close

Вход

Забыли?

вход по аккаунту

?

Computer-Aided Analysis of Organic Mass Spectra.

код для вставкиСкачать
purchased, but also on the later organization and manner
of operation (see Table 3).
In many cases existing computing centers hinder the purchase of a large mixed system. However, it is often overlooked that many of the existing computer installations,
even without scientific or chemical applications, are running
near capacity or will be within a few years. Moreover, at
least in the industrial area, the computing centers are
normally oriented toward commercial applications. Scientific and technical applications, for which other types of
computers and other forms of organization of computing
operation are more suited, often appear as intruders.
5.7. Future Developments
Large mixed systems are desirable not only because of the
requirements of data acquisition and processing for the
directly connected analytical instruments : As previously
mentioned, the early developments of computer programs
for the support of the preparative chemist, the implementation of documentation systems, as well as continuously
improved methods of interpretation in spectroscopy stimulate the purchase of large computers for the field of chemical
research. Moreover, since many documentation systems
and complicated evaluation procedures utilize the data
sets from on-line systems, the most intimate possible
union-which
the mixed system optimally offers-of
on-line and off-line systems is desirable.
Because of the relatively low level of exploitation of a large
system by on-line tasks alone, such a system becomes
economical only with the inclusion of several intensive
off-line computations. A simiiar degree of computer usage
could be attained by using the computer for training
students.
Received: November 18,1971 [A 868 IE]
German version: Angew. Chem. 84, 371 (1972)
Translated by Allan B. Wilson, Miilheimpuhr
[I] E . J . Corey and W T: Wipke, Science 166, 178 (1969).
[2] E. Ziegler, Pittsburgh Conference on Analytical Chemistry and
Applied Spectroscopy, Cleveland 1969 ; E . Ziegler, D. Henneberg, and
G . Schomburg: Pittsburgh Conference on Analytical Chemistry and
Applied Spectroscopy, Cleveland 1970.
[3] E . Ziegler, D.Henneberg, and G . Schomburg, Anal. Chem. 42, 51A
Aug. (1970).
[4] G. Schomburg, F . Weeke, B. Weimann, and E. Ziegler: 8th Internat.
Symposium on Gas Chromatography, Preprints, Paper 15; G. Schoniburg, F. Weeke, B. Weimann,and E . Ziegler,Angew. Chem. 84,390 (1972):
Angew. Chem. internat. Edit. 11, 366 (1972).
[ 5 ] D . Henneberg, K . Casper, B. Weimann, and E. Ziegler, Angew.
Chem. 84, 381 (1972); Angew. Chem. internat. Edit. 11, 357 (1972).
[6] C . Kruger, Angew. Chem. 84,412 (1972); Angew. Chem. internat.
Edit. 11, 387 (1972).
Computer-Aided Analysi's of Organic Mass Spectra
By D. Henneberg, K. Casper, E. Ziegler, and B. Weimann"]
1. Introduction
After introduction of the sample into the spectrometer,
mass spectrometric analysis may be divided into three
phases : Measurement of one or more spectra; evaluation
or work-up of the data; and finally interpretation with a
view to answering the particular question of the analysis.
Evaluation means determination of the masses present
und their relative intensities as well as determination of
metastable ions with the assignment of their possible
origins. Data work-up refers to correction of spectra by
averaging and/or background subtraction and also the
choice of the most suitable representation for the information, i. e. bar-graph spectrum, table, mass chromatogram,
or element map. The last step, that of interpretation, involves identifying molecular fragments, hetero-atoms, and
structural elements responsible for the observed peaks, as
well as the recognition of mixtures. It requires further
quantitative determinations on mixtures or isotopically
labeled compounds and identification or classification by
comparison with calibration spectra.
[*] Dr. D. Henneberg, K. Casper, Dr. E. Ziegler, and B. Weimann
Max-Planck-Institut fur Kohlenforschung
433 Miilheimpuhr, Kaiser-Wilhelm-Platz 1 (Germany)
Angew. Chem. internat. Edit.
Vol. I 1 (1972) 1 No. 5
We wish to demonstrate with a few examples how the use
of a computer system can effectively support all three
phases of mass spectrometric analysis. In order to be able
to show the impressive advantages of computer use, especially in the measurement of spectra, we would first like to
give a few illustrations of the possibilities of mass spectrometric analysis with special emphasis on the more recent
instrumentation and methods.
Three techniques have made the application of mass spectrometry particularly interesting, and in some cases indispensable, in the analysis of organic compounds.
1. The combination of a gas chromatograph (GC) and a
mass spectrometer (MS) is of particular value in the analysis
of mixtures of easily vaporizable compounds. In this technique the eluate from a GC is continuously analyzed by
mass spectrometry. In the course of a chromatographic
analysis it is possible to obtain the mass spectra of all
resolved components (see Fig. 4). It is assumed that the
spectra are recorded rapidly so that distortion due to
changes in the partial pressures is minimized.
2. The direct vaporization of samples having low vapor
pressures results in a considerable increase in the types of
analyzable compounds. Modern mass spectrometers allow
357
spectra to be recorded at sample pressures between
and 10- torr in the ion source. Systems are available where
the sample temperature may be chosen as a parameter between - 100 and + 500°C and thus for many substances a
vapor pressure suitable for the ion source can be chosen.
By this method it is nowadays possible to obtain spectra of
compounds ranging from peptides, with molecular weights
of more than 1000, to volatile transition metal complexes
which decompose well below room temperature.
