close

Вход

Забыли?

вход по аккаунту

?

Modern Analysis.

код для вставкиСкачать
The catalysts consist of octacarbonyldicobalt or cobalt
salts and compounds such as ketones or nitriles, which
can form complexes with C O ~ ( C O )e~. g, . with the
following composition (191:
The reaction proceeds a t 40-100 “C with a dilactone
yield of 15-40 % based on acetylene; coumarin is
formed as a by-product. Substituted acetylenes give
correspondingly substituted dilactones. It has also
been suggested [201 that the reaction be carried out at
50-3000 atm and 60-175 “C.
4. Closing Remarks
Some of the cyclic compounds described may prove
interesting as starting materials and intermediates.
The condensation of trimethylhydroquinone with iso1191 W. Hieber, German Pat. 953796 (1954), German Pat.
1083264 (1958), BASF.
[20] J . C. Suuer, German Pat. 1054086 (1955), Du Pont.
phytol, for example, yields DL-cr-tocopherol (vitamin
E). A cheap hydroquinone can also be used for the
preparation of anthraquinone by oxidation to p benzoquinone, Diels-Alder reaction of the latter in a
molar ratio of 1 : 2 with butadiene 1211, and air oxidation of the resulting octahydroanthraquinone in
ethanolic potassium hydroxide solution. The yield
of anthraquinone based on benzoquinone is about
85 %. The synthesis of unsaturated dilactones may
open up a new route to suberic acid if acetylene is
available at a favorable price, since such dilactones
can be readily hydrogenated with platinum catalysts in
polar solvents to give suberic acid.
The cyclization of alkynes shows that if these compounds can be obtained at a reasonable price, they
can be used for the production of cheap cyclic intermediates. The economic aspect becomes even more
attractive if even cheaper materials such as carbon
monoxide and water are used as well as acetylenes.
Received: July 1 1 , 1969
[A 719 IEI
German version: Angew. Chem. 81. 717 (1969)
Translated by Express Translation Service, London
[21] I. G . Furbenindustrie, German Pat. 494433 (1930).
Modern Analysis[lI
By H. Kienitz[*I
Dedicated to Professor B. Timm on the occasion of his 60th birthday
The important advances being made in modern analytical methods are indicative of the
fundamental changes that are occurring in the theory and practice of “analytical chemistry”. “Information optimization” demands a new approach in teaching and research,
and calls for the integration of chemistry with other scientific and technical disciplines.
1. Introduction
Analysis may be defined as the gaining of information
about the qualitative and quantitative composition
and about the steric structure of substances and chemical species. The modem paths to this type of information differ in method (chemical and physical processes), in execution (accuracy, speed, economy, and
possibility of automation), and with regard to acquisition, critical examination, and representation of the
results. The analyst’s knowledge, discretion, and ex[*] Prof. Dr. H. Kienitz
Forschung
Badische Anilin- u. Soda-Fabrik AG
67 Ludwigshafen (Germany)
[l] Based on a lecture read at the General Meeting of the Gesellschaft Deutscher Chemiker at Hamburg on September 18, 1969.
Angew. Chem. internat. Edit. J Vol. 8 (1969)
1 No. 10
perience are applied to the task of choosing the optimum path, which best suits the manifold factors involved and the analytical problems, and provides an
adequate and correct result.
A survey of the problems and sectors of analysis is
given in Table 1.
Although much scientific progress and industrial development has only been made possible by the results
of analytical investigations, analysis has unfortunately
still not received full recognition as a subject in its own
right. The manufacture of inorganic and organic
products can no longer be carried out without constant analytical supervision. Trace and micro methods
of analysis have found a place in every field of applied
natural science, and have helped us to understand
processes that were formerly beyond the reach of
quantitative examination.
733
Table 1.
Problems of analysis:
I . Qualitative analysis of a mixture of substances (for elements, ions, atomic groupings,
Functional groups, or components)
2. Quantitative analysis of parts
or of the whole of a mixture
of substances
3. Qualitative analysis of a chemical species (single molecule,
macromolecule, unit cell)
4. Quantitative analysis of a
chemical species (number and
steric arrangement of the
atoms and atomic groupings)
Sectors of analysis:
1. Sampling and sample preparation
2. Analytical procedure
A. Methods
a) chemical
b) physical, with alteration
of the substance
c) physical, without alteration of the substance
B. Conditions, imposed
a) by the analytical problem
b) by the substance to be
investigated (quantity,
nature, and state)
C. Requirements
a) accurate, cornprehensive, critical
b) fast
c) economical
d) automatic
3. Critical evaluation of the analytical result. Drawing of conclusions from this result
I n this article we shall consider some modern developments selected from the many aspects of the analyst’s
work.
2. Analysis of Inorganic Solids
The sampling of inorganic solids presents interesting
and often difficult problems. Sampling is the first step
in the sequence sampling - sample preparation analysis - data processing. A sample taken from a
quantity of material that is often many orders of
magnitude greater should be representative.
Whether or not sampling is satisfactory depends on
the requirements; thus, the requirements for a preliminary analysis (accuracy 510%) are less rigorous
than those for an analysis for commercial purposes
(accuracy k l or even 0.2%). Sampling errors are in
general fully reflected in the final result. The number
and size of the samples to be taken are determined
statistically by the homogeneity of the material, the
uniformity of its density (separation of the components), and its particle size. The probability of inclusion in the sample should be uniform for the
material as a whole. This is possible, in general, only
if the material is in motion. Mechanized sampling
therefore always uses a moving stream of material,
such as the material transported by a conveyor belt,
the sample being taken on the basis of quantity or of
time. This is followed by gradual mechanical reduction
of the size of the sample, which is also comminuted in
stages. Large-scale units capable of sampling with a
reproducibility of a few tenths of one percent have
been constructed in recent years for the metal and
potash industries.
The influence of inhomogeneities in a solid sample on
the analytical result increases as the region covered by
the analytical process decreases. Thus inhomogeneities
are of practically no importance in neutron activation
analysis, whereas they have an appreciable effect in
734
X-ray fluorescence analysis and a strong effect in mass
and emission spectrometry.
The detection limits and application ranges of some
chemical and physical methods are shown in Figure 1.
10-9
1~-i2
I
I
,
10-3
,*-6
I
/
I
I
Gravimetry,Titrimetry
,
t
I
I
...
Electrochemical methods
Coiorimetry
Spot test
Flame photometry
Atomic absorption spectrometry
Emission speclrometry
c
__
c
-._
-c
-_
X-ray fiuorescence spectrometry
t
Mass spectrometry
_..
_c
___
Neutron activation analysis
/
I
I
,
I
I D - L I ~ ~ ~ Kib-1
-~
10-310-210-’ 1 l0ppm
10-’1 lOppb
Fig. 1 . Detection limits (t)
and application ranges (- - - -)
tant analytical methods.
i
1’0
I~OOL
of impor-
The element specificity of these methods is generally
satisfactory, though masking can occur (e.g. in the
analysis of metals) in both emission and mass spectrometry. The problem of masking can often be overcome by combination with chemical methods.
The detection limits for atomic absorption spectroscopy have
recently been improved by two orders of magnitude by
evaporation of the samples in graphite cells, with the result
that they are now close to the limits for emission and mass
spectrometry.
Wider use of neutron activation analysis has been hindered
by the lack of neutron sources with adequate neutron fluxes
(apart from nuclear reactors) and of detectors with narrow
energy bands. The situation has been improved by neutron
sources that produce neutron fluxes of up to 1010 njcm2 by
the T(d,n)He reaction; fast neutrons are particularly desirable for the analysis of elements of t h e first short period,
which are only weakly activated by thermal neutrons in the
reactor because of their small cross section. However, activation with fast neutrons has the disadvantages that the detection limits for activation in reactors (neutron flux 1014
n/cm2) cannot be reached for heavier elements, and that very
short-lived nuclides are formed in most cases. The detection
limits for the light elements nitrogen, oxygen, fluorine, and
phosphorus, on the other hand, are between 10 and 100 pg
for a neutron flux of l o 9 njcm2.
The resolving power of scintillation detectors with NaI crystals and photomultipliers, such as are used for the measurement of the radiation from the activated elements, is only
about one tenth of that of the semiconductor detectors that
have recently come into use. Thus the relative energy halfwidth of the 6QCoyline at 1.33 MeV is 5.9% (80keV)
when measured with an NaI(T1) crystal, but only 0.45 %
(6 keV) with a Ge(Li) detector (Fig. 2).
In neutron activation analysis, in which several nuclides are
activated during the irradiation, many individual nuclides
can be both identified and used for very selective quantitative determinations on the basis of the typical maxima of
their gamma spectra and the rate of decay of their radiation,
corresponding to the half lives. Wet separations of the nuclides formed on irradiation can thus be largely avoided.
