Determination of Unsaturation by Near-Infrared Spectrophotometry GODDU
ROBERT F.
Research Center, Hercules Powder Co. , Wilmington The recent commercial availability of high-resolution spectrophotometers which cover the near-infrared region of the spectrum ( 1 to 3 microns) has provided the analyst with a new tool. This paper evaluates this region for the determination of unsaturation in organic compounds. Both terminal methylene groups and cis double bonds may be selectively determined in a wide variety of materials. Other centers of unsaturation d o not interfere with terminal methylene deter-
99, Del.
Tvhen this resolution has been an important factor and has necessitated changing instruments. Molar absorptivities determined on both instrunients usually agree to about + l o % on bands TT ith a half-intensity band width of about 10 mp. I n this and subsequent papers, the near-infrared region is referred to as that between 1
and 3.1 microns, as none of these instruments is capable of recording spectra beyond 3.1 microns with resolution superior to conventional infrared instruments. The vibrations observed in the nearinfrared region are almost exclusively hydrogenic vibrations and combinations of hydrogenic vibrations with other
minations. The sensitivity for )C=CH2 can b e pushed to as low as 100 p.p.m. in many cases. The precision of direct measurements i s generally within ? 1 to 2% of the amount present. Mixtures of cis, trans, and terminal unsaturation may be analyzed for their cis double bond, and terminal methylene contents. Bands at about 1.62 and 2.10 microns may be used for determining terminal unsaturation and a band at 2.14 microns for cis unsaturation. No method has been developed for trans unsaturation. Because of their speed and selectivity, nearinfrared measurements may b e advantageously substituted for bromination or hydrogenation in determining certain types of double bonds.
D
the past few years, several papers have pointed out some analytical possibilities of the near-infrared region of the spectrum (1-3). This paper is the first of several which mill attempt to evaluate in some detail the practical utility of the near-infrared region to determine various functional groups. It deals with the determination of isolated double bonds in organic molecules. There are currently three coniniercially available double-beam spectrophotometers which cover all or part of the near infrared region: Cary Model 14, Beckman Models DK-1 and 2, and Perkin-Elmer Spectracord Model 4000. Mainly because of convenience and wider range, most of the work reported here was done on the Beckman Model DK-2, although a Gary ;\lode1 14 was also available. A Spectracord Model 4000 was not available. Although the Cary has resolution superior to the DK-2, there are relatively few times
Figure 1.
1 mm. undiluted 1 em. undiluted
_____
URIKG
1790
ANALYTICAL CHEMISTRY
Near-infrared absorption spectrum of heptane
Figure 2.
Near-infrared absorption spectrum of 1 -octene
undiluted _ _ _ _ _11mm. cm. undiluted
modes of vibration of the molecule under consideration ( 2 ) . N o fundamental C-H vibrations (except the =CH stretching vibration a t about 3.1 microns) occur in the region, so t h a t all bands observed are either combinations or overtones of the fundamental bands that occur in thcl 3- to 6-micron region. The spectrum of a typical aliphatic hydrocarbon is shonn in Figure 1. There are four rcgioiis of rather intense absorption: 1.2, 1.4, 1.7, and 2.3 to 2.5 microns. These are mainly due to conibinations and overtones of the C-H stretching vibratioiis in the 3.5-micron region. The first overtone occurs a t 1.7 microns and the second a t 1.2 microns. The 2.3- to 2.5- and 1.4-micron bands are due to conibinatioiis of the C-H stretching and other vibrations.
