Carbon-Hydrogen Groups in Hydrocarbons Characterization by 1.10- to 1.25-Micron Infrared Absorption R. R. HIBBZRD AND A. P. CLEAVES .lhtional .Idvisory C o m m i t t e e f o r .leronautics, Lewis Flight Propulsion Laboratory, Clecelund, Ohio
The aberage numbers of primarj CH3, secondary CH,, and aromatic CH groups in hldrocarbons are determined bj absorption spectroscopy in the 1.10 to 1.25 I* (8000 to 9000 cm-1) near-infrared region. A slightly modified, small glass prism spectrometer is used. Because the amounts of absorption of these functional groups are substantially constant in all hydrocarbons of a given class, calibration with a few- members of a class permits the anal>& of streams containing any number of components of the same class. Estimates on the amount of chain branching in paraffins and the degree of substitution on naphthene and aromatic rings can be obtained.
0
blends, and means oere sought to minimize the deviations observed hv Rose in the analysis of naphthenes and aromatics.
VER the past 25 years a considerable amount of research has been devoted to the development of methods for the determination of aromatics, olefins, and naphthenes plus paraffins in hydrocarbon samples. Procedures are now available 11-hich yield satisfactory accuracies in the determination of these classes in gasoline range samples and considerable progress has been made in developing methods for qimilar analyses of heavier stocks. To characterize hydrocarbon samples more completely it would be desirable to have, in addition, information on the amount of chain branching in the paraffin portions and on the degree of paraffinic substitution on the naphthene and aromatic rings. The degree of branching in paraffins is of especial interest in predicting the performance of this class of fuel in reciprocating engines. Sachanen ( 1 2 ) has revielyed the physical and chemical methods proposed for the determination of paraffin branching and, although some appear to be of value in limited applications, he states that “there are no accepted methods for evaluating the cxtent of branching of paraffins present in petroleum fraction..” One of the most promising approaches to the determination of the anioiint of branching appears to lie in the application of nearinfrared absorption spectroqcopy of the 1.10- to 1.25-micron region. In this region the primary, secondary, and tertiary paraffinic and the aromatic C-H groups have absorption bands which are second overtones to the 3.2- t o 3.6-micron fundamentals. Rose ( I I ) , in summarizing the results from a series of investiga?ions a t the Kational Bureau of Standards on the nearinfrared absorption spectra of hydrocarbons. has shown that the specific extinctions of the various functional groups are nearly constant in different hydrocarbons. Thiq near constancy allows analyses to be made for these groups in a variety of hydrocarbons after calibrating with a few typical hydrocarbons, and differs from the usual quantitative absorption spectroscopy in that prior calibration with the actual compounds being determined is not required. It also permits the determination of average concentrations of these functional groups in mixtures containing any number of components of a single hydrocarbon class. The data analyzed by Rose showed serious deviations in the determination of primary, secondary, and tertiary C-H groups in naphthenes of low molecular weight. The Bureau of Standards’ studies were made a i t h a large, custom-built, instrument ( 2 ) of a type not available to more modcstly equipped laboratories; it had tlvo glass prisms of 15-cm. base and 20 em. high in a Littrow mounting %ith a focal length of 2 meters and an effective slit width of 6 A. The present investigation was made to determine the value of a commercially available, smallprism spectrometer in analyzing hydrocarbons for their content groups, using measurements made in of the various C-H the 1.10- to 1.25-micron infrared. Paraffins, naphthenes, and aromatics were examined, both as single components and in
I\STRUMEYT A Y D MATERI4LS
9 model 12-A\Perkin-Elmer spectrometer ( 1 ) with the photoelectric amplifier of AIcAlister, Matheson, and Sweeney ( I O ) was modified for work in the 1- t o 2-micron region. A 60” prism, 60 X 75 mm. base, of DF-2 glass (obtained from the PerkinElmer Corporation) x a s installed with the temperature compensator for the rock salt prism left in place. The General Electric tungsten ribbon filament lamp T-10 operating a t 6 volts and 18 amperes used as the light source produced greater thermocouple response than did the Perkin-Elmer globar or a Western Union 25-watt concentrated arc. This lamp was clamped in an inverted position about 2.5 em. (1 inch) distant from the front left (facing the source unit housing) corner of the monochromator housing, and the latter was shielded by a water-cooled, sheet aluminum reflector. The spherical mirror was adjusted to bring the image of the filament on the entrance slit. A cuvette holder for two 2.54 X 2.54 em. cylindrical cells and a shutter were attached to the monochromator housing in front of the entrance slit.
