Determination of Oxygenated and Olefin Compound Types by Infrared Spectroscopy J. A. ANDERSON, JR., AND W. D. S E Y F R I E D Humble Oil and Rejining Company, Raytown, Texas A laboratory investigation of t h e relationship between molecular s t r u c t u r e of organic compounds a n d their infrared absorption spectra h a s resulted i n development of procedures for t h e quantitative to semiquantitative determination of individual compound types i n complex mixtures, employing infrared absorption spectroscopy i n t h e liquid phase. I n m a n y cases, analyses which heretofore have n o t been possible by other m e a n s can be m a d e by these methods. A s t h e m e t h o d s can be applied to samples as small as 1ml., samples can be analyzed by these procedures t h a t could n o t be handled formerly even when adequate chemical m e t h o d s were available. I n general, infrared absorption procedures for t h e determination of compound types i n complex mixtures are based u p o n t h e facts t h a t , usually, t h e presence of certain functional groups in t h e molecules causes each compound of a given t j pe to absorb
T
HERE has been an increasing need in the petroleum industry in the past few years for methods of analysis that may be applied to the determination of a large variety of compounds in products from refinery processes and from laboratory and pilot unit experimental studies. Although adequate procedures are available for the individual determination of nearly all constituents of Cg and lighter fractions and for certain compounds or groups of compounds in mixtures of higher molecular weight, the lack of pure calibration standards (particularly olefins, diolefins, naphthenes, and oxygenated c.ompounds) in high molecular weight ranges, plus the inherent complexity of such samples, has to date prevented the development of adequate analytical procedures for the determination of individual components in such mixtures. The analytical situation has been further complicated by the fact that, in many research studies, the scale of operations is so small that adequate quantities of material are not available for analysis by conventional means. In conventional methods of analysis by absorption spectros- . copy (1, 3, 4,8, 13, 14) absorption measurements are made on a given sample a t the key wave lengths of all the absorbing compounds present in the sample, and the percentages of the individual components are obtained by solving simultaneous equations written for the absorption a t each wave length involved. This type of analysis for individual compounds is usually limited, however, to samples containing not more than four to eight absorbing substances, although in some cases (8) as many as twelve components have been handled, and is not generally applicable without supplementary separation techniques to complex mixtures above the CS t o COrange, particularly if these mixtures contain olefins or oxygenated compounds. There are frequent references in the literature (a, 6, 7 , 9-11) to the fact that various classes of compounds possess so-called “characteristic” absorption bands, caused by the presence of certain functional groups in the molecule, which occur a t essentially the same wave-length location (or frequency) for all the compounds in a given homologous series. Although (2) the wavelength location varies with the molecular structure of each compound in a series, and the intensity of the characteristic absorption bands (expressed as optical density under standard condi-
infrared radiation a t a constant wave-length characteristic of t h e functional group; a n d when expressed on t h e proper basis, t h e absorption coefficient for each functional group is essentially cons t a n t , regardless of molecular weight. Although there are i m p o r t a n t exceptions to these general phenomena, it has been possible i n most cases t o develop procedures t h a t correct for t h e observed deviations. A t present, i t is possible to make determinations for t h e functional groups of alcohols, aldehydes, carboxylic acids, esters, ketones, a n d five different olefin types i n complex samples such a s hydrocarbon synthesis n a p h t h a s i n a b o u t 4 hours. For oxygenated compound types, average accuracy (deviation from t r u e composition) ranges from *10 to *209& depending upon type a n d complexity of mixture. For olefin types, t h e average deviation is about *3 equivalent bromine n u m b e r units.
tioris) is not constant, it wab felt that by further study a method might be developed whereby, through proper use of characteristic absorption bands caused by functional groups, complex mixtures of compounds of high molecular weight could be analyzed for the concentrations of various types of molecules contained therein. Recently Johnston et al. ( 7 ) have reported an infrared absofption technique whereby a-olefins and total p-, y-, and 8-olefins can be determined in gasolines by employing an analytical scheme based upon the use of the characteristic mono-olefin absorption bands in the 9.5- to 11.5-micron region, using an auxiliary silica gel adsorption technique when necessary for removing interfering saturates and aromatics. For the past several years, an investigation has been in progress in. this laboratory of the relationship between the molecular structure of organic compounds and their infrared or ultraviolet absorption spectra. .4s a result of these studies, a number of procedures have been developed for the quantitative determination of individual compound types present in petroleum fractions, employing infrared absorption spectroscopy in the liquid phase. In general, these procedures are based upon the facts that usually the presence of certain functional groups in the molecules causes each compound of a given type to absorb infrared radiation at a certain wave-length characteristic of the functional group, and when expressed on the proper basis, the absorption coefficient for each functional group is approximately constant, regardless of molecular weight. Although there are important exceptions to these general phenomena, it has been possible in most cases to develop procedures that minimize the observed deviations. The fact that in many cases there are mutual interferences between various functional groups can be overcome by the usual procedure of calibrating for the absorption of each functional group a t each of the wave lengths employed and setting up simultaneous linear equations, utilizing these absorption coefficients in such a way that absorption data obtained on unknown samples may be substituted and the equations solved for the composition of the unknown. In many cases, the “type” analyses are further complicated by deviations from Beer’s law and by background absorption due to the presence of compound types that are not determined by the method. However, all
