3 groups. For gA conformer, however, one
CH3 group is at anti configuration with OH group, thus reduces the energy. For the
calculated barriers along the rotation of θ dihedral angle, the g(g’)→A barrier height is
generally half of the g→g’ barrier. It is suggested that the eclipsed configuration between
H atom and CH(CH3)2 group during the interconversion process leads to the increase of
the barrier. Finally, it is noteworthy to mention that from the potential energy surface
presented in Figure 6.6(a), a large tunneling splitting of Ag(Ag’) conformer is expected,
which further complicates the assignment.
In conclusion, the room temperature spectrum of isobutanol was collected in a
waveguide-based chirped-pulse Fourier transform microwave spectrometer over the
range from 7.5-18.5 GHz. Ground state rotational transitions for four conformers were
assigned. However, due to the complexity of the spectrum and the large number of states
populated at the room temperature, no rotational transition from vibrationally excited
state has been firmly assigned yet. To simplify the assignment, molecules with only one
196
or two stable conformers and less states populated at room temperature are suggested to
be studied through this RT-CP-FTMW set-up in the future.
197
Figure 6.6. (a) Interconversion barriers among Ag, Ag’ and AA conformers of isobutanol
calculated at B3LYP-D3BJ/6-31+G(d) level of theory. (b) Interconversion barriers
among gg, gg’ and gA conformers of isobutanol calculated at B3LYP-D3BJ/6-31+G(d)
level of theory.
198
6.4
References
1.
Huang, Y. T.; Hotopp, K. M.; Dian, B. C.; Chappell, W. J. Microwave Chemical
Sensing at Room Temperature Using an Overmoded Waveguide Design. IEEE
Trans. Microw. Theory Tech. 2012, 60, 2886-2893.
2.
Vaquero-Vara, V.; Zhang, D.; Dian, B. C.; Pratt, D. W.; Zwier, T. S. Delicate
Balance of Hydrogen Bonding Forces in D-Threoninol. J. Phys. Chem. A 2014, 118,
7267-7273.
3.
Zhang, D.; Bocklitz, S.; Zwier, T. S. Broadband Microwave Spectroscopy of
Prototypical Amino Alcohols and Polyamines: Competition between H-Bonded
Cycles and Chains. J. Phys. Chem. A 2016, 120, 55-67.
4.
Balle, T. J.; Flygare, W. H. Fabry-Perot Cavity Pulsed Fourier-transform
Micorwave Spectrometer with a Pulsed Nozzle Particle Source. Rev. Sci. Instrum.
1981, 52, 33-45.
5.
Grabow, J. U.; Stahl, W.; Dreizler, H. A Multioctave Coaxially Oriented Beamresonator Arrangement Fourier-transform Microwave Spectrometer. Rev. Sci.
Instrum. 1996, 67, 4072-4084.
6.
Suenram, R. D.; Grabow, J. U.; Zuban, A.; Leonov, I. A Portable, Pulsed-molecularbeam, Fourier-transform Microwave Spectrometer Designed for Chemical Analysis.
Rev. Sci. Instrum. 1999, 70, 2127-2135.
7.
Arenas, B. E.; Gruet, S.; Steber, A. L.; Giuliano, B. M.; Schnell, M. Chirped-pulse
Fourier Transform Millimeter-wave Spectroscopy of Ten Vibrationally Excited
States of i-propyl Cyanide: Exploring the Far-infrared Region. Phys. Chem. Chem.
Phys. 2017, 19, 1751-1756.
8.
Kolesnikova, L.; Tercero, B.; Cernicharo, J.; Alonso, J. L.; Daly, A. M.; Gordon, B.
P.; Shipman, S. T. Spectroscopic Characterization and Detection of Ethyl
Mercaptan in Orion. Astrophys. J. 2014, 784.
9.
Liu, X. Z.; Bohn, R. K.; Sorenson, S. A.; True, N. S. Rotational Spectra of Benzyl
Cyanide Assignment of the Plannar Conformer and Evidence of a Low Barrier to
Internal-rotation. J. Mol. Struct. 1991, 243, 325-339.
10.
True, N. S.; Bohn, R. K. Conformers and Internal-Rotation Barriers in Nitrous-Acid
Esters - Low-Resolution Microwave Spectroscopic Studies. J. Phys. Chem. 1982,
86, 2327-2336.
11.
Brown, G. G.; Dian, B. C.; Douglass, K. O.; Geyer, S. M.; Shipman, S. T.; Pate, B.
H. A Broadband Fourier Transform Microwave Spectrometer Based on Chirped
Pulse Excitation. Rev. Sci. Instrum. 2008, 79.
199
12.
Uzuyama, T.; Kawashima, Y.; Hirota, E. Fourier Transform Microwave Spectra of
n-Butanol and Isobutanol, Kanagawa Institute of Technology, the Graduate
University for Advanced Studies (2009).
13.
Reinhold, B.; Finneran, I. A.; Shipman, S. T. Room Temperature Chirped-Pulse
Fourier Transform Microwave Spectroscopy of Anisole. J. Mol. Spectrosc. 2011,
270, 89-97.
14.
Finneran, I. A.; Shipman, S. T.; Weaver, S. L. W. Rotational Spectroscopy of 2methylfuran from 8.7 to 960 GHz. J. Mol. Spectrosc. 2012, 280, 27-33.
15.
Hernandez-Castillo, A. O.; Abeysekera, C.; Hays, B. M.; Zwier, T. S. Broadband
Multi-Resonant Strong Field Coherence Breaking as a Tool for Single Isomer
Microwave Spectroscopy. J. Chem. Phys. 2016, 145.
16.
Mohamadi, F.; Richards, N. G. J.; Guida, W. C.; Liskamp, R.; Lipton, M.; Caufield,
C.; Chang, G.; Hendrickson, T.; Still, W. C. Macromodel - an Integrated Software
System for Modeling Organic and Bioorganic Molecules Using Molecular
Mechanics. J. Comput. Chem. 1990, 11, 440-467.
17.
Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.;
Cheeseman, J. R.; Scalmani, G.; Barone, V.; Mennucci, B.; Petersson, G. A., ; et
al. Gaussian 09, revision C.01; Gaussian, Inc.: Wallingford, CT, 2010.
18.
Grimme, S. Accurate Description of van der Waals Complexes by Density
Functional Theory Including Empirical Corrections. J. Comput. Chem. 2004, 25,
1463-1473.
19.
Grimme, S. Semiempirical Hybrid Density Functional with Perturbative Secondorder Correlation. J. Chem. Phys. 2006, 124, 034108.
20.
Pickett, H. M., SPFIT/SPCAT. http://spec/jpl.nasa.gov.
200
VITA
Di Zhang was born in Hefei, Anhui province of People’s Republic of China on
April 19th, 1989. He was raised in Beijing since his father is a professor in metallurgical
engineering in a university in Beijing, where his mother serves as an administrative staff
there. Di spent 6 years in middle/high school attached to Tsinghua University and
graduated in 2006. Then he attended University of Science and Technology of China
(USTC), where he earned a B.S. degree in Chemistry (chemical physics) in June 2010.
During his time there, he worked as an undergraduate research assistant in laboratories of
Professor Hongfei Wang (Chinese Academy of Science, Beijing) and Professor Shilin
Liu (National Synchrotron Radiation Laboratory, USTC, Hefei). He used infrared, laser
Raman and sum frequency generation (SFG) vibrational spectroscopy to study the spectra
of o,m and p-xylenes in Prof. Wang lab. In prof. Liu lab, he wrote a Matlab code to fit
the breakdown diagram of CH2BrCl probed by synchrotron radiation based threshold
photoelectron-photoion coincidence (TPEPICO) spectroscopy and calculated the
dissociation energy from curve fitting. After that, he began his graduate school for
physical chemistry at Purdue University and joined Dr.Zwier's research group in the Fall
of 2010. In the years after, Di’s research was mainly focused on single conformation
spectroscopy of biomolecules and chirped-pulse Fourier transform microwave
spectroscopy. After completion of his Ph.D. at Purdue University in May 2017, Di will
move back to China and join the research group of Professor Mingfei Zhou in Fudan
university as a post-doctoral researcher.
201
PUBLICATIONS
Article
pubs.acs.org/JPCA
Single Conformation Spectroscopy of Suberoylanilide Hydroxamic
Acid: A Molecule Bites Its Tail
Di Zhang, Karl N. Blodgett,† Xiao Zhu,†,‡ and Timothy S. Zwier*,†
†
Department of Chemistry, Purdue University, West Lafayette, Indiana 47907-2084, United States
Rosen Center for Advanced Computing (RCAC), Purdue University, West Lafayette, Indiana 47907-2084, United States
‡
S Supporting Information
*
ABSTRACT: Suberoylanilide hydroxamic acid (SAHA) is a histone deacetylase inhibitor that
causes growth arrest and differentiation of many tumor types and is an approved drug for the
treatment of cancer. The chemical structure of SAHA consists of formanilide “head” and a
hydroxamic acid “tail” separated by an n-hexyl chain, C6H5NH(CO)-(CH2)6-(C
O)NHOH. The alkyl chain’s preference for extended structures is in competition with tailto-head (T-H) or head-to-tail (H-T) hydrogen bonds between the amide and hydroxamic acid
groups. Laser desorption was used to bring SAHA into the gas phase and cool it in a supersonic
expansion before interrogation with mass-resolved resonant two-photon ionization spectroscopy. Single conformation UV spectra in the S0-S1 region and infrared spectra in the hydride
stretch and mid-IR regions were recorded using IR-UV hole-burning and resonant ion-dip
infrared spectroscopy, respectively. Three conformers of SAHA were distinguished and
spectroscopically characterized. Comparison of the experimental IR spectra with the predictions of density functional theory
calculations (DFT, B3LYP D3BJ/6-31+G(d)) leads to assignments for the three conformers, all of which possess tightly folded
alkyl chains that enable formation of a T-H (conformer A) or H-T (conformers B and C) hydrogen bonds. A modified version of
the generalized Amber force field was developed to more accurately describe the hydroxamic acid OH internal rotor potential,
leading to predictions for the relative energies in reasonable agreement with experiment. This force field was used to generate a
disconnectivity graph for the low-energy portion of the potential energy landscape of SAHA. This disconnectivity graph contains
more than one hundred minima and maps out the lowest-energy pathways between them, which could then be characterized via
DFT calculations. This combination of force field and DFT calculations provides insight into the potential energy landscape and
how population was funneled into the three observed conformers.
1. INTRODUCTION
Suberoylanilide hydroxamic acid (SAHA) is a histone
deacetylase (HDAC) inhibitor that binds directly to the
catalytic site of the enzyme, thereby blocking substrate access.
SAHA is known to inhibit class I and class II HDACs and
arrests cell growth of a wide variety of transformed cells.1
SAHA has demonstrated significant anticancer activity in both
hematologic and solid tumors.2,3 Receiving approval by the U.S.
Food and Drug Administration (FDA) for the treatment of
cutaneous T-cell lymphoma (CTCL)4,5 in 2006, SAHA has
become the lead compound in a promising new class of
anticancer drugs.
One of the most striking features of the structure of SAHA is
its linear juxtaposition of nonpolar aromatic, polar amide,
nonpolar alkyl chain, and polar hydroxamic acid groups. The
two polar groups contain hydrogen bond donor and acceptor
groups, enabling hydrogen bonds to be formed either internal
to the molecule or with its surroundings, involving both the
amide group “head” and the chelating hydroxamic acid group
“tail”. The hexyl chain that links them gives great flexibility to
the molecule in interacting with its surroundings. For instance,
the crystal structure of SAHA itself involves an array of “linear”
SAHA molecules in which the six-carbon chain is extended in
an all-trans structure that enables head-to-tail (H-T) and tail-to© 2017 American Chemical Society
head (T-H) H-bonds between SAHA molecules in adjacent
layers.6
More importantly, in the crystal structure of the complex of
SAHA with the histone deacetylase-like protein (HDLP)
complex,7 SAHA also adopts an extended conformation, but
this time with the aromatic head sitting at the entrance to a
long, cylindrical HDLP pocket. The 6-carbon alkyl chain
extends down the length of the pocket, where its hydroxamic
acid tail can engage in bidentate chelation to a Zn2+ cation
located at the bottom of the polar HDLP pocket.
From a fundamental viewpoint, what is not yet established
are the inherent conformational preferences of the SAHA
molecule in its isolated form. This work addresses that need. In
the gas phase, the many intermolecular interactions with the
binding pocket or other SAHA molecules are removed, leaving
only the molecule’s intramolecular interactions to dictate its
inherent conformational preferences.8−10 With one hydrogen
bonding group adjacent to ring and the other at the end of its
long, flexible C6 hydrocarbon tail, SAHA is able to form headto-tail and tail-to-head hydrogen bonds involving the several
Received: December 11, 2016
Revised: January 9, 2017
Published: January 10, 2017
986
DOI: 10.1021/acs.jpca.6b12464
J. Phys. Chem. A 2017, 121, 986−997
202
Article
The Journal of Physical Chemistry A
2. METHODS
2.1. Experimental Methods. The experimental methods
used in the present study have been described in detail
elsewhere.12 SAHA was purchased from Cayman Chemical at
98% purity and used without further purification. In the present
case, laser desorption was used to vaporize the sample. The
powder sample was rubbed into the surface of a graphite rod to
attain a smooth, visually uniform top surface layer. The graphite
rod was placed directly underneath the nozzle orifice via a loadlock assembly. A linear actuator (NSC 200, Newport) was
applied to move the steel rod linearly to ensure exposure of
new sample to the desorption laser. A Nd:YAG laser
(Continuum Minilite II) operating at 20 Hz (5 mJ/pulse, 2
mm beam diameter) was used for desorption and was aligned
through a window above the pulsed valve directly onto the
graphite rod. Ultrahigh purity helium and ultrahigh purity argon
were used as buffer gases (2−3 bar backing pressure) in the
supersonic jet expansion pulsed at 20 Hz out of a pulsed valve
(General, Series 9) with a 500 μm diameter orifice. The
expansion was skimmed by passing through a conical skimmer
placed ∼2 cm downstream to form a molecular beam which
was subsequently photoionized in the ionization region of a
time-of-flight (TOF) mass spectrometer. Trace water in the
sample or gas handling lines led to formation of the SAHA−
water complex.
Monitoring the SAHA parent mass channel (m/z 264), onecolor resonant two-photon ionization (R2PI) was used to
record mass selected UV excitation spectra in the S0-S1 region.
The collimated, frequency doubled output of a Nd:YAG (355
nm) pumped dye laser was used as the ultraviolet light source.
Fluorescein 548 and Rhodamine 6G were used in the dye laser
to cover the wavelength range from 275 to 281 nm at pulse
energies of 0.1−0.3 mJ/pulse at a 20 Hz repetition rate.
Conformation-specific IR spectra were taken using resonant ion
dip infrared spectroscopy (RIDIRS) in the hydride stretch
region (3200−3500 cm−1) and amide I/II (1450−1850 cm−1)
regions. In this double-resonance method, the IR beam was
generated by a Nd:YAG pumped optical parametric converter
(LaserVision) and introduced into the chamber coaxially and
counter-propagating the UV laser beam. To record a spectrum,
the UV laser was fixed on a transition in the excitation spectrum
correlated with a single conformer, while the infrared laser was
turned through the region of interest. The UV laser was pulsed
at twice the frequency of the infrared laser, delayed from the IR
by 200 ns. When the IR frequency is resonant with a transition
which shares the same ground level as the UV laser, the IR
pulse will remove a fraction of the ground state population by
absorption. The difference in ion signals between IR “on” and
IR “off” was monitored by scanning the IR laser and using a
gated integrator (Stanford Research Systems) in active baseline
subtraction mode. To generate IR light in the amide I/II
regions, difference frequency mixing of signal and idler beams
from the OPO was carried out in a AgGaSe2 crystal. IR laser
powers were 3−5 mJ/pulse in the amide NH stretch region and
0.5−1.0 mJ/pulse in the amide I/II region. All RIDIR spectra
were recorded by monitoring the origin transition for each
conformer in their respective UV excitation spectra.
To record conformation-specific electronic spectra, IR-UV
hole burning spectroscopy was employed. The method uses an
identical configuration to RIDIR spectroscopy, except the
wavelength of the IR hole-burn laser was fixed at a unique
ground-state vibrational transition of a particular conformer
donor and acceptor sites in the amide and hydroxamic acid
groups (Figure 1). With a flexible alkyl chain connecting them,
Figure 1. Chemical structure of suberoylanilide hydroxamic acid.
one anticipates the potential presence of several competing
head-to-tail and tail-to-head conformational isomers with
isomerization occurring on a potential energy landscape that
is prototypical in form.
One might anticipate that head-to-tail and/or tail-to-head
cyclic structures will be low in energy due to the hydrogen
bond(s) so formed. As Figure 1 shows, two NH···OC Hbonds are possible that constitute 11-membered rings (denoted
as C11). The OH group offers another H-bond donor site to
which it could bind to the head amide group either alone or in
concert with the hydroxamic acid carbonyl group. These unique
bonding arrangements assume that the hexyl chain is able to
loop back on itself with minimal conformational strain.
However, in pure alkanes, the extended all-trans structure is
most stable for alkyl chains up to 18 in length with each gauche
defect destabilizing the structure by about 2 kJ/mol.11 Thus,
formation of H-bonded cycles in SAHA of necessity occurs
with some conformational strain in the alkyl chain.
Furthermore, the sheer number and increased floppiness of
extended conformations argues for their dominance at higher
temperatures on entropic grounds. Thus, it is fascinating to
explore the inherent conformational preferences of SAHA in
the gas phase.
In this paper, the conformation-specific infrared (IR) and
ultraviolet (UV) spectra of the isolated SAHA molecule are
presented, carried out under expansion-cooled conditions in the
gas phase. Transitions due to three different conformers of
SAHA are observed. Assignments are made for these
conformers based on several pieces of spectroscopic data.
