1 Q0 APPLIED SPECTROSCOPY

Analysis of Organic Reaction Products by Combined Infrared-Gas Chromatography Techniques*

Jeanette G. Grasselli

and

Marcia K. Snavely

Research Department, The Standard Oil Company (Ohio), Cleveland, Ohio

Abstract The combined use of infrared spectroscopy and gas-liquid

chromatography provides a powerful analytmal tool for solwng a variety of problems. A sample trapping techmque is described which allows 0 4 /zl of hqmd sample to be examined in the infrared w~th- out use of beam condensers. Examples are presented to show its usefulness in qualitative analysis. The trapping techmque is effictent enough to allow quant~tatwe collection for analysis by infrared spec- troscopy of two-component, unresolved peaks on the gas chromato- graph. Apphcatmn to ddute water solutmns of acrolein and acetone IS g i v e n

Introduction The combined use of gas-liquid chromatography and

infrared spectroscopy for solution of a variety of analytical problems has been amply described (1-8). The most com- plete discussion is an excellent recent paper by Haslam, et al. (1) covering several collection methods and applica- tions to problems in the plastic industry. There is no doubt that gas-liquid chromatography and infrared spectroscopy are complementary, together forming a powerful analyti- cal team. The chromatograph can be used to provide spec- troscopists with pure compounds for reference purposes or with partial separations of samples so that the infrared in- terpretation is simplified. Also, the chromatograph, being more sensitive to trace concentrations, can detect com- ponents that are buried in the infrared spectrum. On the other hand, infrared analysis gives the chemist fast and positive identification of eluted fractions shortly after they appear. Since they are more specific, infrared methods can be used in some instances where the chromatograph is limited. This is demonstrated when fractionating methods fail to separate closely related components of a mixture or when retention times fail to positively identify the un- known.

Many traps for collecting gas chromatograph fractions have been described (9-13), but most of these are de- signed to examine the vapor phase of the sample in the infrared. Where liquid fractions are collected, beam con- densers are most often required due to the small volume of collected liquid (3, 4, 13). Another means of overcoming the disadvantage of small sample size has been the use of preparative column gas-liquid chromatography (5, 7), and it is this technique which probably promises the greatest advances in the future, especially with the recent intro- duction of more sophisticated and reliable instruments.

A very simple technique for collecting liquid fractions from a conventional gas-liquid chromatograph has been developed which requires only a minimum amount of equipment and has been exceptionally useful and workable in this laboratory. No beam condensers are required, and as little as 0.4/A of a liquid sample can be examined in the infrared without the use of a solvent.

~Presented m part at the Fourth Annual Conference, Cleveland Sec- tion, Society of Applied Spectroscopy, May 20, 1960.

Experimental Apparatus and Procedure

The Perkin-Elmer model 21 infrared spectrophotometer was used in this study. Standard sodium chloride liquid cells of varying thicknesses were employed, as well as a Perkin Elmer sodium chloride microcell with a sample vol- ume of 0.4/zl and a cell spacing of 0.025 mm. The chrom- atographic equipment was a Perkin-Elmer model 154C equipped with a West Marksman stripchart recorder, model M. The recorder is completely transistorized; therefore it requires no warm-up period, and a stable base line is readily obtained. It has a conventional zero to five my scale with 1 sec full scale response.

To collect samples efficiently from the chromatograph, a modification of the exit system was necessary. Serious sample losses resulted from premature condensation of volatile liquids m the three-way solenoid valve and side- exist valve used for conventional trapping from the chro- matographic equipment. Therefore, the exit line was cut immediately after the detector cell, and a ¼ in. glass tube was butted to it with a teflon sleeve as shown in Figure 1. The joint is actually in the oven of the chromatograph so that eluted fractions are heated practically to the point of collection and have a minimum amount of tubing to travel. There is a 5/20 standard taper on the other end of the glass joint, which fits directly the female joint of the micro trap. Other commercial instruments can be readily modified in a similar manner, and in this laboratory every chromatograph, including those which are homemade, is equipped to collect fractions for infrared identification. This modification greatly increases the efficiency and use- fulness of both techniques since fractions may be collected during an initial run and do not have to be submitted for rerunning on a different chromatograph in the infrared section.