V O L U M E 20, N O . 11, N O V E M B E R 1 9 4 8 these factors can be successfully handled in specific cases by the use of proper techniques. To date, the primary emphasis in this investigation has been directed toward the development of procedures for determining various types of oxygenated compounds and olefins in complex mixtures such as the oil products from Fischer-Tropsch type synthesis processes (hereinafter referred to as “hydrocarbon synthesis naphthas” or “synthesis naphthas”). Procedures are now available for the quantitative or semiquantitative determination of the key functional groups in alcohols, aldehydes, carboxylic acids, esters, ketones, and acetals, and for the determination of the olefinic grouping in olefins having the general structures RCH=CH,, RCH=CHR’ (czs and t r a n s ) , RR’C= CH2, and RR’C=CHR”. A complete analysis for all the above conipound types can be made on about 1 ml. of sample in about 4 hours. The reproducibility obtained when this method is used is set by the usual limitations encountered in making absorption measurements and is usually of the order of *5% or less of the equivalent amount of each type of substance present, except for very small concentrations of a given compound type. Unfortunately, lack of sufficient quantities and varieties of pure compounds, especially certain oxygenated compounds and olefin types, for calibration and preparation of synthetic mixtures has prevented a complete development and evaluation of this procedure. For analyses for which an adequate number of pure compounds, covering a fairly wide range of molecular neight, are available, the accuracy (average deviation from the true composition) of the method as determined by analyses of synthetic mixtures is in the order of *lo% of the functional group analyzed for the oxygenated compounds and * 1.5 equivalent bromine number units for the olefin types. At present, it is believed that the accuracies for the determination of alcohols and olefins of the structure RCH=CH, are usually of this order of magnitude. Determinations of other compound types, which in some cases are based on calibration data obtained for only one compound, are usually someivhat less accurate. In these cases, the average deviations from the true composition are usually *20% or less, based on the functional group for oxygenated compounds, and - 3 equivalent bromine number units or less for olefin types. The method in its present form has been applied to the analysis of about 2500 samples and the information so obtained has been found useful in a number of research and pilot plant control projects. APPARATUS, MATERIALS, AND EXPERIMENTAL TECHNIQUES
The work described in this article was performed using a PerkinElmer Model 12A infrared spectrometer equipped with a sodium chloride prism and adapted for the scanning of spectra with a suitable wave-length drive and a contact modulated direct current amplifier of the type developed by the General Motors Corporation (1.2). The instrument was housed in an air-conditioned room with temperature controlled to * l o F. Undqr these conditions, little or no drift was encountered. The conditions of amplification were set so that full-scale deflection on the 0-10 millivolt Brown Electronik recorder was obtained with a 4gnal to the direct current amplifier of the order of 1 microvolt. Under these conditions of relatively low gain and wide slits, noise ievel was of the order of *0.2% of full scale, and noise was an insignificant factor with regard to analytical accuracy. Sample cells employed for the work were of conventional design and consisted of rock salt plates separated by lead gaskets. Cells of approximately 0.1-mm. thickness were used for nearly all measurements. Many of the samples analyzed contained small amounts of water, which caused a slight fogging of the sample cell window over extended periods of use. However, all intensity measurements for the samples, I , and reference intenzity measurements, IO,were made in the same sample cell and calculations were based on differential absorption. For this reason, a slight fogging of the sample cell windows was not a problem. One sample cell was used for over 1000 samples before the fogging became sufficiently serious to warrant changing cells. In cmes where detection of trace amounts of functional group-
999
ings was required, measurements were made in sample cells having thicknesses up to 1 mm. It was necessary that all samples be compared with a reference sample of known absorption; n-heptane (obtained from Westvaco Chemical Corporation) was found highly satisfactory for this purpose. In addition, where it was necessary that the concentration of absorbing material be lowered by dilutihn, nheptane was used as solvent and absorption data for the sample were calculated from absorption data for the blend of sample and n-heptane. The majority of oxygenated compounds employed as calibration standards in this study were obtained from Eastman Kodak Company and used without further purification. Most of the olefins were procured from the Connecticut Hard Rubber Company, the National Bureau of Standards (A.P.I. Project 46), or private sources, and these compounds also were used without further purification. In some cases, corrections were made in observed absorption spectra for interference introduced by the presence of known impurities.
Table I.
Characteristic Band Positions for Various Functional Groups
Functional Group OH (alcohols) CHO (aldehydes) COOH (acids) COO (esters) CO (ketones) -0(acetals and ethers) Olefinic Group RCH=CHa
Wave Length, Microns 2.98 * 0.02 3.63 * 0.01 3.82 (broad band) 5.71 (O.O1)aa 5.78 * (0.01) 8.8
R‘C=CH* R’/ trans-RCH=CHR’ cis-RCH=CHR’
11.24
R‘C=CHRff R‘/ Except for first few members of series.