The conformational specific infrared spectra in hydride and
amide I/II regions serve as diagnostics for structural
determination of individual conformers. The single-conformation ultraviolet spectra also shed light on the conformations
present due to large variations in the S0-S1 origin transition
frequencies. The structures, relative energies, and harmonic
vibrational frequencies for many low-lying conformational
minima of SAHA were calculated using electronic structure
methods, providing several points of comparison between
theory and experiment. Finally, as a means of assessing the
overall form of the potential energy landscape for the molecule,
we modify a common molecular mechanics force field to more
accurately account for the hydroxamic acid OH internal
rotation and use it to calculate a disconnectivity graph for
SAHA. This pictorial summary of the potential energy
landscape provides a useful means of understanding the
observed and missing conformations, aided by predictions of
the isomerization pathways between them.
987
DOI: 10.1021/acs.jpca.6b12464
J. Phys. Chem. A 2017, 121, 986−997
203
Article
The Journal of Physical Chemistry A
observed in the RIDIR spectra while the UV probe laser was
tuned through the wavelength region of interest. For all IR-UV
HB spectra, multiple infrared transitions in the RIDIR spectra
were checked to make sure the hole burn band was unique to a
particular conformation.
2.2. Computational Methods. The long chain present in
SAHA endows the molecule with a high degree of flexibility,
which makes it possible to adopt a large number of stable
conformations and increases the complexity of its potential
energy surface (PES). Early attempts to use the generalized
Amber force field (GAFF) as a screening tool to identify lowenergy minima for optimization via density functional theory
(DFT) calculation showed that this force field was inadequate
to describe the hydroxamic acid functionality. This is not
surprising because the hydroxamic acid moiety is not included
among the test set of molecules (e.g., amino acids, proteins)
used in creating the force field.13 In particular, the weak
stabilization between the OH and CO groups, forming a fivemembered H-bonded ring, is not correctly described.
Figure 2 represents the CNOH dihedral angle scan for one of
the simplest hydroxamic acids, N-hydroxypropanamide, based
transition states connecting them. In such a disconnectivity
graph, the end of each branch identifies the energy of a
particular conformational minimum. The nodal points
represent collections of transition states in the prescribed
energy window that connect the minima below them. In this
way, one can trivially locate the highest energy barrier along the
minimum-energy isomerization pathway between any two
minima on the graph.
The theoretical method for generating disconnectivity graphs
has been described in detail elsewhere.14 In brief, the PES was
surveyed using the GAFF with atomic charges obtained from
the semiempirical AM1 bond charge correction approach. Local
minima on the PES were located using a basin-hopping
algorithm15 within a canonical Monte Carlo simulation carried
out by the GMIN 2.0 program of Wales and co-workers.16 We
carried out up to 100 basin-hopping steps until the global
minimum was found, and the step size was adjusted in each
Monte Carlo step for an acceptance ratio of 0.5.
For each minimum determined by the basin-hopping
algorithm, transition states were located by calculating the
Hessian in GAFF and walking uphill in both directions along
the smallest eigenvalues using a hybrid BFGS/eigenvectorfollowing transition-state search. All stationary points were
converged to a root-mean-squared gradient of less than 4 ×
10−6 kJ·mol−1 Å−1. Then, the minima connected to the
transition states were identified using the DNEB/L-BFGS
method developed by Wales and co-workers.17−19 Previously
unknown minima were added to the growing database of
minima, transition states, and pathways, which were then used
to generate the disconnectivity graph. Finally, we further
systematically expanded the tree by increasing the number of
connections per minimum through single-ended transition state
searches and the overall connectivity of the disconnectivity
graph through parallel double-ended transition state searching.
The disconnectivity graph so generated will be presented and
discussed after the experimental data have been considered.
Fifty unique conformational minima with lowest energies in
the disconnectivity graph were identified and served as starting
geometries for further optimizations using DFT calculations
using the Gaussian 0920 suite of programs. These calculations
involved tight geometry optimizations followed by harmonic
vibrational frequency calculations using B3LYP21 or M05-2X22
hybrid functionals with the 6-31+G(d) basis set. The dispersion
correction from Grimme and co-workers with Becke−Johnson
damping (D3BJ)23,24 was added to the B3LYP functional to
account for the London dispersion energy not correctly
described by standard DFT calculations.
The order of the relative energies of the conformers was not
changed significantly between the M05-2X and B3LYP-D3BJ
calculations. In this paper, we use the B3LYP-D3BJ calculations
to report relative energies and calculated harmonic frequencies,
as its predictions were closer to experiment than the M05-2X
calculations. The harmonic frequency calculations aided in the
assignment of conformational isomers observed in the
experiment. These frequencies were scaled by 0.96 for free
NH stretch, 0.948 for hydrogen bonded NH and OH stretches,
and 0.985 for amide I/II frequencies. These scale factors were
chosen by scaling the calculated IR frequencies of the assigned
structure of SAHA conformer A to the experimental
frequencies, where the corresponding patterns provided an
unequivocal match between experiment and theory.
Figure 2. CNOH dihedral angle scan results for N-hydroxypropanamide.
on the standard GAFF force field (dashed line). The potential
energy curve is nearly flat at dihedral angles within ±60° of
planar. As a consequence, the CNOH dihedral angle has no
strong preference to maintain planarity of the OH group
(dihedral 0°), and this led to artificially low-energy structures in
the force field searches in which the tail OH group points out of
the O(CO)−C−N-O plane to accommodate additional
stabilizing interactions for the OH group, which in SAHA
includes the phenyl ring π-cloud.
To correct this deficiency in the force field, we fit the CNOH
hindered rotor potential to one obtained from a relaxed
dihedral scan of N-hydroxypropanamide carried out at the
B3LYP-D3BJ/6-31+G(d) level of theory, as illustrated by the
blue line in Figure 2. The modified GAFF force field dihedral
angle scan result plotted in Figure 2 as the black line better
describes the dihedral angle preference close to zero degrees
and generates structures with energies in much closer
agreement with both DFT predictions and experiment.
Armed with this modified force field, we pursued a strategy
different than that in previous studies10 for screening structures
for DFT structural optimization. Rather than simply using the
force field to locate minima, the GAFF was used to generate a
disconnectivity graph as a visualization tool to display the
energies of local minima on PES and their connectivity through
988
DOI: 10.1021/acs.jpca.6b12464
J. Phys. Chem. A 2017, 121, 986−997
204
Article
The Journal of Physical Chemistry A
Finally, vertical excitation energies and excited state geometries were computed with time-dependent density functional
theory (TDDFT) with the same basis sets as mentioned above.
2.3. Nomenclature. As discussed above, because the amide
and hydroxamide groups of SAHA are separated by a C6 alkyl
chain, several different intramolecular H-bonding arrangements
are predicted to be possible in the gas phase by the theoretical
calculations. Consequently, the structures are grouped first into
families by H-bonding pattern. The NH and CO groups
adjacent to the phenyl ring are denoted as the head NH and
head CO, while the corresponding groups in the hydroxamic
acid are denoted as tail groups. The size of a hydrogen bonded
ring formed by the NH or OH with the CO groups is
denoted as Cn,25,26 where n stands for the number of atoms
involved in the ring. In 38 out of the 50 structures optimized
through DFT calculations, the hydroxamic acid CO and OH
groups are cis to one another, thereby forming a C5 ring. At the
same time, in 19 of the calculated structures, the head N−H
group forms an additional intramolecular hydrogen bond with
the tail CO, forming a C11 ring. This pattern is described as
an H-T pattern, which is the preferred arrangement in most of
the low-energy conformers. Because the amide groups at the
head and tail are in reverse order to one another, C11 rings can
also be formed through intramolecular hydrogen bonds
between tail NH and head CO groups. In total, nine of
the calculated conformers adopt this T-H pattern. In seven
calculated structures, the tail NH, instead of adopting a T-H
hydrogen bond, points instead to the π-cloud of the phenyl
ring, forming a weak NH-π intramolecular hydrogen bond
(labeled as NH-π). Three structures are predicted to form a
head NH to tail OH intramolecular hydrogen bond (labeled as
H-TOH). For the remaining 12 structures, the C5 rings at the
tail are broken, and the tail OH group points to the phenyl ring,
thereby forming an OH-π intramolecular hydrogen bond.
Notably, these structures still retain the C11 H-T H-bond as
well and are thereby labeled as H-T/OH-π. For all 50 of the
calculated structures, the amide group near the phenyl ring is in
a trans-amide arrangement. The possibility for cis-amide
arrangement is also explored and will be discussed in detail
in Section 4.3.
It is clear from the above description that several conformers
of each H-bonding arrangement (H-T or T-H) can be formed,
which therefore must differ in the configuration of the alkyl
chain that links them. We therefore designate the full
conformational structure by denoting the alkyl chain dihedral
angles along the hydrocarbon chain, numbering the C atoms
C1−C6, respectively, from head to tail. Including the CO
carbons, there are five dihedral angles along the alkyl chain
which are designated by α for C(O/head)-C1-C2-C3, β for
C1-C2-C3-C4, γ for C2-C3-C4-C5, δ for C3-C4-C5-C6, and ε for
C4-C5-C6-C(O/tail). As is standard, we will use gauche+ (g
+), gauche− (g−), anti (a), and eclipsed (e) to describe
dihedral angles around +60°, −60°, ±180° and ±120°,
respectively. Thus, the complete name of a structure would
contain both the H-bonding pattern and C6 chain orientation
(for example, T-H (g+,g+,a,g+,g+)).
Figure 3. R2PI (top trace) and IR-UV HB spectra (lower traces) for
SAHA. Asterisks in the R2PI spectrum are tentatively ascribed to hot
bands of conformer A. Because the electronic chromophore of SAHA
is closely related to that in trans-formaniliide (tFA) and transacetanilide (tAA), their electronic origins are shown in the figure for
reference.29
transitions in the higher wavenumber region are almost three
times stronger than those in the lower wavenumber region.
Using infrared transitions determined from RIDIR spectroscopy, a series of IR-UV hole burning spectra corresponding to
each individual conformation are presented below the R2PI
scan in Figure 3. The R2PI spectrum divides into transitions
due to four unique structures (A−D).
The S0-S1 origin transition of conformer A is at 36095 cm−1
(A) and has short Franck−Condon progressions involving
vibrations of frequency 23 and 67 cm−1 built off of it. The IRUV hole-burning spectrum of A (recorded with the IR at 3258
cm−1) accounts for all the major transitions in the blue part of
the R2PI spectrum, except the small transitions marked by
asterisks, which are tentatively ascribed to hot bands based on
changes in their intensity with desorption conditions. The
background present in the R2PI spectrum but missing in the
hole-burning spectrum is likely due to higher-frequency
vibronic activity arising from B−D.
The corresponding S0-S1 origins for B−D are shifted more
than 400 cm−1 to the red of A, appearing in close proximity of
one another at 35648 cm−1 (B), 35675 cm−1 (C) and 35654
cm−1 (D), respectively. These spectra also exhibit some
Franck−Condon activity in low-frequency vibrational modes,
as summarized in Table 1. On the basis of the fact that the S0-S1
origins of B−D differ from one another by no more than 27
cm−1 and the associated hole-burning spectra display similar
low-frequency vibronic structures, it is clear that the environments for the aromatic ring are similar in all three. This
suggests that B−D are all of one structural type, while A is of a
different type.
The frequency positions of the S0-S1 origins for A−D already
provide some clue to the structures involved. The electronic
origins of B−D are very red-shifted from that of tFA (36004
cm−1) and tAA (35902 cm−1),27 while that for conformer A is
shifted to the blue, as shown in Figure 3. Previous studies28
have identified H2O complexes with tFA in which H2O acts
either as a H-bond donor to the CO site or as a H-bond
acceptor from the NH. Their electronic origins are shifted to
the blue in the former case (36 114 cm−1) and to the red in the
latter (35 783 cm−1). On the basis of these simple arguments,
we anticipate that conformer A will be a T-H structure, while
conformers B and C are both H-T. We will see shortly based on
their infrared spectra that this is indeed the case.
3. RESULTS AND ANALYSIS
3.1. R2PI and IR-UV Holeburning Spectra. The top trace
of Figure 3 presents the one-color R2PI spectrum of SAHA in
the S1 ← S0 origin region, covering from 35590 to 36350 cm−1.
Two distinct groups of transitions are present in the spectrum
with a 450 cm−1 gap between them. The intensities of the
989
DOI: 10.1021/acs.jpca.6b12464
J. Phys. Chem. A 2017, 121, 986−997
205
Article
The Journal of Physical Chemistry A
Table 1. Summary of Calculated (Calc) and Observed (Obs) Vibrational Frequencies and Assignments of SAHA Monomer,
Calculated at the DFT B3LYP-D3BJ/6-31+G(d) Level of Theory.
torsional modes (cm−1)a
ν1
hydride stretches (cm−1)b
ν2
νtail NH
νOH
νhead NH
molecule
obs
calc
obs
calc
obs
calc
obs
calc
obs
calc
SAHA A
SAHA B
SAHA C
23
36
33
23
31
34
67
64
66
70
3256
3497
3497
3259
3492
3490
3306
3351
3373
3310
3342
3348
3457
3351
3390
3459
3335
3352
a
Unscaled harmonic frequencies and experimentally observed spacings of Franck−Condon activity from the IR-UV holeburning scans. bHydride
stretch fundamentals are scaled by 0.96 for free NH stretches and 0.948 for hydrogen bonded NH and OH stretches, respectively. See text for further
discussion.
Figure 4. RIDIR spectra for SAHA conformers A, B, and C in the (a) hydride stretch and (b) amide I/II regions. Calculated IR spectra at the DFT
B3LYP D3BJ/6-31+G(d) level of theory are shown below as stick diagrams in black. These frequencies were scaled by 0.96 for free NH stretch,
0.948 for hydrogen bonded NH and OH stretches, and 0.985 for amide I/II frequencies. Asterisks indicate infrared transitions used to record IR-UV
holeburn scans.
SAHA conformers A, B, and C. The S1 ← S0 origin transitions
of the three conformers were used as monitor transitions for
the RIDIR spectra, resulting in the single-conformer IR spectra
shown. The asterisks in the RIDIR spectra denote the
transitions used to record the IR-UV HB spectra in Figure 3.
The B3LYP-D3BJ/6-31+G(d) predictions for the best-fit
vibrational frequencies and peak intensities of the assigned
structures are shown as stick diagrams immediately below the
experimental RIDIR spectra in Figure 4. Comparison between
observed and calculated vibrational frequencies in the hydride
stretch and mid-IR are given in Table 1.
Because the SAHA molecule possesses one OH and two NH
groups, three hydride stretch fundamentals are anticipated in
the RIDIR spectrum of each conformer. In all three
conformers, there is a single free NH stretch fundamental in
the 3450−3500 cm−1 region. While conformers B and C are at
similar wavenumber positions (3497 cm−1), that for conformer
A is at 3457 cm−1, some 40 cm−1 lower. Thus, as with the UV
spectra, conformer A seems to be of one structural type,
different from those of B and C. On the basis of calculations of
the conformations of SAHA, including extended conformers
where both amide and hydroxamic acid NH groups are free, it
is clear that the frequency of the free NH stretch fundamentals
of head and tail NH groups are indeed different, with the free
amide NH about 40 cm−1 higher in frequency than the free
hydroxamic acid tail. This provides a second piece of evidence
that conformer A is a T-H structure, while B and C are both
from the H-T family.
It is worth noting that, in both tFA and tAA, previous studies
have also observed a minor cis-amide conformer (∼6% of
trans).29 The UV spectrum of the cis-amide conformer of
formanilide has its S1 ← S0 origin band ∼1000 cm−1 red-shifted
from that of the trans-isomer.27 The corresponding transition in
acetanilide has not been detected to date. In the SAHA
spectrum, there are several small peaks in the R2PI spectrum in
Figure 3 around 36000 cm−1 (marked by asterisks) that do not
burn out with conformer A. Their weak intensity prevented
recording RIDIR spectra of these bands; however, the balance
of evidence points to these transitions as hot bands of A or
water complexes rather than other conformers of the SAHA
monomer. Scans taken over the 35590−36350 cm−1 region
revealed no further transitions not accounted for by A−D. As a
result, experimental evidence points toward all observed
conformers arising from the trans-amide structure of SAHA,
the only isomeric form observed in solution or in the crystal.6
We will return to this point in the discussion. Finally, in the
Supporting Information, evidence is presented that structure D
is in fact due to a SAHA-H2O complex because the IR spectrum
contains two more hydride stretch transitions. The ion signal
appears in the monomer mass channel through its efficient
fragmentation following photoionization. Our main focus in
this paper is on the SAHA monomer, with the SAHA-H2O
complex discussed further only in the Supporting Information.
3.2. RIDIR Spectra of Conformers A−C. Figure 4
presents a series of RIDIR spectra recorded in the hydride
stretch (Figure 4a) and amide I/II regions (Figure 4b) for
990
DOI: 10.1021/acs.jpca.6b12464
J. Phys. Chem. A 2017, 121, 986−997
206
Article
The Journal of Physical Chemistry A
The RIDIR spectrum of conformer A also contains two
hydride stretch fundamentals shifted to much lower frequency
(3250−3310 cm−1), which also show an increased intensity and
broadening, all of which are signatures of the formation of Hbonds.30,31 Similar intense, broadened, low-frequency transitions are also present in conformers B and C, indicating that
all three conformers possess significant intramolecular Hbonds. However, when compared with conformer A, all hydride
stretches of B and C are shifted to higher frequencies,
consistent with weaker H-bonds in B and C than those in A.