-
\
Y
z
-3I-n n
z a m
> (3
a W
z w
L
-I
4
I
SLIT WIDTH
[MILLIMETERS)
Figure 1. Resolution of 8860 Cm.-’ Mercury Line as Function of Spectrometer Slit Width 486
V O L U M E 21, NO. 4, A P R I L 1949 Table I.
487
Hydrocarbons with Purity Not Established by Freezing Point Analysis n %o
Hydrocarbon n-Decane n-Tetradecane n-Hexadecane 2,3,4-Triniethjlpentane Cyclopentane Met liylcyclopentane Cyclohexane 3Iethylcyclohexane Eth~-lr)-clohexane Decalin Benzene Tetrnlin
Observed 1.4119 1 ,4294 1 ,4346 1.4032 1.4061 1.4092 1.4261 1.4223 1.4334 1.4757 1.5009 1.5460
n-Octadecane Biphenyl h-apht tialene
Literature
(4)
1.4114 1,4289 1,4343 1,4043 1 ,4064 1,4098 1.4262 1 ,4230 1,4332 1.4811 (cis) 1,4697 (trans) 1,5012 1.5438
3lelting Point, C. 27-28 28.0 fia-:n . " 69 . 9_ 79,5-80 80.3 _ I
Of the 55 hydrocarbons examined, 40 were either Bureau of S h n d a r d s samples or mat'erials prepared and/or purified by the Organic Synthesis Section of the Kational Advisory Committee for Aeronautics, Cleveland laboratory. These were all 97+ % in purity and most were better than 99y0 pure as determined by , The remaining 15 t,ime-temperature freezing point anal hydrocarbons vere obtained from a variety of sources and are listed in Table I n i t h their n2i or melting point values. Comparison of the observed and literature constants s h o w that moht of these compounds were of good quality. The uw of some hydrocarbons of lower purity should yield no significant inaccuracies, for the impurities were probably isomers with the same or nearly the same number of the various funct,ional groups. The carbon tetrachloride used as the solvent was of technical grade and had no measurable absorption in the 1.10-125 0 region. PROCEDURE
\Vave-lerigth calibration was based on the riiercury arc lines $J862, 8860, i112, and ti538 em.-' listed by JIc;2lister (Y), using a type lti200 Hanovia 125-watt, lamp as the source. [Wave numbers in inverse ceiitinieters (cm.?) are used hereafter in place of n-ave lengths in microns ( p ) to conform with the major portion of the previous work published for this spectral region.] Improper temperature compensation of the prism required frequent checking of wave numbers. This was facilitated by setting up a calibration curve in which frequency in wave numbers (cm.?) was plotted a$ a function of microdrive displacement from the 8860 cni. mercury line. I n malting wave nuniber calibrations, readings w r e taken every 0.5 scale division of the microdrive. This corresponds to 10 em.-* and represents the maximum uncertainty in wavc number assignments. The performance of the monochroniator was tested with the procedure used by Hogness, Zscheile, and Sidnell ( 6 ) . At several slit widths the thermocouple response was measured for a series of wave number settings taken across the 8860 cm.? mercury line and the half-energy band width was plotted as a functiou of slit viidth in Figure 1. The prismatic resolution (at zero slit width) is seen to be 13 c m - l and the incremental resolution for 0.10-mm. slit to be 70 cni-'. Absorption studies were made at 0.0125-mm. slit where the half-energy band width is 22 c ~ i i - ~ . 16
.I4
.I 2
lleasurements of hydrocarbon ahorptioiis were made in carbon tetrachloride solution with concentrations of 100 to 300 100 grams per liter to yield extinctions E = log % transmitt;=) in the 0.2 to 0.8 range. Solutions of the hydrocarbons were compared with the solvent alone in matched 2.54-cni. cclln. JIolecular extinctions were calculated on a mole-cni. basis
(
E
where c = concentration of hydrocarbon in moles per liter and With pure hydrocarbons the known molecular weights were used t o calculate molecular extinctions, but with the blends of pure hydrocarbons and with technical misturea the molecular weights were determined cryoscopically in either benzene or nitrobenzene. An average error of 1.170 and a masimuni error of 3.2y0 were obtained in the determination of the molecular weight of 20 hydrocarbon blends. Single readings of extinction made a t the bmve numbers of maximum absorption were not always reproducible, but Tvhen several readings at 0.5 scale division intervals (10 cm.-l) \vere taken acrossthe absorption peak and the three highest extinctions averaged, the average values checked to within 2.070. Where extinctions were measured, not at a wave number of maximum absorption but a t a preselected wave number, this same technique was followed; readings were made a t the desired wave number and at 0.5 scale division both above and below this point and the extinctions were averaged. Concentrations of the various C-H groups were determined from the molecular extinctions by simultaneous solution of equations of the type
00
l-
X W
LL
a
= =
d (moles CHs/mole) C (moles CH3/niole)
+ B (moles CHz/mole) + D (moles CH&nole)
where €1 and €2 are the molecular extinctions of a hydrocarbon a t the two wave numbers used for analysis. m,l-l and ciii.,-l; A and B are coefficients equal t o the estinction per nioie of respective functional group a t em.,-'; and C, D are the same a t cm.2-1 If the extinction per mole c f functional group was rsactly the same in all hydrocarbons of a given class, Coefficients d,R , C, arid D would be constants and could be precisely evaluated from d a t a from any two compounds. Because this is not true, the coefficients were evaluated by application of the method of least squares to estinction data from a number of compounds of each class. Substitution of the molecular extinctions for each conipound in the proper equations yielded concentrations in terms of moles of each group per mole of hydrocarbon. These were compared with known values and individual and average deviations were calculated. Blends were prepared and similarly analyzed. T h e concentrations of functional groups were also determined in several petroleum fractions.