1 ~ OVEN

I

Pressure Gouge

Regulator

FIG. I . MODIFICATION OF P E - 1 5 4 C FOR TRAPPING SAMPLES

VOL. 16, No. 6, 1962 191

~-50mm~ - ~ - 5 5 / 2 0 ~" JOINT mrn 111l

120 I JOINT

5,ram 111 3mm J IIII oo rub,~

7 | - . I l l ~-~r~)nlFub,na __t._~ eserv°lr k,~--Hole above ~'quator

Finger Trap Mtcro Dewar

77 mm Internal Depth

FIG. 2. T R A P FOR C O L L E C T I N G LIQUID FRACTIONS, AND

MICRO DEWAR

Figure 2 is a diagram of the liquid trap.t The 10/20 standard taper joint is fitted with an Asco "Quorn" teflon sleeve, which gives a tight fit and lubricity for disassem- bling and yet avoids the danger of contamination from stopcock grease. The 50/20 standard taper joint which fits the male exit joint of the chromatograph is used without any lubricant. The unique feature of the trap is the inner tube, which ends in a reservoir bulb containing a small opening located above the equator. The carrier gas passes easily through the trap and out at A while the sample con- denses in the reservoir or around the sides of the inner tube. The samples can be transferred directly to a microcell by capillary action through the entrance tube, as illustrated in Figure 3. If such a small amount of liquid is present that It fails to fill a microcell, a useful trick is to push the sample into the light path of the cell with a solvent such as carbon tetrachloride. The solvent will usually not even be detected in the spectrum, because it serves only to fill

FIG. 3. F I L L I N G OF M I C R O C E L L F R O M F I N G E R T R A P

tManu£ac tu red by E u c h d Glass Engineer ing Co., 11310 Wade Park Ave., Cleveland, Ohio.

the micro dipper of the cell and allows the sample to fill the light path of the cell ahead of it. I f the sample collects as droplets on the sides of the inner tube and will not flow down to the reservoir with gentle tapping or use of a low- pressure stream of nitrogen, a solvent may be used to flush it through to the outer tube where it is transferred to micro or macro liqmd cells with disposable micropipets. It is preferable not to use a solvent, however, since solubility of the unknown, even m standard solvents such as carbon tetrachlonde or carbon disulfide, is not assured. No diffi- culty has ever been experienced with sample loss using this system, even when trapping components which boil only slightly above room temperature (Table I) .

The finger trap is cooled in a micro dewart, Figure 2, containing crushed dried ice. Dry ice-acetone should not be used since back condensation of acetone into the trap m- variably occurs and contaminates the sample spectrum. Dquid nitrogen may also be used, but extreme care must be exercised in warming to avoid flash vaporization of very volatile materials.

TABLE I. Q U A N T I T A T I V E R E S U L T S OF A C E T O N E -

A C R O L E I N MIXTURES

Fed to Chroma- Analyzed by tograph, wt. % IR, wt. % Dewatlon

Acetone Acrolem Acetone Acrolem Acetone Acrolem

1.11 2.98 1.20 2.98 + 0 . 0 9 0.00 1.86 1.97 1.82 2.12 - - 0 . 0 4 + 0 . 1 5 1.98 2.35 1.70 2 10 - - 0 . 2 8 - - 0 15 2.64 1.74 2.42 1.44 - - 0 . 2 2 - - 0 . 3 0 1.71 2.67 1.60 2 43 - - 0 . 1 1 - - 0 . 2 4 1.16 2.13 1 22 2.10 + 0 . 0 6 - - 0 . 0 3 1.32 1 74 1.20 1 70 - - 0 . 1 2 - - 0 . 0 4

Std. dev. 0 18 0 18

The traps may be inserted quite rapidly when identifi- cation of a peak is desired. When it is also of interest to check quantitative distribution or homogeneity of a peak, one trap can readily be attached to collect the up-side of a peak, then quickly removed whale a second trap is inserted to collect the down-side of the same peak. Using a conven- tional 0.02 ml feed to the chromatograph, a peak consti- tuting as httle as 2% of the total sample can be trapped, transferred to the microcell, and examined in the infrared with no difficulty. When the concentration of the desired component is less than 2% in the sample mixture (with a 0.02 ml feed), successive rapid injections may be used if there is no interfering peak in a reasonable time interval; also, trappings from several runs may be collected.