For conformer B, scans that extend down to 3200 cm−1
revealed no additional transitions, suggesting that the
broadened band at 3351 cm−1 may arise from an overlap of
two H-bonded hydride stretches32 that are more closely spaced
than in conformer C.
In the amide I (1650−1750 cm−1) region, two transitions are
resolved for all conformers, corresponding to the two CO
stretches in the molecule. The 1450−1600 cm−1 region is
somewhat more complicated, containing transitions due to the
NH bends of amide and hydroxamic acid groups and several
weak benzene CH bend fundamentals. Again, the patterns for
conformer B and C resemble each other, while the pattern for
conformer A is different. The sharp transitions in all three
spectra around 1620 cm−1 are due to aromatic CC stretching
modes.
Comparison of the experimental spectra with the scaled
harmonic frequencies and IR intensities of low-energy conformers of SAHA (stick diagrams in Figure 4) leads to the
structural assignments for conformers A−C presented in Figure
5. The match between experiment and calculation relative to
under the assumption that electronic excitation will change
these vibrational frequencies only modestly from their ground
state values. This latter comparison adds confirming evidence
to the assignments but could not be considered diagnostic on
its own.
Conformer A is assigned to a T-H structure (Figure 5A)
labeled as T-H (g+,g+,a,g+,g+). This structure is also the
calculated global minimum at the B3LYP-D3BJ/6-31+G(d)
level of theory, consistent with the large intensity of its
transitions in the UV spectrum (Figure 3). The free head NH
stretch of A appears at 3457 cm−1, similar to the NH stretch
fundamentals of trans-formanilide (3463 cm−1) and transacetanilide (3472 cm−1).33 The slight shift to lower wavenumber is potentially due to anticooperativity8 in which the
head NH bond is weakened by formation of an intramolecular
H-bond to its amide CO group, which acts as an acceptor in
the T-H intramolecular H-bond, forming an 11-membered Hbonded ring (C11, Figure 5). The tail-to-head intramolecular
H-bond is short (1.90 Å, Figure 5A), producing a tail NH
stretch fundamental at 3256 cm−1. The tail OH group of
conformer A is cis to the CO group, forming a C5
intramolecular H-bond with the CO oxygen, which places
the OH stretch fundamental at 3306 cm−1.
In the amide I region, the head CO is an acceptor for the
C11 T-H hydrogen bond. As a result, its CO stretch
fundamental is shifted down in frequency to 1697 cm−1 when
compared to that of trans-fromanilide (1742 cm−1) and transacetanilide (1728 cm−1). The tail CO of the hydroxamic acid
group appears at 1681 cm−1 due to the unique chemical
environment of this group and the C5 ring it forms with the tail
OH. The amide II band for the head NH group is at 1540
cm−1, slightly higher in frequency than in trans-formanilde
(1529 cm−1) and trans-acetanilide (1528 cm−1), reflecting the
same indirect effect of the strong H-bond to its CO group.
For the tail NH, the predictions of the calculations are that its
NH bend fundamental should be shifted to even higher
frequency (∼1570 cm−1). However, this transition is predicted
to have very weak intensity and was not observed
experimentally.
The structure assigned to conformer B (Figure 5B) is
calculated to be 5.1 kJ/mol higher in energy than conformer A
at the B3LYP-D3BJ/6-31+G(d) level of theory. Notably, this
structure is determined to be the lowest-energy member of the
H-T family. Conformer C, with its many spectral similarities to
B, is assigned to a conformer in the same family shown in
Figure 5C with an energy only 1.2 kJ/mol higher than that of B.
The difference between these structures lies largely in the
folding of their C6 alkyl chains, as reflected in their dihedral
angles, with B labeled as H-T (g+,g+,e,a,g−) and C labeled as
H-T (a,g−,g−,a,g−). In general, conformer B adopts a tighter
loop for the C6 hydrocarbon chain, while a somewhat more
extended chain is favored by conformer C. This leads to a
distance between the head NH and tail CO groups in C
slightly larger than that in B and a different approach angle for
the NH···OC H-bond.
As anticipated, for both conformer B and conformer C, the
free tail NH stretch fundamentals appear at around 3497 cm−1,
about 40 cm−1 up from the position of the free head NH
stretch observed in conformer A. The tail OH retains the same
structural preference in forming a weak C5 H-bond with the tail
CO. However, in this H-T family, the tail CO also acts as
an acceptor to the donor head NH group in the H-T H-bond,
forming a bifurcated double ring structure with both C5 and
Figure 5. Calculated optimized structures assigned for SAHA
conformers A−C and structure D, assigned to the SAHA−H2O
complex at the DFT B3LYP-D3BJ/6-31+G(d) level of theory. The
zero-point corrected relative energies are included.
that of alternatives (see Figure S4 and Table S1) is sufficiently
good to make these assignments secure. As anticipated,
conformers adopting both H-T and T-H patterns are observed
in the gas phase.
Table 1 compares the observed and calculated vibrational
frequencies for conformers A−C in the hydride stretch and
mid-IR regions. Low-frequency vibrations that appear in the
R2PI spectrum are also compared with the assigned structures
991
DOI: 10.1021/acs.jpca.6b12464
J. Phys. Chem. A 2017, 121, 986−997
207
Article
The Journal of Physical Chemistry A
Table 2. Dihedral Angles along the C6 Alkyl Chain of the Three Observed Conformers of the SAHA Monomer from Head-toTail
dihedral angle (degrees)
conformer
NC12
C123, α
1234, β
2345, γ
3456, δ
456C, ε
56CN
A
B
C
cis-1
cis-2
+86
−130
−95
+130
+144
+65
+62
+165
−79
−44
+75
+56
−58
+62
−46
−164
−111
−58
+64
+171
+72
+164
+164
−179
−63
+61
−65
−67
+59
−64
−118
+131
+133
+63
+126
C11 H-bonds sharing the same tail CO group. As a result,
both H-bonded tail OH and head NH stretches are shifted up
in frequency relative to those in A, revealing another
anticooperative effect in that one H-bond to the same acceptor
group weakens the other. For conformer B, as the calculated
stick diagram suggests, those two transitions are overlapped
with each other and thus form a broadened peak at 3351 cm−1.
According to the calculation, the C5 tail OH is much weaker
than the H-bonded head NH stretch.
In conformer C, calculations predict that the relative
wavenumber positions of the H-bonded NH and OH stretch
are reversed, appearing at 3390 and 3373 cm−1 respectively. In
the amide I region of both B and C, the tail CO stretch is
shifted to lower wavenumber relative to A, reflecting its
character as a double H-bond acceptor. The head CO
stretch, which is now free, shifts up to around 1724 cm−1,
similar to the value in trans-acetanilide (1728 cm−1). As
expected, the head NH bending fundamentals are also shifted
slightly up in frequency to 1560 cm−1 for both conformers in
the amide II region as a result of engaging in a H-T
intramolecular H-bond.
In arriving at the assignments, a large number of alternative
low-energy structures and structures in other conformational
families were compared with experiment, including those
engaged in NH-π, H-TOH, and H-T/OH-π H-bonded
architectures (Figure S4). However, they were either
completely inconsistent with experimental patterns or were
significantly poorer matches. In addition, most of the structures
belonging to different families suffer from energies significantly
higher than those of the assigned structures, consistent with
their absence in the expansion.
hydroxamic acid groups. The unique wavenumber positions
of the free NH groups of head (3457 cm−1) and tail (3497
cm−1) provide characteristic spectroscopic signatures of H-T
and T-H structures. In all three conformers, the hydroxamic
acid OH group is cis to the CO group, forming a 5membered H-bonded ring (C5) that is opposite to its
orientation when chelating the Zn2+ cation in the HDAC
binding pocket, which uses the oxygen lone pairs of the CO
and OH groups.7
In conformer A, the global minimum, the hydroxamic acid
tail engages as H-bond donor via its NH group to the CO
group of the amide head, forming an 11-membered H-bonded
ring (C11) in a T-H structure. This H-bond is strong with a Hbond length of 1.90 Å, leading to a broad and intense NH
stretch fundamental at 3256 cm−1. In conformers B and C, the
direction of the H-bond is reversed, with the molecule’s head
NH acting as H-bond donor to the tail CO group. These TH hydrogen bonds also form C11 rings due to the reversal in
order of the NH and CO groups in the amide and
hydroxamic acid moieties of SAHA. Conformers B and C have
R2PI transitions about one-third the size of those of A,
consistent with their calculated relative energies about 5 kJ/mol
higher than that of A. They also have somewhat weaker headto-tail H-bonds in the 3350−3400 cm−1 region with calculated
hydrogen bond lengths of 2.11 and 2.14 Å, respectively.
Each of these structures incorporates a turn in the alkyl
chain. In previous studies from Luttschwager et al.,11 Raman
spectra provided spectral evidence that the straight-chain nalkanes prefer an extended structure up to n = 17−18, with
each gauche defect providing a destabilization of ∼2 kJ/mol.
For pure alkyl chains longer than this threshold length, the alkyl
chain folds back on itself using a turn composed of four gauche
defects, configured as (gm‑2, gm‑1, a, gm+1, gm+2). By positioning
this turn midway through the alkyl chain (m = n/2), the two alltrans segments on either side are antiparallel to one another,
where dispersive attractions can act in concert along these
segments to stabilize the folded structure.
In SAHA, the 6-carbon alkyl chain would by itself
energetically prefer an extended structure. However, as the
labeling scheme in Figure 5 indicates, the presence of the amide
and hydroxamic acid groups at either end of the alkyl chain
cause it to fold into a 4-gauche (conformer A), 3-gauche
(conformer C), or 3-gauche, 1-ecclipsed (conformer B) turn
that positions the tail and head groups where they can engage
in a H-bond that stabilizes the fold. The turn in A is just as in
the pure alkyl chains, in this case (gα, gβ, a, gδ, gε), and this ideal
turn does indeed position the head and tail for a strong, nearlinear T-H hydrogen bond. In Table 2, we listed not only the
dihedral angles along the alkyl chain (α−ε) but also those on
either side (NC12 and 56CN), which denote the orientation of
the first and last C−C bond relative to the amide or hydroxamic
acid planes. It is noteworthy that the alkyl chain prefers an out-
4. DISCUSSION
4.1. Inherent Conformational Preferences of SAHA
Monomer. A primary motivation for the present study was to
understand the inherent conformational preferences of SAHA
monomer in the gas phase where environmental effects are
removed. Whether in crystalline form or in its binding to the
HDAC enzyme pocket, the alkyl chain is extended to facilitate
interactions with other SAHA molecules or the binding pocket.
Yet, in the absence of these environmental effects, the C6 alkyl
chain, which prefers an extended conformation, is countered by
the stabilizing effect of intramolecular H-bond(s) between the
molecule’s head amide group and tail hydroxamic acid.
Our study seeks to determine how many and which
conformers are present in the gas phase. Using laser desorption
to bring SAHA into the gas phase and cooling the molecules in
a supersonic expansion, we recorded single-conformation UV
and IR spectra that led to the identification and assignment of
three conformers of the SAHA monomer, shown in Figure 5.
All three conformers are tightly folded structures that contain
intramolecular H-bonds between head amide and tail
992
DOI: 10.1021/acs.jpca.6b12464
J. Phys. Chem. A 2017, 121, 986−997
208
Article
The Journal of Physical Chemistry A
Figure 6a shows an overview of the GAFF disconnectivity
graph for SAHA, while Figure 6b focuses on the section
enclosed by the dotted red line where conformers A and B
reside. The disconnectivity graph places stable conformational
minima at the end of vertical branches that denote their relative
energies. These minima are connected by branches that group
together transition states of the same energy into energy bins
that are user-determined. In the present case, transition states
are grouped into bins separated by 1.00 kcal/mol (4.18 kJ/
mol). The numbering of the conformational minima denotes
the order in which they were found in the search process.
Note, first, that conformers A, B, and C are all predicted by
the modified GAFF force field to be among the low-energy
conformers. This gives confidence that the overall structure of
the graph is meaningful and that it can aid in a deeper
understanding of the conformational landscape of SAHA. At
the same time, the relative energies calculated by the force field
are not in perfect agreement with DFT B3LYP-D3BJ
calculations, as can be seen from the positions of the symbols
(X) and Table 3. Therefore, in the discussion that follows, we
use the disconnectivity graph to provide an overview of the
potential energy surface and isomerization pathways but use the
DFT B3LYP-D3BJ/6-31+G(d) calculations to refine our
arguments. Table 3 contains relative potential energies both
of-plane orientation for these two ancillary dihedral angles. The
end result is that the amide and hydroxamic acid planes are
nominally perpendicular to one another in all three conformers
(Figure 5) with different approach angles for the H-bond so
formed.
4.2. Disconnectivity Graphs, Isomerization Pathways,
and Observed Conformers. Single-conformation IR and UV
spectroscopy provides a powerful tool for dissecting a
complicated spectrum into its constituent components due to
individual conformational isomers. Assignments are made by
comparing the observed single-conformer spectra with ab initio
or DFT calculations. To make these assignments, however,
classical force field searches are typically used as a screening
tool to locate conformers and to order them by force field
energy as a means of prioritizing the quantum chemical
calculations, which are much more computationally intensive.
The field of single-conformation spectroscopy is now at a
point where it is capable of serving as the basis for refining
these force fields, especially in their applications to isolated, gasphase molecules using the assigned structures as benchmarks
for doing so. Accurate force fields would then open whole new
classes of problems for exploration, both in the size of the
molecules which could be explored, and in going beyond
spectroscopy to understand the dynamics of conformational
isomerization.34 In this latter context, the disconnectivity graph
serves as a powerful tool for summarizing the entire potential
energy landscape for the molecule, with all its conformational
minima, transition states, and linked pathways between
individual minima.
In the present work on SAHA, we took steps in this
direction, first in modifying the hydroxamic acid CNOH
dihedral potential within GAFF based on B3LYP-D3BJ/631+G(d) calculations, as shown in Figure 2. Then, armed with
these parameters, we used the modified version of GAFF to
create a disconnectivity graph for gas-phase SAHA monomer,
shown in Figure 6. We were motivated to construct the
disconnectivity graph for SAHA because of the prototypical
nature of the conformational landscape, with H-T, T-H, and
extended conformers all possible and their interconversion
pathways and their energetics difficult to intuit.
Table 3. Calculated Relative Energies and Free Energies of
10 Lowest Energy Conformers of SAHA in the
Disconnectivity Diagram at the B3LYP-D3BJ/6-31+G(d)
Level of Theory. Assigned Structures are Marked in Bold.
The Two Lowest Energy cis-SAHA Structures are Added at
the End of the Table Below the Dashed Line.
ΔE (kJ/mol)
free-energy
correction
(kJ/mol)d
ΔG
(kJ/mol)e
0.0
2.1
0.0
2.9
0.0
5.0
3.9
7.6
3.6
6.9
6.1
3.8
9.7
10.7
7.2
5.9
7.9
8.7
7.3
7.7
1.7
0.7
9.0
8.4
6.5
3.8
9.0
11.1
8.6
8.7
4.7
1.5
13.3
10.2
10.2
9.5
8.8
5.2
14.0
4.1
9.9
8.9
-2.7
6.2
12.5
16.8
−8.3
−4.6
−6.7
−2.9
12.3
12.7
5.6
9.8
B3LYPD3BJb
isomer
GAFFa
SAHA A
Min52 (cool
to A)
Min373
Min56 (cool
to A)
SAHA B
Min139
(cool to
A)
Min74
Min42 (cool
to B)
Min107
(cool to
B)
SAHA C
7.5
0
0.0
2.4
3.9
9.5
cis-SAHA1
cis-SAHA2
B3LYPD3BJc
General Amber force field. bCalculated relative energies (kJ/mol) at
the B3LYP-D3BJ/6-31+G(d) level of theory without harmonic zeropoint energy correction. cCalculated relative energies (kJ/mol) at the
B3LYP-D3BJ/6-31+G(d) level of theory including harmonic zeropoint energy correction. dGibbs corrections calculated at preexpansion sample temperatures (taken as T = 300 K) at the B3LYPD3BJ/6-31+G(d) level of theory. eΔG = free-energy correction +
ΔE(B3LYP-D3BJ).
a
Figure 6. (a) Disconnectivity graph for SAHA using the modified
GAFF. Red asterisks indicate the locations of SAHA A and SAHA B.
(b) Close-up view of the dashed rectangle region of the SAHA
disconnectivity graph where the assigned structures for SAHA A and
SAHA B are located. The zero-point energy corrected relative energies
calculated at the DFT B3LYP-D3BJ level of theory are also indicated
(X), taken from Table 3.
993
DOI: 10.1021/acs.jpca.6b12464
J. Phys. Chem. A 2017, 121, 986−997
209
Article
The Journal of Physical Chemistry A
Figure 7. Stationary points along the lowest-energy isomerization pathway predicted by GAFF between SAHA A, B, and C, calculated at the DFT
B3LYP-D3BJ/6-31+G(d) level of theory. Molecular geometries and optimized parameters for stationary points are included in the Supporting
Information.
red-dashed rectangle. This branch also contains the GAFF
global minimum (Min52), while SAHA C, also a H-T structure
like B, is in a separate branch with a different dihedral angle
pattern.
According to GAFF, the global minimum for SAHA
monomer is Min52, a conformation that adopts the same
dihedral angle pattern (g+,g+,a,g+,g+) as that of SAHA A but
with a H-T rather than T-H H-bond. An obvious question,
then, is how the isomerization between A and Min52 occurs.
Notably, after DFT reoptimization at the B3LYP-D3BJ/631+G(d) level of theory, SAHA A becomes the global
minimum on the disconnectivity graph with Min52 calculated
to be 2.1 kJ/mol higher in energy.