In general, after introduction of the sample the temperature is set so that the sample has an insignificant vapor
pressure. The temperature is then increased until the vaporization process is complete. In this way, a certain amount
of fractionation occurs for mixtures which can afford
valuable information for use later in the interpretation.
In special cases, the composition of the mixture can be
quantitatively determined (Fig. 1).
3. High resolution mass spectrometry allows determination of the empirical formula of molecules and their fragments by means of very precise mass identification. Computers have been successfuIIy used in conjunction with
high resolution mass spectrometry for some time"]. That
the method is almost unthinkable without the help of computers is clear from the wealth of data that must be analyzed. Large volumes of many-digit numbers are produced
whose processing culminates in the choice of the possible
empirical formula from a large number of atomic combinations.
In the following article we shall discuss only the GC-MS
combination and direct vaporization techniques with online computer operation and direct our attention solely to
low resolution spectra. This area is still in its infancy. The
application of high resolution mass spectrometry to the
sample introduction techniques discussed in 1 and 2 above
remains limited to a few special cases due to the particularly
large amounts of equipment required.
In order to highlight the effect of computerization on experimental, analytical, and last but not least organizational
aspects, a comprehensive description of the available software has been intentionally avoided.
magnetic-scan spectrometers. For a 30-minute chromatogram in which a spectrum is taken every 5 seconds a total
of 360 spectra will result. Experience has shown that of
these 360 possible spectra only 150 to 200 are necessary to
obtain the maximum amount of information. Further reduction assumes prior knowledge of which chromatogram
peaks are of interest to the analysis. Two-hundred spectra
correspond to about 40 m2 of paper. It would be simply
impossible in an analytical service laboratory to evaluate
so many spectra manually, even to the point of deciding
which spectra are important and which are not.
However, with data processing this problem is easily solved
with a minimum of time and effort. For the cost of recording
a single spectrum on UV-paper with a galvanometric recorder it is possible to record and store an entire series of
spectra with a computer.
..
_
t ..'.
110°C
1686911
12
90°C
I
-
10
0
c-t
6
2
L
fmtnl
Fig. 1. Fractional vaporization of a mixture of n-paraffins.
and total ionization (pressure)
Changes in the temperature (------I
(- - -)were recorded with an X,-X2 recorder. During the total vaporization 30 spectra were measured at intervals of ca. 25 s. The four vaporization curves given for the separate components are obtained from
the intensities of their respective parent peaks in the 30 spectra. For
chemically similar compounds, the areas under the vaporization curves
agree well with the known composition of the mixture in mol-percent.
31
391
412
432
450
23
25
21
2. Measurement of Mass Spectra in Conjunction
with an On-line Computer
Typically, the partial pressures of the components in the
ion source change as a function of time for the GC-MS
combination and the direct vaporization techniques. This
is particularly obvious for gas chromatographic peaks. The
partial pressure of the sample changes rapidly, especially
during measurements with good chromatographic separation. The first peaks in Figure 4, for example, have widths
at half-maximum of about one second. Under normal conditions (fast scan of the magnetic field) it is not possible to
obtain good spectra with such peaks.
For the analysis o f a total chromatogram, at Ieast one mass
spectrum per chromatographic peak is necessary. If overlapping of several components occurs and the background
is not constant, then many more spectra must be recorded.
At best, 10 to 20 spectra per minute can be recorded with
358
The direct vaporization of a mixture of solids is shown in
Figure 1. The change in the composition of the mixture
with time is clearly recognizable. Although the partial
pressure of each component changes relatively slowly a
rapid-scan recording of the spectra is recommended. Many
substances vaporize irregularly, which can result in a considerable distortion of a single spectrum. For qualitative
evaluation of the analytical results shown in Figure 1, it is
sufficient to record four spectra provided they are measured
at favorable times for the separate components. In order
to know when the spectra are best recorded and which
corrections must be carried out, a prior knowledge of the
shape of the vaporization curve is required. This is obtained
by plotting the intensity of a chosen mass from a series of
successive spectra. This technique has long been in use
with the GC-MS combination and is applied in several
Angew. Chem. internat. Edit.
1 Vol. I 1
(1972)
1 No. 5
variations : mass chromatograms measured directly with
fixed massesr2.31 or extracted by the computer from a series
of complete spectra for arbitrary massesE4]; alternating
mass (AVA-techniq~e[~]);
or isotope scan[61.
The evaluation of an analysis such as that shown in Figure 1
takes several hours by hand, whereas a computer requires
but a few minutes to deliver the corresponding picture.
In order to obtain such a picture for an unknown mixture,
it is first necessary to know which masses are characteristic
of only one component. An experienced investigator determines this by examining spectra from the beginning, middle,
and end of the vaporization. We will attempt, however, to
make the optimum choice with the aid of a computer programf41.
The first great advantage of measuring spectra with a computer is the ease with which a large number of spectra can
be recorded and stored. Although perhaps only a few
spectra are actualIy utilized for interpretation of the analysis,
the recording of unnecessary spectra requires less effort
than the repetition of an entire measurement if unsufficient
spectra are available to allow interpretation.
The second advantage of using a computer is that the
measurement is performed automatically and does not
require supervision by a technician. In the case of a
GC-MS combination without automatic measurement,
the undivided attention of a technician is required to follow
the chromatogram in order to measure each desired spectrum at the right moment, i e . at a chromatogram peak
maximum. In practice the situation is often simplified by
determining the rough form of the chromatogram in a trial
run. With direct vaporization prior information cannot
be obtained. In this case the vaporization process is followed in successive spectra recorded at short intervals-for
example with a storage oscilloscope. In order to record the
“right” spectra during the fractionation of a mixture, it is
necessary to recognize and interpret continuous changes
in the spectra. This is, of course, only superficially possible
during an experiment and requires much experience and
concentration. On the other hand, with automatic recording of many spectra the technician can devote time during
the measurement to other tasks such as the evaluation of
previous experiments or preparation for new measurements.