Special mention should be made of the extreme difficulty of the preparation of calibration standards for
trace determinations. Conflicting results, often differing by several orders of magnitude, depending on the
method (particularly in the ppm range), are not unAngew. Chem. internat. Edit. / Vol. 8 (1969) No. I0
1173
1333
J
.1
'x
,'
no separation
Separation by phases
Separation methods
eg Flotatlon
Air classification
Magnetic separation
Mechanical selection
Anodic oxidation
u
'Obi
loz
1
I
0
1
L
I
1200
1000
N o ~ o ~ e ~ i z a t i ~ n Polished section and
thin transparent slices
Identitication
Anaiysis by scanning
X-ray diffraction
tight and electron
X-ray fluorescence
microscopic investigation
Electron soft X-ray
Electron probe X-ray
Macro analyzer
Micro analyzer
Thermal analysis
Ion microprobe mass analyzer
Laser micro emission spectral
analysis
IR absorption
Electron diltraclion
Identitication
Chemical analysis
Physical analysis
Atomic absorption
Spectral analysis
X ray fluorescence
Electron soft X-ray
Macro analyzer
X-ray diffraction
Polarization optical investigation
Mineralogical investigation
rn
Fig. 3.
1LOO
Analysis of solids.
Quantum energy ikeV1 -3
Fig. 2. Comparison of the energy spectrum of 60Co recorded with a
germanium detector (top) and with a scintillation detector using an NaI
crystal (bottom).
common. Absolute values might be obtained by mass
spectrometry if the ion sources had fixed ion yields for
known ionization probabilities without interelement
effects; the results are, however, at best only semiquantitative. Reliable calibration standards are therefore essential in trace analyses, whether by mass
spectrometry or by any other method. Zone melting
can be used for the purification and subsequent doping
of samples whose melting points are not too high
(m800 to 1000°C). Since it is very difficult to avoid
inhomogeneities and contamination during the mixing,
fusing, or sintering of samples, the only way of obtaining quantitative standards for trace concentrations is
by chemical analysis, though some uncertainty remains even here because of possible contamination
during concentration.
For compact inhomogeneous samples, where we are
interested not in an average analysis but in the investigation of individual phases and their distribution
in an area or volume, the examination can be carried
o u t after mechanical or chemical separation of the
phases, or otherwise by point analysis of small surface
or volume elements (on ground surfaces or their sections) without phase separation.
The principal methods for the analysis of solids are
surveyed in Figure 3.
Point examination of solid samples can be achieved
with the aid of electron, ion, or laser beams. A region
about 30 pm in diameter is evaporated from the sample
e.g. with a laser beam to form a crater. The vapor is
ionized with auxiliary electrodes and analyzed by
emission or mass spectrometry. The electron beam of
a "microprobe" has a diameter of w2 pm. It produces
element-specific secondary X radiation at the surface
of the sample, and the spectrum and intensity of this
radiation are determined with proportional counter
tubes. The scanning of the surface may also be carried
Angew. Chem. internat. Edii.
1 Vol. 8 (I969) f No. I0
out with an ion beam, the secondary ions produced
being led to a small mass spectrometer. The electron
and ion microprobes give only semiquantitative
analytical ififormation for local areas; the distribution
of the phases in a surface region can, however, be
found very accurately by measurement of the relative
areas.
With the macroprobe, the area irradiated with electrons (about 2cm2) is several orders of magnitude
greater than with the microprobe. The X-rays produced at the surface of the sample are again used for
qualitative and quantitative analysis. This is a good
method for the determination of the light elements
boron to sulfur. Macroprobe analysis is more sensitive
to surface impurities than X-ray fluoresccnce analysis;
moreover, its use is confined to electrically conducting
samples (or samples with a conducting coating).
hv I t
tlhv
elastic scattering
inelastic scattering
\\\ '>-
X-ray tluorescence. hv
=K-t
Auger effect, E,,,=K-ZL
Fig. 4.
Interaction of X-rays (or of electrons) with matter.
735
X-ray fluorescence, microprobe, and macroprobe
analyses are based on atomic processes in which
electrons or X-ray quanta interact with the electrons
of an atom (Fig. 4).
The photoeffect involves the ejection of electrons f r o m
the inner shells by the incident X-ray quanta. The
bonding state of the emitted electron can be deduced
from its kinetic energy and the energy of the absorbed
quantum. The vacancy left by the ejected electron is
then filled by outer electrons, the energy liberated
being released as an X-ray quantum (X-ray fluorescence) or in the form of a secondary emitted electron
(Auger electron). For the ejection of inner electrons,
the X-ray quanta may be replaced by electrons of
corresponding energy.
The possibilities offered by spectroscopy based on the
emission of Auger electrons will be merely mentioned
here. This method can be used for the analysis of surfaces (the topmost one or two layers of atoms), and
has a high detection sensitivity, particularly for light
elements. It is possible in this way to detect partial
coverages of 0.1 (and in special cases down to 0.01)
layer of atoms. The method is therefore very promising
for use in problems of catalysis, adsorption, corrosion,
and surface diffusion.
The photoelectric effect, in which electrons are released
from the inner (K or L) shells of a n atom by X-ray
quanta ( A k a or MgK, radiation) has recently acquired growing analytical importance. The emitted
electrons are separated in a magnetic field according
to their energy distribution (ESCA, Electron Spectroscopy for Chemical Analysis). The element-specific
energy distribution of these secondary electrons is
shown in Figure 5.
I
LI
Be
B
C
100
200
300
N
0
F
I
m
LOO
500
Bond energy lev1
600
700
Fig. 5 . Photoelectron signal of the elements of the second period
(K shell).
Since the kinetic energy of the emitted electrons varies
by several electron volts (1 eV A 29 kcal/mole), according to the bonding state of the atoms, conclusions
may be drawn, for example, regarding the bonding
state of the two differently bound sulfur atoms in thiosulfate or of the differently bound iodine atoms in
iodide or iodate. This method, which possesses considerable potential for the analysis of organic compounds, will be dealt with again in Section 4.
Development trends in the analysis of inorganic substances, taken as a whole, point to the need for the
determination of traces, non-destructive methods of
analysis, and methods that allow at least semiquanti-
736
tative analysis of the smallest possible areas or volumes. There is also an increasing demand for methods
that are capable of being automated.
3. Chromatography
F. F. Runge, who discovered aniline in coal tar, separated the
components of dyes into brightly colored ring-shaped zones
by placing a drop of a solution in the middle of a sheet of
blotting paper. He mentioned this process as early as 1822,
in his dissertation, and in 1850 he published a book “Zur
Farbenchemie, Musterbilder fur Freunde des Schonen und
zum Gebrauch fur Maler” (Color chemistry, patterns for
lovers of beauty and for artists). Though he had discovered
the basis of paper chromatography, and even introduced the
“development” of colorless into colored zones, he had no
thought of a n anarytical method. C. F. Schonbein, F. Goeppelsroeder, and E. Fischer also used this method during the
last century for the separation of ions of inorganic salts,
as well as of dyes, and it was finally introduced into analysis
by M . Tswett as “chromatography” (derived from the Greek
~b v.pi$or, the color). H e resolved the dye leaf-green into
green and yellow zones in separating columns packed with
an adsorbent.
Separations of this type (on “two-dimensional” thin layers in
thin layer or paper chromatography, and on “one-dimensional” columns in liquid or gas chromatography) are still
referred to as “chromatography”, though they are no longer
used only for the separation of colored components. Chromatography has grown into an essential analytical tool,
particularly gas chromatography, which can be traced back
to A . T. James and A . J. M. Martin, and even to A . Eucken
and E. Cremer.
Efforts are being made in all chromatographic methods
to improve the separation and the detection. In the
two-dimensional thin layer and paper chromatography, the separation and the specificity for certain
groups of substances are improved by the successive
use of diffusion effects in two directions or by the
simultaneous use of different physical effects, as in
electrophoresis. The detection is often based on specific color reactions. Thin layer chromatography, and
to some extent also paper chromatography, have recently been developed into quantitative methods, and
variances of less than 3 % have been obtained for repeated determinations. Refinements are continuously
being made to the technique of these methods in order
to increase their basically rather limited separatory
efficiency.
In the one-dimensional separation methods of column
and capillary (liquid and gas) chromatography, the
detection of the eluted substances is almost ideally
solved for gas chromatography. The detectors are
characterized by very high sensitivity over a very wide
dynamic range, and Combination of this with substance-specific methods (particularly with mass spectrometry, as a combination of two trace and micro
methods) is particularly effective. Though liquid
chromatography is also capable of giving very good
separations, the problem of the detection of the
eluates is still unsolved. Physical effects in the liquid
phase cannot be expected t o give detector signals of
the same order of magnitude as those obtained in the
gas phase; transfer detectors, which transfer the liquid
Angew. Chem. internat. Edit. ] Val. 8 (1969) ] No. 10
eluate of liquid chromatography into a flame ionization detector, side-step the problem instead of providing a true solution.