The lowest-energy pathway between Min52 and SAHA A
identified by the GAFF disconnectivity graph was recomputed
by DFT methods and is shown in Figure 7. The transition
states between individual minima were confirmed by intrinsic
reaction coordinate (IRC)36 calculations. Note that the highest
energy barrier between Min52 and A is 13.1 kJ/mol, a barrier
small enough that population can be funneled from Min52 to A
during the collisional cooling process in the early portions of
the expansion. Note that the isomerization from Min52 to
Min56 involves breaking the H-T hydrogen bond by
reorienting the tail hydroxamic acid group, with the C5−C6−
C(O/tail)-N(-H/tail) dihedral angle changed in so doing
from 103° to −122° in the intermediate state (Min56). In the
second step, the entire formanilide head rotates about its N(H/head)-C(O/head)-C1-C2 dihedral angle from −141° to
85°, forming a short (1.83 Å) tail-to-head intramolecular Hbond in SAHA A. The second barrier (TS297) is located only 6
kJ/mol above the intermediate state. Thus, the pathway from
Min52 to A involves breaking the H-T and forming the T-H Hbond without changing the alkyl chain configuration and does
so over a surprisingly small barrier. In fact, Rice−Ramsperger−
Kassel−Marcus (RRKM) rate constants for both steps are
calculated to be around 1010−1011 s−1 at the average internal
energy of SAHA monomer at 300 K (⟨Evib⟩ = 46 kJ/mol),
much faster than the cooling rate. Therefore, it is anticipated
that the population of Min52 originally present in the laserdesorbed monomer is largely interconverted to SAHA A in the
with and without zero-point correction and relative free
energies calculated at 300 K.
To focus our analysis on the lowest-energy conformations
and pathways, the disconnectivity graph in Figure 6b is
restricted to minima connected to transition states within 33
kJ/mol of the global minimum, which captures the pathways
connecting A, B, and C. In total, 159 minima were identified for
SAHA, connected by 207 transition states.
One of the insights to be gained from the disconnectivity
graph is the way in which it divides the potential energy surface
into basins containing sets of low-lying minima connected by
transition states that are also comparatively low in energy. All
the low-lying minima on the potential energy surface have an
intramolecular H-bond, whether NH···OC H-T or T-H Hbonds, and/or those involving a π-H-bond (e.g., OH···π). The
fully extended structure in which the alkyl chain is all-trans is
calculated to be 44 kJ/mol higher than the global minimum in
the GAFF force field, 27 kJ/mol from DFT B3LYP-D3BJ/631+G(d). In this sense the conformational minima that have no
interactions between SAHA’s head and tail are just off the top
of the disconnectivity graph in Figure 6.
In SAHA, the low-energy isomerization pathways involve
breaking/reforming H-bonds and reconfiguring the alkyl chain
to bring the head and tail groups into contact with one another.
Our initial expectation was that the disconnectivity graph would
have two major basins involving H-T and T-H minima
separated by a comparatively large barrier. However, this is
not the case. In actuality, the energetics of breaking H-bonds
and reconfiguring the alkyl chains are similar in size, leading to
a disconnectivity graph with appearance more like a banyan
tree,35 with many low-lying minima separated by barriers of
approximately the same height. We use the disconnectivity
graph to understand why conformers A, B, and C appear in the
expansion-cooled gas phase sample and why other low-lying
minima are absent.
According to the GAFF force field, the SAHA minima are
grouped more by having common or similar dihedral angle
patterns along the C6 hydrocarbon chain rather than via their
intramolecular H-bonding patterns. This is immediately evident
from Figure 6b, where conformers A and B, which are T-H and
H-T structures, respectively, are in the same branch inside the
994
DOI: 10.1021/acs.jpca.6b12464
J. Phys. Chem. A 2017, 121, 986−997
210
Article
The Journal of Physical Chemistry A
Figure 8. (a) Structures, labels, and zero-point energy corrected energies (relative to SAHA A at the DFT B3LYP-D3BJ/6-31+G(d) level of theory)
of the two low-energy conformers of cis-amide SAHA. (b) Stick diagrams of the calculated frequencies and IR intensities in the hydride stretch region
using the same scale factors (0.948) for hydrogen bonded NH and OH stretches. The experimental RIDIR spectra in the hydride stretch region are
presented in Figure 4a.
supersonic jet expansion and is thus absent in the R2PI
spectrum.
Similar arguments can be used to rationalize the absence of
conformers Min56/139 and Min42/107, which are in the same
sub-branches as SAHA A and SAHA B, respectively (Figure
6b). As Table 3 shows, these four minima have very low
isomerization barriers (4−5 kJ/mol) to the lowest energy
minima in their sub-basin (A or B). Thus, like Min52, structural
relaxation into A or B through interconversion over low-energy
barriers is postulated to account for their not being observed
experimentally.
For SAHA B, although it resides in the same general branch
of the disconnectivity graph as SAHA A with the same dihedral
angle pattern at the beginning of the C6 hydrocarbon chain, the
differences in the last three dihedrals places the two conformers
in different sub-branches with a larger isomerization barrier that
likely hinders their interconversion. The barrier from B to A is
predicted to be 26 kJ/mol by DFT (Figure 7), slightly higher
(∼3 kJ/mol) than the barrier between SAHA B and SAHA C
molecules. Both the A → B and B → C isomerization pathways
are included in Figure 7 and were verified by intrinsic reaction
coordinate (IRC) calculations.
The other way in which the populations of the conformers
could be modified is if their free-energy corrections are quite
different from one another. Table 3 also includes the relative
free energies of the 10 lowest energy conformers of SAHA at
the B3LYP-D3BJ/6-31+G(d) level of theory. Free-energy
corrections were made at 300 K, not knowing the internal
energy of the laser desorbed molecules prior to cooling in the
expansion. Note first that the relative free energies for A, B, and
C are three of the five lowest, consistent with their large
populations in the expansion. In free energy, they are matched
only by Min52 and Min139, which have already been argued to
lose their population during the collisional cooling process due
to small barriers to A. Indeed, the negative free-energy
correction for conformer C is consistent with its presence
among those observed.
Only two of the seven nonobserved minima in Table 3 lack a
low-energy cooling pathway to A−C. Of these, Min373
incorporates a H-T intramolecular H-bond like that in B and
C; however, the tail OH, instead of engaging in a C5 ring with
the tail CO group, points to the phenyl ring and forms an
additional OH···π intramolecular H-bond. This conformer has
an energy only 4 kJ/mol above the global minimum and a free
energy close to SAHA B. With a high isomerization barrier to
assigned conformers (>20 kJ/mol), it is less likely that
population initially in this structure would be lost by collisional
energy transfer. Min42 is somewhat higher in both ΔE (8.6 kJ/
mol) and ΔG (5.0 kJ/mol) and thus less of a concern. The
present data cannot determine whether inaccuracies in the
calculated relative energies or barrier heights or some anomaly
of the laser desorption process led to their not being detected.
The significant free-energy corrections of the low-lying
conformers listed in Table 3 raise the question of whether these
free-energy corrections are sufficient to bring the fully extended
conformers into contention with those in Table 3. To test this
possibility, free-energy corrections were calculated for a set of
extended-chain conformers. In every case, their internal
energies are so high (>25 kJ/mol at B3LYP-D3BJ/6-31+G(d)
level of theory) that the final ΔG value (>15 kJ/mol) is still
much higher than the assigned conformers. As a result, these
extended chain conformations of SAHA are not observed in our
experiment, as would be obvious from their two free hydride
stretch fundamentals.
4.3. cis-Amide Structures and Laser Desorption. The
discussion in Section 4.2 neglected one possibility that must
still be considered. The headgroup of SAHA is an alkylated
formanilide, Ph-NH-CO-R. The amide group’s conjugation
with the phenyl ring produces a smaller than typical energy
difference between trans-amide and cis-amide isomers, and in
formanilide itself, both cis- and trans-isomers were observed in
995
DOI: 10.1021/acs.jpca.6b12464
J. Phys. Chem. A 2017, 121, 986−997
211
Article
The Journal of Physical Chemistry A
the expansion.27 As a result, it was important also to explore
this possibility in SAHA, where the cis-amide structure could
potentially engage in hydrogen-bonding arrangements with the
hydroxamic acid tail that are not available to trans-SAHA.
However, the standard force field parameters in GAFF places
an artificial barrier at the cis-amide configuration because in
peptides, trans-amide structures are preferred over cis nearly
exclusively. As a result, we explored the cis-amide conformational space by fixing the amide group dihedral angle in the cisconfiguration and carrying out a force field search that
identified over 400 cis-amide minima for SAHA. Subsequent
optimization of the low-energy structures at the DFT B3LYPD3BJ/6-31+G(d) level led to the identification of two
conformers shown in Figure 8a, which are calculated to be
lower in potential energy and free energy than any trans-amide
SAHA structure so far considered. All other cis-amide minima
are significantly higher in energy than these two structures.
Both of these conformers engage in a pair of strong, bridging
H-bonds, T-H NH···OC and H-T NH···OH with different
alkyl chain configurations. The cis-1 conformer has H-bonds
(1.99 Å H-T, 1.89 T-H) shorter than those of cis-2 (2.02 Å HT, 1.95 Å T-H), consistent with its lower energy. Both of these
conformers have calculated IR spectra (Figure 8b) that are
completely inconsistent with any of the observed conformers
A−C because no free NH group is present in them, and both
NH stretch fundamentals are shifted below 3350 cm−1.
We searched over the 35590−36700 cm−1 region in R2PI for
additional transitions not yet accounted for. The TDDFT
calculations at B3LYP D3BJ/6-31+G(d) level of theory predict
that cis-1 will have its S1 ← S0 origin slightly blue (+108 cm−1)
of SAHA A, but IR-UV hole-burning proved that all transitions
blue of the SAHA A origin are due to A. Thus, there is no
experimental evidence for cis-SAHA structures in the expansion.
Furthermore, the barrier to interconversion from cis to trans is
characteristically high (>94 kJ/mol), so that interconversion
during collisional cooling is out of the question.
With no experimental evidence for cis-SAHA in the
expansion, we seek an explanation for their absence. It seems
most likely to us that the laser desorption process produces
only trans-amide structures. As a crystalline solid, SAHA exists
exclusively in the trans-amide form. Because laser desorption
occurs out of a thin film of the solid directly into the gas phase
and is done under gentle conditions designed not to
decompose the sample, we postulate that the initial internal
energy of the desorbed SAHA molecules is well below the
barrier to trans → cis isomerization. As a result, no cis-amide
conformers are formed, and full thermal equilibrium of the gas
phase molecules does not occur. As a result, the SAHA
monomer is a clear example in which the observed conformers
are affected by the means used to bring them into the gas phase.
that bring the hydroxamic acid functional group(s) into close
spatial proximity with the head amide group.
A disconnectivity graph created using GAFF provides insight
into the overall shape of the potential energy landscape for
SAHA. The basins on the potential energy surface have similar
alkyl chain conformations, and interconversion pathways from
H-T to T-H structures can occur by twisting the head amide
and tail hydroxamic moieties successively through hindered
rotations that break and then reform new H-bonds between
them without reconfiguring the alkyl chain. The calculated
barriers for doing so are somewhat smaller than the energetic
cost for breaking a H-bond in the absence of other
compensating attractions. Alkyl chain reconfiguration is more
energetically costly, preventing isomerization between A, B, and
C during the cooling in the expansion. Calculations predict that
two cis-amide structures have potential and free energies below
any of the trans-amide structures, but no experimental evidence
exists for their presence in the expansion. We postulate that the
laser desorption of the solid, in which SAHA exists exclusively
in a trans-amide configuration, leads only to trans-amide
structures in the gas phase, which cannot interconvert to cisamide on the time scale of the experiment.
While we have not focused attention on it here, we also
observed a single structure for the SAHA-H2O complex (see
the Supporting Information). This is the first step in what could
be a useful follow-up study to probe the evolution in the
conformational preferences of SAHA with increasing number of
water molecules in SAHA−(H2O)n clusters. It could also be
useful to search for cis-amide structures when SAHA is
dissolved in nonpolar solvents.
■
ASSOCIATED CONTENT
S Supporting Information
*
The Supporting Information is available free of charge on the
ACS Publications website at DOI: 10.1021/acs.jpca.6b12464.
RIDIR spectra of the SAHA-H2O complex (structure D),
effect of water on the conformational preferences of
SAHA (Figures S1−S4) and Table S1, molecular
geometries and optimized parameters for stationary
points (PDF)
■
AUTHOR INFORMATION
Corresponding Author
*E-mail: zwier@purdue.edu.
ORCID
Di Zhang: 0000-0002-8642-8694
Timothy S. Zwier: 0000-0002-4468-5748
Notes
The authors declare no competing financial interest.
■
■
5. CONCLUSIONS
The present study of the conformational preferences of the gasphase SAHA molecule revealed the presence of three
conformers under jet-cooled conditions, all of which are tightly
folded conformers involving tail-to-head or head-to-tail Hbonds. This is in striking contrast to the extended structure
SAHA takes up in crystalline form or when binding to HDAC
in its function as an anticancer drug. The characteristic
frequencies of the free NH groups of the amide head and
hydroxamic acid tail provide a clear diagnostic of the structural
type. The 6-carbon alkyl chain that serves as chemical linkage is
sufficiently long that several structural turns can be formed by it
ACKNOWLEDGMENTS
The authors gratefully acknowledge support for this work from
the National Science Foundation (NSF Grant CHE-1465028).
REFERENCES
(1) Johnstone, R. W.; Licht, J. D. Histone Deacetylase Inhibitors in
Cancer Therapy: Is Transcription the Primary Target? Cancer Cell
2003, 4, 13−18.
(2) Kelly, W. K.; Marks, P. A. Drug Insight: Histone Deacetylase
Inhibitors - Development of the New Targeted Anticancer Agent
Suberoylanilide Hydroxamic Acid. Nat. Clin. Pract. Oncol. 2005, 2,
150−157.
996
DOI: 10.1021/acs.jpca.6b12464
J. Phys. Chem. A 2017, 121, 986−997
212
Article
The Journal of Physical Chemistry A
(3) Kelly, W. K.; Richon, V. M.; O’Connor, O.; Curley, T.;
MacGregor-Curtelli, B.; Tong, W.; Klang, M.; Schwartz, L.;
Richardson, S.; Rosa, E.; et al. Phase I Clinical Trial of Histone
Deacetylase Inhibitor: Suberoylanilide Hydroxamic Acid Administered
Intravenously. Clin. Cancer Res. 2003, 9, 3578−3588.
(4) Salmi-Smail, C.; Fabre, A.; Dequiedt, F.; Restouin, A.; Castellano,
R.; Garbit, S.; Roche, P.; Morelli, X.; Brunel, J. M.; Collette, Y.
Modified Cap Group Suberoylanilide Hydroxamic Acid Histone
Deacetylase Inhibitor Derivatives Reveal Improved Selective Antileukemic Activity. J. Med. Chem. 2010, 53, 3038−3047.
(5) Grant, S.; Easley, C.; Kirkpatrick, P. Vorinostat. Nat. Rev. Drug
Discovery 2007, 6, 21−22.
(6) Griffith, D. M.; Szocs, B.; Keogh, T.; Suponitsky, K. Y.; Farkas, E.;
Buglyo, P.; Marmion, C. J. Suberoylanilide Hydroxamic Acid, a Potent
Histone Deacetylase Inhibitor; its X-ray Crystal Structure and Solid
State and Solution Studies of its Zn(II), Ni(II), Cu(II) and Fe(III)
Complexes. J. Inorg. Biochem. 2011, 105, 763−769.
(7) Finnin, M. S.; Donigian, J. R.; Cohen, A.; Richon, V. M.; Rifkind,
R. A.; Marks, P. A.; Breslow, R.; Pavletich, N. P. Structures of a
Histone Deacetylase Homologue Bound to the TSA and SAHA
Inhibitors. Nature 1999, 401, 188−193.
(8) Kusaka, R.; Zhang, D.; Walsh, P. S.; Gord, J. R.; Fisher, B. F.;
Gellman, S. H.; Zwier, T. S. Role of Ring-Constrained gamma-Amino
Acid Residues in alpha/gamma-Peptide Folding: Single-Conformation
UV and IR Spectroscopy. J. Phys. Chem. A 2013, 117, 10847−10862.
(9) Vaquero-Vara, V.; Zhang, D.; Dian, B. C.; Pratt, D. W.; Zwier, T.
S. Delicate Balance of Hydrogen Bonding Forces in D-Threoninol. J.
Phys. Chem. A 2014, 118, 7267−7273.
(10) Zhang, D.; Bocklitz, S.; Zwier, T. S. Broadband Microwave
Spectroscopy of Prototypical Amino Alcohols and Polyamines:
Competition between H-Bonded Cycles and Chains. J. Phys. Chem.
A 2016, 120, 55−67.
(11) Luttschwager, N. O. B.; Wassermann, T. N.; Mata, R. A.; Suhm,
M. A. The Last Globally Stable Extended Alkane. Angew. Chem., Int.
Ed. 2013, 52, 463−466.
(12) Dean, J. C.; Buchanan, E. G.; Zwier, T. S. Mixed 14/16 Helices
in the Gas Phase: Conformation-Specific Spectroscopy of Z-(Gly)(n),
n = 1, 3, 5. J. Am. Chem. Soc. 2012, 134, 17186−17201.
(13) Davis, Z. S. Exploring Conformational Preferences of Flexible
Biomolecules Utilizing the Instrument for Cold Ion Spectroscopy and
Force Field Methods. Ph.D. Dissertation, Purdue University, West
Lafayette, IN, 2015.
(14) Shubert, V. A.; Baquero, E. E.; Clarkson, J. R.; James, W. H., III;
Turk, J. A.; Hare, A. A.; Worrel, K.; Lipton, M. A.; Schofield, D. P.;
Jordan, K. D.; et al. Entropy-Driven Population Distributions in a
Prototypical Molecule with Two Flexible Side Chains: O-(2acetamidoethyl)-N-acetyltyramine. J. Chem. Phys. 2007, 127, 234315.