By using the windons a t 1.25 to 1.35microns, 1.45 to 1.65 microns, and 1.8 to 2.2 microns, it is possible to determine numerous impurities or functional groups in predominantly hydrocarbon media. The spectrum of a typical terminal unsaturate, 1-octene, is shown in Figure 2. The large number of sharp bands due to the addition of a terminal methylene group is evident and has also been shon-n by Holman and Edmondson ( 1 ) . Of these bands, the two of most value are a t 1.63 and 2.11 microns. These bands are both intense and are in regions n here the background due to C-H absorption is relativelylon . The 2.22-micron band is intense, but somewhat less useful because it occurs on the side of a n intense C-H combina-
I CG 90
60
20 IC
C
Figure 3. Near-infrared absorption spectrum of trans-l-methyl2-pentene
___ 1 mni. undiluted _ - _ _ _ 1 em. undiluted
-
tion band. The discussion of the determination of terminal unsaturation in this paper is limited to the combin at’ion band a t 2.1 microns and the first overtone of the =CH, stretching vibration a t 1.6 microns. N o s t of the emphasis is on the use of the lattcr. The spectra of t n o compounds with internal unsaturntion are shonn in Figures 3 and 4. Thc bands clue to the unsaturatiori are conbiderably weaker than in a terminally unsaturated compound. As a mattcxr of fact. the trans isomer has no intense and unique bands. The cis compound has a relatively strong band a t 2.14 microns. This is the band n.ith which the subsequent n-ork in this paper is concerned. From eraluation of the region t o date, it has not been thought desirable to attempt to determine trans double bonds, because of the lack of a unique and intense band and because of the intense infrared band a t 10.34 microns which is adequate for most analytical purposes. Holman and Edmondson ( 1 ) recently suggested that cis unsaturation in nonconjugated fatty acids could be determined a t 2.19, 2.14, or 1.18 microns. Kork in this laboratory has shown that although cis compounds h a r e stronger bands a t 1.18 microns than trans compounds, there is appreciable interference from trans a t that wive length. Furthermore, this band in okic acid is extremely weak, having a molar absorptivity of about 0.02 liter per mole-cni. A band a t 1.62 microns is also potentially useful for cis compounds, but again the absorptivity is small, limiting the utility of the band. The 2.19micron band, probably that occurring a t 2 . l f 5 microns in Figure 3, is 11-eaker and less intense than that a t 2.14 microns and therefore n-as not investigated. Although it u-as not investigated, i t may be possible to develop a method for trans double bonds, using the band a t 1.77 microns. Because absorption bands in this region are due to hydrogenic vibrations, i t might be expected that fully substituted double bonds n-ould exhibit no absorption bands and partially substituted ones would h a r e proportionally loner molar absorptirities. Compounds nere not readily available to test this hypothesis.
= 90 Y
f
EXPERIMENTAL
3c
20
0 C
Figure 4. pentene
Near-infrared absorption spectrum of cis-4-methyl-2-
undiluted _ _ _ _ _11nun. cm. undiluted
Beckman Model DK-2 Spectrophotometer. Foi qualitative work and for recording complete nearinfrared spectia as in Figures 1 t o 4, the instrument n-as set u p as follons: scanning time, 5 ; scale eupansion, 2 X ; time constant, 0.1 or 0.2; sensitivity, 30; range, 0 t o 100% transmitt anc e. Carbon tetrachloride TTas used in the ieference beam. For quantitative TT ork the follorring settings were used: scanning time, 50; VOL. 29, NO. 12, DECEMBER 1957
1791
scale expansion, expanded 2 x ; time constant, 0.2; sensitivity, 50 (or 100 if reference standard was used); range, appropriate absorbance scale. Cary Model 14 Spectrophotometer. T h e Cary was used only for quantitative work. T h e settings were: slit control, 30 (or 45 if a reference standa r d was used); slit height out; speed 10 8.per second. T h e 1- and 5-cm. cells were of fused silica or quartz and matched t o within 0.003 em. in length and t o within 0.01 absorbance unit over t h e region covered. Corex cells might be used equally well for work below 2.6 microns. For some of t h e spectra of pure cis or trans hydrocarbons, 0.9mm. spacers were used in 1-em. cells; this gave rise t o somewhat larger errors t h a n t h e work in 1-em. cells. The sources of most of the chemicals used are indicated in the tables. The solvents, with the exception of carbon tetrachloride, were neitherdried nor puri-
fied beyond a simple distillation. As it was shown in this work that the bands under investigation are insensitive to solvent effects, extensive purification was considered unnecessary. Because a double-beam instrument was used, minor absorption bands due to the solvent have no effect on the spectrum of the solute, except in so far as slightly wider slits might cause a decrease in the molar absorptivity of the absorption bands. In obtaining some of the spectra in the 2.1-micron region, the technique of differential spectrophotometry was used to cancel much of the background absorption due to standard C-H bands. For example, 5 cm. of 0.633M palmitic acid in carbon tetrachloride was used in the reference beam when 1 cm. of pure oleic, linoleic, or other fatty acid was run. Thus, both sample and reference had roughly the same number of CHa, COOH, and CH2 groups in the light
Table I. Effect of Structure on 1.62-Micron Terminal Methvlene Absorotion -
Compound 1-Octene 1-Octadecene 3-Chloro-2-chloromethyl-1-propene 2-1Iethylacrolein Acrolein Styrene or-Methylstyrene Allyl acetate Ethyl acrylate Acrylic acid Methyl methacrylate Acrylonitrile Vinyl isobutyl ether Vinyl n-butyl ether Vinyl or-terpinyl ether Vinyl acetate
Source Solvent Phillips Petroleum, pure grade cC14 Matheson, Coleman & Bell CC14 Distilled lab sample Matheson, Coleman & Bell (corr. to 90%) Matheson, Coleman & Bell Distilled lab sample Distilled lab sample Distilled lab sample Rohm & Haas (stabilized) (Corr. to 88% purity) Rohm & Haas (stabilized) Dist. Products (practical) Carbide & Carbon (redistilled) Carbide & Carbon (redistilled) Distilled lab sample Distilled lab sample
Amax, i~
I
Molar Absorptivity, e, Liter/ Mole-Cm.