Though the substance specificity of the flame ionization detector is poor, the value of gas chromatography is that it can tell us that a substance originally
thought to be pure actually contains at least as many
impurities as there are additional signals in the chromatogram, apart from the principal component. It is
now known that even the “purest substances” are
generally accompanied by traces of numerous other
substances. The uncontrolled use of analytical methods
in which a substance is examined “en bloc” is thus no
longer justifiable, and a chromatographic analysis
should always be carried out either beforehand, simultaneously, or afterwards to enable the interpretation
of the results to be completed. This is true even of the
so-called highly specific methods, apart from the
purely element-specific ones.
Whereas further appreciable progress in gas chromatography is to be expected only in fast and high-efficiency
separations and in the ultra-trace and maxirnumselectivity fields, the development of classical column
chromatography into modern liquid chromatography
in columns has just begun.
In gas chromatography, the techniques used in
present-day practice are still lagging appreciably
behind the latest developments (Fig. 6).
t
P
The present stage of development can be summed up
as follows. While retaining its extremely broad scope,
gas chromatography has become the most accurate
quantitative method for the analysis of volatile or decomposable organic substances. The substance to be
detected enters the detector at a low concentration,
i.e. in the ideal state from the physical and measurement points of view; the matrix, ie. the carrier gas, is
extremely pure. Variances of down to 0.01 % are quite
possible. Gas chromatography allows analyses to be
carried out over the concentration range of from 10-5
to 100%; in special cases, volatile stable substances in
gases can be detected down to a concentration of
10-10%. One sample can be resolved into as many as
400 (and in special cases up to 1000) chemical species.
Quantitative determinations take about 1 to 20 min
and require about 10-3 g of sample, though 10-6 to
10-8 g would be sufficient if it were possible to prepare
and transport such small quantities of material. More
than 100000 chromatographs were in use in laboratories and factories in 1968; only a few percent of
these are used in process chromatography, though gas
chromatography is excellently suited for automation.
The existing practical achievements are not yet enough. It is
expensive to obtain information that can be evaluated directly
(on-line) with a computer. Mechanization has not yet gone
far enough; though semi-automatic gas chromatographs are
already commercially available, their operation is unchecked
and uncritical. The separating power, sensitivity, and speed in
routine analyses are generally lower by more than two orders
of magnitude than they need be, and in some cases than the
levels required on economy grounds. For example, up to 3000
separation plates per instrument are often accepted where
up to 100000are needed to detect the components in question.
The sensitivity of ordinary commercial instruments is so low
that it is just possible to recognize traces at a concentration
of l O - 4 % (ppm), though traces in the lO-7% (ppb) range
should be detectable, and these are the very components that
are often found to be of immense importance both in chemical
and in biochemical processes.
Gas chromatography is today making a decisive
contribution to the utiIization of the capabilities of
structure-specific and structure-identifying methods,
and the combination of chromatographic methods is
becoming increasingly important (Fig. 7).
I
1952 1956 1960 1964 1968 1972
10.6 10.~-
1000 -
f
- 10’8oo-g
z
I
p
10-9-
c
600-2
-
z lp+
2
‘D
d LOO -
5
c
200 -
1
)0-131
10.1‘-
Angew. Chern. internat. Edit. J Vol. 8 (1969) 1 No. I0
Fig. 7. Possibilities for the combination of gas chromatography with
other analytical methods.
Little can be said yet about the state of modern liquid
chromatography, except that the expected broad
development has probably started this year. The first
instruments allowing fast column chromatography
with high entry pressures and high separation efficiencies with programmed gradient elution have
737
Table 2.
Survey of chromatographic separation methods
Mobile phase
Stationary
phase
Temperature
( “C)
Liquid [a]
Geometry
Name of method
Up to about
500 [dl
Tube dia. 0.01 to
10 mm, length a
few cm to km
Gas partition
chromatography
-210 to 1700
Tube dia. 0.1 to
10 mm, length a
few cm t o 0.1 km
Gas adsorption
chromatography
Gas
Solid b 1
Gas
Solid Ic1
-210 to 1700
As above
Gas diffusion
chromatography
Vapor (in the
supercritical state)
One of the
above three
[el
As above
Dense gas
chromatography
Liquid
Liquid [a]
[d, fl, depends
on P
Tube dia. 0.1 to
10 mm, length a
few cm to 0.1 km
Column
chromatography
with secondary
names indicating
the nature of the
separation process,
of the stationary
phase, or of the
material or
physical
programming
0.1 to 10
(2000)
[d, fl, depends
on T
Liquid
Solid [bl
As above
As above
As above
As above
Liquid
Solid [c]
As above
As above
On lavers
Thin layer
chromatography
On paper
Paper
chromatography
In a thin film of
the stationary
phase on an inert
rough-surfaced
support
Thin film
chromatography
Applications and
capabilities
General:
gas chromatography.
For volatile or
decomposable
substances that can be
converted into volatile
or decomposable
derivatives.
Qualitative and
above all quantitative
analysis of all
substances that are
soluble in the mobile
phase in seconds to
minutes
General:
liquid chromatography.
For all substances that
are soluble in the
mobile phase.
Qualitative and
quantitative analysis
in seconds to hours
[a1 On porous support or wall. Ibl Absorbent. Icl Microporous, inert (e.g. glass, metal, ceramics, molecular sieve, clathrate former).
[dl Possibly programmed.
[el Depends on the supercritical data for the mobile phase.
[f] Depends on the separating system.
already been constructed; however, they are still much
more complicated than modern gas chromatographs.
In a typical example, almost 8000 separation plates are
obtained per meter of column length (diameter 2 to
3 mm) at a rate of 40 separation plates per second.
An intermediate position between gas chromatography
and liquid chromatography is occupied by “dense gas
chromatography”, which is chromatography in the
supercritical fluid state, in which elution is carried out
at carrier gas pressures of up to 2000 atm. Intermolecular forces in polymers having molecular weights of
up to several hundred thousand can be broken down,
thus permitting separations far superior to those
achievable by gel permeation chromatography. During
this kind of chromatography in the supercritical region
of the carrier gas, a temperature of 40 “C is sufficient
to volatilize substances which would hardly have been
thought to behave in such a manner, e.g. carbohydrates
such as ribose, arabinose, xylose, and glucose at pressures of about 1400 atm in carbon dioxide.
All chromatographic methods require at least one
mobile and one stationary phase.The separation of the
substances in question is based on differences in the
migration rate, solubility, complex formation, adsorption, and includability, and the pressure, temperature, and geometry of the system play a very
738
important part. All possible intermediate stages and
combinations are possible between “solution”, “chemical bonding”, “adsorption”, and “sieve effect”, and
in general, none of these principles acts alone. It is
therefore problematic to use names that involve the
mechanism of action. Modern chromatographic methods are not only a decisive part of modern analysis;
they also, per se, provide a constant feedback to the
theory and practice of physicochemistry and chemistry.
4. Structural Analysis of Organic Compounds
It has been found convenient to classify organic compounds
according to the manner in which the C atoms are linked.
They are described according to their carbon skeleton, according to whether the C chains are linear or branched, according to whether they are single carbocyclic or heterocyclic rings or ring systems, and according to the size of the
rings and the bonding of the members. Functional groups are
inserted in or attached to the carbon skeleton or the heterocyclic nucleus and give rise to the great variety and complexity of organic compounds. The ordering principle of
“skeleton” and “functional group” is also followed in the
nomenclature. It would be particularly useful if the information and results of a structural analysis also fell into these
categories.
Angew. Chem. internat. Edit. [ Vol. 8 (1969)/ No. I0
Table 3.
Comparison of some methods of structural analysis.