(15) Wales, D. J.; Doye, J. P. K. Global Optimization by BasinHopping and the Lowest Energy Structures of Lennard-Jones Clusters
Containing up to 110 Atoms. J. Phys. Chem. A 1997, 101, 5111−5116.
(16) Wales, D. J. GMIN2.0; University of Cambridge: Cambridge,
U.K. (http://www-wales.ch.cam.ac.uk/software.html).
(17) Wales, D. J.; Walsh, T. R. Theoretical Study of the Water
Pentamer. J. Chem. Phys. 1996, 105, 6957−6971.
(18) Trygubenko, S. A.; Wales, D. J. A Doubly Nudged Elastic Band
Method for Finding Transition States. J. Chem. Phys. 2004, 120, 2082−
2094.
(19) Liu, D. C.; Nocedal, J. On the Limited Memory BFGS Method
for Large-Scale Optimization. Math. Prog. 1989, 45, 503−528.
(20) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.;
Robb, M. A.; Cheeseman, J. R.; Scalmani, G.; Barone, V.; Mennucci,
B.; Petersson, G. A., et al. Gaussian 09, revision C.01; Gaussian, Inc.:
Wallingford, CT, 2010.
(21) Becke, A. D. Density-Functional Thermochemistry. III. The
Role of Exact Exchange. J. Chem. Phys. 1993, 98, 5648−5652.
(22) Zhao, Y.; Truhlar, D. G. Density Functionals for Noncovalent
Interaction Energies of Biological Importance. J. Chem. Theory Comput.
2007, 3, 289−300.
(23) Grimme, S. Accurate Description of van der Waals Complexes
by Density Functional Theory Including Empirical Corrections. J.
Comput. Chem. 2004, 25, 1463−1473.
(24) Grimme, S. Semiempirical Hybrid Density Functional with
Perturbative Second-Order Correlation. J. Chem. Phys. 2006, 124,
034108.
(25) Gord, J. R.; Walsh, P. S.; Fisher, B. F.; Gellman, S. H.; Zwier, T.
S. Mimicking the First Turn of an alpha-Helix with an Unnatural
Backbone: Conformation-Specific IR and UV Spectroscopy of
Cyclically Constrained beta/gamma-Peptides. J. Phys. Chem. B 2014,
118, 8246−8256.
(26) Buchanan, E. G.; James, W. H.; Choi, S. H.; Guo, L.; Gellman, S.
H.; Mueller, C. W.; Zwier, T. S. Single-Conformation Infrared Spectra
of Model Peptides in the Amide I and Amide II Regions: ExperimentBased Determination of Local Mode Frequencies and Inter-Mode
Coupling. J. Chem. Phys. 2012, 137, 094301.
(27) Manea, V. P.; Wilson, K. J.; Cable, J. R. Conformations and
Relative Stabilities of the cis and trans Isomers in a Series of Isolated
N-phenylamides. J. Am. Chem. Soc. 1997, 119, 2033−2039.
(28) Dickinson, J. A.; Hockridge, M. R.; Robertson, E. G.; Simons, J.
P. Molecular and Supramolecular Structures of N-phenyl Formamide
and its Hydrated Clusters. J. Phys. Chem. A 1999, 103, 6938−6949.
(29) Cabezas, C.; Varela, M.; Caminati, W.; Mata, S.; Lopez, J. C.;
Alonso, J. L. The Two Conformers of Acetanilide Unraveled Using
LA-MB-FTMW Spectroscopy. J. Mol. Spectrosc. 2011, 268, 42−46.
(30) James, W. H., III; Buchanan, E. G.; Guo, L.; Gellman, S. H.;
Zwier, T. S. Competition between Amide Stacking and Intramolecular
H Bonds in gamma-Peptide Derivatives: Controlling NearestNeighbor Preferences. J. Phys. Chem. A 2011, 115, 11960−11970.
(31) Walsh, P. S.; Buchanan, E. G.; Gord, J. R.; Zwier, T. S. Binding
Water to a PEG-Linked Flexible Bichromophore: IR Spectra of
Diphenoxyethane-(H2O)(n) Clusters, n = 2−4. J. Chem. Phys. 2015,
142, 154303.
(32) Dean, J. C.; Buchanan, E. G.; James, W. H., III; Gutberlet, A.;
Biswas, B.; Ramachandran, P. V.; Zwier, T. S. Conformation-Specific
Spectroscopy and Populations of Diastereomers of a Model
Monolignol Derivative: Chiral Effects in a Triol Chain. J. Phys.
Chem. A 2011, 115, 8464−8478.
(33) Miyazaki, M.; Saikawa, J.; Ishizuki, H.; Taira, T.; Fujii, M.
Isomer Selective Infrared Spectroscopy of Supersonically Cooled cisand trans-N-phenylamides in the Region from the Amide Band to NH
Stretching Vibration. Phys. Chem. Chem. Phys. 2009, 11, 6098−6106.
(34) Dian, B. C.; Longarte, A.; Winter, P. R.; Zwier, T. S. The
Dynamics of Conformational Isomerization in Flexible Biomolecules.
I. Hole-Filling Spectroscopy of N-acetyl tryptophan Methyl Amide and
N-acetyl tryptophan Amide. J. Chem. Phys. 2004, 120, 133−147.
(35) Despa, F.; Wales, D. J.; Berry, R. S. Archetypal Energy
Landscapes: Dynamical Diagnosis. J. Chem. Phys. 2005, 122, 024103.
(36) Gonzalez, C.; Schlegel, H. B. An Improved Algorithm for
Reaction-Path Following. J. Chem. Phys. 1989, 90, 2154−2161.
997
DOI: 10.1021/acs.jpca.6b12464
J. Phys. Chem. A 2017, 121, 986−997
213
Article
pubs.acs.org/JPCA
Broadband Microwave Spectroscopy of Prototypical Amino Alcohols
and Polyamines: Competition between H‑Bonded Cycles and Chains
Di Zhang,† Sebastian Bocklitz,‡ and Timothy S. Zwier*,†
†
Department of Chemistry, Purdue University, West Lafayette, Indiana 47907-2084, United States
Institut für Physikalische Chemie, Universität Göttingen, 37077 Göttingen, Germany
‡
S Supporting Information
*
ABSTRACT: The rotational spectra of the amino alcohols D-allo-threoninol, 2-amino1,3-propanediol, and 1,3-diamino-2-propanol and the triamine analog, propane-1,2,3triamine, have been investigated under jet-cooled conditions over the 7.5−18.5 GHz
frequency range using chirped-pulsed Fourier transform microwave spectroscopy.
Microwave transitions due to three conformers of D-allothreoninol, four conformers of
2-amino-1,3-propanediol, four conformers of 1,3-diamino-2-propanol, and four conformers of propane-1,2,3-triamine have been identified and assigned, aided by comparison
of the fitted experimental rotational constants with the predictions for candidate
structures based on an exhaustive conformational search using force field, ab initio and
DFT methods. Distinctions between conformers with similar rotational constants were
made on the basis of the observed nuclear quadrupole splittings and relative line
strengths, which reflect the direction of the permanent dipole moment of the conformers.
With three adjacent H-bonding substituents along the alkyl chain involving a combination
of OH and NH2 groups, hydrogen-bonded cycles (3 H-bonds) and chains (2 H-bonds)
remain close in energy, no matter what the OH/NH2 composition. Two families of H-bonded chains are possible, with Hbonding substituents forming curved chain or extended chain structures. Percent populations of the observed conformers were
extracted from the relative intensities of their microwave spectra, which compare favorably with relative energies calculated at the
B2PLYP-D3BJ/aug-cc-pVTZ level of theory. In glycerol (3 OH), D-allothreoninol (2 OH, 1 NH2), 2-amino-1,3-propanediol (2
OH, 1 NH2), and 1,3-diamino-2-propanol (1 OH, 2 NH2), H-bonded cycles are most highly populated, followed by curved
chains (3 OH or 2 OH/1 NH2) or extended chains (1 OH/2 NH2). In propane-1,2,3-triamine (3 NH2), H-bonded cycles are
pushed higher in energy than both curved and extended chains, which carry all the observed population. The NH2 group serves
as a better H-bond acceptor than donor, as is evidenced by optimized structures in which H-bond lengths fall into the following
order: r(OH···N) ≈ r(OH···O) < r(NH···N) ≈ r(NH···O).
I. INTRODUCTION
Amino alcohols and their derivatives are widely used in organic
synthesis and medicinal chemistry. Reduced from a chiral pool
such as the L-amino acids, β-amino alcohols serve as chiral
auxiliaries or ligands for asymmetric catalysis.1 Various amino
alcohol derivatives also exhibit antimicrobial2 and antifungal3
activity. As a result, the amino alcohol group has been adopted
in several antibiotics, including ethambutol4 prescribed for
treatment of tuberculosis and other infections.
With a combination of amino and hydroxyl functional groups
distributed along an alkyl chain, the amino alcohols can engage
in intramolecular H-bonds that dictate the conformational
preferences of the molecules. One useful strategy for probing
the hydrogen-bonding architecture of conformationally flexible
molecules is to study them in isolated form in the gas phase,
where supersonic expansion can be used to collisionally cool
the sample, thereby trapping the population in the low-lying
conformational minima where the conformations can be
interrogated by a range of spectroscopic methods.5−7 When
the molecule of interest incorporates a UV chromophore, IR/
UV double resonance methods can provide single conformation
© 2015 American Chemical Society
IR spectra that report directly on the hydrogen bonding
architecture via the hydride stretch fundamentals.8 However,
they require the presence of an aromatic chromophore, and
provide less detailed structural characterization than might be
ideal.
Chirped-pulse Fourier transform microwave spectroscopy
(CP-FTMW)9 is a powerful alternative for the assignment and
structural determination of different conformational isomers in
the gas phase. It allows for the acquisition of broadband spectra
at high resolution of molecules that possess permanent dipole
moments. Moreover, the intensities of the rotational transitions
can be related in a straightforward way to the relative
populations. Thus, it is a well suited technique to examine
the conformational properties of the small amino alcohol
molecules in the gas phase.
In a recent study from our group,10 CP-FTMW spectroscopy
was used to determine the conformational preferences of a first
Received: October 30, 2015
Revised: December 9, 2015
Published: December 10, 2015
55
DOI: 10.1021/acs.jpca.5b10650
J. Phys. Chem. A 2016, 120, 55−67
214
Article
The Journal of Physical Chemistry A
included on the basis of harmonic vibrational frequency
calculations.
The experimental methods for the CP-FTMW measurements have been described in detail elsewhere.22 Briefly, the
solid sample was wrapped in cotton and inserted into a stainless
steel sample holder that was heated to ∼130 °C to obtain
sufficient vapor pressure and entrained in neon carrier gas at a
backing pressure of 0.7 bar. The sample holder was located
immediately behind a pulsed valve (Parker General Valve,
Series 9) with a 1 mm diameter nozzle orifice, operating at 10
Hz. One microsecond long frequency-chirped microwave
pulses spanning the 7.5−18.5 GHz range interrogated the jetcooled molecules. Upon interaction with the microwave field, a
macroscopic polarization was induced in the sample, and a 20
μs free induction decay (FID) was collected and down
converted for display and processing on a 12 GHz digital
oscilloscope operating at a sampling rate of 40 GS/s. A total of
10 000 free-induction decays (FIDs) were averaged in the time
domain, both with (signal + background) and without
(background only) spectra. Limited sample sizes, especially
for propane-1,2,3-triamine, prevented longer averages. The
background contains resonances arising from reflecting surfaces
in the chamber, which were identified and removed from the
spectrum.
Two related methods were employed to extract relative
populations of the observed conformers from the microwave
spectra. Intensities of transitions were tabulated without an
attempt to normalize for changes in microwave power with
frequency.9 A simple, approximate method employed in
previous work,23 involved taking the sum of the intensities of
each type of microwave transition (a-, b-, or c-type) for a given
conformer (e.g., conformer X), dividing by the square of the
associated component of the dipole moment squared, and
summing:
amino alcohol, D-threoninol. With functional groups on three
adjacent alkyl carbons (one NH2 and two OH groups),
cooperative hydrogen-bonded networks could be formed, with
examples of both cyclic and chain structures represented.
In what follows, we expand our study of the amino alcohols
to include D-allothreoninol, 2-amino-1,3-propanediol, and 1,3diamino-2-propanol, and supplement these studies with the
triamine analog 1,2,3-triamino propane. D-Allothreoninol is a
diastereomer of D-threoninol, differing in the chirality at a single
site. In the other two, the terminal methyl group is removed to
simplify the potential energy surface. In so doing, it is possible
to concentrate more directly on the changes induced by the
position and number of NH2/OH groups along the alkyl chain.
This series has as its trialcohol analog glycerol, HOCH2−
CH(OH)−CH2OH, whose microwave spectrum was recently
studied in detail by Ilyushin et al.11 The three adjacent OH sites
in glycerol are reminiscent of the sugars.12 As we shall see, the
molecules in our series have low-energy conformers that form
both H-bonded chains and cycles, much as occurs in glycerol.
As a result, we can compare and contrast the ways in which the
chains and cycles compete with one another as a function of
OH/NH2 makeup.
II. EXPERIMENTAL AND COMPUTATIONAL METHODS
To identify the possible conformational minima associated with
each molecule in the series, an exhaustive conformational
search was carried out using the Amber* force field in the
MACROMODEL13 suite of programs. Depending on the
molecule, anywhere from 20 to 100 structures were found
within the 50 kJ/mol energy window prescribed for the search.
These structures served as starting geometries for full
optimizations using ab initio and density functional theory
(DFT) calculations via the Gaussian 09 suite of programs.14
For initial structure prediction, optimizations were carried out
at the MP2/6-311++G(d,p) level of theory. Rotational
constants at this level of theory have been shown in previous
work to be in close agreement with experiment.15
Although this level of theory is useful for structure
determination, the calculated energies are known to contain
significant basis set superposition error.16 Therefore, further
calculations were carried out, using the redefined structures of
the MP2/6-311++G(d,p) method as starting point, for further
optimization and energy determination. MP2/aug-cc-pVTZ
calculations explored the effects of increasing the basis set on
the MP2 energies. DFT calculations employing the B3LYP17 or
M05-2X18 hybrid functionals, or B2PLYP19 double-hybrid
functional, all with the aug-cc-pVTZ basis set, provided a
range of methods between which relative energies could be
compared. Dispersion correction from Grimme and co-workers
with Becke−Johnson dampening was added to the B3LYP and
B2PLYP calculations.19,20 Both of these levels of theory were
shown to give good results for the relative energies in a recent
study of monoglyme (CH3OCH2CH2OCH3), a molecule
similar in size to the aminoalcohols of interest here.21 Also,
the M05-2X hybrid functional has performed well in predicting
relative energies in previous work.18 Tight convergence criteria
were employed, and zero-point energy corrections were
PXα
∑i Ia(i)
μa
2
+
∑j Ib(j)
μb
2
+
∑k Ic(k)
μc 2
Alternatively, a least-squares fit to the microwave spectra for
each conformer was obtained by first varying the temperature
to obtain a best-fit temperature (Trot = 1.5 K), and then
extracting the relative population of each conformer from the
normalization constants for each conformer obtained from the
best-fit. The two methods yielded similar percentage
populations, consistent with estimated errors of about ±5%
on these percentages.
III. RESULTS AND ANALYSIS
A. Nomenclature. Despite the small size of these
molecules, with three adjacent hydrogen bonding substitutents
involving a combination of OH and NH2 groups, several
different combinations of H-bonding architecture (e.g., cycle
versus chain) and backbone dihedral angles are possible. In
each case, about 10 different conformations of each molecule
are predicted to have energies within 500 cm−1 of the global
minimum. D-Allothreoninol (2S,3R), is a diastereomer of Dthreoninol (2S, 3S). The other amino alcohols and the triamine
all lack the terminal methyl group present in threoninol, giving
them greater symmetry and eliminating one of the chiral
centers. Nevertheless, a similar nomenclature to the one used
for D-threoninol,10 which is adapted from one that is often
employed for small aliphatic amino acids, was used here.
56
DOI: 10.1021/acs.jpca.5b10650
J. Phys. Chem. A 2016, 120, 55−67
215
Article
The Journal of Physical Chemistry A
Two examples are shown below for 2-amino-1,3-propanediol.
After rotating the molecules so that as many heavy atoms as
possible are placed above the plane of an upside-down Vshaped carbon framework, the substituents are listed as α-OH,
β-NH2, and γ −OH from left to right. Roman numerals I, II,
and III denote the direction of the H-bonds along the carbon
framework, α → β → γ (I), γ → β → α (II), and γ → α → β
(III). A subscript (2 or 3) is then used to denote the number of
the intramolecular hydrogen bonds within the molecule, 2 for
H-bonded chains and 3 for H-bonded cycles. Finally, the
dihedral angles between adjacent heavy atom substituents are
given in parentheses, with angles within ±10° of +60° and
−60° labeled as gauche (g+ and g−, respectively) whereas those
in the range between ±90° and 180° are labeled anti (a). The
dihedral angle between Oα and Nβ is listed first, whereas that
for Nβ with respect to Oγ is listed second.
Similar bonding patterns are also present in propane-1,2,3triamine and 1,3-diamino-2 propanol. In propane-1,2,3triamine, because the β(NH2) has two hydrogen bond donors,
a unique H-bonding pattern (labeled IV) is possible in which
the β amine donates both to α and γ substituents, α(NH2) ←
β(NH2) → γ(NH2). Finally, in 1,3-diamino-2-propanol, several
structures with only a single H-bond were also predicted within
500 cm−1 of the global minimum. However, as we shall see,
these structures were higher in energy, and not observed in the
supersonic expansion.