.IO
i
g
-
d = cell thickness in em.
€1
F
0
100
70transinittarice CLi
€2
w
=log-----
06
J
0
z
.04
.02
C RESULTS
WAVE NUMBER, Crn-l
Figure 2.
Typical Absorption Spectra 8000 to 9000 cm.-l
Twent,y hydrocarbons were scanned between 8000 and 9000 cm. -I; Figures 2 and 3 show the spectra of a few typical examples. Absorption maxima for the paraffinic CH, group fell between 8220
488
ANALYTICAL CHEMISTRY
Table 11. Determination of CHI and CH, Groups in Paraffins on Basis of Extinctions Measured at Wave Number of Maximum Absorption Paraffin n-Heptane n-Octane n-Decane n-Dodecane n-Tetradecane n-Hexadecane n-Octadecane 2,3-Dimethylhexane 2,4-Dimethylhexane 2,5-Dimethylhexane 2.2,4-Trimethylnontana r_._"_.__
2,2.5-Trimethylhexane Average error in number of groups
Number of CHn Groups Found True Difference
Number of CHz Groups Found True Differencr
2.17 4.06 3.68 3.92 4.93
2 2 2 2 2 2 2 4 4 4 5
-0.13 0.03 -0.01 -0.01 -0.14 0.00 0.17 0.06 -0.32 -0.08 -0.07
4.99 6.24 8.16 10.22 12.05 13.68 15.82 2.37 2.64 2.46 1 .oo
5.20
5
0.20
1.91
1.87 2.03 1.99 1.09 1.86
2.00
0.10
1
-0.01 0.24 0.16 0.22 0.05 -0.32 -0.18 0.37 0.64 0.46 0.00
2
-0.09
3
6
8
10 12 14 16 2 2 2
0.23
Table 111. Determination of CHI and CH, Groups in Paraffins on Basis of Extinctions Measured at 8235 and 8360 Cm.-' Parafin n-Pentanr n-Heptane n-Octane n-Decane n-Dodecane n-Tetradecanr n-Hexadecana n-Octaderane .~. ~ - ~ ~ . . 3-hlethylhexane 3-Ethylhexane W-Dimet hylbutane 2,3-Dimethylhexane 2,4-Dimet hylhexane 2,5-Dimethylhexanr 2.2,3-TrimethyIoentane .~ 2,2,4-Trimethylpentane 2,3,3-Trimrt hylpentane 2,3,4-Trirnethylpentane 2,2 5-Trimethyldexane Average error in number of groupc ~~~
~
Number of CHa Groups Number of CHr Groups Found True Difference Found True Difference 2.17 2.02 2.20 2.20 1.90 1.87 2.16 ~ 1 . 6 .4 3.00 3.51 4.04 4.26 3.84 4.16 4.67
2 2 2 2 2 2 ~
4 4 4 5
0.17 0.02 0 20 0.20 -0.10 - 0 13 -0.36 0.00 0.51 0.04 0.26 -0.16 0.16 -0.33
3.43 5.42 6.05 7.97 10.32 12,15 13.74 16.12 3.09 3.76 0.97 1.98 2.52 2.28 0.57
10 12 14 16 3 4
0.20
0. iS
3 5 6
8
1
2 2 2 1
0.43 0.42
0.05 -0.03 0.32 0.15 -0.26 0.12 0.09 -0.24 -0.03 - 0 02 0.j2 0.28 -0 43
~~
.5.20
5
0.85
1
- 0 . 15
4.73
5
-0.27
0.67
1
-0.33
4.24
.i
-0.26
0 9.3
0
5.55
5
1..54
2
0.55
0 . 93
--0.46
and 8280 cm.-l, the parafink C'& group betiveeii 8360 and 8390 cm.-l, and the aromatic CH group between 8670 and 0 24 0.28 8775 cm.? Although Rose reported analyses for the tertiarv __ _CH group based on measurements at 8160 cm.-l, no maxima were Table IV. Determinatioll of CH:1 and CHz Groups in observed at this wave number and no determination of this group Blends of Paraffins was made. Absorption maxima for the paraffinic CH, and C H I Number of CHBGroups S u m b e r of CH? Groups groups showed small, although definite, variations, and the Parafin Blend K O . Found True mfference Found T r u e Difference aromatic CH band showed a considerable shift towards lowrr 2.83 P-1 2.81 0.02 4.35 4.08 0.27 wave numbers when the amount of ring substitution increased. P-2 3.43 3.53 -0.10 3.70 3.49 0.21 P-3 3 . 4 4 3 . 3 4 0.10 3 . 1 7 3 . 0 3 0.14 Molecular extinctions of 12 paraffins, taken at the Paraffins. P-4 3.66 3.81 -0.75 2.15 2.15 0.00 wave numbers of maximum absorption or at the inflection points P-5 4.10 4.ER -0.05 1.72 1.76 -0.04 nere first used t o evaluate where no maxima were resolved, Average error in number of groups 0.08 0.18 coefficients A , B , C , and D . Using these coefficients the molei of CH3 and CH2 groups per mole were determined for the 12 Table V. Uncorrected Det-ination of (ZH, and CHZ compounds and are shown in Table 11. The average deviations Groups in Naphthermes of 0.10 CHa group and 0.23 CH, are the bame as those shown by (On ba-ii of extinctions measured a t 8235 and 8860 cm.