Qualitative Analysis

Fusel Oil

A sample of fusel oil was submitted for identification of alcohol components. The infrared spectrum was well resolved and the alcohol type quickly recognized as primary from the strong 9.4 micron band. A search of reference C4 and C0 alcohols yielded a good spectral match with 3- methyl- 1-butanol. There were only a few minor differences between the hterature reference spectrum and the sample spectrum in the 7.2, 7.8, and 10.6 micron regions. A chromatogram was obtained by running the sample through a three m PEG 200 (40% on firebrick) column. The chro- matogram showed at least six components in the sample, the major one identified as 3-methyl-l-butanol from its infrared spectrum. With further detailed examination of infrared references and comparisons with the spectra of the

192 A P P L I E D SPECTROSCOPY

2O ? 7C 3 6 9 12

FIG. 4 . T R A P P E D F R A C T I O N F R O M A L K Y L A T E

15

hterature reference spectrum of methacrolein and the infra- red spectrum of the sample trapped peak closely matched with two striking exceptions: the carbonyl band at 5.8 microns and a large band at 8.6 microns were split in the methacrolein reference spectrum but not in the sample. A sample of methacrolein was obtained and distilled, and when the &stilled fraction was examined in the infrared, an identical match with the trapped sample was obtained. The literature reference spectrum of methacrolem con- tamed an impurity.

various trapped peaks, the sample was identified as a mix- ture of 3-methyl-l-butanol, 2-methyl-l-butanol, 2-methyl- 1-propanol, and traces of n-propanol and n-butanol. All of the components were primary alcohols with similar struc- tural features and would have been very difficult to identi- fy positively without the reformative and highly important chromatographic data.

Plant Alkylate

Another interesting analysis involved the bottom frac- tion from a distxllation of a sample of plant alkylate. Char- acterization of this bottom fraction was requested, and the sample was run on a 4 ½ m Apiezon L column on the gas chromatograph. The zso-octane components were readdy identified with only one exception, a small peak on the shoulder of another band. The area of this peak indxcated that this component was less than 2% of the total mix- ture. It was therefore trapped, &luted in carbon disulfide, and its infrared spectrum obtained (Figure 4). Since the unknown was probably only a single compound and a hydrocarbon, an IBM sort (14) was made on the infrared reference deck of API standard hydrocarbons. Sorting only on the bands from 7.3 to 10 microns, the unknown was identified in 4 mln as 2,2,5-trimethylhexane, a nonane ex- hibiting a boiling point reversal and therefore confusing the identification in the chromatographic trace.

Reaction Product of Methyl Propionate and

Formaldehyde

A product from the reaction of methyl propionate and formaldehyde gave an interesting large chromatographic peak on the gas chromatograph, whose retention time matched exactly that of methyl methacrylate, the desired and expected product. The infrared spectrum of the trapped peak and the infrared spectrum of methyl methacrylate were compared, and it was immediately apparent that the trapped peak was not methyl methacrylate, as retention time had erroneously indicated. The sample was identified as diethyl ketone, again using an IBM sort on bands in the fingerprint region. This example strengthens the com- mon knowledge that relative retention times must be used with great caution as bases for identification and analyses of samples on the gas-liquid chromatograph.

Reaction Product of Acrolein and Trioxane

Infrared identifications often lean heavily on exact qualitative matching with a good reference spectrum. Sometimes, however, even the literature spectra must be doubted. A sample of the reaction product of acrolein and trioxane was identified, again by the gas chromatographic retention time, as methyl acrylate. However, when the peak was trapped and examined by infrared spectroscopy, it was observed that it was definitely not methyl acrylate. The spectrum was compared by sorting with a IBM machine, using intense bands in the 9 and 10 micron regions, and the reference spectrum of methacrolein was obtained. The