Finally, it is worth noting that the three propyl derivatives
that form the main sequence studied here (2-amino-1,3propanediol, 1,3-diamino-2-propanol, and propane-1,2,3-triamine) are all symmetric, and thus have two identical minima on
the potential energy surface related by reflection of the standard
configuration through a vertical plane. This changes I3(g−g+)
into II3(g+g−). In principle, tunneling can interconvert these
minima, but in practice, no such tunneling splittings were
observed. Because all minima of these molecules come in such
identical pairs, the relative populations are unaffected by their
presence.
Figure 1. (a) Experimental rotational spectrum of jet-cooled Dallothreoninol from 7.5 to 18.5 GHz. (b) Close-up of the 15.85−15.95
GHz region with calculated stick spectra due to conformer A (red), B
(blue), and C (green) below, showing the quality of the fit. (c) Further
expansion of 5 MHz regions around the 322−212 transitions of
conformers A, C are shown in (b), to compare the experimental and
calculated nuclear quadrupolar splittings for the two assigned cyclic
conformers. The F′−F″ labels are included in the figure.
program26 to refine the rotational constants and expand the
number of fitted transitions. Transitions due to three
conformers were identified in the spectrum, with calculated
stick spectra for the three shown in red (conformer A), blue
(conformer B), and green (conformer C) below the
experimental spectrum. The number of fitted transitions N,
the average standard deviation for each fit (σ), and the resulting
sets of fitted rotational constants (A, B, C), and inertial defect
(Δ) are listed in Table 2. For the range of rotational transitions
observed, centrifugal distortion constants provided only
marginally better fits with experiment, and so are not included
in the table. Figure 1b includes an expanded view of the 15.85−
15.95 GHz region, demonstrating the quality of the fit.
The three assigned conformers for D-allothreoninol are all
near-prolate asymmetric tops with the Ray’s asymmetry
parameter κ = (2B − A − C)/(A − C)27 lying between −0.5
and −0.8. Two of them have similar κ values around −0.6, and
the third one has κ = −0.75, indicating two different structural
classes based on the dihedral angles between adjacent heavy
atom substituents. Compared with calculation, structures II and
III in Table 1 have κ values around −0.6, a value consistent with
a g+g− configuration. A κ near −0.75 is consistent with any of
the structures I, IV, and V, which all share a g−g−
configuration, or structure VIII with its g+g+ configuration.
Structures with similar κ values typically also have similar
rotational constants because they share the same heavy atom
configuration. For example, the experimental rotational
constants for structures II and III in the g+g− family are
3132, 2180, and 1920 MHz and 3167, 2129, and 1920 MHz,
respectively. These differences are not large enough to
discriminate between them on the basis of rotational constants
alone.
B. Microwave Spectra. 1. D-Allothreoninol. The experimental microwave spectrum of D-allothreoninol over the range
of 7.5−18.5 GHz is presented in Figure 1a. With hundreds of
lines observed with a range of intensities, we anticipate that, as
in the case of D-threoninol, separate contributions from
different conformers will contribute to the spectrum. Because
each conformer exhibits a unique pattern of lines, transitions
from each are intermingled with one other.
To analyze the spectrum, we focused attention on the nine
predicted lowest energy conformers within 500 cm−1 of the
global minimum, as summarized in Table 1. The calculated
rotational constants for these conformers were used as a
starting point in providing predictions for the rotational
spectrum, using a semirigid rotor Hamiltonian.24 The resulting
spectrum was then plotted using the JB95 software developed
by Plusquellic25 as a visualization and analysis tool. Tentative
assignments were made on the basis of the close match-up of
several strong lines in the spectrum. These experimental
frequencies were then put into Pickett’s SPFIT/SPCAT
57
DOI: 10.1021/acs.jpca.5b10650
J. Phys. Chem. A 2016, 120, 55−67
216
Article
The Journal of Physical Chemistry A
Table 1. Calculated Rotational Parameters and Relative Energies of the Nine Most Stable Confirmations of D-Allothreoninol
a
Calculated rotational parameters at the MP2/6-311++G(d,p) level of theory. Rotational constants at the B2PLYP-D3BJ/aug-cc-pVTZ level of
theory are also included after the slash (/). bCalculated relative energies (cm−1) at the B2PLYP-D3BJ/aug-cc-pVTZ level of theory, including
harmonic zero-point energy correction.
Conclusive evidence for structural assignments comes from
the different 14N nuclear quadrupole splitting patterns, which
are produced by the interaction between the nuclear spin
angular momentum I (I = 1 for 14N) and the electric field
gradient created by the rest of the molecule at the site of each
nucleus. Figure 1c shows 5 MHz regions around the 322−212
rotational transitions of the two g+g− conformers, and
compares the experimental nuclear quadrupolar splittings
with those calculated for the assigned structures of the two
cyclic isomers. The patterns are different enough to make a
clear assignment for conformer A as structure II, the II3(g+g−)
conformer, whereas conformer C is assigned to structure III,
the I3(g+g−) conformer. This fitting procedure determines a
set of nuclear quadrupole coupling constants χ, which are very
sensitive to the orientation of −NH2 group with respect to the
principal inertial axis system.28 The fitted values for χ for the
assigned conformers are listed in Table 2 and match nicely with
those predicted by the ab initio calculations.
Structures I and IV have similar quadrupole coupling
constants due to the similar orientations of the amino group
in these two conformations. To distinguish between them, we
Table 2. Experimental Rotational Parameters of the Three
Assigned Conformers of D-Allothreoninol
a
Errors in parentheses are expressed in units of the last digit. bNumber
of fitted lines, including nuclear hyperfine components. cStandard
deviation of the fit.
Table 3. Calculated Rotational Parameters and Relative Energies of the Eight Most Stable Confirmations of 2-Amino-1,3propanediol
A (MHz)a
B (MHz)
C (MHz)
μa (D)
μb (D)
μc (D)
μT (D)
χaa (MHz)
χbb (MHz)
χcc (MHz)
κ
ΔE (cm−1)b
I (B)
II (A)
III (C)
IV
V (D)
VI
VII
VIII
I2(g−g−)
I3(g−g+)
II2(g−g−)
II2(g+g+)
II2(g−g+)
II2(g−g−)
III2(g−a)
II2(g−g+)
6083/6103
2279/2268
1997/1988
−1.7
0.3
0.9
1.9
−0.29
2.64
−2.36
−0.86
0
4242/4224
3134/3146
2550/2549
0.3
2.7
1.7
3.2
−2.87
1.69
1.18
−0.31
55
5996/6017
2273/2263
1977/1973
4.4
0.9
1.4
4.7
−4.67
2.75
1.91
−0.85
204
6042/6058
2258/2251
1983/1979
1.2
1.4
3.2
3.7
−0.66
2.67
−2.02
−0.86
139
7735/7727
1977/1975
1699/1695
3.5
−1.1
1.8
4.1
−3.94
1.79
2.16
−0.91
201
6033/6009
2235/2239
1951/1954
2.0
0.1
0.9
2.2
−4.48
2.79
1.69
−0.86
264
4173/4239
31673015
2196/2137
2.1
2.2
0.6
3.1
2.15
0.17
−2.32
−0.02
400
7642/76622
1952/1950
1686/1681
1.9
−0.5
−0.1
2.1
−3.83
1.81
2.02
−0.91
304
a
Calculated rotational parameters at the MP2/6-311++G(d,p) level of theory. Rotational constants at the B2PLYP-D3BJ/aug-cc-pVTZ level of
theory are also included after the slash (/). bCalculated relative energies (cm−1) at the B2PLYP-D3BJ/aug-cc-pVTZ level of theory, including
harmonic zero-point energy correction.
58
DOI: 10.1021/acs.jpca.5b10650
J. Phys. Chem. A 2016, 120, 55−67
217
Article
The Journal of Physical Chemistry A
make use of the direction of the dipole moment in the
molecular frame, whose projections on the inertial axes
determine the relative intensities of the a-, b-, and c-type
microwave transitions in the spectrum. The rotational spectrum
of conformer B shows strong a-type, medium strength c-type,
and relatively weak b-type transitions. These data are consistent
with the assignment of conformer B as structure I, which is
predicted to have μa > μc > μb. C-type transitions are predicted
to be strongest in structure IV, counter to experiment. Thus, we
assign conformer B to structure I, labeled as II2(g−g−).
2. 2-Amino-1,3-propanediol, 1,3-Diamino-2-propanol,
and Propane-1,2,3-triamine. Similar procedures were followed
in arriving at assignments for the rotational spectra of 2-amino1,3-propanediol, 1,3-diamino-2-propanol, and propane-1,2,3triamine molecules, respectively. The results are summarized in
Tables 3−8. These molecules possess higher symmetry than D-
the carbon framework. As a result, Ray’s asymmetry parameter
gains in importance over the dihedral angles of the functional
groups as a means of grouping them into different structural
classes.
For 2-amino 1,3 propanediol, the predicted rotational
constants and asymmetry parameters for the eight lowest
energy structures are summarized in Table 3, with κ values
ranging between −0.02 and −0.91. Conformers fall into
subgroups around four characteristic κ values (κ = −0.02,
−0.3, −0.85, −0.91), with the κ = −0.91 family characteristic of
the most extended structures, whereas κ = −0.02 family is most
compact.
When compared with the experimentally derived κ values, a
clear assignment for conformer A to structure II, I3(g−g+), can
be made, because it is the only one of the low-energy structures
with κ values around −0.3.
Conformer D is fit with rotational constants that give a κ =
−0.91. Two κ = −0.91 conformers, structure V and structure
VIII, are labeled as II2(g−g+), as they are extended H-bonded
chains, possessing similar rotational constants, dipole moment
projections, and nuclear quadrupole coupling constants. A close
look at their structures reveals that they are related to each
other by a rotation of the free α(OH) group in the chain
structure. Comparing the measured percent populations with
calculated relative energies (Table 9), we find that structure V is
favored as the structure observed for conformer D. The fact
that we see only one of these structures suggests the possibility
that conformational interconversion may be happening during
the collisional cooling in the expansion. We will consider this
possibility further in the Discussion.
There are two observed conformers (B and C) with κ =
−0.85, and four calculated structures in this category to choose
from (structures I, III, IV and VI). Conformer B is
unambiguously assigned to structure I through a comparison
calculated and experimental quadrupole coupling constants and
electric dipole moments. Conformer C is then assigned to
structure III over structure VI, which differ once again only in
the orientation of the free α(OH) group in the chain structure.
Exactly analogous arguments are used to assign conformer C to
Table 4. Experimental Rotational Parameters for the
Assigned Conformers of 2-Amino-1,3-propanediol
A (MHz)
B (MHz)
C (MHz)
Δ (kHz)
χaa
(MHz)
χbb
(MHz)
χcc (MHz)
Nb
σc/kHz
κ
μa:μb:μc
conformer A
conformer B
conformer C
conformer D
4208.577(8)a
3130.694(1)
2527.345(1)
0.59(7)
−2.37(2)
6049.922(1)
2265.006(3)
1981.188(4)
0.33(8)
−0.31(3)
5981.55(7)
2257.073(1)
1965.833(9)
0.32(3)
−3.93(2)
7679.43(7)
1968.92(9)
1689.01(1)
1.22(3)
2.35(4)
2.28(6)
1.64(4)
1.14(3)
20
5.5
−0.28
b>c>a
−2.03(4)
15
15.2
−0.86
a>c>b
1.64(6)
13
4.4
−0.85
a>c>b
1.59(4)
9
3.1
−0.91
a>c>b
−3.23(9)
a
Errors in parentheses are expressed in units of the last digit. bNumber
of fitted lines, including nuclear hyperfine components. cStandard
deviation of the fit.
threoninol and D-allothreoninol molecules because the methyl
group at the end of the threoninol structure is removed from
Table 5. Calculated Rotational Parameters of the Eight Most Stable Confirmations of 1,3-Diamino-2-propanol
A (MHz)a
B (MHz)
C (MHz)
μa (D)
μb (D)
μc (D)
μT (D)
N1χaa (MHz)
N1χbb (MHz)
N1χcc (MHz)
N5χaa (MHz)
N5χbb (MHz)
N5χcc (MHz)
κ
ΔE (cm−1)b
I (A)
II (C)
III (B)
IV (D)
V
VI
VII
VIII
II3(g−g+)
I2(g+g−)
II2(g+g−)
I2(g−g−)
S(g+ anti)
I2(g+g+)
I2(g+g+)
S(anti g+)
4314/4323
3019/3002
2488/2480
0.3
1.9
1.7
2.6
−1.99
−0.35
2.34
2.17
−0.52
−1.65
−0.42
0
8073/8071
1969/1969
1711/1707
−2.1
−2.3
0.5
3.1
2.39
−4.55
2.16
2.92
2.24
−5.16
−0.92
112
8062/8053
1947/1948
1695/1692
3.3
−1.7
1.5
4.0
−1.93
0.72
1.21
2.39
−4.58
2.18
−0.92
95
6132/6119
2239/2211
1981/1960
−3.3
−2.4
0.7
4.1
2.79
−3.98
1.19
−4.95
2.28
2.67
−0.88
239
5633/5640
2279/2269
1764/1758
−0.3
−2.2
0.5
2.3
−0.04
−2.19
2.23
2.42
−4.83
−2.41
−0.73
405
6209/6228
2219/2214
1968/1965
−2.4
−2.9
−0.9
3.8
1.61
−2.22
0.61
1.89
2.44
−4.33
−0.88
405
6103/6135
2206/2200
1962/1959
−2.8
−1.7
1.2
3.5
1.59
−2.14
0.55
0.69
−1.08
0.39
−0.88
377
4303/4484
2889/2729
2117/2061
1.4
−2.1
0.2
2.6
1.63
−1.71
0.08
0.97
−3.35
2.38
−0.29
490
a
Calculated rotational parameters at the MP2/6-311++G(d,p) level of theory. Rotational constants at the B2PLYP-D3BJ/aug-cc-pVTZ level of
theory are also included after the slash (/). bCalculated relative energies (cm−1) at the B2PLYP-D3BJ/aug-cc-pVTZ level of theory, including
harmonic zero-point energy correction.
59
DOI: 10.1021/acs.jpca.5b10650
J. Phys. Chem. A 2016, 120, 55−67
218
Article
The Journal of Physical Chemistry A
Table 6. Experimental Rotational Parameters for the Four Assigned Conformers of 1,3-Diamino-2-propanol
A (MHz)
B (MHz)
C (MHz)
Δ (kHz)
N1χaa (MHz)
N1χbb (MHz)
N1χcc (MHz)
N5χaa (MHz)
N5χbb (MHz)
N5χcc (MHz)
Nb
σc/kHz
κ
μa:μb:μc
A
B
C
D
4304.640(5)a
2985.145(8)
2465.185(1)
0.71(5)
−1.66(7)
−0.35(7)
2.01(7)
1.79(7)
−0.45(8)
−1.34(8)
11
10.7
−0.43
b>c>a
7983.566(2)
1941.489(2)
1687.348(2)
1.68(1)
−1.55(6)
0.73(7)
0.81(7)
2.09(3)
−3.73(4)
1.64(4)
21
15.9
−0.92
a>b>c
8000.795(8)
1961.541(9)
1701.152(5)
0.25(2)
2.14(3)
−3.78(3)
1.64(3)
2.57(4)
1.79(4)
−4.37(4)
17
9.6
−0.92
b>a>c
6056.698(1)
2250.505(9)
1996.815(1)
1.77(5)
2.16(5)
−3.19(5)
1.03(5)
−4.37(1)
2.19(1)
2.18(1)
13
14.5
−0.88
a>b>c
Errors in parentheses are expressed in units of the last digit. bNumber of fitted lines, including nuclear hyperfine components. cStandard deviation
of the fit.
a
Table 7. Calculated Rotational Parameters of the Eight Most Stable Confirmations of Propane-1,2,3-triamine
A (MHz)a
B (MHz)
C (MHz)
μa (D)
μb (D)
μc (D)
μT (D)
N1χaa (MHz)
N1χbb (MHz)
N1χcc (MHz)
N3χaa (MHz)
N3χbb (MHz)
N3χcc (MHz)
N5χaa (MHz)
N5χbb (MHz)
N5χcc (MHz)
κ
ΔE (cm−1)b
I(A)
II(B)
III
IV(C)
V
VI(D)
VII
II2(g+g+)
IV2(g−g+)
II3(g−g+)
II2(g−g−)
I2(g−g+)
II2(g−g+)
I2(g+g+)
5805/5810
2259/2240
1969/1950
0.4
0.1
1.8
1.9
2.51
−3.65
1.15
−0.43
2.32
−1.88
0.62
−2.47
1.86
−0.85
0
7575/7518
1929/1924
1686/1680
0
−2.6
1.6
3.0
2.31
−4.37
2.06
1.64
2.59
−4.24
2.31
−4.37
2.06
−0.92
64
4143/4126
3007/2988
2452/2428
0.1
−0.1
1.4
1.4
1.71
−0.18
−1.52
−3.59
2.01
1.58
−1.82
−0.28
2.10
−0.34
201
5837/5814
2227/2219
1930/1919
2.5
−1.1
0.1
2.8
1.69
−1.63
−0.06
−3.68
2.41
1.27
2.23
1.75
−3.98
−0.85
221
7405/7356
1945/1939
1688/1682
−1.7
−1.6
1.6
2.8
2.60
1.42
−4.02
−2.64
1.19
1.44
2.35
−4.24
1.89
−0.91
171
7393/7346
1920/1915
1669/1664
−2.9
−1.4
0.5
3.3
2.35
−4.27
1.92
−2.39
1.15
1.24
−1.42
0.84
0.59
−0.91
188
5729/5716
2218/2208
1924/1912
−2.9
−0.0
2.0
3.6
0.99
−1.21
0.22
−3.78
2.35
1.43
1.65
−1.49
−0.15
−0.85
266
a
Calculated rotational parameters at the MP2/6-311++G(d,p) level of theory. Rotational constants at the B2PLYP-D3BJ/aug-cc-pVTZ level of
theory are also included after the slash (/). bCalculated relative energies (cm−1) at the B2PLYP-D3BJ/aug-cc-pVTZ level of theory, including
harmonic zero-point energy correction.
unambiguous assignment of conformers A, B, C, and D to
structures I [II3(g−g+)], III [II2(g+g−)], II [I2(g+g−)], and IV
[I2(g−g−)], respectively.