3. Programs for Measurement and Evaluation
of Mass Spectra
Two programs, MSDAT and MACO, developed by us, are
employed in the recording and evaluation of spectra. The
functions of MSDAT and MACO are discussed briefly
below.
The program MSDAT facilitates the recording, reduction,
and temporary storage of spectra to be evaluated. For recording of spectra it is possible to choose from three modes
of operation :
1. Push-button measurement of a single spectrum.
3. Measurement of a single spectrum in a 5-s cycle of the
spectrometer consisting of 2 s spectrum and 3 s pause. In
Angew. Chem. internat. Edit. / VoI. I 1 (1972) / No. 5
this mode the start of computer measurement is synchronized with the spectrometer scan after pushing a button.
3. Automatic computer measurement in a 5-s cycle. By
means of a parameter word it can be dictated whether
every spectrum in the cycle or every second, third, ere.
should be measured ; in this way, spectra may be recorded
every 5, 10, IS s, etc.
Each spectrum is composed of IOOOO digital values (2s
recording time for m/e=20 to 450, 5 kHz digitization
frequency), which during the course of the measurement
are reduced by MSDAT to the maxima contained in the
spectra. Positions, intensities, and form factors of the
maxima together with values for the base line and the total
ionization are stored. The latter is a sum of all intensities
(ITOT) which is related to the pressure in the ion source.
By comparison of the ITOT value of a spectrum with that
of previous spectra MSDAT can, for example, determine
whether the spectrum in a GC-MS analysis was measured
on the base line or on the tail of a previous peak. In the
latter case, temporary storage is pointless so that further
handling of this superfluous information is eliminated (see
Fig. 4)[’’I. The possibility of not evaluating a spectrum can
be overridden by means of the parameter word previously
mentioned.
The variable operational mode of MSDAT may be adjusted
to fit the particular problem :
1. For a normal GC-MS analysis, automatic measurement in 5-s cycles is chosen. Spectra for which the value of
ITOT is the same as that of the immediately preceding
spectrum are suppressed.
2. For vaporization of a solid sample, recording of spectra
at 10-s or f5-s.intervals is usually sufficient. The spectrum
suppression is eliminated by the appropriate parameter
word in order to obtain the complete vaporization curve
from a uniform series of data points.
3. In order to obtain a good calibration spectrum of a pure
substance, a number of spectra are averaged. For this
purpose one chooses the fastest repetitive scan, 5 s. In this
mode the spectra suppression must be turned off since it
would prohibit recording repetitive spectra of a calibration substance at a constant inlet pressure.
At this point it should be clear how a single program can
be flexibly adapted to perform many different tasks.
Program MACO evaluates and summarizes all spectra
belonging to a particular analysis. First, the data pertinent
to identification of the analysis are fed in, i.e. analysis
number and sample designation (Fig. 2). Following this,
MACO works-up the data stored by MSDAT. Normal
mass peaks and metastables are recognized, noise peaks
removed, and the masses calculated. In addition, MACO
carries out a series of control functions which test the reliability of the mass determinations, signal intensity, position of the base line, and peak shape. Without a computer,
it would be impossible to carry out all these checks for each
spectrum. From each spectrum MACO stores the following list of information : integral masses ; deviations from
integral masses ; intensities and peak type (i. e. sharp peak,
metastables, unresolved doublets, or undetermined type
of peak).
359
For each individual spectrum evaluated some preliminary
information is printed out immediately containing the
number of the spectrum, intensity and mass of the base
peak, the total ionization, the largest significant observed
mass, and a vaiue which characterizes the reliability of
mass determination (Fig. 2). If one of the above-mentioned
control functions has discovered an abnormal situation a
corresponding comment appears in addition. In Figure 2,
for example, a reference is made to overloading for spectra
10 and 11.Thus it is possible to judge the correctness of the
measurement from the data given in the control information immediately after the measurement of each spectrum
RlJN
VERS
DSK MAC0
71p1310
ANALYSIS-FILE:857
COMMENT
:SOYA
TYPE NNN C F I L F - N R . )
FX
fll0
a R PUTO:F\UT
8 1 N T PMPS
ITOT
6p1
27
?R
P
56
77
2R
3
64
P7
2R
4
5R
27
28
5
61
?7
28
6
5R
26
28
7
25
58
28
R
59
27
28
9 1114
5p19
45
OVERSHOOT
In 4366 1p173
45
CHECK E A S F L I N E HEPD:
OVFRSHOOT
1 1
61x1
1073
45
I?
299
115
45
99
29
28
13
1.4
1x3
33
41
15
75
29
2P
I 6
59
P7
28
NO
LSM
1
355
74
2x1
147
6R
68
71
66
6R
74
701
51
I82
51
183
281
355
355
354
355
51
54
64
57
63
69
355
281
355
355
355
281
-.16
MSH
EXECUTION T I M F :
?@.Rm
SFC.
4 P I W . 4P.3X
TOTAL F L P P S F D T I W E :
M3 FXFCUTION ERROQS D F T F C T F D
SEC.
FX17
?C
Fig. 2. Example of a print-out from program MACO. The underlined
numbers are input data. R U N DSK MACO starts the program. 875 is
the analysis number, SON BA 010 is the sample identification number.