Oxidation Products of Alpha Diisobutylene

The infrared spectrum of alpha di-iso-butylene oxida- tion products is quite banded and complex, showing hy- droxyl, carbonyl, and several olefin absorptions. To identify individual compounds from such a mixture would have been difficult by infrared methods alone. The sample was examined in the gas chromatograph on a 3 m PEG col- umn, and 5 peaks were obtained, all of which were trapped and identified by infrared examination. The chromatogram is shown in Figure 5. Peaks Nos. 1, 2, and 3 were alpha di-iso-butylene, acetone, and tertiary butanol, respectively. Peaks Nos. 4 and 5 were not completely resolved. Yet by trapping the up and down sides of the double peak, peak No. 4 was ~dentified as 4-methyl-4-pentene-2-one, and peak No. 5 was 1,2-epoxy-2,4,4-trimethyl pentane. These re- sults were obtained using a 0.03 ml sample injected into the chromatograph and illustrate mcely what can be done with this combination of techniques.

laJ o0 Z O o. bJ tic

tic

I - bJ ¢:1

I

I I I o 5

Time (minutes)

FIG. 5. C H R O M A T O G R A M OF ALPI~IA-DI- / /SO-BUTYLENE

O X I D A T I O N PRODUCTS

5

I I IO

Quantitative Analysis

The combination of infrared and gas chromatography can be used not only for qualitative identifications but for quantitative data as well. To determine the efficiency of this trapping system, a quantitative analysis was attempted on dilute aqueous, synthetic solutions of acrolem and ace- tone. These components were selected because of their volatility, which would be a severe test on the system and demonstrate adequately its expected sensitivity. Also, these two materials were not resolved on the chromatograph using a standard PEG column. Therefore, the system was ideal for this test. Separation of the water was accomplished on the chromatograph, the "common" acrolein-acetone peak was quantitatively collected, and the respective amounts of acrolein and acetone in the trapped peak were analyzed by infrared methods.

VOL. 16, No. 6, 1962 193

100 oGI , ol I I I I I I I I

J'UIL/ I ' / I I Itll t l\ f l V I II I I I I I I'll I/,11 II1\ / I k/I /I I I / / I I I I

o/ IVl' lL l \ 1 / I IVl uI I I - I I I I 2 3 4 5 6 T 8 9 I0 11 12 13 14 15 16 IT

Wan length in Microns

FIG. 6. INFRARED SPECTRUM OF ACROLEIN AND ACE- TONE IN ACETONITRILE

For the analysis, a 3 m 40% PEG on firebrick column was used, operated at 100°C and employing 0.04 or 0.02 ml of sample. The common acrolein-acetone peak at 3 rain retention time was trapped and diluted with standard spectro-grade acetonitrile for infrared examination. The resulting spectrum is shown in Figure 6. The acetone band at 8.18 microns and the acrolein band at 8.65 microns, both corrected for basehne, were used for the analysis, although it would be possible, at low concentrations, to use the split carbonyl bands.

The general method for trapping and collecting previ- ously described was used, but great care and precise tech- nique had to be exercised in working with quantitative solutions. Recahbration was required if a new chromato- graphic column was prepared. The standard solutions for calibration were prepared by carefully weighing acetone and freshly distilled acrolein into volumetric flasks before dilu- non. The chromatograph was calibrated using distilled- water solutions of acetone and acrolein. A range of one to four wt. % total concentration of the two components was plotted versus the area of their common peak. The infra- red was cahbrated using acetonitrlle solutions in a 0.10 mm sodium chloride, macro-hqmd cell. The standards were prepared in concentrations of from 0.5 to 3.0 wt %. The conventional Beer-Lambert plot was a straight hne for both the acetone band at 8.18 microns and the acrolem band at 8.65 microns. The slope of the acetone line was 0.292 and that for acrolein was 0.126.

A series of acetone solutions were run as a control to determine the error in the method. (Acetone was used since it is easier to handle than acrolein.) Known acetone solu- tions in water were transferred to the chromatograph. The acetone peak was trapped and diluted with 0.1 ml of ace- tonitrile, carefully measured with a cahbrated micro pipet. The results were calculated both from the area under the gas chromatographic peak and from infrared absorption. The standard deviation of the values obtained by the infra- red analysis from the known concentrations was 0.055. Reproducibili ty checks were also made on several solutions. Results on these standards gave a cross-check on the method and showed that the sampling injection to the chromatograph was quanntat ive and reproducible and that the trapping was efficient with no hold-up in the lines or trap.