10.05,10.98
* 10.36 *
f
(0.02)
(0.02)
(0.02) 14.0-14.6 (variable) 11.9-12.7 (variable)
DEVELOPMENT OF METHOD
Characteristic Band Positions. Study of the spectra of various compounds has shown that characteristic band positions in the infrared region can be found for nearly all the major olefin and oxygenated compound types present in hydrocarbon synthesis naphthas. Band positions for various functional groupings are shown in Table I. The wave lengths shown are nominal band positions for aliphatic compound types, and do not necessarily apply to all other types of materials. Hence, a method of analysis based on the constancy of these band positions is dependent to a large extent upon the type of sample analyzed. Experimental work has shown that these bands are reproduced from compound to compound in the straight-chain aliphatic homologous series with sufficient constancy to allow their use in many types of analytical work. As has been found by other investigators (8, 7 ) who have employed characteristic absorption bands, some deviation from the nominal band position is usually encountered for the first and sometimes for the second members of a given series. Such deviations are usually insignificant in the case of total synthesis products because of the high proportion of compounds of higher molecular weight present in such samples. In the analysis of low boiling fractions, however, where there may be only one or two members of a compound type present, consideration must be given t o possible band shifts. It has not been possible to measure band positions for many compounds representative of each of the compound types shown in Table I, largely because of lack of availability of pure compounds. A qualitative check, however, on the constancy of the band positidns in hydrocarbon synthesis naphthas was made by inspection of band positions of the functional groups in distillation fractions of different boiling ranges. Band positions were determined for narrow distillation cuts covering the boiling range from 100 O to 600 ’ F., and in nearly every case the band positions of the functional groups in these fractions were found
ANALYTICAL CHEMISTRY
1000
to correspond closely with those found for the total sample. In order to check on the validity of the assignment of these bands to given functional groups, infrared analyses based on these bands for the individual fractions were checked against chemical methpds of analysis in cases where chemical methods were available. This was done extensively for the oxygenated compound functional groupings, and the data so obtained completely justified the selection of the indicated band positions. Functional Group Absorption Coefficients. Study of the absorption spectra for several types of olefins and oxygenated compounds has shown that the absorption coefficient (optical density per unit thickness of sample) decreases as the molecular weight of the compound type increases. In general, for the compound types investigated, the variation of absorption coefficient is, to a f i s t approximation, proportional to the inverse of the molecular weight. Hence, it appears that if absorption coefficients were expressed on the basis of a particular absorbing group and not on the entire molecule, one might obtain a constant coefficient for the functional grouping-Le., independent of the molecular weight. This has been found true in a number of instances by experimental check of a number of compounds of a certain compound type. In other cases constancy of coefficient was checked by assuming that the coefficient was constant, analyzing a sample on this basis, and cross-checking the infrared analysis against chemical methods for the given functional grouping. It is conventional practice in the analysis of liquid samples by the infrared technique to express absorption coefficients for a given sample cell thickness, using volume fraction-Le., effective thickness-of the absorbing molecule for the concentration term in the familiar Beer’s law. In other cases, however, concentration is expressed on a molecular basis-Le., moles per unit volume. If, however, instead of using either of these terms for concentration, one substitutes the concentration of the functional grouping, then the calculated absorption coefficient becomes the absorption coefficient for the functional group. The method of expression for the concentration of functional grouping selected as most convenient for analytical purposes expresses the concentration on a weight basis-i.e., weight fraction or weight per cent functional grouping in the sample. In order to meet the requirements of the Beer-Lambert law (that concentration either must be expressed on a molecular basis or must be proportional to the effective thickness), it is necessary to add another term to compensate for the deviation encountered by the use of the expression of concentration on a weight basis. I t can be shown by derivation that the following relationship is true, providing that absorption coefficients for functional groups are constant. K0.1’’ = functional group absorption coefficient =
( E (wt. fraction functional group) where P = density of mmple in grams per ml., KO.]’= Ko.1 K ( B G ) Oand . ~ ,KO., = d / c .
d = log
(2)
where IO = intensity of radiation transmitted through n-heptane in 0.1-mm. sample cell, Z = intensity of radiation transmitted through sample or blend of sample with n-heptane in 0.1-mm. sample cell, c = volume fraction of sample in solvent, and K(~o10.l= differential “background” absorption coefficient = differential in absorption between n-heptane and the “background material” in a 0.1-mm. sample cell (see discussion for explanation of this term). The value K O . ] ”is referred to hereinafter as the functional group absorption coefficient. Values have been calcu’ated for each of the functional groups at their own characteristic band positions and at the band positions of other functional groups. In order to illu$rate the con-
Table 11. Infrared Absorption Data for Alcohols a t 2.99 Microns
Compound Methyl alcohol Ethyl alcohol n-Propyl alcohol Isopropyl alcohol n-Butyl alcohol see-Butyl alcohol Isobutyl alcohol Blend of Cs-Clr alcohols Av.
Relative Absorption Coefficient for Alcohols“ 1.000 0.692 0.541 0.513 0.445 0.406 0.436 0.254
0.539
Relative Absorntion ~
%
Deviation from Av. $85.5 +28.4 0.4
+- 4.8 -15.6 -24.7 -19.1 -52.9 t28.9
coe-&-iient
F
for OH Group Devlatiotl [Wt. Basisla from Av. 1.000 4-1.7 0.998 +1.5 1.002 +1.9 0.970 -1.3 1.015 +3.3 0.926 -5.8 0.983 0.0 0.973 -1.0 0.983 k2.4
Absorption measurements made for blends of indicated alcohol(s) in n-heptane a t optical density range 0.4 to 0.5 in each case. 5
Table 111. Infrared Absorption Data for RR’C=CH$ Olefins a t 11.24 Microns Relative Absorption Coefficient for Olefins 1.000
Compound 2-Methyl-1-butene 0.676 2-Ethyl-1-butene 2,3-Dimethyl-l-butene 0.625 2,4-Dimethyl-l-pentene 0.400 2,3,3-Trimethyl-l-butene 0 , 5 8 1 2,3,3-Trimethyl-l-pentene 0.516 2,4,4-Trimethyl-l-pentene 0.418 3,3-Dimethyl-2-isopropyl1-butene 0.555 2,6,6-Trimethyl-l-heptene0.514 2,4,4,5-Tetramethyl-lhexene 0.530 2-Ethyl-1-hexene 0.581 Av. 0.581
%
Deviation from Av. +72.1 1-16.4 7.6 -31.2 0.0 -11.2 -28.1
+ -
Relative Absorotion Coeffikent for C=C Group
1.000 0.764 0.714 0.524 0.749 0.730 0.607
”(.