1.636 1.636
0.31
%-Decane 1.629 cc14 1.629
0.31 0.20
0.27
CCl, CCl, CCl4 CCl4 CClr Dioxane CC14 CCl4
ccI4
cc14 cc14 CC14
Table II. Effect of Solvent on Terminal Methylene Absorption of Ethyl Acrylate at 1.62 Microns Molar . Absorptivity, E , Liter/ Mole-Cm. 0.327 0.321 0.314 0.289
Solvent Range D 0.007 2 Toluene Decane 0.030 2 Carbon tetrachloride 0.014 3 Methyl ethyl ketone 0.000 3 tert-Butyl alcohol -0.280 .. Diethyl Cellosolve 0.279 o:oio 2 2 Ethyl orthoformate 0.2’79 0.002 Tetrahydrofuran 0.2i3 ... 1 Dioxane 0.265 1 Diethylene glycol dimethyl ether 0: 002 0.263 2 Tri-n-butyl phosphate 0.262 4 0.006 Tetraethylene glycol dimethyl ether 0.251 0.006 .. Triethylene glycol dimethyl ether 0.250 0.006 .. Dimethylformamide 0.003 4 0.239 Hexamethylphosphoramide 4 0.014 0.237 Butyrolactone 3 0.009 0.236 Dimethyl sulfoxide 3 0,009 0.228 q Number of determinations or curves run t o obtain E . Absorptivity varies with concentration from 0.30 to 0.26 when absorbance varies from 0.3 t o 0.7. This is probably a dit-width effect, but is reproducible. It is also true to a lesser extent with toluene.
1792
ANALYTICAL CHEMISTRY
path, so that the unsaturation constituted the major difference in the samples. The matching of sample and reference need not be exact with respect to the number of CHI or CH2 groups-for example, hexane would be a suitable blank for hexene, as use of butane as a reference would present some problems. Differential work with infrared is not new, a recent article on this technique being by Powell (4). However, the idea of matching compounds to references, with the exception of the functional group in which one is interested, may be somen-hat novel. It also may be of more use in near-infrared, as over the 1.0- to 2.6-micron region there is more energy with which to work than in the infrared region, and as a result even relatively concentrated blanks can be used with fairly narrow slits. TERMINAL METHYLENE ABSORPTION
Sear-infrared absorption data for the first overtone of the terminal methylene C-H stretching vibration are ineluded in Tables I and 11. Less complete data on the 2.1-micron combination band are included in Table 111. Table I shows the effect of structure on the wave length of absorption and molar absorptivity of the terminal methylene group. I n general, the more active the double bond, the lon-er the wive length of the absorption maximum. However, the molar absorptivities of the bands are relatively constant within the known purities of the compounds involved. Thus, it is possible to determine the total terminal methylene content of samples containing a number of different compounds of similar structure. Both vinyl and vinylidene structures are presented in Table I; the wave length of absorption and absorptivity are about the same for both types of compounds, which is different than in the conventional infrared region. I n the 10- to 12-micron region vinyl and vinylidene groups have very different absorptions. This similarity of vinyl and vinylidene in the near-infrared is advantageous in functional group analysis and allows near infrared to complement the regular infrared. The effect of various solvents on terminal methylene absorption is shown in Table 11. There seem t o be three distinct types of solvents: those which contain no oxygen and give the highest absorptivity; those which contain oxygen, but are not particularly polar, such as the ethers and esters, and which give intermediate values; and those which have more polar groups, such as dimethyl sulfoxide and dimethylformamide, and which give the lowest absorptivities. The variations in absorptivity are not due t o differences in spectral slit of the instrument with various solvents, as all the solvents listed in