Quanti ta t ive results on known mixtures of acrolein and acetone are shown in Table I. The same conditions were used. The standard deviation of the infrared results for both acrolein and acetone from the known values was 0.18.

An interesting alternate way of performing this analysis serves to illustrate how accurate and quantitative this combination of tools can be. Standard water solutions of acrolein were transferred to the gas chromatograph, and the absorbance of the infrared bands as obtained from the trapped peaks was plotted against the concentrations

of the solutions. In other words, the infrared is cali- brated with the feed to the chromatograph. The calibra- tion was a straight line having a slope of 0.050 for one to four wt. % acrolein in water. The same type of cali- bration was obtained for acetone. A solution containing 1.32 wt. % acetone and 1.74 wt. % acrolein was ana- lyzed in this way. From the peak area, the total percent acrolein and acetone in the water solution was analyzed as 2.95 wt. %; from the infrared absorbance of the trapped peak the concentration of acrolein was calculated as 1.75 wt. %, and acetone, by difference, was 1.20 wt. %. The agreement between the known and analyzed values was good and again gave a check on the analysis.

For all of this work, of course, utmost precautions were taken to avoid contamination of the collected sam- ple in the lines, the block housing of the detector, and especially by the slow decomposition of column material. The instrument lines were thoroughly cleaned with a suitable solvent and flushed with air before and after each run. A pipe cleaner was used to remove any residue from the exit line. This cleaning procedure had to be followed conscientiously, even though somewhat tedious and time consuming because in some instances materials from a previous run or the column material itself had been trapped and identified.

None of the various sources of contamination such as stopcock grease or cleaning solvent could be over- looked. An interesting example of this was the identifica- of a phthalate ester in a trapped peak, which could not have been a component of the sample injected into the chromatograph. Upon investigation i t was discovered that tygon tubing containing a phthalate ester as a plasticizer had been used in one of the laboratory chromatographs. To safeguard against this type of contamination, blank runs had to be made frequently by at taching the trapping apparatus and running pure carrier gas through the in- strument. I f any material was collected, further cleaning or removal of the contaminating source was required.

Summary

A few of the apphcations i l lustrating the versatility of analysis introduced by combming infrared and gas chromatography have been mentioned.

A very simple trap is employed that is suitable for use on any commercial instrument. Samples can be col- lected quickly and run, often without the use of a sol- vent. Beam condensers are unnecessary with this system, and standard liquid cells may be employed.

Even though the methods have been very successful thus far, the existing possibilities have just begun to be reahzed. Using preparative columns would reduce the re- striction on concentrations of the unknown. This would provide much larger samples for infrared analysis and eliminate completely the need for solvents.

The sensitive detection of trace components in com- plex infrared samples can be accomplished through use of the chromatograph. Chromatograph peaks can be positively identified qualitatively by infrared, and un- resolved peaks from the chromatogram can be detected qualitatively and applied to quantitative analysis by in- frared methods. Further experimentation along these lines will certainly increase the already substantial contribu- tions of these two instruments in solving difficult ana- lytical problems.

194

A c k n o w l e d g m e n t

The authors gratefully acknowledge Mrs. Connie N. Jones and Miss V. Frances Gaylor, who designed the micro liquid trap for adaptation to a commercial gas- liquid chromatograph.

Literature Cited

(1) J. Haslam, A. R. Jeffs, and H. A. Willis, ANALYST 86, 44 (1961)

(2) D . M . W . Anderson, IBID. 84, 50 (1959) (3) J. E. Stewart, R. O. Brace, T. `johns, and W. F.