Devlation from Av. +31.1 0.1 - 6.4 -31.3 1.8 4.3 -20.4
+ -
4.5 -11.5
0.882
0.889
+15.6 +l6.5
-
0.876 0.660 0.763
+14.8 -13.5 t14.2
8.8 0.0 t17.4
stancy of these coefficients, absorption data for one oxygenated compound type (alcohols) and for one olefin type (RR’C=CH2) are given in Tables I1 and 111, respectively. In Table 11, absorption data are listed for a number of alcohols believed to be representative of the types found in hydrocarbon synthesis naphthas. Absorption coefficients are given for the alcohols and for the OH functional group. It will be noted that the OH functional group coefficient for the series of alcohols listed is relatively constant (+2.4%), whereas the absorption coefficients for the alcohols themselves decrease progressively as the molecular weight of the alcohol is increased. The alcohols listed were selected on the basis of their similarity to alcohols believed to be present in hydrocarbon synthesis naphtha, and therefore do not cover a wide range of molecular structures of aliphatic alcohols. I t is expected not only that certain alcohols with a centrally located OH group and with considerable branching of alkyl side chains would possess characteristic absorption bands somewhat displaced from 2.99 microns, but also that their absorption coefficients might differ appreciably from those observed for the alcohols shown. That this may be true has been indicated by Coggeshall(5), who has shown that the presence of large alkyl groups on carbon atoms adjacent to the OH group of phenols lessens the degree of hydrogen bonding and thereby causes shifts in the position and in the intensity of the OH band. The band used in the present studies for OH analysis is believed to be the one corresponding to the hydrogen-bonded
OH. A second illustration of the constancy of the functional group absorption coefficieht is given for the RR’C=CHs olefin type in Table 111. The functional group absorption coefficient approaches constancy with much less precision than was observed for the alcohols, and the average deviation of the functional group absorption coefficient is only slightly less than that observed for the average coefficient for the olefinic compounds themselves. However, the maximum deviation is somewhat
V O L U M E 20, N O . 11, N O V E M B E R 1 9 4 8 less for the functional group coefficients and the deviations are more random in nature than in the case of the average olefin coefficients; furthermore, the data suggest that much larger deviations would be experienced for the olefin coefficients for olefins of higher molecular weight than those studied. Nonconstancy of functional group absorption coefficients is, of course, a fairly serious problem in functional group type analysis and often limits the accuracy that can be obtained. There are numerous reasons why functional group absorption coefficients are not constant. Although it is not considered within the scope of this paper to discuss the reasons for these deviations, it appears desirable to mention several examples of deviations that have been encountered and handlcd successfully in analytical L\ork. Figure l illustrates a correlation of the absorption of alcohols a t the OH band position a t 2.99 microns with OH concentration. In the higher OH concentration ranges a linear rclationship is found, whereas in the lower OH concentration ranges there is a distinct departure from linearity. Thia deviation has been observed for numerous alcohols of the type encountered in hydrocarbon synthesis naphtha and is believed t o be associated with diminution in hydrogen bonding with increased dilution. As this effect is common to the alcohols used in this investigation, one can develop a general correlation of the type illustrated in Figure 1 and take account of the departure from Bccr’s law bv the use of a working curve ( 2 ) .