-l) Rose for 9 normal paraffins and are somewhat less than his deviaNumber of CHI Groups Number of CHr Groups tions of 0.3 CH3 group and 0.5 CH2group obtained on 18 branched Kaphthrne Found True Difference Found T r u e Differcncr paraffins a t the same wave numbers. These average deviations 0 3.17 Cyciopentane 3.17 1.72 D -3.28 show the order of accuracy that can be expected when the Methylcyclopentane 3 . 4 1 1 2.41 r.78 4 -2.22 I 2.05 Ethylcyclopentane 3 ,05 2.94 5 -2.0ii number of CH, and CHI groups is determined in a relatively pure 2.04 4 - 1 !(ti Isopropylcyclo3.76 2 1.76 uentane paraffin in the near infrared and where extinctions can be measCyclohexane 1,5? 0 1.59 5.41 - 0 . .iY ured at wave numbers of maximum absorption for thr single hfethylcyclohexane 2.20 I 1.23 4.28 -0.72 Ethykyclohexane 2.21 1 1.21 0.22 6 -0.78 components. 3-Methylbicyclo3.34 1 2.34 7.53 9 - I 4i However, the observed variation i i i nave numbers of absorphexyl 0 2.13 30.20 12 - 1.80 tion maxima for different niembera of a gjven class made it 1,2-Dicyclohexyl2,l3 ethane necessary to preselect the analytical wave numbera for the 10.89 13 -2 I I 1,3-Dicyclohexyl2.32 0 2.32 propane determination of CH, and C'H2 groups in blends. This also 0 1.40 7.69 8 -U.31 Decalin 1 40 simplified the determination of extinctioni, as measurementi 6.80 ? --l)2el Isopropyl Decalin 3 14 2 1.14 mere required only at each analytical wave number rather than Average error in niinihrr of group-. 1 .00 1.45 the larger number needed to e-tablish the maximum values used in Table I. As thc average wave numbers for maxima absorption in 12 paraffins nere found to be 8235 and 8360 cm.-', these were' bet\veeri 0.7 and 2.4 and the group analyses show but minor chosen for the determination of CH, and CH, groups, respectively, systrmatic deviations as a function of this variable. These and extinctions for 19 paraffins were measured. The coefficient* deviations also show the order of accuracy that might be expected nere again established by least squares and the number of CH? in the analysis of technical paraffin mixtures. and C H I groups was calculated. The results are shown in Table Naphthenes. Because no method for the quantitative separaI11 and, as expected, the deviations are somewhat greater than tion of naphthenes from paraffins is known, and no characteristic those shown in Table 11, for the extinctions of each component naphthene bands can be reFolved, it is desirable to analyze for were not necessarily measured a t the wave number that would be naphthcnes at the same wave number and using the same coefoptimum for that component. ficients as for paraffins. Therefore, ext,inctions were measured Because the trend of the deviations as functions of the amount for 12 C; t o C1, naphthenes at, 8235 and 8360 cm.? and the of branching or of molecular weight is small and less than the concentrations of CH3 and CH2 groups were determined using the average deviations of the individual compounds, the errors should same coefficients as those used for the determination of these largely cancel out in the analysis of blends of several components. groups in paraffins. As shown in Table V, the number of CH, This was found to be true, as shown by the data in Table I V groups so determined is always high by 1.1 t o 3.2 groups and the where 5 paraffin blends, each containing 9 or 10 components, were number of CH? groups low by 0.2 to 3.3 groups. Deviations are analyzed t o give average deviations of 0.09 and 0.13 CH3 and CIL groups, respectively. The blends were made up of CSto CIO greatest. for simple naphthenes and decrease with increasing alkyl substitution. paraffins with the ratio of CH, to CH2 (degree of branching) varied