Sixteen spectra were evaluated, two of these showing warning comments. The column headings are:
ITOT: = Total ioni~ation
BINT: = Base peak intensity
BMAS: = Mass of base peak
LSM : = Last significant mass in the spectrum
MSH : = Scale factor for mass determination
because the time-sharing operating system allows MACO
to run parallel to recording of spectra. In addition to this,
the teletype print-out provides a convenient record of the
whole experiment.
The mass numbers are obtained with the aid o f a mass/time
calibration. Slow drift and small fluctuation errors are,
within certain limits, corrected by MACO. The calibration table is produced automatically (9 s of computing
time) by Program MASCAL from a PFK spectrum. Calibration twice a day is generally sufficient.
360
The program combination MSDAT and MACO is capable
of recognizing peaks of the same order of magnitude as
background noice (for fast scan spectra, depending upon
the photomultiplier voltage, 3-5 mV). That yields a
dynamic range of 3 x f04 relative to the saturation limit
of the amplifier (IOOV) which can be accepted by the computer. Mass peaks smaller than 0.03% of the base line are
suppressed in the case of intense spectra.
In addition, MACO recognizes and evaluates metastable
ions. Thus with computer techniques it is possible to obtain
and correct all the available information from a spectrum,
even on fast scan operation. In our experience both the
masses and the intensities are more accurately determined
by computer than from typical oscillographic records. This
is attributable to the quality of the quartz-controlled time
base and the linearity of the analog-digital converter.
We have recently installed a device which automatically
switches sensitivity scales and thereby increases the dynamic range to 3 x 10’. This is necessary for GC-MS analysis
of mixtures whose components are present in widely differing concentrations. Otherwise, several runs with various
amounts of the sample must be made.
4. Print-out of Mass Spectra
In the print-out of computerized mass spectra the most
suitable form is chosen for the particular application in
question. For quantitative determinations in which the
composition of a mixture and/or isotope ratios are to be
displayed a table is most suitable. For a qualitative interpretation, for exampIe determination of the type of spectra
and the characteristic fragmentation pattern, representation as a simple bar-graph is to be preferred. For documentation purposes, however, a bar-graph is not suitable since
the dynamic range is often too small and metastable ion
peaks are not easily reproduced. Analog spectra are not
particularly suitable for any type of further evaluation. It
is, therefore, unnecessary to record analog spectra parallel
to computer measurement. Analog spectra are possibly of
importance only in special analyses dealing with shapes of
peaks and metastables.
Spectra evaluated by MACO, as well as any other suitably
stored spectra, can be printed-out as tables with Program
REMAS. For reasons of clarity and ease of filing, it is important that the print-out has a standard form. An example
of such a print-out in tabular form is shown in Figure 3.
The uppermost field contains all pertinent identification
information. The main field (left center) gives masses and
intensities, which may be normalized to the base line or
total ionization as desired. If the signal is not a normal,
sharp mass peak, a symbol designating the type of peak
appears (see legend of Fig. 3). The metastable ions are tabulated again to the right of the main field along with assignments of their possible origins where possible.
The capability of the program group MSDAT, MACO,
and REMAS is clearly evident from the mass peaks nz/e = 86
and 28 in Figure 3. The star alongside m/e=86 indicates
that the peak is broader than normal and hence may oriAngew. Chern. internat. Edit.
Yol. I I (1972) 1 No. 5
The profile of the reduced spectrum“] in the lower right
field provides certain information important to the qualitative interpretation of the spectrum, since it is characteristic of certain classes of compounds.
ginate from a metastable ion or unresolved doublet. The
m/e = 86 peak therefore appears once again among the
metastable peaks (Fig. 3, center right). There one sees the
exact mass of the ion (86.02), a possible formation route
,N,LIsIS-NR.:l
SIIIPLE:
W E
r
15
OATE:
Tlw: 11’48
18-ri8-71
tNT
!LlT
HIE
lhll
.*..**
0.47
10.72
1.87
15.95
0.37
50
57
5s
32
n.04
e .Fa
34
U’d3
28 0
29
369
31
35
37
38
39
.
0.35
111
112 L
113
0.Bb
114
b.31
19.53
115
0.57
71
,2
14-31
0.1”
77
*.IS
78
*
183 s
0.11
185
0 .05
(1.06
191
0 .#4
193
0.Q5
0.*4
51
v.33
*.1C
*.?a
124
125
126
127
1.na
8.d5
2.21
0.32
131
0.24
135 L
0.04
137
0.09
0.04
0.07
100.0n
3.18
45
0.11
:::2
84
84.5
a5
80 *
9: L
0.04
8.03
0.04
46
Q.03
0.n4
91
@.or
50
0.16
93 i
51
0.53
0.19
d.08
0.65
0.27
141 L
7-12
149
0.nb
L.28
4.35
151
0.m
157
153
15 4
8.Bb
0.05
94
9.60
V5
9h
97
98
56
57
8.99
90
1PO
59
8
9.3.62
4.15
1:04
E.‘?