Information
Sample
Method
~~
Structural elements and ring systems of atomic
skeletons consisting of carbon and hetero atoms,
functional groups, hydrogen bonds, symmetries,
isomerism
Infrared spectroscopy (microwave spectroscopy)
Solid, liquid, solution, gas or vapor
Absorption of infrared light (or microwaves) due to
excitation of vibrations and internal rotations
of molecules
1-10 mg, in nlicrocells from 10 ug
Raman spectroscopy
Solid, liquid, solution
Scattered radiation whose frequency is the
difference between the frequency of the incident
visible light and the frequency of the excited
molecular vibrations
50 mg - 10 g. with laser
excitation from 1 mg
As infrared spectroscopy; selection rules of the
normal vibrations make the two methods
complementary; valuable for symmetrical
molecules
Electronic spectroscopy
(Solid), liquid, gas, vapor,
dilute solution
0.1-10mg
Double bonds and single electrons, particularly
conjugated systems, e.g. aromatic compounds,
heterocyclic systems with aromatic character
Nuclear magnetic resonance
Solution
Absorption of radio waves due to resonance of
nuclear spin moments in a magnetic field
5-100 mg
Structural elements and functional groups on the
basis of the environment of protons or of nuclei that
exhibit nuclear magnetic resonance (e.g. 13C, 19F,
29Si, "P); determination of the hydrogen (proton)
content of chemical compounds and groups
Electron spin resonance
Solid, liquid, solution
Detection and determination of free radicals,
their structure and electron distribution
Absorption of radio waves due to resonance of
electron spin moments in a magnetic field
1-200 mg
5-100 mg
Optical rotation, circular dichroism
Liquid, solution
Rotation of the plane of polarized light, difference
in absorption of right-handed and left-handed
polarized light as a function of wavelength
1-200 mg
Mass spectrometry
Gases, liquids, and solids that can
reach a vapor pressure of 10-5 t o n
a t 300 "C without decomposing
Absorption of ultraviolet and visible light according
to the electron levels in molecules
Mass and relative abundance of molecular ions
and fragment ions formed on ionization of molecules
usually by electron bombardment or strong electric
fields
Relative and absolute configuration of centers of
asymmetry, conformation of asymmetric molecules
Molecular weight, elemental composition,
detection of hetero atoms on the basis of the natural
isotope abundances, structural analysis by specific
breakdown of molecules on electron bombardment
4 0 . 1 mg
~~~
X-ray diffraction, neutron diffraction
Single crystal
interference of X-ray or neutron beams as a result
of diffraction by the electron clouds o r nuclei of the
atoms of the oriented molecules in the crystal
1-10 mg and 0.1-1
respectively
g
Determination of the complete molecular structure,
steric arrangement of the atoms (distances, angles);
hydrogen atoms can also be located by neutron
diffraction
Some methods of structural analysis are listed in
Table 3. Only those most commonly used at present
will be considered in this section.
of four rings, the two acetoxy groups, and the p-bromobenzoyloxy group, which, with its heavy bromine atom, was
introduced to facilitate the processing of the X-ray diffraction data (fixing the phase angle).
The desire to identify directly the structural units and
functional groups in a molecule is satisfied to some
extent by X-ray analysis (Table 4).
The method can in principle be used for all compounds that exist as single crystals, whether inorganic
or organic and with ionic or molecular lattices. An
X-ray structure analysis always gives a picture of the
complete electron density distribution in a unit cell of
a crystal. It is consequently necessary to include all the
atoms of the unit cell in the calculations, so that one
cannot, for example, confine one's attention to a
specific functional group and ignore the rest of the
molecule. The method is a particularly important tool
for the determination of partial conformations and
absolute configurations, e.g. of natural products, or
for the investigation of solid-state reactions of inorganic compounds for topochemical analysis by
determination of intermolecular and intramolecular
distances in the crystal.
Owing to the advances in automatic data acquisition
and processing, it is no longer necessary to use a derivative containing a heavy atom, provided that the molecule consists of not more than 50 atoms (apart from
hydrogen atoms). Measurements on many isomorph-
Table 4. Advantages and disadvantages of X-ray analysis,
Advantages
Disadvantages
Information about the entire
intramolecular and intermolecular geometry
No elucidation of partial structures
possible; it is always necessary to
consider all the atoms in the molecule
Substance (-0.1 mg) must exist as a
single crystal (-0.1 mm3); time
required for a structural analysis
4-12 weeks
Substance is still intact after
the analysis
Fieure 8 shows the result of an investigation by Hoppe,
Hecker, er al. [21 on a neophorbol derivative; the perspective
I
representation of the centers of gravity of the atoms (to the
right of the electron density diagram) clearly shows a system
[2] W. Hoppe, F. Brand], I. Strell, M . Rohrl, J. Gassmann,
E. Hecker, H . Barrsch, G . Kreibich, and Ch. Y. Szczepanski,
Angew. Chem. 79, 824 (1967); Angew. Chem. internat. Edit. 6 ,
809 (1967).
Angew. Chem. internat. Edit.
Vol. 8 (1969) No. I0
739
x
Fig. 8.
Fourier analysis of neophorbol 13.20-diacetate 3-p-bromobenzoate
ous derivatives are necessary in the case of large molecules; the structure of these molecules can be conclusively established if there are enough derivatives,
personnel, and computer time available. The time required for a medium-large molecule containing 30
atoms apart from hydrogen is 4 to 12 weeks; the time
increases roughly linearly with the number of nonhydrogen atoms up to 50 atoms, but much more
rapidly beyond 50.
As can be seen from Table 3, IR, Raman, NMR, and
mass spectrometry, in complete contrast with X-ray
analysis, give only structural elements and environmental and symmetry relations of atomic groupings in
molecules, as well as their fragmentation on electron
and photon bombardment and in very strong electric
fields.
The manner in which the results of the various methods
can be combined to find the structure of a molecule is
shown below; the information provided by the various
methods is indicated, and their effectiveness is compared with that of other methods.
In connection with automatic data processing and
combined interpretation, it is necessary to realize that
there are basically two types of analytical information,
i.e. information from which conclusions can be drawn
directly and information from which structural elements can be recognized with the aid of empirical
tables.
IR spectroscopy yields only information of the second
type. The precise information obtained depends on
the region of the spectrum. Very characteristic and
relatively reliable information about functional groups
(such as OH, NH, C r N , C=O, C=C, C=N) is found
between about 3700 and 1550cm-1, while types of
substitution on aromatic rings and heavy atom substitution can be recognized from the wagging vibrations between about 900 and 400 cm-1. In the region
between these two, which corresponds mainly to the
skeletal vibrations of the molecule, considerable frequency changes occur as a result of changes in struc-
740
ture, and this region cannot be used directly as a
source of structural information. This region, however,
is very substance-specific (fingerprint region), and is
therefore particularly useful for the identification of
organic compounds.
For example, the following indirect information can be obtained from the IR spectrum of ethyl trans-cinnamate:
1
number
wave
fcm-1)
685/715
1
770
I
1271
1495
1580
1712
Assignment
}
I
C6HSR
a,Bunsaturated ester (non-aromatic)
C=C-C-0-R
C6H6
C = O (acid or ketone or a,&unsaturated ester)
We must now ask ourselves whether the information
can be improved by the use of more sophisticated
apparatus, e.g. by a n increase in resolution. The
answer is negative as far as the spectra of liquids and
also of most solids are concerned, since the limit of
resolution in these two cases is the natural band width.
For gases, on the other hand, increased resolving
power can lead to resolution of broad bands into
narrower component bands (rotational structure),
though this fact is not generally used in normal determinations of the structures of organic compounds.
The greater accuracy of the position of the absorption
bands, however, can in some cases be of additional
value (groups with a C=C function between 1650 and
1800 cm-1).
Raman and I R spectroscopy are based on the internal
vibrations of the molecules, the so-called normal
vibrations, which appear in accordance with selection
rules determined by the physical principles of the
Angew. Chem. infernat. Edit. J Vol. 8 (1969)/ No. I0
methods; thus a change in the dipole moment during
vibrations leads to IR absorption bands, while a
change in polarizability gives Raman lines.
Raman spectra, like I R spectra, give only empirical
information. The two methods are complementary,
since according to the selection rules, which are based
on molecular symmetry, vibrations may be e.g. IRactive and forbidden in the Raman spectrum or vice
versa, or they may be allowed or forbidden in both.
A recent important development in Raman spectroscopy is the use of lasers (He/Ne) as Raman light
sources. Smaller quantities [a few pl (mm3)] can be
used for measurements of this type, even when the
substance is colored or slightly turbid; crystalline
powders can also be used, with the result that the
lattice vibrations of solids can now be readily measured. New possibilities for the determination of the
crystallinity of high polymers have appeared, and it
seems to be possible to determine the length of polymer chains. Finally mention should be made of the
ease of measurement of the degree of polarization of
Raman lines, and hence of determining the symmetry
of normal vibrations, which is also of value in structural analysis, by excitation with lasers (Fig. 9).
L59
Direct stafements:
-
Peak of highest mass
number at m / e 176
Even molecular mass
number
Mol. wt.: 176
+ No N atom or even number
of N atoms
Absence of isotope
multiplets
M + / M +-1- 1 = 100/10.5
+ No halogen atoms
+ 10-11 C atoms
Empirical s f a f e m e n f s :
Key fragments mle
+ CH,
15
29
77
Peak differences m / c 176
131
77
+ CzHs
--f
C6H5 (monosubstituted
aromatic compound)
0
> 4 5 + C2HzO-
>
>
28 +
!I
-C-
26 +
-CH=CH-
+
C6Hs
The value and the direct applicability of the information obtained from isotope multiplets is demonstrated
in Figure 11. Such information shows conclusively the
nature and number of certain hetero atoms.
I
x + 4 +8
Fig. 9.
Polarization measurements in the Raman spectrum of CCI4.
Unlike infrared spectroscopy, mass spectroscopy gives
information that permits direct conclusions as well as
providing empirical information (cf. Fig. 10).