09 FIGURE
CALIBRATION
I
CURVE
/I
WEIGHT PERCENT OH I N SAMPLE
h second type of deviation is the nonconstancy exhibited by RCH=CH2 type olefins, especially the straight-chain olefins. Although insufficient pure compounds are available to test this correlation exhaustively, it has been shown that there is an increase in functional group absorption coefficient with molecular weight for the Cs to Cg a-Olefin5; as the molecular weight is further increased, the functional group coefficient for this olefin type approaches a constant value. This type of deviation can be handled for narrom-boiling fractions by using a functional group absorption coefficient appropriate for the fraction analyzed, or, in the case of wide-boiling mixtures, by the use of a weighted average functional group coefficient determined by proportioning
1001
the coefficient on the basis of the relative amounts of olefins of different molecular weights normally present in samples of the type analyzed. In Table IV are listed summarized absorption coefficient data for all the oxygenated and olefin compound types included in this investigation, together with uncertainty values (average deviation from the average) for each type; the uncertainty values are in some cases based on a limited number of determinations and are not coilsidered too firm, especially when the doubtful purities of some of the calibration standards are taken into account. The uncertainty values given are not indicative of the total uncertainty of a given determination, but only of that portion of the total uncertainty introduced by deviations of a given functional group a t the band under consideration. Interferences. In many cases mutual interference is encountered between different functional groupings, and these interferences must be taken into account in quantitative analytical work. It will be noted from the data in Table IV that in many cases these interfering absorptions are small and have a negligible effect upon the over-all accuracy of the analysis. Tn other caqes, hon-evrr, the interferences are large, and serious crrors can result in the analysis if the interference coefficients are not properly evaluated. There are several limitations of the method attributable to the interfering absorption of other groups. One of these limitations is in the determination of the CHO grouping of aldehydes, in that absorption attributable to the COOH group of carboxylic acids is very high at the 3.63-micron CHO band position. Not only must the COOH content of a sample be accurately known in order that proper correction for its absorption can be made, but it is necessary for accurate analysis that the concentration of COOH grouping not be large with respect to the concentration of CHO grouping. I t has been found in practice that if the concentration of COOH grouping is equal to or greater than that of the CHO grouping, direct analysis for the CHO grouping is not feasible and can be performed only after the acids have been removed by chemical treatment. The interference functional group absorption coefficient of the OH grouping a t the 10.36micron band of tmns-RCH=CHR’ is high, and the average deviation for this interference coefficient is also high. For mixtures in which the alcohol concentration is not high, the approximate average interference absorption coefficient can be used without introducing serious error. (This is usually the case for hydrocarbon synthesis naphthas.) In the analysis of distillation fractions wherein specific alcohols are concentrated relative to the olefins, consideration has to be given to the particular alcohol present in order that errors due to the interference of the alcohols may be minimized. One of the most difficult phases of the analysis is concerned with the determination of the COO group of esters and the CO group of ketones. The lower molecular weight components of these two compound types possess functional group absorption bands a t slightly lower wave lengths than the nominal band positions. Although this effect is negligible in wide-boiling mixtures, it must be given consideration in the analysis of narr&-boiling fractions of relatively low molecular weight. For wide-boiling mixtures, the separation between these two bands is sufficient so that little error in the determination of one grouping in the presence of the other grouping is encountered unless one group is present in concentrations of 5 to 10 times that of the other grouping. However, it has been observed in the literature (2) and found also in this investigation, that the CHO grouping of aldehydes and the COOH grouping of acids also possess strong bands in the same wave-length region that is used for the COO and CO analysis. All those bands are attributable to the conimon CO grouping possessed by all these functional groups. The CHO band of aldehydes in this wave-length region is located a t about 5.75 microns, or intermediate between that of the COO group (5.71 microns) and the CO group (5.78 microns).
ANALYTICAL CHEMISTRY
1002
Table IV.
Functional Group
Av. Dev. Ka.1"
Ko.1" A.
OH CHO COOH
35 1.0 4.1
coo co c=c
0.2 0.2
Si1
Summarized Absorption Coefficients Uncertainty %OH/% Indicated Groupa
No.of Compounds Tested
OH Band a t 2.99~ 10.7 8 10.48 4 3 10.23 t0.04 5 10.03 3
...
G.
*0.001 *0.001
..
Indicated GroupC
coo co c=c
10.05 * O . 73 10.13
... ... ...
2 4 4
*0.006
*0.09 10.02
..
... ...
, .
coo co c=c
1.6 0.4 8.05 Nil Nil Nil
10.2
D.
COO Band a t 5.71~8
10.33 10.23
... ...
...
0.7 31.9 18.0 77.9 10.4 Nil
coo co c=c
E.
10.05 17.1 11.3 13.2 10.13 Nil
10.03 10.04 10.03
2 4 4 5 3
...
...
Indicated Groupf 1.0.001 10.09 10.02 10.04 t0.002
2
4 3 5 3
...
..
CO Band a t 5.78~9
5% OH CHO COOH
0.8 22.7 94.7 14.8 69.3 Xi1
coo co c=c
10.1 18.45 11.6 *0.6 13.0
...
GO/% Indicated Grouph 1.0.001 10.12 10.02 +0.01 *0.014
2 4 3 5 3
..
)C=CHn
0.8
... ... ... 10.05
1
1 1 2
...
...
...
13.6
1 1 12
+0:0i
1.7
10. 4
2
*o.oos
0.5
...
;C=CHR"
1.3
Y
Functiowl Group OH CHO COOH
coo co
8.6 2.6 5.3 1.5 1.9
...
14.5 *l.O A0.4 *l.? .to. 9
I
.
Functional Group OH CHO COOH
coo co
Br No. RCH=CHi/% Indicated Groupm 4.2 1.9 7.9 0.8 0.6
13.3 10.9 *o. 1 t0.4
11 4 3
!
10.2 R -"
H.
Summarized Absorption Coefficients for
1
11 4 3 5 3
\
10.44 10.12 10.01 h0.05
10.03
C=CHz Band a t 11.24~
R"
Olefin Type trans-RCH=CH,R' cis-RCH=CHR RCH=CHz
R
\C=CHz
Br No. RR'C=CHx I Br No. Indicated Typen 0.8 0.4
...
11 1
...
2 5
+0.'6
7
4:01
43.1
16.1
11
10.14
0.5
...
1
...