1 61.3”
.29
0.06
3.10
107
63
0.P7
65
0.13
108 L
109
111)
138
139
195
0.03
I97 L
0.03
199
0.536
200
0.03
2@1
0.07
219
0.10
231
B.06
261
262
263
0.03
0.11
0.05
0.77
0.15
-1
2>.16
29
2
25.12
29.47
32.81
57.11
39.17
69.01
e.10
70
71
57
99
28
28
16
*2
0.88
41
0 .14
2.22
E.EA
43
71
2
2
*2
ni
27
q2
4
I
m .I4 0
7
*
*
*
0.32
0.04
111
11,
1
16
15
2
43
41
37
39
41
70
98
99
0
3
5
B.11
0.05
155
0.05
0.04
277
278
143
3.05
281
0.05
165
0.04
3.31
0.04
s.07
0.B6
0.17
a.e7
i-bop
2. 91
3-
3
4-
0
i
0
I
56789-
4
127
13-126
14-182
/ 15
........
..........
.. .....*)......
1
10.
2
11- 23
LNALYSIS-NR.:l
UI
IN!
0.85
M.
_---- - - . _ _--_
-----__
- -.---__---
84.28
86 0 2
9 a:37
183.00
R.11
47
52
53
55.5
0.16
8.05
0.04
B.11
80
44
0.m
02 7
70
43
0.05
121
122
0.27
6.44
1.72
17
80
9
119
123
*
f.
0.23
181
0.P8
0.08
92
83
*
8.16
7P
69
1s’
R/i
IN1
E.21
#.I6
2.22
e.06
1.17
40
::
0.h5
*/E
-..*
70-iYO~i
BISL-PEIK
.........*........ .... ....... ...... ....... .......
*/E
.....*...*.......*.....I........
25
27
I ON,-YOLTdtE:
R E F E R R E D TO:
2-.ltTH”L“FPT99
*
.*..*.
.........
Fig. 3. Standardized print-out of a mass spectrum produced by program REMAS on a line printer. The print-out contains
all values from the measured spectrum, as well as some derived information and information on the origin or measurement.
Peak type: D = unresolved doublet
.5 =doubly charged ion
* =possibIe metastable ion
L =small peak, type undetermined
(i. e. 114+99), and the deviation of the measured from the
theoretical mass (5/100 mass units). This mode of degradation is resonable (loss of a methyl group from 2-methylheptane). Due to the isotopic distribution, a peak having
an intensity of about 0.14% should appear at rn/e=86 (i. e.
9 % of m/e= 85). It is then clear the m/e = 86 peak arises
from the superposition of this isotopic peak and the peak
of the metastable ion resulting from the 114-99 decomposition. Without this superposition, the metastable ion
would appear only in the metastable ion field ofthe REMAS
print-out. At m/e=28 the partially resolved doublet for N,
and C,H, is shown.
Figure 3, bottom left, shows additional information such
as the contribution of the base peak to total ionization, and
the absolute magnitude of the base peak. The latter allows
assessment of spectrum quality. If a given spectrum is the
resultant of an averaging or corrective operation, this field
will contain information about the way in which the spectrum was arrived at.
Should one of the control functions have responded in
program MACO, the corresponding comment would be
given in the blank lower center portion of Figure 3.
Angew. Chem. internat. Edit.
Vol. I1 (1972) 1 No. 5
If a bar-graph is desired, the spectrum can be drawn by a
suitable plotter. Two examples of bar-graph spectra are
shown in Figure 6. The plotter requires about 1-2 minutes
per spectrum and is therefore not suitable for the reproduction of series of spectra. For this reason we will soon employ
a Statos recorder which requires 4 s to plot a spectrum.
5. Treatment of Spectra and Analyses
In the previous sections we have discussed how a series of
spectra may be obtained as starting material for a mass
spectrometric analysis. It will now be illustrated how the
necessary spectra are chosen and corrected during a
GC-MS analysis. It is first necessary to obtain an overall
view of the chromatogram drawn by the plotter according
to program DIPLO. Figure 4 shows the first part of analysis No. 875.
The times at which spectra are recorded and stored are
indicated by crosses on the chromatogram. It is evident
[*] “Reduced spectrum” after Crawford and Morrison [7].
361
875
Fig. 4. Overview of the GC-MS analysis No. 875, drawn by program DIPLO with a plotter: -,
chromatogram, measured as the total current from
a pressure measwing ion source at 20 eV ; f , signifies measurement and storage of a mass spectrum. The number of the spectrum is printed every fifth
total ionization values of the mass spectra.
measurement; 0,
The scale of the chromatogram is so chosen that the smaller peaks are clearly visible. Peaks too large for the chart paper are shown in reduced scale so
that the largest signal appears as a full scale deflection. Its height is given alongside the scale.
that spectra recorded in peak-free regions or in flat tailing
portions of the chromatogram (for example following
spectrum 62) are neither stored nor further evaluated. As
seen for spectra 21 and 22, however, if the ITOT value is
approximately the same in a region of one or two successive
GC peaks, both spectra are stored['5! The enlargement
scale for the illustration of the chromatogram as well as
for the ITOT value can be chosen by a dialog after calling
the program. With such a diagram, it is possible to find the
spectra which are important for a particular analysis, even
for a long, complicated chromatogram with many recorded
spectra.
An example of a simple, straightforward trace analysis will
now be described ; the corresponding MACO print-out
was depicted in Figure 2. In Figure 5 a portion of the
DIPLO picture of the analysis is shown.
During the total running time of 250 s, spectra were measured at intervals of 5 s. Of the resulting 50 spectra, only 16
were stored and evaluated as shown in Figure 5. In this
case, the GC-MS combination was connected to the computer only during the time 3 minutes before the elution of
the main component until shortly after the appearance of
362
the trace component. This limitation of the measurement
was possible due to prior knowledge of the chromatogram.
The trace component is best represented in spectrum 14.