For propane-1,2,3-triamine, the asymmetry parameters of the
calculated structures divide into three groups with values near κ
= −0.34, −0.85, and −0.91 (Table 7). Unfortunately, due to
limited sample size, only a small number of the most intense
rotational transitions could be definitively assigned at the
present signal level, making our assignments somewhat more
tentative. Nevertheless, transitions due to two structures in each
of the last two families are tentatively assigned. These are
confirmed and strengthened by nuclear quadrupole splittings
associated with these transitions. Detailed comparison of the
values of the 14N quadrupole coupling constants lead to the
identification of conformer A and C as structure I and IV in the
κ = −0.85 family, whereas conformers B and D are assigned to
structures II and VI in the κ = −0.91 family, respectively.
structure III over structure VI, the two unassigned conformers
in the κ ≈ -0.85 category.
On the basis of the CP-FTMW spectrum (Figure S1),
transitions due to four conformers of 1,3-diamino-2-propanol
have been assigned. A wider variety of κ value groups is
observed in this case than in 2-amino-1,3-propanediol. Five
groups are identified (κ = −0.29, −0.42, −0.73, −0.88, and
−0.92) in the eight conformations with energies within 500
cm−1 of the global minimum. Among the higher-energy
structures are two (structures V and VIII) that incorporate
only a single hydrogen bond (κ = −0.29 and −0.73).
More complex hyperfine splitting patterns are observed with
the presence of the second nitrogen atom, leading to a new set
of nuclear quadrupole coupling tensors χaa, χbb, χcc for each
structure as summarized in Table 5. In this way, within each
subgroup determined by different κ values, the predicted 14N
quadrupole coupling constants are different enough to make an
60
DOI: 10.1021/acs.jpca.5b10650
J. Phys. Chem. A 2016, 120, 55−67
219
Article
The Journal of Physical Chemistry A
percent population. It is unlikely that collisional removal of the
cyclic conformer population during the cooling process was
occurring, because the calculated barrier between the cycle and
the global minimum chain, II2(g+g+), is calculated to be around
1300 cm−1, large enough to trap population behind it during
the cooling process.
A small number of very weak transitions remain unassigned
in each spectrum. They might come from other conformers
that lie at higher energies or 13C isotopes of more stable
structures. However, due to their weak intensities, no further
assignments were possible.
Table 8. Experimental Rotational Parameters of the Four
Assigned Conformers of Propane 1,2,3-triamine
A (MHz)
B (MHz)
C (MHz)
N1χaa (MHz)
N1χbb (MHz)
N1χcc (MHz)
N3χaa (MHz)
N3χbb (MHz)
N3χcc (MHz)
N5χaa (MHz)
N5χbb (MHz)
N5χcc (MHz)
Nb
κ
μa:μb:μc
A
B
C
D
5769.44(4)a
2233.75(9)
1943.57(1)
3.56(1)
−4.46(4)
0.90(4)
−0.14(2)
2.17(4)
−2.02(4)
0.15(2)
−1.94(2)
1.78(2)
14
−0.85
c>a>b
7630.95(5)
1932.55(1)
1663.83(8)
3.61(6)
−4.26(8)
1.64(8)
1.28(3)
3.49(1)
−3.20(1)
3.64(6)
−3.33(5)
1.31(5)
16
−0.91
b>c>a
5777.45(2)
2195.76(1)
1907.41(1)
1.65(1)
−1.70(3)
0.04(3)
−3.60(9)
2.02(7)
1.58(7)
2.10(1)
1.37(2)
−4.53(2)
18
−0.85
a>b>c
7309.08(8)
1910.06(1)
1657.83(7)
2.42(6)
−4.87(1)
1.459(1)
−2.30(6)
1.27(2)
1.03(2)
−1.46(6)
0.49(1)
0.97(1)
9
−0.91
a>b>c
IV. DISCUSSION
One of the primary motivations for studying the microwave
spectroscopy of the present series of amino alcohols and
triamines is the opportunity they afford for probing in some
detail the inherent conformational preferences of alkyl chains
decorated with amino and alcohol functional groups. Building
on previous results for glycerol, with 3 OH groups,11
prototypical propyl derivatives with two OH/one NH2, one
OH/two NH2, and three NH2 groups were studied. Firm
assignments were made for a total of 15 conformers of four
molecules were made, and their relative populations
determined.
A. Comparing Predicted Energies and Observed
Populations. Before considering the conformational preferences of the molecules in this series, it is important first to test
the levels of theory that are best used in making comparison
with experiment. The principal experimental results from the
microwave spectra are the fitted rotational constants, dipole
moment directions, and nuclear hyperfine splitting parameters
used for conformational assignments. The experimental results
(rotational constants, dipole moment components, and nuclear
a
Errors in parentheses are expressed in units of the last digit. bNumber
of fitted lines, including nuclear hyperfine components.
Interestingly, structure II has a central double-donor NH2
group (H2N ← HNH → NH2), and is formally a IV2(g−g+)
structure in our nomenclature, forming a bifurcated chain.
The lone calculated κ = −0.34 structure (structure III)
adopts a cyclic conformation with three connected NH···N
intramolecular hydrogen bonds. The dipole moment of this
structure is rather small but is directed almost exclusively along
the out-of-plane c-axis. Given the signal-to-noise ratio on the
spectrum (Figure S2), we anticipated being able to observe this
conformer experimentally; however, no c-type transitions were
observed near their predicted frequencies. The present signal
levels leads to an upper bound of 20% (for S/N = 3) on its
Table 9. Percent Populations of the Observed Conformers and Comparison of Their Relative Energies Calculated at the
B2PLYP-D3BJ/aug-cc-pVTZ Level of Theory
molecule
D-allothreoninol
2-amino-1,3-propanediol
1,3-diamino-2-propanol
propane-1,2,3-triamine
glycerold
conformer
population (%)a
population (%)b
ΔE(B2PLYP)c (cm−1)
II (A)/cycle/II3(g+g−)
I(B)/curved chain/II2(g−g−)
III (C)/cycle/I3(g+g−)
II (A)/cycle/I3(g−g+)
I (B)/curved chain/I2(g−g−)
III (C)/curved chain/II2(g−g−)
V(D)/extended chain/II2(g−g+)
I(A)/cycle/II3(g−g+)
III(B)/extended chain/II2(g+g−)
II(C)/extended chain/I2(g+g−)
IV(D)/curved chain/I2(g−g−)
I(A)/curved chain/II2(g+g+)
II(B)/extended chain/IV2(g−g+)
IV(C)/curved chain/II2(g−g−)
VI(D)/extended chain/II2(g−g+)
1a/cycle/II3(g−g+)
2b/curved chain/II2(g+g+)
3b/curved chain/II2(g+g+)
5c/extended chain/II2(g+g−)
7/curved chain/I2(g+g+)
58
37
5
52
37
8
3
34
31
21
14
59
22
17
2
48
44
8
56
33
6
5
40
29
22
9
58
20
18
4
0
58
188
55
0
204
201
0
95
112
239
0
64
221
188
0
57
205
235
394
large
large
small
small
small
a
Population percentage determined by summing the total intensities of a-, b-, and c-type transitions and dividing by the square of the theoretically
predicted electronic dipole moments. bPopulation percentage determined by using the scale factors obtained from the best fit to the intensities for
each conformer in rotational fitting program (JB95). cCalculated relative energies (cm−1) at the B2PLYP-D3BJ/aug-cc-pVTZ level of theory,
including harmonic zero-point energy correction. dTaken from ref 11 (Ilyushin et al.).
61
DOI: 10.1021/acs.jpca.5b10650
J. Phys. Chem. A 2016, 120, 55−67
220
Article
The Journal of Physical Chemistry A
hyperfine splittings) are compared to the different calculations
that were used. The comparison revealed a good agreement
among the MP2and B2PLYP methods with the experiment,
strengthening the conclusion10 that MP2/6-311++G(d,p) is a
good level of theory for matching experimental rotational
constants with theoretical predictions.
Having made conformational assignments, the other major
experimental findings of the present work are the percentage
populations of the assigned conformations, extracted from the
relative intensities of the microwave transitions. The percent
populations of the observed conformers are summarized in
Table 9 for the four molecules in this series. The table also
includes a comparison of their calculated relative energies at the
B2PLYP-D3BJ/aug-cc-pVTZ level of theory which has proven
to be most consistent among the methods across the series
compared to experiment. In the Supporting Information (Table
S5−S8), further comparisons of the calculated relative energies
at the MP2, B3LYP-D3BJ, and M05-2X levels with the aug-ccpVTZ basis set can be found. In the Discussion that follows, we
use the B2PLYP-D3BJ calculations as the principal point of
comparison with experiment. Although the precise energy
differences vary from one method to the next, all the
dispersion-corrected functionals perform well with this large
basis set, and in general appear to be converging toward similar
relative energies. In most cases, the percent populations
correlate remarkably well with the calculated relative energies,
given the small energy differences involved.
To cite some examples, in D-allothreoninol, 1,3-diamino-2propanol, and propane-1,2,3-triamine, the measured percent
populations in the supersonic expansion follow the calculated
energy ordering and even the magnitudes of the energy
differences in most cases (Table 9). In 2-amino-1,3-propanediol, the lowest energy cyclic [I3(g−g+)] and curved-chain
[I2(g−g−)] structures carry most of the population, also as
predicted by theory. Even in glycerol, where previous studies
had not extracted relative populations quantitatively, the
qualitative population sizes match the calculated energies
quite well. At the same time, the correlation between
experimental populations and conformer energies is not perfect.
For instance, in 2-amino-1,3-propanediol, B2PLYP calculations
predict that the curved chain conformer B [I2(g−g−)] is the
global minimum, with the cyclic conformer A [I3(g−g+)]
slightly higher in energy, by 55 cm−1 . However, the
experimental percent population of A (56%) is almost twice
that of B (33%), pointing to the cycle as the global minimum.
More importantly, some structures predicted by calculation
to have low energies are “missing” in the expansion. For
instance, in D-allothreoninol, structure IV in Table 1 (II2(g−
g−) configuration) was not observed experimentally although
its energy is nearly isoenergetic with structure III [I3(g+g−)],
assigned to conformer C, which carries 8% of the observed
population. Structural relaxation29 into a lower-energy conformation through interconversion over a low-energy barrier is
suggested. This process can occur in the early stages of
supersonic expansion as a result of collisions with buffer gas.
Figure 2a shows the interconversion barrier between structure
IV and structure I listed in Table 1 for D-allothreoninol. Both
these structures belong to the II2(g−g−) structural family and
differ from each other by a rotation of the free OH group in the
curved-chain structure, effectively switching which lone pair on
the oxygen atom is used as H-bond acceptor site. Because a
low-energy interconversion barrier of 275 cm−1 is predicted, the
higher-energy conformer is able to relax into the lower-energy
Figure 2. (a) Interconversion barrier between structure IV to structure
I of D-allothreoninol at the B3PLYP-D3BJ/aug-cc-pVTZ level of
theory. (b) Interconversion barrier between structure IV to structure I
for 2-amino-1,3-propanediol at the B3PLYP-D3BJ/aug-cc-pVTZ level
of theory.
minimum during the collisional cooling in the expansion. This
process will increase the population of structure I (conformer
B) and decrease the population of structure IV, which is not
found in the expansion.
In 2-amino-1,3-propanediol, a similar low-barrier pathway
exists between structure IV and structure I (322 cm−1 barrier,
Figure 2b), or structure VIII and structure V (324 cm−1
barrier), explaining the inability to detect population in either
structure IV or VIII experimentally. Finally, in propane-1,2,3triamine (Table 7), the cyclic structure III [II3(g−g+)] is not
observed experimentally, despite being lower in energy than
two others that are observed (structures II and IV). The fact
that the cycle is not observed is due in part to its lower dipole
moment, which points almost exclusively along the c-axis;
however, given the small energy difference between structures
III and I, a low-energy pathway may be implicated as well.
Taken as a whole, these results provide further evidence that
the relative populations of conformers observed in the
expansion are determined not only by their relative stabilities
but also to some extent by the sizes of the isomerization
barriers between them.30 Where differences occur in the
ordering of populations by energy, a low-energy isomerization
pathway is likely responsible for the observed discrepancy,
draining population from a higher-energy minimum into a
lower-energy one during the collisional cooling process. Indeed,
Ruoff et al.29 have deduced on the basis of small molecule
cooling in expansions that collisional removal of a higherenergy conformer is possible when the barrier to isomerization
is less than 400 cm−1, consistent with our deductions.
62
DOI: 10.1021/acs.jpca.5b10650
J. Phys. Chem. A 2016, 120, 55−67
221
Article
The Journal of Physical Chemistry A
Figure 3. Calculated energy level diagrams for glycerol, 2-amino-1,3-propanediol, 1,3-diamino-2-propanol, and propane-1,2,3-triamine through
B2PLYP-D3BJ/aug-cc-pVTZ level of theory.
B. Preference for Cycles versus Chains as a Function
of NH2/OH Content. The small amino alcohol molecules
studied in this work all possess three adjacent H-bonding
substituents as a combination of NH2 and OH groups along the
carbon framework. As a result, one similarity in their energy
landscapes is that both cyclic and chain HB patterns are
present, and in close energetic proximity. The hydrogenbonded chains incorporate two H-bonds linking adjacent
substituents (α → β → γ or γ → β → α) along the carbon
backbone. The cycles have three hydrogen bonds, closing the
cycle by forming a hydrogen bond between the α and γ
carbons.
Figure 3 shows a set of energy level diagrams for 2-amino1,3-propanediol (2 OH/1 NH2), 1,3-diamino-2-propanol (1
OH/2 NH2), and propane-1,2,3-triamine (3 NH2), calculated
at the B2PLYP-D3BJ/aug-cc-pVTZ level of theory that best
matches with experiment. As a point of comparison, the relative
energies of different structures of glycerol are also calculated at
the same level of theory and are included in the figure. In all
four members of the series, the lowest energy cyclic and chain
structures are within around 200 cm−1 of one another. 2Amino-1,3-propanediol has the smallest gap, with chain more
stable than the cycle (ΔE = Ecycle − Echain = 55 cm−1), whereas
the largest gap is found in propane-1,2,3-triamine, with chain
preferred over cycle again (ΔE = 201 cm−1).
The calculated structures for the full set of observed
conformers in the series 2-amino-1,3-propanediol, 1,3-diamino-2-propanol, propane-1,2,3-triamine, and glycerol are shown
in Figures 4−7, respectively. The structures are divided into the
three major H-bond structural types: cycles, curved chains, and
extended chains. The observed structures of 2-amino-1,3propanediol are prototypical (Figure 4), with one cycle, two
curved chains, and one extended chain detected in the
expansion. As Table 10 summarizes, the OCCC and CCCO
dihedral angles (e.g., G, T) configure the terminal heavy atoms
in positions where they can form cyclic, curved chain, or
extended chain structures. These dihedrals are both gauche (but
of opposite sign) in the cycle, the curved chain has one gauche
and one trans, and the extended chain has two trans.
To form H-bonds, adjacent OH/NH2 functional groups
must be gauche with respect to one another, with XCCY
dihedrals of approximately ±60°, as summarized in Table 10.
The H-bonded cycles have XCCY dihedral angles that change
sign along the carbon framework [(g+g−) or (g−g+)]. When
combined with two gauche XCCC and CCCY dihedrals, the
Figure 4. Calculated structures for the full set of observed conformers
of 2-amino-1,3-propanediol with calculated HB distances (in Å) at the
B2PLYP-D3BJ/aug-cc-pVTZ level of theory. Assigned structural types
and their relative energies calculated at the same level of theory are
included. The global minimum is shaded in red, and the lowest energy
local minimum is shaded in blue.
Figure 5. Calculated structures for the full set of observed conformers
of 1,3-diamino-2-propanol with calculated HB distances (in Å) using
B2PLYP-D3BJ/aug-cc-pVTZ level of theory. Assigned structural types
and their relative energies calculated at the same level of theory are
included. The global minimum is shaded in red, and the lowest energy
local minimum is shaded in blue.
three OH/NH2 groups are positioned on the same side of the
plane of the carbon framework, where three H-bonds can be
formed. Interestingly, this combination of dihedrals built off the
triangular carbon framework produces nearest-neighbor (α−β,
63
DOI: 10.1021/acs.jpca.5b10650
J. Phys. Chem. A 2016, 120, 55−67
222
Article
The Journal of Physical Chemistry A
XCCC/CCCY dihedrals in the trans configuration, minimizing
steric strain in the carbon framework at both ends. XCCY
dihedrals of opposite sign, (g−g+) or (g+g−), produces two Hbonds of medium strength. In propane-1,2,3-triamine, the
lowest energy extended chain structure is a bifurcated chain,
with the central NH2 group acting as a double donor to the two
terminal amino groups.
Table 11 compares the key structural parameters of the
lowest energy chain and cyclic structures for each molecule in
the series of glycerol, 2-amino-1,3-propanediol, 1,3-diamino-2propanol, and propane-1,2,3-triamine. These structures are
highlighted in red (global minimum) and blue (second lowest)
in Figures 4−7. Anticipating that a short, near-linear H-bond is
best, we looked for a correlation between the XH···Y−H-bond
distance and bond angle (180° = linear). Indeed, this
correlation is evident in Figure 8, which plots the H-bond
distance versus angle of the lowest-energy cyclic and chain
structures of each molecule, grouped by H-bond type.