R/ R 'C=CHR"
Functional Group OH CHO COOH
coo co
3.8 1.5 4.9
0.9 0.6
12.5 10.7 10.3 t0.7 +o. 4
11 4 3 5
3
*O.Oh
\C=CHR"
Band a t 1 2 . 1 7 ~ ~
R"
... Br No. transRCH=CHR'/% Indicated Groupk *0.8 10.2 1.0.1 10.2
Br XO. RR'C=CHz/ % Indicated Group* t0.39 10.11 10.05 10.1 1
R I. Summarized Absorption Coefficients for
\C=CHRn
(
R
+0.001
R' R
R'
0. 7 0.8 50.3
R/
Br No. transRCH=CHR'/ Br No. Indicated Type! 35.6 1.1 1.6
Br No. RCH=CHs/ Br No. Indicated Type1
\
R';C=CH2
...
F. trans-RCH=CHR' Band a t 10.36~'
Olefin Type trans-RCH=CH,R' cis-RCH=CHR RCH=CHz R
Band a t 10.98~
R/
% COO/% OH CHO COOH
Summarized Absorption Coe5cients for RCH-CHz
Olefin Type trans-RCH=CH,R' cis-RCH=CHR RC H= C HZ R
...
C. COOH Band a t 3.82~
OH CHO COOH
No. of Compounds Tested
Av. D,ev. Ko.1
*O.OOi
70 CHO/% 3.0 8.0 10.2 Nil Nil Nil
Ko.~"
10.02b 10.01
B. CHO Band a t 3.63~
OH CHO COOH
Functional Group
Olefin Type trans-RCH=CH,R' cis-RCH-CHR RCH=CHt R 'C=CHz
10.2
Br No. RR'C=CHR"/ Br, No. Indicated Typeq 1.3 1.0
...
0.7
10.2
...
1 1 7
10:01
...
1.4
==0.2
2
-0.01
...
1
...
R/ R 8 Nominal band position. Methyl acetate and ethyl acetate absorb a t slightly lower wave lengths. I Average deviation of &.I" iKO.,"of COO group. Nominal band position. Acetone and methyl ethyl ketone absorb a t slightly lower wave lengths. FAverage deviation of Ko.i" iKo.1" of CO group. i Band position for trans-RCH=CHR' olefins found to be constant to 10.02 micron for synthesis naphtha fractions corresponding to carbon number ranges on the olefins of Cs to CIS. 160 i Average deviation of KOJ"X 24 t Ro.1" of trans-RCH-CHI. k Average deviation of Ko.1" i K0.i" of trans-RCH=CHR'.
'C=CHR"
14.4
R/ RR'C=CHR"/% Br No. Functional Group OH CHO COOH
coo co
3.4 1.6 2.5 0.9 0.6
*0.4 11.0 10.5
10.3 10.4
2 4 3 5 3
Indicated Group7 10. 19 +0.47 10.23
1.0.14 10.19
V O L U M E 2 0 , NO. 11, N O V E M B E R 1 9 4 8
1003
Table IV (Continued) Funotional Group
Ko.1"
Av. Dev. &.I"
No. of Compounds Tested
Functional Group
Summarized Absorption Coefficients for cis-RCH=CHR' Band a t 14.1389 Br No. ciaRCH=CHR'/ Br No. Indicated Olefin Type Tywf ... trans-RCH=CH,R' 0.0 6s-RCH-CHR 8.6 ... RCH=CH: 0.2 J.
R
Ko.1"
Av. Dev. Ke.1"
No. of Compounds Tested
K.
Summarized Absorption Coefficients for cis-RCH=CHR' Band a t 14.3480 Br No. cisRCH=CHR'/ B r No. Indicated Olefin Type Typew trans-RCH=CHR' 0.8 1 cis-RCH-CHR' 11.2 1 ... 1 RCH=CH* 0.6 ...
...
t.o.15
2
1
Functional Group OH CHO COOH
coo co
Br No. ciaRCH=CHR'/% Indicated Group"
6.8 0.3 0.9 0.1 0.2
t0.5
2 4
*0.2 ==o. 1 10.2
3
t o . 15
4 3
Functional Group OH CHO COOH
coo co
7.1 0.6 1.2 0.2 0.2
t0.6 to.2 tO.5
10.1 t0.03
2
4 3 4 3
to.01
... Br No. cisRCH=CHR'/% Indicated Group t0.36 to.12 10.30 10.06 t0.02
Average deviation of Ke.1" iKO.,' of RCH=CHt. 160 Average deviation of 9 0 . 1 " X - i Ko.1" of RCH=CHs. 24 Average deviation of K,.i" iKe.1" of RR'C=CHg. Average deviation of Ko.1" X ' O iKe.1" of RR'C=CHz. 24 Nominal band podtion. This band is variable from 11.9 to 12.7 microns. Average deviation of Ko.1" iKi.1" of RR'C=CHR". Average deviation of Ko.1' X iKe.1" of RR'C=CHR".