As can be seen in Figure 5, about 40% of the total ionization in spectrum 14 results from the constant background
and the tail of the main component. In this case, the desired
correction is obtained by subtracting the arithmetic mean
of spectra 13 and 15. The effect of this correction is shown
in Figure 6.
After this correction, a pure hydrocarbon spectrum remains.
Apart from the peaks at rn/e=28 and 32 (due to air) the
peaks removed in the corrected spectrum are mainly due
to the major component, diisopropylbenzene (rn/e= 45,59,
73, and 87).
This example shows clearly how it is possible to achieve an
optimum stepwise reduction of data and applied operations: Prior knowledge of the chromatogram allowed limitation of the length of the measurement to about 4 minutes.
During this time span, at 5-s intervals, a total of 50 spectra
were recorded by MSDAT. Of these 50 spectra, 16 were
temporarily stored and later evaluated by MACO. For
solution of the analytical problem, i. e. determination of
Angew. Chem. internat. Edit. 1 VoI:Il (1972) No. J
ARJALYS,
-
NR.
It
Fig. 6. Mass spectra displayed as bar-graphs drawn by a plotter. Top:
Spectrum 14 of Analysis 875. Bottom: Spectrum 14 after subtraction
of arithmetic mean of spectra 13 and 15 (Fig. 5).
Addition of spectra yields an average which is employed
in cases where several weak spectra are available for one
component or where a very accurate calibration spectrum
of a pure substance is desired. (It is not necessary to divide
by the number of spectra added together since this is already accounted for in the output program normalization.)
m
Fig. 5 Excerpt from the DlPLO display of Analysrs 875 (cf. Fig. 2). In
order to show more clearly how spectrum 14 of the trace component
can be corrected the course of ITOT between the squares during the
analysis was later added by hand.
the trace component, only 3 of the 16 spectra were necessary; apart from the uncorrected spectrum only one spectrum was finally printed out.
Only 3 operations are generally necessary for the correction of spectra : addition ; weighted subtraction ; and correction of intensity distortions due to pressure changes
during the course of measurement (TI-correction). For
example, TI-correction would be desirabIe for spectra 38
and 39 of Figure 4. However, before each TI-correction the
constant portion of the background must be subtracted.
After correction, the spectra may be compared. If the corrected spectra are not identical, then the chromatogram
peak to which they belong is due to more than one component. In the case of spectra 38 and 39, a practically undistorted spectrum is obtained by simple addition since
the distortions of the separate spectra tend to cancel when
added.
A weighted subtraction must be applied, for instance, in
order to obtain the best spectrum of the substance represented by the middle peaks of the group centered at 220 s
in Figure 4 : certain fractions of spectra 15 and 17 must be
subtracted from spectrum 16.
Angew. Chem. inremat. Edit.
Vol. I 1 (1972) 1 No. 5
Without the computer, these arithmetic operations for
“correction” of spectra are normally omitted since they
are time-consuming and unpleasant tasks. An experienced
analyst recognizes part of the “impurities” of a spectrum
and disregards them during the interpretation. Since computer programs for interpretation or comparison of spectra are easily disturbed by impurities, computer interpretations therefore require background-corrected spectra.
6. Computer-Aided Interpretation of Mass Spectra
The last step of an analysis, interpretation, is particularly
interesting and important. Requisite for the development
of suitable methods is the availability of a comprehensive,
reliable collection of computer-compatible spectra. The
establishment of a collection of required magnitude will
be possible in those laboratories in which spectra are directly measnred and corrected with a computer. Model studies
have already been carried out for limited collections at
Stanford University in California[*]with programs attempting to reconstruct the decision-making process employed
by an experienced analyst in spectra interpretation
(Artificial Intelligence). However, this method soon runs
up against the same difficulties which limit non-computerized evaluation of spectra, i. e. insuKcient knowledge of
reliable fragmentation rules.
363
RUN DSK ISOTOP
An obvious type of interpretation is identification by comparison with a collection of stored spectra. Descriptions
of this method have been given in some detail[”. ‘’I. The
trace component analysis problem discussed in the previous
section would be largely soluble with this method since
practically all C,H, hydrocarbons are tabulated. The
spectrum shown in Figure 6 could only belong to a compound of this formula. There remain, however, three alternatives whose spectra are all quite similar, 2-methyl-4pentene, 4-methyl-2-pentene, and 2,3-dimethyl-2-butene.
ISOTOPE PROBABILITY DISTRIBUTION
STATE TEXT:
TABLE AND GRAPH OF
SN
BR
C
C AND 18 H
BR,8
1 SN.1
1
1
8
18
x
MASS
PROBABILITY
1.7
0.1
2.8
0.2
25.8
15.4
66.9
33.6
100.0
23.3
65.3
5.8
18.4
1.6
10.1
0.9
305
306
307
308
30 9
310
31 1
312
313
314
31 5
31 6
31 7
318
319
320
A final decision in favor of 2-methyl-4-pentene is possible
only when additional results from another analytical method are considered- for example, chromatographic retention times. This decision could also be made by our computer system. In the signal from the pressure-measuring
ion source, the GC-MS combination delivers a chromatogram comparable with a normal chromatogram. The
existing programs for evaluation of chromatograms and
for determination of retention times can be called for use
by the GC-MS analysis, since in the Mulheim computer
system it is possible to make use of information stored by
other users. These possibilities are exploited between the
gas chromatography and mass spectrometry divisions and
are especially advantageous for the GC-MS combination.