Table I1 had a comparable nominal slit width of 0.025 to 0.026 mm., or halfintensity band width of 2.4 mp, except tert-butyl alcohol, for which the slit was 0.050 mm., and toluene, for which the slit was 0.040 mm. The terminal methylene absorption bands bein, measured a t 1.6 microns have a half-intensity band width of 8 t o 12 mp, which is independent of solvent. These data indicate t h a t terminal methylene groups can be accurately determined in the presence of an extremely wide variety of other compounds or functional groups. Even other types of unsaturation such as terminal acetylenic hydrogens or aromatic compounds do not influence the terminal methylene determination. The wave lengths of the absorption maxima in the 2.1-micron region and molar absorptivities of several compounds with terminal unsaturation are 0.G
L
included in Table 111. These data indicate that the absorptivities of this combination band vary much more widely than those of the same compounds at the 1.6-micron overtone band. This variation can be both a n advantage and a disadvantage. Greater sensitivity may be obtained for some compounds at 2.1 microns than at 1.6 microns. -4 disadvantage is that these data indicate t h a t the 2.1-micron band is a poor one to use for functional group analysis of unknown compounds. A t 1.6 microns one can estimate the molar absorptivity of a compound t o j=lOyG. This is not so a t 2.1 microns, unless the structure of the unknown is well defined. The ratio of the heights of the 1.6- and 2.1-micron bands may be of definite use in the qualitative identification of compounds which contain terminal unsaturation. Because of the large amount of light
Table 111. Molar Absorptivities of Terminal Methylene Absorption Maxima at 2.1 Microns in Carbon Tetrachloride
Molar Absorptivity, e, Liter/ Mole-Cm.
-
0.30
-
0.20
-
Compound trans-Dichloroethylenea cis-Dichloroethylenea Acrolein 1-Octene illlyl acetate 3-Chloro-2-chloromethyl-1-propene Ethyl acrylate Vinyl n-butyl ether Vinyl acetate
i i
z c
0
0 e
available at 1.6 and 2.1 microns, it would be possible to do very high precision assay work on terminal unsaturation, using a compound of known purity in the reference beam. Trace work is also possible, especially at 1.6 microns, where most media absorb very little and the spectral slit remains small with a long-path-length cell. From the absorptivities in Table I, it can be calculated that 100 p.p.m. >C=CH, (about 4 mmoles per liter) should give an absorbance of about 0.01 in a 10-cm. cell. Experimentally this has been found very close to true in several cases.
4
*“T
-, ,,,,
a,’o[
-3
/-.J-->
,
-,*,*% ’
\,
--A’
0.03
2.10
2.15
MICRONS
2.n
2.10 M I C R O N S 2.15
2.n
Figure 5. Near-infrared absorption spectra of cis- and frans-4-methyl2-pentene in 2.1 -micron region a t high resolution
___ 20YG by volume cis-4-methyl-2-pentene in carbon tetrachloride
- _ _ _ _ 207, by volume trans-4-methyl-2-pentene in carbon tetrachloride 20Yc by volume 2-methylpentane in carbon tetrachloride used in reference beam
Amax
2,137
0.118
2.136 2.117 2.112 2.105
0.116
2.104 0.315 2.103 0.63 2.093 0.170 -0. 12b 2.089, 2.121 a Compounds from Matheson, Coleman 8: Bell. They cannot be considered truly terminal methylene compounds, but are unique, as their spectral data indicate. Vinyl acetate has no sharp band in this region and thus is unlike other compounds in this table.
~
Table IV.
0.47
0.49 0.43
~~
Near-Infrared Absorption Data at 2.14 Microns on Compounds with Internal Double Bonds hIolar
. ~ .