11
103
1-1
Fig. 10.
mle
-
Mass spectrum of ethyl cinnamate.
Angew. Chem. internat. Edit.
/ Vol. 8 (1969) / No. 10
x
ti,
I,
x+2
rn
Fig. 1 1 .
r+2
r+2
x+2
x + 4
Characteristic halogen and sulfur isotope multiplets.
To see how the ions other than the molecular ion are
interpreted in the spectrum of ethyl cinnamate (Fig.lO),
let us consider briefly the fragmentation processes that
take place in the ion source of the mass spectrometer.
These processes depend both on the energy balance of
the bond cleavages involved in the fragmentation and
above all on the stability of the positive and neutral
particles formed. Thus the fragmentation of ethyl741
benzene, for example, is primarily controlled by the
formation of the stable tropylium ion:
73
80
70
:[
LO
m/e = 9 1
M.150
lI=021
30 -
The fragmentation paths and mechanisms are therefore molecule-specific rather than group-specific,
since the formation of certain fragment ions depends
energetically on the competition of different atomic
groupings. In addition to the fragmentation of single
bonds, rearrangements frequently also occur, subject
to certain steric conditions; though these further
complicate the fragmentation mechanisms, they are
readily recognizable in many instances, and are then
very specific. An example is the “McLafferty rearrangement”:
29 l+3
86
151
LO
29
30
20 10 -
43
i
,
I,
L
-I
61
74
133
Ill1
IBI
II
II
Fig. 12. Mass spectra of D-ribose a) with electron bombardment ionization [3] and b) with field ionization 141.
When X = 0 and Y = OH, the ion of mass 60, which
is characteristic of acids with more than three C atoms,
is formed.
A rough interpretation of a mass spectrum can be
made with the aid of empirically compiled tables of
characteristic “key fragments” (e.g. ethyl cinnamate:
mass numbers 15, 29, 77 correspond to CH3, CzH5,
and C6H5) and typical mass differences (45 & CzH50;
characteristic of ethyl esters).
T
.
Specific syntheses of derivatives of the substance under
examination and partial chemical degradation can
often be of great value in structural determinations.
Strongly polar substances that cannot be evaporated
without decomposing are converted into derivatives
that can be vaporized without decomposing (silyl
ethers, trifluoroacetates, etc.).
1
2%
I
11.
99 113 127 141 155 169 183 197 211 225
J, /I
.I , l
.,
.
100
150
200
L,
1
250
254
100 -
bl
50 -
It is sometimes difficult to recognize the molecular
peak. For aliphatic alcohols, ethers, and carboxylic
acids, as well as for branched aliphatic hydrocarbons,
the intensity of this peak is usually less than 2 % of the
sum of the intensities of all the other signals, whereas
aromatic and alicyclic compounds give very intense
molecular peaks. If, however, the ionization method
is changed from electron bombardment to field ionization, or if the ionization voltage in electron bombardment ionization is reduced, the peak of the molecular
ion becomes quite predominant, despite the lower overall intensity of the spectrum (Figs. 12 and 13).
achieved by conventional electron bombardment ion
sources 153.
Here again we are faced with the question whether the
information value of mass spectrometry could be im-
Wider use of field ionization mass spectrometry has
been hindered in the past by the low overall intensity
of the spectra. With carbon whiskers as electrodes in
a field emission ion source, however, it has recently
become possible to obtain sensitivities as high as those
[3] K. Biemann: Mass Spectrometry. McGraw-Hill, London
1962, p. 350.
[4] H. D . Beckey in W. L. Mead: Advances in Mass Spectrometry. Vol. 2. Institute of Petroleum, London 1966, p. 48.
[5] H. D . Beckey, Lecture read to the GDCh Section “Analytische Chemie” at Freiburg in April 1969.
742
[
57 71 85 99 113 127 1L1 155 169 183 197 211 225
i ./
i ii /I K I1 i l 1 il i
50
100
150
200
m l
rnle
-
I
250
Fig. 13. Mass spectra of octadecane with ionization energies of a) 70 eV
and b) 17 eV.
Angew. Chem. internat. Edit.
/ Vol. 8 (1969)1 No. I0
proved e.g. by an increase in the resolving power.
Even instruments having an average resolution m ] A m
of 2000-3000 resolve a signal of mass number 28 due
to Nz and CO. A tenfold increase in resolution allows
the separation of lines with higher mass numbers
(Fig. 14), so that the elemental compositions of fragment ions and molecular ions can now also be given.
The usual accuracy of mass determinations is about
When arranged schematically in the form of an ''element map", they can be interpreted more easily than
when they are simply listed according to their mass
numbers.
L
C,H,N'
hH7'
In principle, the elemental compositions can be calculated
and an element map can be constructed with the aid of
tables; since it is convenient in any case to use computers in
the processing of high-resolution mass spectra, this task
may also be incorporated in the program.
To illustrate the information obtainable from a nuclear magnetic resonance spectrum, let us again take
the example of ethyl trans-cinnamate (Fig. 15).
Mass 103
Mass 57
18722.141
Mass 73
IC?!
Fig. 14. Ion pairs separated by high-resolution mass spectrometry.
+0.0025 to &0.003 mass units. An unknown compound e.g. having a molecular weight of 587.3444
i 0.0029 could have any of the formulas C28H4gN3010,
C2&45N706, C33H49N08, and C34H45N504. An elemental analysis giving the results C: 70.65, N: 4.90,
and 0: 16.65%, with a cryoscopically determined
molecular weight of 290 i 5, would give the empirical
formulas C17H23NO3, C17H25NO3, and C17H21NO;
a mass-spectrometrically determined molecular weight
of 289 would rule out the second and third of these
formulas.
Combination with elemental analysis is particularly
important if a molecule also contains other hetero
atoms, such as F, S, and P, that are either isotopically
pure or have no characteristic isotope ratios.
Styrene, CsH8, is practically indistinguishable from
pentanethiol, C5HlISH, (AM = 0.0034), and benzene,
cannot be distinguished from fluoropropanol,
C3H7OF ( A M = 0.0021). The same is true for the
molecular ions and the fragments:
80
70
lel Id!
Chemical
KO
Multiplicity
shift
[PPml
The advantage of high-resolution mass spectrometry,
despite certain limitations, can be seen e.g. in the mass
spectrum of ethyl butyrate; instead of mass numbers,
elemental compositions are now given for the peaks of
the spectrum:
Angew. Chem. internat. Edii.
/ Vol. 8 (1969)J No. I0
60
30
Couphng
:onstant IHzI
Intensity
ssignment
~-
--
3 ITripletI
I
3
@CHz-
4 23
4iQuartetl
I
2
CH,@O-CH
K 40
PlDoubiet 1
16
1
7 35
nlMultiplet
7 70
2 [Doublet1
5
16
Oppm
10
la1
20
Ibl
120
Fig. 15.
(AM = 0.0013)
50
If1
6
0
1
IH-NMR spectrum of ethyl trans-cinnamate.
The area under the multiplet having a given shift corresponds
to the number of protons, and the ratios of the areas are thus
equal to the ratios of the numbers of protons in various environments. The chemical shifts and coupling constants for
various atomic groupings can be found in empirical tables.
The triplet a and the quartet b in the spectrum correspond to
an ethyl group, and the two doublets c and e to a -CH=CHgroup. The two hydrogen atoms could also be cis, but this
would be shown by the coupling constant: cis w 8 Hz,
trans w 16 Hz. The higher-order multiplet d, in which five
non-equivalent protons couple with one another, is more
complicated; it can be interpreted (cf. Fig. 16), though the
interpretation is not necessary for the structural analysis,
particularly in this case.
743
TMS-SB
457075 Hz
TMS-SB
~ 3 e . m ~ ~
artho - Protons
TMS-SB
418.760 Hz
mefa -and para- Protons
I
I
I]
”’
1
Spectral parameters for phenylacetylene
[solution in CCLL1
jP72z161
Fig. 16. Experimental and calculated NMR spectrum of the
0,
3bGC-E
4
m,
5
and p protons of phenylacetylene
i
40
30
LO
40
40
30 ppm I61
0
Fig. 17. Spin-spin coupling for the A2X2 system.
An example of the transition from lower-order to
higher-order NMR multiplets is shown in Figure 17.
Neighboring CH2 groups carry groups that can induce different intramolecular fields. The multiplet
structure is characterized by the coupling constant J,
the shift difference A s v , and their ratio JIAsv, i.e. the
coupling factor. In the case of the A2X2 system, i.e.
the two CH2 groups, we should expect two triplets in
the first order (Fig. 17, left). In the spectra from left to
right, not only do the signals move closer together
(A~Y
becomes smaller), but their structure also becomes more complicated, until they finally degenerate
into a singlet. The structure of the multiplet is determined by the coupling factor which is (0.25 for the
first order of the multiplet. If, on the other hand, it is
>0.25, complicated multiplets of higher order are obtained.