The OH group is the better H-bond donor, forming all seven
of the shortest H-bonds (r(OH···Y) < 2.25 Å), whereas eight of
the ten longest have the poorer NH as H-bond donor (r(NH···
Y) > 2.30 Å). The anticipation that NH2 groups would be
better acceptors than OH groups is not clearly borne out by the
present data, in part because the range of H-bond distances of a
given type is substantial. Thus, along an alkyl chain, nearestneighbor (α−β or β−γ) and next-nearest-neighbor (α−γ) Hbonds follow the order r(OH···N) ≈ r(OH···O)< r(NH···N)≈
r(NH···O).
Although the cycle seems preferable over the chain by virtue
of its extra H-bond, the triangular structure leads to H-bonds
that are quite strained, and of unequal strength. As Figure 8
shows, the H-bonded chains (red symbols) are intermediate in
length, pitting two medium-strength H-bonds against what is
often a combination of strong and weak H-bonds in the cycle.10
Interestingly, at the B2PLYP-D3BJ/aug-cc-pVTZ level of
theory, H-bonded cycles are preferred in glycerol and 1,3diamino-2-propanol, whereas a curved-chain structure is lowest
in propane-1,2,3-triamine. In 2-amino-1,3-propanediol, the
cycle and lowest-energy curved chain are close in energy,
with calculation predicting the curved-chain lower, but
experimental populations pointing toward the reverse. As the
preceding discussion has amply illustrated, several counterbalancing factors contribute to the relative stability of cycle and
chain in any one molecule, leaving no simple explanation of
these trends. It is noteworthy, however, that the extended
chains are quite high in energy in the molecules with two
terminal OH groups (glycerol and 2-amino-1,3-propanediol)
but drop in energy in the diamino and triamino analogs with
one or more NH2 group on a terminal carbon.
C. Effect of the Methyl Group on Structural
Preferences. Figure 9 illustrates the effect of lengthening
the carbon framework by addition of a methyl group at one end
of 2-amino-1,3-propanediol. Depending on the chirality of the
newly formed chiral center, either D-threoninol studied in
previous work10 or D-allothreoninol studied here is formed. The
preference for cycles over chains is retained in both molecules.
This methyl group breaks the original symmetry of 2-amino1,3-propanediol, thus generating two unique conformers on the
basis of each parent conformation of 2-amino-1,3-propanediol,
as illustrated in Figure 9.
The relative energies of the conformers in D-threoninol and
D-allothreoninol are reasonably close to those in 2-amino-1,3propanediol, indicating that the addition of the methyl group
Figure 6. Calculated structures for the full set of observed conformers
of propane-1,2,3-triamine with calculated HB distances (in Å) using
B2PLYP-D3BJ/aug-cc-pVTZ level of theory. Assigned structural types
and their relative energies calculated at the same level of theory are
included. The global minimum is shaded in red, and the lowest energy
local minimum, a bifurcated extended chain, is shaded in blue. Note
that the cyclic structure was not observed experimentally (see text for
further discussion).
Figure 7. Calculated structures for the full set of observed conformers
of glycerol with calculated HB distances (in Å) using B2PLYP-D3BJ/
aug-cc-pVTZ level of theory. Assigned structural types from ref 11 and
their relative energies calculated at the same level of theory are
included. The global minimum is shaded in red, and the lowest energy
local minimum is shaded in blue.
Table 10. Summary of the Sets of Dihedral Angles
Associated with Each of the Prototypical H-Bonded
Structural Types (X, Y = OH or NH2)
dihedral angles
structure type
cycle
curved chain
extended chain
XC(α)C(β)Y
XC(β)C(γ)Y
XCCC
CCCX
g−
g+
g−
g+
g−
g+
g+
g−
g−
g+
G+
T
G
T
T
G−
G
T
T
T
β−γ) and next-nearest-neighbor (γ−α) X···Y heavy-atom
distances in the cyclic structures that are nearly equal (2.8 ±
0.1 Å, Table 11), so that an H-bond that closes the cycle could
be similar in strength to a nearest-neighbor H-bond in a chain.
The curved chains combine XCCY dihedrals of the same sign
[(g+g+) or (g−g−)] with one gauche XCCC and one trans
CCCY dihedral, thereby directing the H-bonded chain from
above the plane of the carbon framework to in-plane, or vice
versa. This trans configuration relieves strain along the carbon
backbone relative to the cycle. The extended chains have both
64
DOI: 10.1021/acs.jpca.5b10650
J. Phys. Chem. A 2016, 120, 55−67
223
Article
The Journal of Physical Chemistry A
Table 11. Summary of Calculated XH···Y HB Distances (in Å) and Bond Angles (in deg) of the Lowest Energy Chain and Cyclic
Structures for Each Molecule in the Series Glycerol, 2-Amino-1,3-propanediol, 1,3-Diamino-2-propanol, and Propane-1,2,3triamine at the B2PLYP-D3BJ/aug-cc-pVTZ Level of Theory
cyclic
OH···O
OH···N
NH···O
NH···N
glycerol
2-amino
2.11/134
2.19/117
2.62/97
2.10/136
2.30/109
2.40/106
chain
diamino
triamino
glycerol
2-amino
diamino
2.23/115
2.33/107
2.17/117
2.53/102
triamino
2.33/108
2.21/112
2.03/123
2.63/96
2.36/118
2.29/113
2.30/123
2.54/99
2.48/104
2.38/109
does not change the energy landscape appreciably. More
striking is the fact that there are significantly fewer conformers
of D-allothreoninol detected in the expansion compared to Dthreoninol. This is in keeping with the energy level diagram
calculated for this pair of disastereomers, as presented in Figure
10. As described in previous work,10 15 conformations of Dthreoninol were predicted by calculation to reside within 500
cm−1 of the global minimum at the MP2/6-311++G(d,p) level
of theory, whereas only six conformations were found for its
allo form in the same range. As a result, the seven lowest energy
conformers (2 cycles and 5 chains) are assigned for Dthreoninol in the densely populated energy level diagram
whereas only three (2 cycles and 1 chain) are found for Dallothreoninol. Calculations at the B2PLYP-D3J aug-cc-pVTZ
confirm this conclusion (Figure 10).
The lowest cyclic and chain structures of D-threoninol and Dallothreoninol are shown in the figure, together with Newman
projections along the Cβ−Cγ bond. When viewed along the
Cβ−Cγ chemical bond, the CH2OH group attached to the Cβ
Figure 8. Calculated H-bond distance (Å) versus XH···Y bond angles
(degrees), at the B2PLYP-D3BJ/aug-cc-pVTZ level of theory, grouped
by H-bond types.
Figure 9. Structural evolution from 2-amino-1,3-propanediol (center) to D-threoninol (left) and D-allothreoninol (right). Zero-point corrected
relative energies (in cm−1) calculated at the B2PLYP-D3/aug-cc-pVTZ level of theory are included. Mirror image pairs in 2-amino-1,3-propanediol
are included in the boxes.
65
DOI: 10.1021/acs.jpca.5b10650
J. Phys. Chem. A 2016, 120, 55−67
224
The Journal of Physical Chemistry A
■
Article
AUTHOR INFORMATION
Corresponding Author
*T. S. Zwier. E-mail: zwier@purdue.edu.
Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS
■
REFERENCES
Support for this work by the National Science Foundation is
gratefully acknowledged [NSF CHE-1213289 and CHE1465028 ].
(1) Soai, K.; Niwa, S. Enantioselective Addition of Organozinc
Reagents to Aldehydes. Chem. Rev. 1992, 92, 833−856.
(2) Chennakesava Rao, K.; Arun, Y.; Easwaramoorthi, K.;
Balachandran, C.; Prakasam, T.; Eswara Yuvaraj, T.; Perumal, P. T.
Synthesis, Antimicrobial and Molecular Docking Studies of
Enantiomerically Pure N-alkylated beta-amino Alcohols from Phenylpropanolamines. Bioorg. Med. Chem. Lett. 2014, 24, 3057−63.
(3) Hazra, B. G.; Pore, V. S.; Dey, S. K.; Datta, S.; Darokar, M. P.;
Saikia, D.; Khanuja, S. P. S.; Thakur, A. P. Bile Acid Amides Derived
from Chiral Amino Alcohols: Novel Antimicrobials and Antifungals.
Bioorg. Med. Chem. Lett. 2004, 14, 773−777.
(4) Mavrov, M. V.; Simirskaya, N. I. New Synthesis of Ethambutol
and Related alpha,beta-acetylenic Amino Alcohols. Pharm. Chem. J.
2013, 46, 730−735.
(5) Zwier, T. S. Laser Spectroscopy of Jet-cooled Biomolecules and
Their Water-containing Clusters: Water Bridges and Molecular
Conformation. J. Phys. Chem. A 2001, 105, 8827−8839.
(6) Liu, K.; Brown, M. G.; Saykally, R. J. Terahertz Laser Vibrationrotation Tunneling Spectroscopy and Dipole Moment of a Cage Form
of the Water Hexamer. J. Phys. Chem. A 1997, 101, 8995−9010.
(7) Chin, W.; Piuzzi, F.; Dimicoli, I.; Mons, M. Probing the
Competition Between Secondary Structures and Local Preferences in
Gas Phase Isolated Peptide Backbones. Phys. Chem. Chem. Phys. 2006,
8, 1033−1048.
(8) Kusaka, R.; Zhang, D.; Walsh, P. S.; Gord, J. R.; Fisher, B. F.;
Gellman, S. H.; Zwier, T. S. Role of Ring-Constrained gamma-Amino
Acid Residues in alpha/gamma-Peptide Folding: Single-Conformation
UV and IR Spectroscopy. J. Phys. Chem. A 2013, 117, 10847−10862.
(9) Brown, G. G.; Dian, B. C.; Douglass, K. O.; Geyer, S. M.;
Shipman, S. T.; Pate, B. H. A Broadband Fourier Transform
Microwave Spectrometer Based on Chirped Pulse Excitation. Rev.
Sci. Instrum. 2008, 79, 053103.
(10) Vaquero-Vara, V.; Zhang, D.; Dian, B. C.; Pratt, D. W.; Zwier, T.
S. Delicate Balance of Hydrogen Bonding Forces in d-Threoninol. J.
Phys. Chem. A 2014, 118, 7267−7273.
(11) Ilyushin, V. V.; Motiyenko, R. A.; Lovas, F. J.; Plusquellic, D. F.
Microwave Spectrum of Glycerol: Observation of a Tunneling Chiral
Isomer. J. Mol. Spectrosc. 2008, 251, 129−137.
(12) Carcabal, P.; Jockusch, R. A.; Hunig, I.; Snoek, L. C.; Kroemer,
R. T.; Davis, B. G.; Gamblin, D. P.; Compagnon, I.; Oomens, J.;
Simons, J. P. Hydrogen Bonding and Cooperativity in Isolated and
Hydrated Sugars: Mannose, Galactose, Glucose, and Lactose. J. Am.
Chem. Soc. 2005, 127, 11414−11425.
(13) Mohamadi, F.; Richards, N. G. J.; Guida, W. C.; Liskamp, R.;
Lipton, M.; Caufield, C.; Chang, G.; Hendrickson, T.; Still, W. C.
Macromodel-an Integrated Software System for Modeling Organic and
Bioorganic Molecules Using Molecular Mechanics. J. Comput. Chem.
1990, 11, 440−67.
(14) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.;
Robb, M. A.; Cheeseman, J. R.; Scalmani, G.; Barone, V.; Mennucci,
B.; Petersson, G. A.; et al. Gaussian 09, revision C.01; Gaussian, Inc.:
Wallingford, CT, 2010.
(15) Alonso, J. L.; Perez, C.; Sanz, M. E.; Lopez, J. C.; Blanco, S.
Seven Conformers of l-Threonine in the Gas Phase: a LA-MB-FTMW
Study. Phys. Chem. Chem. Phys. 2009, 11, 617−627.
Figure 10. Energy level diagrams for all conformational minima of Dthreoninol and D-allothreoninol within 500 cm−1 of the global
minimum, calculated at the B2PLYP-D3/aug-cc-pVTZ level of theory.
Conformers that are assigned in the expansion are shown in red.
Structures of the lowest energy chain and cyclic conformers are also
plotted with their Newman projections.
atom is in an anti configuration relative to the CH3 group on
the back carbon in D-threoninol for both cyclic and chain
structures. However, in D-allothreoninol, with its opposite
chirality on the Cγ atom, a gauche conformation is adopted in
both cases. We infer that D-allothreoninol experiences a larger
steric hindrance brought on by the different chiral center, which
raises the energy of the structures, and leads to a more sparsely
populated energy level diagram, with fewer conformers with
measurable population, as observed.
V. CONCLUSIONS
The conformational preferences of a prototypical set of
trisubstituted aminoalcohols (2-amino-1,3-propanediol, 1,3diamino-2-propanol, and D-allothreoninol), and the triamine
analog 1,2,3-triaminopropane have been explored using a
combination of broadband microwave spectroscopy and
theoretical calculations. Rotational constants, dipole moment
directions, and nuclear quadrupolar splittings are used to make
firm assignments of a total of 15 conformations of the four
molecules. By placing NH2 and OH functional groups on
adjacent carbons along a propyl chain, one can form hydrogenbonded networks. The low-energy structures show a remarkable variety of H-bonding architectures, including cycles, curved
chains, extended chains, and bifurcated chains. These
architectures are in close energetic proximity, with subtle
differences between molecules depending on the NH2/OH
makeup and steric effects. Extending the sequence of NH2/OH
groups to four or more promises an even richer variety of
possibilities worth exploring.
■
ASSOCIATED CONTENT
S Supporting Information
*
The Supporting Information is available free of charge on the
ACS Publications website at DOI: 10.1021/acs.jpca.5b10650.
Structures, energy differences, and rotational constants
for all conformational minima (within 500 cm−1 of the
global minimum) of D-allothreoninol, 2-amino-1,3propanol, 1,3-diamino-2-propanol, and propane-1,2,3triamine; Figures S1 and Figure S2 of experimental
rotational spectra (PDF)
66
DOI: 10.1021/acs.jpca.5b10650
J. Phys. Chem. A 2016, 120, 55−67
225
Article
The Journal of Physical Chemistry A
(16) Grimme, S. Accurate Description of van der Waals Complexes
by Density Functional Theory Including Empirical Corrections. J.
Comput. Chem. 2004, 25, 1463−1473.
(17) Becke, A. D. Density-functional Thermochemistry. III. The Role
of Exact Exchange. J. Chem. Phys. 1993, 98, 5648−5652.
(18) Zhao, Y.; Truhlar, D. G. Density Functionals for Noncovalent
Interaction Energies of Biological Importance. J. Chem. Theory Comput.
2007, 3, 289−300.
(19) Grimme, S. Semiempirical Hybrid Density Functional with
Perturbative Second-order Correlation. J. Chem. Phys. 2006, 124,
034108.
(20) Grimme, S.; Ehrlich, S.; Goerigk, L. Effect of the Damping
Function in Dispersion Corrected Density Functional Theory. J.
Comput. Chem. 2011, 32, 1456−1465.
(21) Bocklitz, S.; Suhm, M. A. Constraining the Conformational
Landscape of a Polyether Building Block by Raman Jet Spectroscopy.
Z. Phys. Chem. 2015, 229, 1625−1648.
(22) Shirar, A. J.; Wilcox, D. S.; Hotopp, K. M.; Storck, G. L.; Kleiner,
I.; Dian, B. C. Impact of Molecular Conformation on Barriers to
Internal Methyl Rotation: The Rotational Spectrum of m-Methylbenzaldehyde. J. Phys. Chem. A 2010, 114, 12187−12194.
(23) Neill, J. L.; Douglass, K. O.; Pate, B. H.; Pratt, D. W. Next
Generation Techniques in the High Resolution Spectroscopy of
Biologically Relevant Molecules. Phys. Chem. Chem. Phys. 2011, 13,
7253−7262.
(24) Watson, J. K. G. Vibrational Spectra and Structure; Durig, J. R.,
Ed.; Elsevier: New York/Amsterdam, The Netherlands, 1977; Vol. 6,
pp 1−89.
(25) Plusquellic, D. F. JB95 Spectral Fitting Program; NIST:
Gaithersburg, MD. http://www.nist.gov/pml/electromagnetics/
grp05/jb95.cfm.
(26) Pickett, H. M. SPFIT/SPCAT; NASA: Pasadena, CA. http://
spec.jpl.nasa.gov.
(27) Ray, B. S. Ü ber die Eigenwerte des Asymmetrischen Kreisels.
Eur. Phys. J. A 1932, 78, 74−91.
(28) Blanco, S.; Sanz, M. E.; Lopez, J. C.; Alonso, J. L. Revealing the
Multiple Structures of Serine. Proc. Natl. Acad. Sci. U. S. A. 2007, 104,
20183−20188.
(29) Ruoff, R. S.; Klots, T. D.; Emilsson, T.; Gutowsky, H. S.
Relaxation of Conformers and Isomers in Seeded Supersonic Jets of
Inert Gases. J. Chem. Phys. 1990, 93, 3142−50.
(30) Florio, G. M.; Christie, R. A.; Jordan, K. D.; Zwier, T. S.
Conformational Preferences of Jet-cooled Melatonin: Probing transand cis-amide Regions of the Potential Energy Surface. J. Am. Chem.
Soc. 2002, 124, 10236−10247.
67
DOI: 10.1021/acs.jpca.5b10650
J. Phys. Chem. A 2016, 120, 55−67
Документ
Категория
Без категории
Просмотров
0
Размер файла
8 322 Кб
Теги
sdewsdweddes
1/--страниц
Пожаловаться на содержимое документа