Compound cis-2-Pentene trans-2-Pentene cis-3-Hexene cis--l-~Iethyl-2-pentene
trans-4-hIethyl-2-pentene
Cyclopentene
Cyclohexene Methyl oleate Methyl linoleate Oleic acid
Ln,*, Source Am. Petroleum Inst. Am. Petroleum Inst. Am. Petroleum Inst. Phillips (957,) Phillips (99%) Phillips (99%)
Solvent Kone ?;one Sone Sone Cc14 CCI, Yone
CCI4 Kone CCl, Emory KO.2301 Sone Pacific Vegetable and Oil None (bleached) Hormel Foundation Sone Phillips (99%)
Reference Solution %-Pentane (4m. Petroleum Inst.) n-Pentane (Am. Petroleum Inst.) n-Heptane (Phillips, purified) 2-hlethylpentane (Phillips, 997c) Above in CC14 A4bovein CCI, Cyclopentane (Matheson, Coleman & Bell) iibove in CCL Cyclohexane (llatheson) Above in CC1, Methyl palmitate (Eastman) in CC1, Methyl palmitate (Eastman) in CCl,
LL
2 133 Sone 2 138 2 138 2 138 Tone 2 138
Absorptivity, Nominal E, Liter/ Slit,a Mole-Cm. Mm. 0.125, 0.121 0.028 0.028 n. _ 128 ~ . 0.028 0.129, 0.135 0.028 0.131 0.030 -0,004 0,030 0.186 0,028
2 138 ~.~0 199 ~
2.136 2.136 2.142 2 141
Palmitic acid in CCI, (Purified lab 2.141 sample) 2.143 Palmitic acid in CCI4 2.141 .-.
0.028 0.027 0.027
0.k
0.153 0,108 0,178
-
0.079 0.079
0 071 (DK-2) 0.25 0,109 (Cary) 1 . 8
Linoleic acid Hormel Foundation Sone 0.123 (DK-2) 0.25 z.. 145 0 206 (Cary) 1 . 8 Elaidic acid Purified lab sample CCI, Palmitic acid in CCl, 2.143 0 008 (Cary) 1 . 8 Crotonaldehyde hfatheson, Coleman & Bell CCI, CCI, 0 028 2.108 0 33 a A nominal slit of 0.028 mm. on the Beckman DIG2 is equivalent to half intensity band width of 2 mp. 0.079 mm. equals 5.5 mp, and 0.25 mm., 17 mp. On Cary Model 14 a nominal slit of 1.8 mm. is equivalent to half intensity band width of 4.7 mp. Half intensity band width of these bands is 10 to 11 mp. . I n
VOL. 29, NO. 12, DECEMBER 1957
1793
CIS A N D TRANS UNSATURATION ABSORPTION
The absorption curves of cis- and trans-4-methy1-2-pentene in the 2.1micron region are shown in Figure 5 , These curves were obtained a t high resolution and high sensitivity as described above. The spectrophotometric data on several cis and trans compounds are included in Table IT'. The absorptivities of all of the cis straightchain hydrocarbons are equivalent within the probable experimental error of measurements. The cyclic hydrocarbons might be expected to h a m somewhat unusual absorptivities because of their rigid configuration. The r a v e length of the absorption mavimum is also approximately the same for all of the cis hydrocarbons, n-ith the exception of cis-Zpentene. On the Becknian DK-2, the absorptivities for the fatty acids and fatty acid esters are lower and nonlinear with respect to the number of double bonds because the per cent unsaturation is much smaller, thus requiring more hydrocarbon to be in both the sample and reference beam. This causes an increase in the spectral slit width, which, in turn, leads to a lower absorptivity. When the same solutions are measured on the Cary Model 14, which operates with narron-er spectral slits, the molar absorptivities are much higher and h a r e a linear dependence on the number of cis double bonds. As a matter of fact, the nbsorptivities as determined on the Cary for these compounds approach those for the cis double bonds in the lower moleculariveight hydrocarbonson the DK-2. S o n e of the compounds with internal t r a m double bonds have absorption a t the wave length of the comparable cis compound in eyeess of that which could be accounted for b y cis impurities. [Crotonaldehyde, reported to have trans configuration (a), is anomalous in absorbing in this region and its molar absorptivity is unusually high]. It should thus be possible to determine readily the cis double bond content of mixtures containing cis, trans, and saturated compounds. For example, a known mixture of the follonk,0' coniposition mas prepared.