As in IR and mass spectroscopy, it is reasonable to
hope that the structure of the NMR multiplet might
become clearer if the “resolution” were increased.
This would require a decrease in the signal width,
744
which would in turn be possible only by an increase in
the constancy of the magnetic fields (fundamentally an
increase in the constancy of the temperature) and of
the measuring frequency. One order of magnitude
(from v / A v = 2 x 108 to 2 x 109) might be gained in
this way, but a natural limit is imposed by the relaxation of the internal motion of atomic groups in the
molecules. It is commonly thought that the “resolution” can be improved by increasing the magnetic
fields, e.g. from 14100 gauss and a measuring frequency of 60 MHz currently used in routine work to
23500 gauss and 100 MHz; instruments working with
51 700 gauss and 220 MHz have been available for
two years. A further increase in the field strength and
measuring frequency still seems possible.
In connection with this development of nuclear magnetic resonance instruments, one speaks of an increase in the ‘‘resolution” of the spectra with the
magnetic field strength. This is not, however, true of
the resolution as usually defined in spectroscopy. On
the other hand, the chemical shift increases very appreciably with increasing magnetic field strength. In
the parlance of optical spectroscopy, which should be
Angew. Chem. internat. Edit.
1 Vol. 8
(1969) No. 10
used to describe this effect, the angular dispersion of a
220 MHz instrument is greater by a factor of 3.7 than
that of a 60 MHz instrument. The multiplet groups
move apart as the field strength and measuring frequency are increased, but the multiplet structure simultaneously becomes simpler. This is understandable if
one considers that the coupling constants are independent of the field strength, whereas the shift difference increases with increasing field strength. The
coupling factor J/AGv, which determines the structure of the multiplicity, accordingly becomes smaller,
and the multiplets become simpler.
This effect is shown in Figure 18. The same spectrum (60
MHz) of 2-chloroethanol is shown twice on the left, and the
spectrum of the phenyl ether is shown (top right) for comparison. A greater difference in the shielding constants leads
to a greater shift difference, and hence t o a smaller coupling
factor, so that the structure of the multiplet becomes simpler.
In the lower right-hand diagram, which shows the 220 MHz
spectrum of 2-chloroethanol, the external field is greater and
the same effect is thus observed.
16)22181 C
60 MH,
60 M H z
60MHZ
2ZOMHz
I
W
O
H
CO-iH
Fig. 18. 1H-NMR spectra of 2-chloroethanol and its phenyl ether at
various measuring frequencies.
To sum up, therefore, we can say that an increase in
the field strength and measuring frequency moves the
multiplets apart, and so disentangles them and makes
them simpler; however, this is accompanied by a loss
of information. It should also be mentioned that the
Angew. Chem. internat. Edit.
1 Vol. 8
(1969) No. I0
signal-to-noise ratio is appreciably improved (by a
factor of 3-5) by the change from 60 to 220 MHz.
It is convenient to leave the data processing and all
the steps in the evaluation of information from IR,
Raman, NMR, and mass spectroscopy to a computer,
which can also carry out searches and comparisons in
stored empirical tables of the structural elements, and
can take over the functions of memory, learning, and
searching:
1. Automatic data collection and conversion into a
form suitable for interpretation (e.g. determination of
mass numbers and elemental compositions in mass
spectra, separation of overlapping bands);
2. Comparison with catalogs of spectra;
3. Input of reliable preliminary information;
4. Output of structural data in the form of partial
structures and the arrangement of structural parts;
5. Comparison of the results obtained by different
methods.
As a result of these combined integrated interpretations, the computer yields structural elements, partial
structures, o r complete structures, depending on the
difficulty of the problem. It will provide self-consistent
proposed structures as well as structures that still
involve some uncertainty, and it will also list any conflicting results and negative findings. Decisions between various possibilities and the selection of the
eventually proposed structure (possibly after further
investigations based on intermediate results) remain
the work of the chemist. The most reliable confirmation of an analysis is generally a total synthesis.
The substance whose structure is to be analyzed must
be substantially pure. Separation methods, particularly
chromatographic, are of great value in this connection,
both preparatively and analytically as a purity check.
The combination of separation and determination
methods has already been described for the case of
gas chromatography and mass spectrometry.
A method that should be of great importance for
structural analyses in the future is one reported two
years ago by Siegbahn161, who named it ESCA. This
method makes use of the photo-ionization of atoms by
relatively soft X-ray beams (cf. Section 2). Siegbahn
used a magnetic separation system in his electronic
spectrometer. The Varian Company, which undertook
the development of instrumentation for the method
(also known as induced electron emission spectroscopy, IEES), introduced electrostatic separation with
a multichannel store; it is possible in this way to
resolve spectra with a sharpness of separation of
1eV. Figures 19 and 20 show the corresponding spectra
of acetone and ethyl trifluoroacetate, in which substitution effects on the carbon are clearly shown by a
change in its bond energy ( E B ) (according to Sirgbuhn). Manufacturers have recently announced the
production of instruments, which no longer separate
the photoelectrons in a magnetic field but by a time-offlight principle in a n electric field at a resolutions o f
161 K . Siegbahn et a/., Nova Acta Regiae Societatis Scientiarum
Upsaliensis, Ser. IV, 20, 1 (1967).
745
1 eV; these instruments will probably ensure that this
new method will soon take its place alongside other
methods of molecular spectroscopy. Moreover, in this
case, a n integrating interpretation is also possible with
the aid of a computer.
!
i
L
ANALOGUE
SIGNAL
Central processing unit
,
+ + +
Execution of logical steps
f
Data
Combination of
reduction logical operations
Comparison Selection
Result
I
,
950
955
Emetle’JI
Fig. 19.
1195
1200
YesiNo Greater,
Valuation and criticism
less
ESCA spectrum of acetone.
Fig. 21.
Use of a computer for analytical problems.
0
more values per second by an analog-to-digitalconverter, which passes the resulting digital values to
the core memory of the computer. The data can be
further processed directly in time-sharing operation
during the data collection. Where the information
flow is high, the incorporation of a temporary store in
the peripheral equipment is often advisable in order to
relieve the central unit. The information can then be
recalled later, or may be passed on, after packing, to
a larger processor. Using data and programs from
storage units, the computer can compare, select, condense, and relate the data by logical operations, in
such a way as to obtain an intermediate or final result.
I
1190
1195
Elmet lev1
295
-
+
290
285
E,leVI
Fig. 20. ESCA spectrum of ethyl trifluoroacetate.
X-ray structure analysis provides the structure of the
entire molecule, though the demands on time and work
are still relatively severe. The molecular spectroscopy
methods can never yield direct information about a
carbon skeleton in structural analyses; they cannot
show directly the presence of rings, chains, or branching. Recognition of typical ring vibrations is possible
only by modern laser Raman spectroscopy employing
polarization measurements, so that, with the exception of the aromatic six-membered ring, methods
of molecular spectroscopy can recognize functional
groups and the environments of atoms, as well as
isolated steric arrangements, which can often be combined t o give partial or total structures only on the
basis of a well founded chemical knowledge.
5 . Data Processing in Analysis
Some possibilities offered by the use of electronic data
processing equipment in analysis are shown in Figure
21. Analog signals are accepted at a rate of 10000 or
746
An obvious step is to employ calculating machines,
from small desk models to large computers, for pure
calculating tasks in the processing of analytical results.
With skilled programming, not only of analytical parameters, but also of standardization factors, names of
substances, operating terms, etc., a small computer
can prepare a complete analytical account, thus easing
considerably secretarial work in the laboratory and
largely eliminating errors involved in calculation,
transfer of data, and writing; it can, moreover, carry
out comparisons with theoretical values, summation
checks, calculation of standard deviations or errors in
multicomponent systems. However, such a machine
cannot be expected to perform special tasks corresponding to the capability of a full-scale computer.
In the examples mentioned so far, the computer is
used merely as a calculating machine whose program
includes a comparison with a reference value. However, it becomes much more directly involved in the
analysis if it is also used for data acquisition and
processing. As a calculator it simply derives an analytical result on the basis of existing experimental
values, whereas in data acquisition and processing
the computer itself obtains values that can be used in
analysis from a large number of individual values,
and manipulates these further to obtain an intermediate or final result.
This can be illustrated for the case of a titration. The end
point of a titration is usually determined visually on the basis
Angew. Chem. infernat. Edit.
1 Vol. 8 (1969) 1 No. 10
of a color change; as soon as this color change occurs, the
tap of a buret is closed and the value is read off. In electrochemical titration, the titration curve (volume of titrant used
against the potential difference) can be recorded, and the end
point of titration is then determined graphically as the point
of inflection of the curve. It is well known that the points of
inflection are indistinct, particularly when there are several
potential steps, so that it is difficult to determine these points
accurately. If the potential difference as a function of time
is fed into the computer as a large number of individualmeasurements (e.g. every 1&15 ms), care being taken to ensure
that the addition of titrant is exactly linear with time (the
volume per unit time should also be accurately measurable),
the computer can find the point of inflection by differentiation
(formation of the first, and if necessary also the second,
derivative) with much greater accuracy than is possible by
hand.