It is possible to envision a general Laboratory Information
GRAPH:
X
305
306
301
30 8
309
310
31 1
312
313
31 4
31 5
316
31 7
318
319
326
X
XXXXXXXXXXXXX
XXXXXXXX
xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
xxxxxxxxxxxxxxxxx
xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
xxxxxxxxxxxx
xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
xxx
xxxxxxxxx
X
xxxxx
?
?
7
?
?
?
t
?
?
0
10
20
30
40
50
60
70
80
t
90
T
100
STATE TEXT:
EXIT
-
Fig. 7. Print-out from program ISOTOP. Top: Input data are underlined: program start and desired
form ofthe print-out (TABLE AND GRAPH) of the isotopedistribution for the formula C,H,,Br,Sn,.
Center: Mass table and bar-graph spectrum. Bottom : Molecular peak group whose hetero-atom
content was to be determined.
seems especially
The employment of decision vector^[^^
attractive where results are expected which would be inaccessible without the aid of a computer. The main task of
the computer is to “train” a vector to a set of known spectra, using an iteration process within a learning program.
A vector so trained is applied to a spectrum to be evaluated
and a “yes-no” decision is obtained. In this way, it would
be possible to determine the presence of oxygen, a vinyl
group, a ring, etc.
364
System which would facilitate the exchange of various
analytical results between analysts as well as allow access
of chemists to analytical information.
The complete computer interpretation of an analysis is
possible in the case of quantitative gas analyses. A program
Califor such analyses has been in use over three
bration spectra of 45 inert gases, hydrocarbons, and volatile
solvents are stored. Starting with the mass spectrum of a
mixture the program calculates the composition of the
Angew. Chem. internal. Edir. / Vol. I I (1972) / N o . 5
mixture and prints out a control slip for documentation as
well as the final analysis results for the researcher. The program utilizes the method of regression analysis.
A final example of computer assistance is provided by a
program which computes the isotope distribution associated with a given empirical formula (Fig. 7).
Simple isotope distributions with 0, N, and up to 30 C
atoms as well as the distribution of Cl and Br may be found
in tables. Complicated distributions are not tabulated and
cannot be calculated by hand without unreasonable effort.
Therefore, such a program is indispensable in cases where
isotope-rich hetero-atoms are combined with many carbon
or several halogen atoms.
The future holds the promise of particularly interesting
developments in the area of computerized spectra interpretation.
7. Organization of Mass Spectrometric Data
Processing within the Time-sharing System
In the preceding sections it was shown that measurement
and evaluation of spectra are largely automated by the
computer. Interpretation of spectra, on the other hand, is
accomplished on the teletype in a preprogrammed dialog
using a series of programs with which the user is able to
carry out only suitable and necessary operations. As a prerequisite for such a mode of operation, all spectra, regardless of origin and previous manipulating processes, must
be filed in the same uniform format so that they are available for any kind of further treatment, interpretation, or
output.
The logical interrelationships of the operations applied to
mass spectra from the initial measurement to the final interpretation are depicted schematically in Figure 8.
In the computer system employed by us1141all data and
programs are stored on a magnetic disk, i.e. in a large
memory with random access. All operations may be started
MS
GC
I
1
Lineprinter
Plotter
-- Statos-Recorder
--Display
4
Teletypes
Fig. 8. Organization scheme for processing of mass spectrometric data.
The file includes actuai parameters, spectra, and programs, to which
simultaneous, rapid access from several locations is possible in timesharing operation.
Angew. Chem. internat. Edit. / Vol. I 1 (1972)
1 No. 5
at arbitrary times, several operations such as measuring,
processing, storing, and retrieving of spectra, may be started
simultaneously from different teletypes. Such a system
makes for optimum flexibility and all tasks, including program development, can be economically distributed.
From the schematic diagram, and above all from the examples presented here, it is clear that the computerization
of mass spectroscopic analysis does not aim at becoming a
completely automated (black-box) process. Rather, the
computer is used as a controllable instrumental aid whose
operation is guided by the analyst and thereby adapted to
fit each facet of the analytical problem. A completely automatic, combined application of all possibilities for spectral
correction and interpretation for each recorded spectrum
would quickly lead to the situation where the computer is
overloaded with unnecessary work-not to mention the
huge expenditure of programming effort. A fully automatic
analysis up to the print-out of the final results would only
be possible if essentially all conceivable results and associated complications were foreseeable in advance. A special
case of this is the quantitative gas analysis already mentioned, i. e. a routine analysis the evaluation procedure of which
is precisely defined and which occurs frequently enough to
make the programming effort worthwhile.
Of decisive importance in the procedure described here,
namely, the successive application of several programs or
program modules with variable parameters, is the dialog
between the program-user on the one side and the program
processing the data sets on the other. This dialog takes place
at the terminal from where it is possible to start and interrupt programs. The dialog, i. e., the questions and answers,
are appropriately preprogrammed. A simple example of
such a dialog is the choice of the scales for the drawing made
by program DIPLO (Figs. 4 and 5). The program determines and prints the maximum peak height in the chromatogram and the largest value of the total ionization in a
given series of mass spectra. From this it is possible to
specify the desired scales for the plotter diagram.
Warnings against attempts to carry out illogical operations
and the possibility of terminating the program in such a
case are of particular importance for a real dialog. If, for
example, large negative values were produced by subtraction of background, or during an averaging an incorrect
spectrum were used, warnings are issued and modifications
of the processing procedure are allowed. This kind of dialog,
which permits changing of the program flow during execution, is not to be confused with the input of parameters at
the start of a program. These permit variation of the program flow before execution, but do not allow user interactions after the program has started.