a Side of nominal band a t 14.34 microns. This position used to minimize aromatic interference. L Average deviation of K0.t" iKe.1" of cis-RCH= CHR'. 160 u Average deviation of Ke.1" X - ; i ~ i Ko.l" of cis-RCH=CHR'. &* Kominal band position. Little information available on constancy of position. However, analyses of pilot unit samples containing large amounts of RCH=CHR' olefins have checked well with chemical bromine number. w Average deviation of Ke.1" i Ke.1" of cis-RCH=CHR'. 2 Average deviation of Ko.1' X 160 i KO.,"of cis-RCH=CHR'. 24
The band position for aldehydes is also variable for the first few members of the straight-chain aliphatic homologous series. The band position for the COOH group of acids in this region
is located a t a slightly higher wave length (about 5.79 microns) than that for the CO group of the ketones. The bands exhibited by each of these functional groupings are strong, and in order to make successful determinations of the COO grouping of esters and the CO grouping of ketones, it is necessary that proper correction be applied for the interference of the aldehydes and the acids. Considerable success has been obtained through the use of band width curves. It has been found that the bands exhibited by all the esters investigated are similar in shape. This is also true for the ketones, acids, and aldehydes; furthermore, there is a great degree of similarity between the bands for the given groupings. Figure 2 shows the band width curves for several esters superimposed on one graph. Although the wave length of peak absorption for each of these esters was not quite the same, the band shapes are reasonably constant from ester to ester. The constancy of band shapes makes possible the analysis of mixtures containing all these functional groupings, providing the peak location for each group can be found. For samples of hydrocarbon synthesis naphtha, nominal band positions have been developed for each group and, through the use of band width curves, estimates of interference coefficients for each of these functional groupings a t the other band positions give satisfactory results when used in the analysis of mixtures containing all the groupings. Background Absorption. In addition to correcting for mutual interference effects, it is also necessary in determining compound types to take into account "background" absorption caused by the presence of substances other than those determined by analysis. In general, these materials can be divided into two classes: paraffins and aromatics. I t is known that hydrocarbon synthesis naphthas consist largely of straight-chain molecules. Consequently, a straightchain paraffin containing none of the functional groupings analyzed for serves as a suitable material for estimating paraffinic
L
m
n o P II
r
q4
1004
ANALYTICAL CHEMISTRY
Table V.
Background Absorption Data for Total Hydrocarbon Synthesis Product
Band Position, 2.99 3.63 3.82 5.71 5.78 10.36 10.98
p
Functional Group OH CHO COOH
COO
CO trans-RCH=CHR' RCH=CHz
n
)C=CHz
11.24
Background Density for 0.l-Mm. Cell 0.000 -0,032 -0,007
+0.030 +0.030 -10,074 T O 023
absorptions a t this wave length and a t the two olefin band positions are known. The relative absorptions for the aromatics in hydrocarbon synthesis naphtha a t these three band positione are illustrated in Table VI. A convenient method for determining the relative absorptions of the aromatics a t the three band positions consists of measuring the differential absorption between n-heptane and a sample of deoxygenated, selectively hydrogenated synthesis naphtha; the differential absorption corrected for paraffinic background absorption is attributable to aromatics.
f0.064
R' R 12.17 14.13 14.34
'c=cm), R/ c{s-RCH=CHR; czs-RCH=CHR
0.000
Table VI. Aromatic Interference Data for Total Hydrocarbon Synthesis Product Relative Absorption CoefFicient
0.000 0.000
Band Position,
background absorption-Le., the differential absorption between a straight-chain paraffin and a synthesis naphtha sample gives, to a first approximation, the absorption attributable only to olefinic or oxygenated functional groupings. With certain exceptions, n-heptane can be used to approximate the paraffin background absorption in synthesis naphthas. The exceptions are discuased in the following paragraphs. The absorption of the background a t the oxygenated compound functional group bands in some cases differs appreciably from that for n-heptane. Although the differential absorptions between n-heptane and the background at the OH band position and a t the COO and CO band positions are usually negligible, the differential absorptions a t the CHO band position and the COOH band po4tion are large enough to be significant in the analysis of samples containing small amounts of these functional groups. Furthermore, the differential background absorption a t these two latter band positions has been found to vary with the molecular weight and type of naphtha fraction analyzed. Variation in differential background absorption is considered one of the most serious limitations of the analysis, and can usually be corrected only by experience with the particular type of samples analyzed. Determination of differential background absorption coefficients at the oxygen functional group bands is usually made by measurement of the differential absorption of a deoxygenated synthesis naphtha against n-heptane in a sample cell of known thickness. The background absorption a t the olefin band positions can be measured conveniently by measurement of the differential absorption between n-heptane and a sample of deoxygenated, hydrogenated synthesis naphtha. In Table V are listed background absorption data for a typical hydrocarbon synthesis product. Hydrocarbon synthesis naphthas contain aromatics in addition to olefins, oxygenated compounds, and paraffins. &4lthoughthe aromatic content of synthesis naphtha samples is usually low, it is a variable quantity; therefore, absorption of the aromatics cannot be considered as a constant background term. The most serious interference from aromatics is encountered a t the band positions for RR'C=CHR" and cis-R'CH=CHR type olefins. Corrections for the aromatics a t other wave lengths used in the analysis me small and may be considered negligible. As the aromatic content is usually small, a simple correction method for this interference has been developed. The aromatic material in synthesis naphtha possesses a characteristic band a t 14.3 microns. On the assumption that relative aromatic distribution is fairly constant (although total aromatic content may vary), it is possible to estimate aromatic corrections a t the two olefin band positions mentioned above by making a reference absorption measurement a t 14.3 microns, provided the relative