The computer can also take over t h e data processing in the
evaluation of gas chromatograms, particularIy where signals
overlap or where the base line is not constant. The earlier
planimetric method of manual evaluation, which was subject to considerable errors (1-3%),
has been steadily replaced by electric integrators, which give very good reproducibility ( ~ 0 . %).
3 However, these offer only an incomplete soIution to the problems of base-line drift and the
separation of overlapping signals. A computer, on the other
hand, can fully correct for the base-line drift by constructing
a tangent at the beginning and end of a signal, or of a group
of overlapping signals, and determining the area above the
tangent. To separate overlapping signals, an approximate
gaussian curve is determined from the signals, in a chromatogram, and for the overlapping signals, the two gaussian
curves are sought whose superposition best reproduces the
measured signal. The simple procedure of dividing superimposed signals into area elements by vertical lines is sufficiently accurate in many cases if t h e assignment of the area
elements is correct.
All these computer evaluations depend on the recording of
the true experimental values. This is seriously affected by
noise in the measurement signals and their disturbance by
isolated signals due to sudden voltage or current surges, and
considerable expense is involved in t h e elimination of these
interfering factors.
As in gas chromatography, a computer can also be used in
mass spectrometry for data acquisition, processing, and reduction and for obtaining analytical results.
The question of economy is also important. As is
shown in Figure 22, the times taken for the various
steps of the analysis differ considerably.
If the preparation of the sample and the measurement
take a long time in relation to the evahation, the use
Timerequired i n %
........ ......
of a computer is not normally justified. It is advisable
in such cases to modify the procedure, rationalize or
mechanize individual steps, or improve the measuring
method. If the number of samples for routine examination is very large while the amount of calculation
is small, a programmable desk calculator or digital
data collection, e.g. with a tape punch for off-line
evaluation on a larger computer, may be advantageous.
The direct connection of smaller or larger data processing units is recommended whenever the collection and
evaluation of the experimental values accounts for a
medium or major part of the total work of analysis.
In physical methods it is usually the collection of the
experimental values that involves a great deal of
manual work, which is simple, but requires care and
concentration (gas chromatography, mass spectrometry); the use of medium sized computer systems is
profitable in such cases.
If the times for measurement and for evaluation are
roughly the same without a computer, the measuring
process becomes rate-determining with automatic data
processing. It is then desirable to speed up the measuring process. An example of this is gas chromatography,
where the operating speed could be considerably increased, but where this has been of no interest until
now because of the long times taken for the evaluation.
This is a case in which the use of computers has had
repercussions on the method itself.
I n view of its cost, the fullest possible use should be
made of a data processing system. This condition is
satisfied in particular in routine operation, e.g. in
round-the-clock industrial analyses, particularly when
the data How is matched to the capacity of the computer by connection of a large number of analytical
instruments. Extremely high data flows from individual measuring instruments also justify the use of a
computer for these instruments alone, provided that
they are used often enough (e.g. in high-resolution
mass spectrometry). These computers, which are intended for a single field of use, and which operate for
long periods with the same program, can be profitably
used with the minimum equipment for this purpose,
particularly if they also take over the duties of supervision (regulation) of the measuring instrument and
optimization of the measured values (improvement of
the signal-to-noise ratio) (e.g. in nuclear magnetic
resonance).
~
I
Registration
t+3
*- 60rn'n
Caiculation
1
I
Fig. 22.
Time taken for the various steps in analysis.
Angew. Chem. internat. Edit. J Vol. 8 (1969) J No. 10
20-70
The preparation and control of the measuring processes can be carried out by a small computer, The experimental data are reduced by this computer, and
may be passed to a large computer for further processing. In X-ray structure analysis, for example, the intensities of several thousand reflections from one crystal
are measured with a diffractometer, which is controlled
by a small computer. This computer optimizes the
measuring process for each reflection, corrects the experimental values, and passes them in a reduced form
to a large computer. An attempt is now made, with the
aid of a suitable program, t o assign phases to the
measured structure amplitudes, in order to obtain a
747
model of the structure by a three-dimensional Fourier
analysis. This model is then refined by the computer
by optimum matching with the experimental data.
The control of the diffractometer by the computer
may be compared t o automatic process control in
industry, and is probably the first process t o reach this
level of perfection in the field of measurement.
This opens u p whole new vistas for t h e role of computers i n analytical work, though the applications a t
present a r e still rather trivial. Its role as a pure calculator is widened, a n d it also assumes control functions
over the analytical instruments. I n emission spectroscopy, for example, one measurement may be repeated
several times; the mean of the individual results is
used to calculate the standard deviation, which is then
compared with the expected standard deviation of the
analytical method. If the result is greater than the expected value, the computer decides t o repeat the analytical measurement. The new result, together with the
standard deviation, provides the basis of a criticism
that either the sample may be inhomogeneous or the
analytical method is unsatisfactory. The analyst himself must now draw his conclusions, but the basis of
his decisions is provided by the computer.
Other duties of the computer are distinguishing between
systematic and random errors a n d verifying the correctness of results. It is not difficult for the computer
t o store a large number of repeated measurements,
find their statistical mean, and check their statistical
distribution. If the distribution differs from the gaussian or other suitable distribution, the errors in the
measuring process are not only purely random ones.
The error of the method is found by repeating the
entire measuring process several times, and occasional
measurements o n calibration samples confirm o r
refute the correctness of the measuring process. All the
intermediate calculations involved in this procedure
a r e carried out by the computer, which provides the
analyst with the quantities required as a basis for
decisions, or may even itself indicate the satisfactoriness or uncertainty of analytical findings.
I t is now beyond all doubt that the resulting speed of
the whole measurement a n d evaluation operation and
the increase in the accuracy and correctness of the
results that is generally possible are leading t o a new
era in analytical work, with the scientific and economic
consequences that follow from it. I n recognition of the
economic importance of analysis, it has become our
duty t o encourage and force through the decisions that
are necessary t o our entry into this new era.
6. Automation in and with Analytical Chemistry
The possibility of influencing an analytical process by
decisions suggests a comparison with control and
regulating functions in chemical processes (Fig. 23).
Three stages of process control are known in industrial production, i.e. control (manual control), regulation (semi-
748
Starting materials,
Utilities
Process and controi
Product and
quality control
Manual control
Fully automatic
control
Fig. 23.
Control, regulation, and automation. Stages in the unloading
of the human operator in chemistry (symbolic: two-stage process).
automatic control), and automation (fully automatic control). In manual control, information provided by measuring
instruments on the basis of the instantaneous state of the
process is translated into commands by human agents. The
process is thus under human supervision, and the quality of
the product and the economy of production depend on the
efficiency and availability of labor.
In semiautomatic process control, the same information as is
used in manual control is fed into data converters. These are
regulating instruments with no automatic checking or critical
functions, which convert the information supplied into commands in accordance with a fixed program or simply in
such a way as to maintain a predetermined state. An important characteristic of a good regulator is the ability to
convert the information into commands in such a way that
the desired action occurs as quickly and as accurately as
possible. Data feedback allows optimum matching of the
commands to the prevailing state. However, all these features
cannot be compared to the specific characteristics shown by a
human agent controlling a process. Though the regulator
does the mechanical operations faster, better, and more continuously than a human operator, it lacks the checking and
decision-making functions and the ability to adapt to different states. As the demands made by process control become
more stringent, therefore, the usefulness of semiautomatic
control diminishes. The stage of full automation is then
necessary.
Full automation means a high degree of replacement of
human activities by machines, and includes checking,
decision-making, and adaptation functions. Checks can be
made only by comparison of data. Decision-making and
adaptation require the observation and evaluation of a
change in data with time. Tasks of this nature are within the
capabilities of computers. Automation in the chemical industry is thus characterized by the use of a computer, which
compares data, follows their trend, and uses this as a basis for
commands.
I t can be seen from Figure 24 that the various steps
leading from sampling, via the measuring process, t o
the analytical result can be considered in the same
scheme as was used for process control (cf. Figure 23).
Manual analysis, in which a sample is subjected to transformation and measurement, has reached only the semiautomatic stage even when complex instruments are used for the
production and collection of data, since the state of the samAngew. Chem. internat. Edit. 1 Vol. 8 (1969) No. 10
Initial sample,
analytlcal process
accessory materials
Product information
+-Lt7+yb
Sample preparation
Cahbralicn
Control
n
Automated analysis
Fig. 24. Stages in automatic analysis: manual analysis, fully mechanized
analysis (with checking, correction, and evaluation by human operators),
and automated analysis.
ple is (unconsciously) checked by a human operator and
corrected if necessary, and since the human operator must
also check and correct the information produced, e.g. by unconscious comparison or with the aid of a calibration analysis. Transformation of the results of the measurements is
sometimes necessary. It is always necessary to check in some
way whether the information is “correct enough”.