Ulrich, NATURE 186, 629 (1960) (4) W. S. Gallaway, T. `johns, D. G. Tipotsch, and

W. F. Ulrich, Pittsburgh Conference on Analytical Chemistry and Applied Spectroscopy, March, 1958

(5) N. Brenner and V. J. Coates, Pittsburgh Confer- ence on Analytical Chemistry and Applied Spec- troscopy, March, 1957

APPLIED SPECTROSCOPY

(6) R. V. Helm, D. R. Latham, C. R. Ferrin, and J. S. Ball, ANAL. CHEM. 3 2 , 1765 (1960)

(7) J. S. Matthews, F. H. Burow, and R. E. Snyder, IBID. 32, 691 (1960)

(8) H. H. Hausdorff and N. Brenner, OIL GAS J. 56, 89 (August 4, 1958)

(9) H. Szymanski, R. Povinelli, D. Stamires, and G. Lynch, ANAL. CHEM. 31 , 2110 (1959)

(10) .J.M. Lesser, IBID. 31, 484 (1959) (11) H. E. Belhs and E. J. Slowinsky, Jr., J. CHEM.

PHYS. 25, 794 (1956) (12) H. J. Gold, CHEMIST ANALYST 4 9 , 112 (1960) (13) CIC Newsletter, Connecticut Instrument Corpora-

tion, No. 6 and 9, (1960) (14) Bool~ of Codes and Instructions for Wyandotte-

ASTM Punched Cards Indexing Spectral Absorp- tmn Data, Amer. Soc. Testing Materials, Phila- delphia, Pa.

Submitted November 1, 1961

T

Notes

Note on the Conversion Factor for the Constant Temperature D.C. Arc Method

M. Malinek:l:

Metallurgical Institute, Czechoslovak Academy o f S c i e n c e ,

Prague, Czechoslovakia

When the Constant Temperature D. C. Arc Method (1, 2) was apphed, following the procedure of Eichoft and Addink (5), to prism spectrographs not considered by these authors, distinct anomalies were found. The instru- ments in question were Soviet medium quartz prism ISP- 22 and Soviet Littrow KSA-1 spectrographs. (For a brief description of these Russian instruments, see (4)). The conversion factors (C. F.) were found not to be constant over the wavelength range of 2500 to 3500 A, and the instruments behaved as if the conversion factor was a function of wavelength. Because both spectrographs were prism instruments one would expect the conversion factor to be equal to umty over the whole wavelength range being examined (5). However, it was experimentally shown that the conversion factor increased as the wave- lengths became shorter.

In the case of the medmm quartz ISP spectrograph, the plate tilt was at first considered responsible for such a shape of the conversion factor curve. The plate tilt of this instrument is 42 ° at 2600 A (due to employing a collimating mirror instead of the usual collimator lens) compared to about 24 ° for other medium instruments, as, e.g., Q 24. With the latter instrument, the C. F. was found to be unity as with other spectrographs employed by the authors (5), and the expected differences between

TABLE I. APPROXIMATE VALUES FOR d 2 sin e/f 2 FOR Two

SPECTROGRAPHS

Ins t rument 2400 A 3000 A 3600 A

Q 24 19 15 13 ISP-22 14 12 11

~Present Address Institute of Geochemistry and Mineral Resources, Czechoslovak Academy of Sctence, Prague, Czechoslovakm

3.2

2.8

~ 2 4 C3

2O

z

~2

08

el~ o o

25 26 27 28 29 30 3.t 32 33 ~4 WAVELENGTH A.lO ~

FIG. 1. C. F. CURVE BEFORE (O) AND AFTER (O) THE MODIFICATION OF THE AUTOCOLLIMATING INSTRUMENT

the Q 24 and ISP-22 may be verified by the expression for the radiant power falhng on the plate. For the radiant power F falling on the plate, we can write (diffraction effect of the slit neglected) :

F ~ d 2 sin ~/f2 [1]

where d is the diameter of the camera lens, f is the focal length of the camera lens, and e is the plate tilt.

The approximate numerical values of the right side of Eqn. [1] for the two instruments are given in Table I. If the condition of equal integrated intensity of the select- ed iron lines (3) is satisfied for both instruments at, e.g., 3000 A, equal spectral energies will not cause the same measured intensity, as the intensity values for the two in- struments will differ by 8% at both extremes of the wave- length range employed by the method (1, 2). However, the influence of the 22 ° difference in the plate tilts of the two prism spectrographs on the radiant flux of the mono- chromatic sht image does not account for the intensity deviations found experimentally. Moreover, similar results