2.99 3.63 3.82 5.71 .5.78
10.36 10.98
p
F'unctional Group OH CHO COOH
coo co
trans-RCH=CHR' RCH=CHz
for Aromatioc Nil Xi1 Nil
Nil Nil
Nil Xi1
R Nil
11.24
R'
R 'c=CHR"
12 17
0.185
R" 14 13 14.34
cw-RCH=CHR' cis-RCH=CHR'
0 210 1 000
SUMMARIZED TECHNIQUE FOR ANALYSIS AND CALCULATION
The general technique employed for the analysis of hydrocarbon synthesis naphtha samples for oxygenated and olefin compound types is conventional, except that analysis is made for functional groups instead of compounds and background corrections are made. %-Heptane is used for reference intensity (I,) measurements and corresponding intensity measurements are made on the sample a t the same wave lengths in the same cell. In cases where excessive absorption is encountered a t any functional band position, the sample is diluted with n-heptane to a proper optical density range. Absorption coefficients for the sample are then calculated from the data for the blend and corrected to a 0.1-mm. cell basis. Background corrections and aromatic interference corrections are made when necessary. The corrected absorption coefficients are then those coefficients representing the contribution of the functional groupings alone. The absorption coefficients, divided by the density of the sample (grams per liter) are substituted into linear simultaneous equations written for the absorption a t all the wave lengths involved, and the equations are solved for the percentages of the individual functional groupings. Some complication is introduced by the fact that the absorption of the OH groups a t the OH band position does not follow Beer's law, and determination of this component is handled by the use of a working curve. I t has been found convenient in most cases to solve the simultaneous equations involved by the method of successive approximations. Results for each of the oxygenated functional groupings are usually reported on a weight per cent basis-i.e., weight per cent functional groupings in the total sample. Results for the olefin functional group determinations are usually calculated over to equivalent bromine numbers; in order to do this, one simply multiplies the value obtained for the weight per cent C=C of each olefin type by 6.67. For routine samples approximately 4 hours are required for complete analysis and calculation.
V O L U M E 20, NO. 11, N O V E M B E R 1 9 4 8
1005
several research samples; olefin compound types in several synthetic and research samples; A . Oxygenated Compound Types, Weight Per Cent and both oxygenated and olefin Functional Synthetic Sample Synthetic Sample Synthetic Sample Hydrocarbon Synthesis Product Group A= Ca B -C B“ ___ compound types in a synthetic Type Present Found Present Found Present Found Chemical Infrared Chemical Infrared mixture that was submitted “blind” during a cross-check program. The analyses give an indication of the accuracies and reproducibilities to be B . Olefin Compound Type8 expected of the method under Hydrocarhon Synthesi3 routine operating conditions. Synthetic Sample D Synthetic Sample E Synthetic Sample Product A , Some of the analyses include Wt. $%” . Wt. % C Br N0.c Br No. Olefin Type Present Found Present Found Present Found Chemical Infrared determinations of olefins of rrans-RCH=CH,R‘ 6.1 6.6 4.7c 7.3 2.7 4.7 , .. 7.2 the type RR’C=CR”R”’. 2.2 ... 4.2 2.4 12.OC 16.2 cis-RCH=CHR 15.9 21.0 67.4 ... 56.1 70.1 RCH=CHs 44.4 34.0C 35.1 43.4 Compounds of this type are R\ not determined by direct 33.6 29.0 25.6 21.2 1.4 2.6 ... 5.3 measurement in the infrared method but can be estimated €7 by difference between the total ... 23.7 20.2 0.7 0.9 , . . 3.3 ‘c=c / H bromine number and the sum R‘ ‘R” of the calculated bromine __ ___ numbers for the other olefinic Total (sum) 100.0 100.0 100.0 100.0 77.3 77.8 72.8 76.1 ., , 99.2 , , , 108.4 ... ... ... ,.. Total before norm. types. The analysis shown C. Analysis of Synthetic Blendd for Olsdns and Oxygenated Compound Typea in Table VII, C, included Weight % Oxygen Br N o . - a determination of the funcFunctional Group Synthesis Anal>sis Olefin Group Synthesis .halyzir tional group of acetals. This 57.5 56.2 RC H= C H2 OH 1.42 1.96 analysis has not brcn disR CHO 1.16 0.82 cussed in detail becauae such COOH 1.14 I 06 ‘C-=CH> 7.7 11.1 compounds have not been coo 2.42 2 73 dl found in significant concenco 44 1 czs-RCH=CHR tranu-RCH=CH,R’ } 9’9 7 0 -001 92 1 54 1 trations in the types of samR 7.6 9.8 ples analyzed ; however, the \C=CHR” R _. general procedure for the R \ /R” determination of acetals or C=C 9.1 ( 5 , l )e ethers is essentially the R‘ R”’ -same as discussed above for 90.5 90.5 the other oxygenated coma Blends prepared from amyl alcohol butyric acid, ethyl acetate, methyl ethyl ketone, butyraldehyde. b Sum of values for aldehydes and keiones. pounds. and 2-methyl-2-butene. “2-Octene“ used for C Blends prepared from 1-ootene, 2-octene, 2-ethyl-1-hexene Although the method in a nthesis contained 34% of 1-octene; this factor was taken intd consideration in computing composition of syntEetic mixtures. its present form is useful, d This synthetic contained 60 volume % olefins (14 components), 30 volume % oxygenated compounds (13 components), 2 volume % aromatics (1 comoonent). and 8 volume c; Daraffins and naphthenes (2 components). it has definite limitations e Value obtained b y differenre. and is not generally applicable to a wide variety of Table \ - I l l . Olefin Types in C6 Fractions of Synthesis complex samples without apNaphthas and Refinery Cracked Products propriate modification