The measurement of spectra, evaluation and interpretation
of analyses, and program development are separate tasks
assigned to different persons within the laboratory. Only
a multi-user time-sharing system will permit the execution
of these tasks simultaneously from several teletypes, or
temporally independently of one another, and in the form
of a dialog. Thus an optimized and economical use of a
modern mass spectrometer is possible.
The time-sharing operation has worked especially well for
program development. Since the use of the computer
365
normally does not require an operator and programs can
be changed, compiled, and started at any time, it is possible
to directly test and verify new ideas. Thus the computer becomes a thinking tool for the scientist.
In this connection two special types of programs should be
mentioned. First, programs which serve as support for the
development of optimum methods. For instance, one of
these programs is written so that certain regions or peaks
from the original 100o0 data points of a spectrum can be
quickly found and printed out via a teletype. For instance,
this allows a user to investigate the effect of various smoothing functions or to test which algorithm or parameters give
the optimum signal to noise ratio or optimum differentiation between metastable ions and partially separated
doublets.
The second type of program is the test version of a regular
measurement and evaluation program. Such programs
print out all important intermediate results and thereby
permit the location of errors in the program or spectrum.
The construction of our program system for measurement
and evaluation is already substantially complete, having
begun about three years ago. Two years ago automatic
measurement was introduced, and automatic GC-MS
analyses have been performed on a routine basis since then.
In some cases, automatically measured analyses using
direct vaporization are performed with the same program
system and the same spectrometer.
On-line connection of additional mass spectrometers is
now rapidly accomplished since the basic programs already developed can be used with only minor changes.
With the exception of MSDAT, all programs are presently
written in Fortran. This is feasible because of the size and
speed of the computer system and has the advantage that
program improvements may be easily made. Such improvements are always necessary in scientific laboratories
since the problems to be solved are constantly changing
and it is necessary to take instrumentational advances into
account as quickly as possible.
Received: November 18,1971 [A 869 IE]
German version: Angew. Chem. 84, 381 (1972)
Translated by Dr. L. C. Glasgow, Miilheimmuhr
[I] See literature cited in H . Kienitz: Massenspektrometrie. Verlag
Chemie, Weinheim 1968.
121 D. Henneberg and G . Schomburg in van Swaay: Gas Chromatography, Hamburg 1962. Butterworths, London 1962, p. 191.
[3] D. Henneberg and G. Schomburg, Z. Anal. Chem. 215,424 (1965).
[4] R. A . Hites and K. Biemann, Anal. Chem. 42, 855 (1970).
[ S ] C. C. Sweeley, W A . Elliot, and R . Ryhage, Anal. Chem. 38, 1549
(1966).
[6] G. Schomburg and D.Henneberg, Chromatographia 1968,23.
[7] L. R. Crawford and J . D. Morrison, Anal. Chem. 40, 1465 (1968).
[8] A . Buchs et al., Helv. Chim. Acta 53, 1394 (1970); see also earlier
literature cited therein.
191 P . C. Jurs, B. R . Kowalski, ?: L. Isenhour, and C. N . Reilley, Anal.
Chem. 42, 1387 (1970); see also earlier literature cited therein.
[lo] P . C . Jurs, Anal. Chem. 43, 22 (1971).
[I11 B. A . Knock, 1. C. Smith, D.E. W i g h t , R . G . Ridley, and W Kelly,
Anal. Chem. 42, 1516 (1970).
[I21 H . S . Hertz, R. A . Hites, and K. Biemann, Anal. Chem. 43, 681
(1971), and further literature cited therein.
[I31 K . Casper, N . Kreitz, and D. Henneberg, unpublished.
[14] E. Ziegler, D. Henneberg, and G . Schomburg, Anal. Chem. 42,51 A
(1970).
[15] D.Henneberg, K . Casper, and E. Ziegler, Chromatographia 1972,
209.
Data Processing in Gas Chromatography
By Gerhard Schomburg, Fritz Weeke, Bruno Weimann, and Engelbert Ziegler I*’
1. Introduction
Data processing in gas chromatography is usually viewed
as the execution of a large number of routine analysesprimarily analyses of mixtures whose quantitative and
qualitative compositions are subject to only slight variation. However, the interest shown by industry in GC data
processing has additional, “natural” motives : The large
number of instruments frequently encountered in industrial
laboratories and the corresponding demand for operating
personnel make rationalization and automation essential
for uniformity in routine analytical work.
[*] Dr. G. Schomburg, F. Weeke, B. Weimann, and Dr. E. Ziegler
Max-Planck-Institut f i r Kohlenforschung
433 Miiiheimmuhr, Kaiser-Wilheim-Platz 1 (Germany)
366
In contrast, data processing with GC instruments employed
in research or development has hitherto attracted less
attention or interest. This is probably due to the lower
number of instruments connected per laboratory, especially
if dedicated systems for gas chromatography only are
considered. Gas chromatography, however, is rarely used
on its own in research and development but usually in
conjunction or even in direct combination with other
physical methods of analysis such as mass, NMR, and IR
spectroscopy.
In these circumstances, larger computer systems having
more peripheral equipment and higher efficiencies should
be considered from the viewpoint of higher total cost and a
better price/performance ratio. Each of the analytical
methods mentioned will undoubtedly make use of electronic data processing in the future, so that each would
Angew. Chem. internat. Edit. 1 Vol. I 1 (1972) 1 No. 5
Документ
Категория
Без категории
Просмотров
1
Размер файла
1 144 Кб
Теги
mass, aided, organiz, analysis, computer, spectral
1/--страниц
Пожаловаться на содержимое документа