Analytical measurements with “automatic” instruments are
comparable to semiautomatic process control. These instruments should, however, be described as automatic only if the
criteria of automation (checking, correction, decisionmaking, and adapting) are satisfied. An automatic vending
machine is much more automatic than an “automatic”
titrating instrument, since it checks a coin inserted and decides whether to deliver the goods or not (e.g. if foreign
money or a button has been used instead of the correct coin).
Though an automatic vending machine can be deceived, or
may refuse to deliver the goods even when a gold coin has
been inserted, this error characteristic does not alter the
difference between automation, as a high level of checked,
corrected, and adapted mechanization, and mechanjzation
without checking and decision-making functions.
The instruments usually referred to as automatic analysers,
as well as “automatic” combustion equipment for element
analyses and “automatic” titrating instruments, are instruments that function in accordance with a control program,
but which contain no idea of automation. Phrases like “automatic laboratory” refer at best to the evaluation, and never
to the analytical process. An automatic apparatus should
be 171 an assembly of mechanism and instruments that forms
an informationally closed system. The quality of being “informationally closed” is essential to the process taking place
in a truly automatic apparatus. One can also say that an
automatic apparatus must carry out this process in accordance with the “input/output” principle, which states that
material and/or informative input quantities are processed
internally with no help from external agencies, and that the
results are again definjte material and/or informative output quantities. The process is often specified expressly as in
automatic “computer”. Unlike in a machine or instrument,
the individual steps in the operation of an automatic apparatus need not be predictable either in their nature or in
their duration, but are varied within preset limits by internal regulating and control processes. For this purpose control signals and/or response signals are used to vary the
actuating variables of the mechanisms and to match them
[ 7 ] Z. anal. Chem. 237, 81 (1968).
Angew. Chem. internat. Edit. / Vol. 8 (1969)J No. I0
to one another. The characteristics of the object to be acted
upon must also be determined directly and introduced into
the data cycle.
Where the input quantities are fixed, we speak of programmed
automatic systems. Units for the elimination of variable input quantities in the form of random variables, deviations,
or perturbing quantities are also known as automatic regulators or controllers. However, many units correspond to both
concepts.
For true automation, the individual steps in analytical
chemistry (sampling, measuring, and data processing)
must be automated, i.e. provided with automatic
checking, correction, and adaptation to changed
conditions.
These requirements are already largely satisfied in data
processing, and perfectly practicable proposals have
been made for automatic analytical instruments. In
sampling, on the other hand, the requirements can
scarcely be met even in the simplest cases. Systems of
two or more phases and components with different
densities and volatilities raise problems that have not
even been adequately thought out, far less solved.
Requirements that must be satisfied by analytical instruments for automatic process control are given
below. It should be noted that a n analytical method,
however important it may be to the process, is incomplete unless it incorporates a checking system corresponding to the human criticism that always intervenes between sampling and evaluation of the measurements in analysis in the laboratory.
This checking system involves the storage, comparison, and
checking of changes and rates of change within the analytical
system. The calibration functions and the reference standards
must be supervised and automatically corrected in case of
deviation, as must the input of the correct measurement
data.
The internal checking is a special characteristic, and one that
is obviously necessary for a production process. The instrument itself decides whether the output analysis result is
probably correct or whether it should be rejected. The process control learns from this decision. An automatic system
must therefore constantly make yes-or-no decisions that are
often difficult for a human worker.
These checking and decision-making functions are natural in
automatic computers. The calculating programs always
contain a loop that checks whether the calculation and logic
steps carried out are correct. It is also usual with punched
tapes to carry out this check by cross footing of the punched
holes. Two internal checks must be distinguished for the
analytical instrument:
1. the automatic checking of the function of the instrument
itself and
2. the insertion of analysis of a calibration mixture at
regular intervals and the detection of excessively large nonsystematic errors.
The demands on the sampling system become increasingly
stringent with increasing measuring sensitivity of the analytical method. The sampling reliability decides whether the
(now very high) outlay on analysis and measuring equipment
is justified and whether the conclusions drawn from analytical
results, which are sometimes of great economic importance,
are correct. It is known that the sampling, like the analysis
itself, contains a systematic error that depends on the procedure and a non-systematic error that can be influenced only
by the method. Whereas the magnitude (more accurately the
standard deviation) of the non-systematic error can be determined by a sufficiently large number of repetitions of the
measurements, the discovery and elimination of systematic
749
sampling errors is sometimes very difficult. The automation
of sampling is more difficult than the automation of the
analytical measuring process.
7. Conclusion
The elimination of signals due to temporary strong fluctuations and to periodic fluctuations in the process is another testing problem in the automatic control of plants.
-
It should be clear from the developments outlined
above that modern analysis has little in common with
the analytical chemistry of two or three generations
ago, which is still being taught in technical colleges.
7
Phvsical
Process
control:
.
.
The modern chemist must therefore bring himself up
ideal case
to date, and acquaint himself with this important
Correction
measured value
branch of knowledge, which links together chemistry,
physics, mathematics, biology, and medicine, as well as
>Classical chemical process
many branches of technology. Analysis can now claim
+
control,
I
human supervision
}
to be an independent field of instruction and research.
Correction
Sampling
Evaluation
Isolation
Analytical measurement
analytical process control
I
Fig. 25.
process.
Analytical data for the control or regulation of a chemical
Finally, Figure 25 shows the ideal combination of an
automated analysis for the regulation of a chemical
process.
Analysts with a good training in inorganic and organic
chemistry, with wide knowledge and experience in
analytical methods as a whole, and with a deeper
knowledge of some special fields are required today in
research and industry. They should be familiar with
mathematics, physical measuring techniques, electronics, data processing, information theory, and logic,
t o enable them t o take part in the development of a
rapidly growing new field, i.e. the logical, exhaustive,
and correct evaluation of analytical information from
material processes, and their critical appraisal and
reduction to the essential facts: the optimization of
information.
The author is grateful to the staff of the PhysikalischAnalytisches Laboratorium der BASF, and particularly
Dr. Giinzfer, Dr. Kaiser, and Dr. Seidl, as well as Dr.
Briigel and Dr. Witte, for stimulating discussions.
Received: June 16, 1969
[A 722 IEI
German version: Angew. Chem. 81, 723 (1969)
Translated by Express Translation Service, London
C 0M M U N I CAT10 N S
Malonodithioamide from Dimethylaminomalonodinitrile and Hydrogen Sulfide
onodithioamide ( 3 ) . Both products react with hydrogen
sulfide in the above mentioned solvents to give the compound ( 4 ) .
By H. Eilingsfeld and M. Patsch [*I
Dedicated to Professor B. Timm on the occasion of his 60th
birthday
N--C bonds in quaternary ammonium salts can be cleaved
reductively, e.g. electrolytically[I] or with sodium amalgam [*I.
We have observed that even hydrogen sulfide is able to
cleave an N-C bond, with reduction of the carbon released,
in the case of certain ammonium salts and tertiary amines.
This reaction of dimethylaminomalonodinitrile ( I ) with
hydrogen sulfide leads to the formation of malonodithioamide ( 4 ) in very good yields.
The reaction takes place particularly readily at room temperature in pyridine and in pyridine/glacial acetic acid and
ethanol/tertiary amine mixtures. Since malonodinitrile in
alcohol adds one hydrogen sulfide to only one nitrile group
with formation of cyanothioacetamide 131, while malonodithioamide is formed from dimethylaminomalonodinitrile
under the same reaction conditions, it is assumed that the
cleavage of the dimethylamino group takes place after the
HzS addition. Under certain reaction conditions (e.g. in
benzene at 0 "C)we have been able to isolate dimethylaminomalononitrilemonothioamide (2) and dimethylaminomal-
750
The ammonium salt (5)[41, which can be prepared by Nalkylation of ( I ) , also reacts with hydrogen sulfide to give(4).
In order to explain the course of the reaction ( 3 ) + ( 4 ) we
assume that a nucleophilic substitution by SH at the nitrogen
takes place, similar to that postulated in the alkaline hydrolysis of chloroamines [51. However, results of our previous
investigations do not rule out a free radical mechanism.
The thiohydroxylamine derivative which one would expect
to be formed along with malonothioamide in the cleavage of
(3j is not stable under the reaction conditions, as the behavior of sulfenamides shows; disulfide and amine are formed.
Angew. Chem. internat. Edit. / Vol. 8 (1969) J No. 10
Документ
Категория
Без категории
Просмотров
3
Размер файла
1 730 Кб
Теги
modern, analysis
1/--страниц
Пожаловаться на содержимое документа