cis-4-~Ieth~.l-2-penteiie
trans-4-Methyl-2-pent ene 2-llcthylpentane
33 G 33 8
32.7
The cis-4-methyl-2-pentene content, as determined b y near-infrared. was 32.6%. As this corresponded to an absorbance of 0.17 in 1-em. cells, it is reasonable to assume that about 5% of the cis compound (1.5y0cis -CH=CH-) could be readily determined in 1-cm. cells and that possibly as low as 1% of
1794
ANALYTICAL CHEMISTRY
0.50-
Figure 6. Near-infrared absorption spectrum of carbon tetrachloride solution of 1 octene and cis-4-methyl-2pentene in 2.1 -micron region
-
0.'10-
Reference, 2-methylpentane in CCla
0.300
€2.111
z
4
Q
I-Octene cis-4-Methyl2-pentene
m 0 YI m
0.20-
0.10-
2.05
€2.138
0.49
0.002
0 006
0.131
J 2. IO
2.15 M I C PO',
S
the cis compound could be determined in longer cells. Another mixture containing 35% 1-octene and 657, cis-l-methyl2-pentene !vas analyzed for both conqtituents in the 2.1-micron region and found to contain 35% 1-octme and 607, cis-4-methyl- 2- pentene. Both peaks were n-ell resolved from each other (2.112 and 2.138 microns. Figure 6) and neither compound contributes appreciably to the absorption a t the other wave length. It is thus possible in favorable cases to determine both tc,rminal and internal cis unsaturation n ith fair accuracy on the same saniplc and using the same scan. h known sample of tall oil fatty acids was also run. The cis double bond content of the sample calculated from its olcic, lineoleic, linolenic, and coiijugated dicnoic acid content 17-as4.09M. A war-infrared scan garc a d u e of t . O O M for the cis douhle bond content. Other snniplcs of fractionated tall oil fatty a d s have been also analyzed successfully. I n general, the agrecment betneen knon n and found valucs on all milturps containing cis double bonds is encouraging.
detect and determine unsaturation. Because of the nature of the problems, most applications to date have been on determination of terminal methylene groups rather than cis double bonds, Such analyses vary from the determination of a few hundred parts per million in of 3-chloro-2-chloromethyl-1-propene 3,3-bis(chloroniethyl)oxetane, to following esterification of alcohols rvith methacrylic acid and determining the purity of rinyl ethers. Almost every application tried to date has been successful. Because of its speed and specificity, near-infrared can advantageously replace bromination or hydrogenation for the determination of unsaturation when appropriate double bond types are involved.
ACKNOWLEDGMENT
The author n-odd like to express his appreciation to K l l i a m &I. Bowe, who ran many of the spectra and determinations included in this paper, and to Kescott C. Kmyon for his helpful criticisms during the course of this investigR t'ion.
CONCLUSION
Isolated single double bonds of scvera1 typw may be readily detected arid deterniincd 1,s use of the near-infrared Terminal region of the spectrum. methylene groups may be determined a t 1 . G or 2.1 microns with a sensitivity \\ ithin 0.017,>C==C'H2 and a prccision and accuracy nithin + l to 2%. Cis double bonds may be dr.terminrti a t 2.1 microns with a sensitirity of about 1% -CH=CH--. lli\turcs of tcrniiiial, cis. and trans double honcls may be analyzed rapidly for cis and terminal double bonds in the near-infrared. The author's laboratory has used nearinfrared &ensivrlj for the past year to
LITERATURE CITED
( 1 ) Holmnn, R.
T., Edniondson, P. It.,
L C.H E ~ 28, I.
1533 (1935). IT., Specfrochirn. A c f a 6 , 257 (1951). ( 3 ) Lauer, J. L., Ro?enbauni, E. J., A p p l . Spectroscopy 6, S o . 5 , 29 (1952). (4) Pon.ell, H., J . d p p l . Chem. 6, 488 (1056). (5) Young, IT.G., J . Cheni. SOC.54, 2498 (1932). ~
A
(2) %ye,
RECEIVED for revier May 25, 1957. Acwpted Jill?. 2i, 1967.