489
V O L U M E 21, NO. 4, A P R I L 1949 Of several functions of riaphthenicity examined, the weight pvr oent naphthene ring as calculated by the A - d formula of ripkin, Martin, and Kurta (8) from density and molecular weight \bas found t o yield the best factors for correcting these deviations. I n Figure 4 the deviations in the number of CH, groups are plotted as a function of weight per cent naphthene ring. With Aero correction at 0% naphthene ring and a linear correction as a tunctiori of per cent ring assumed, straight lines were drawn through the origin for each type of ring. A remarkably good relationship for the cyclopentanes is evident, w e n though cyclopentarir vields a value of 114% naphthene ring to this type arialysi-. The single ring cyclohexanes showed somewhat mor? -ratter around a line of about one-half the slope of the cyclopvntants line. The dicyclohexyls fell near the cyclopentane linr arid the Decalins near the single-ring cyclohexane line. Figure ,5 is a Yimilar plot for the CH, deviations and shows the same genvial characteristics but of opposite Pign anti a i t h greater scattrr I [ the points. Hecause no method is known which would differentiate bet i r t'yn these claqses of naphthenri, average lines are drawn on hoth f i g u i ~to yield factors for correcting the number of CH? and C'$l, groups as a function of calculated weight per cent ring. \ v ~ \ h{brse factors the data in Table VI were derived for the 12 naphthrries. The average deviations are 0.55 CH3 and 0.86 ('Ha groups and the sign of the deviations for each type naphthtxnc t i that expected considering the compromise nature of the correction factori If only onr typr of naphthene wrrr prcwnt 01 i t
the relative amounts of two or more types were known, then the use of the appropriate lines from Figures 4 and 5 would yield greatly improved results. I n this case the average deviations for the 12 naphthenes shown in Table VI would be approximately 0.1 CH, and 0.2 CH, groups per mole-Le., the average displacement of the points of each class from the line drawn for that clans.
I
rd'
BENZENE
w
=- 00 0
50
IO0
WEIGHT PERCENT CYCLOPARAFFIN RING
-l
o
IX
0
06
Figure 4. Deviations in Number of CHI Groups as Function of Weight Per Cent Naphthene R i n g
w
The content of (:Ha a i d (,'€I2groups in five naphthcw blrrldc is compared wit,h the known amount in Table 1'11. Blend S-1 was made up of C, to Cil naphtht.nes and \vas rt~lativt~ly rich in 0 z cyclohexanes; 5 - 2 was of Cn to C g naphtherie- and rcllatively rich in cyclopentanes; and N-3 cont:tincd approximately equal 02 S-4 amounts of Cr t o C s cyclohexaries and eyclopt~ritanc~~. was a hlend of c ' , ~to c',, Ikcalins arid 2-ring c and X-,5 contained roughly equal amounts of all t h r a h o w four '00 02 04 06 08 901(103 types of naphthenes. The nveragr deviai ions arc' 0.20 C ' H , WAVE NUMBER. cm-1 and 0.37 ('112 groups. Figure 3. Tjpical Ahsorption Spectra Three blends of 21 coriiporicrits containing hot11 pitraffini arid riaphthc,iir* were aiialyzed by the same procedure arid the results 8000 to 9000 rm. 1 are shown in Table T'III. The avarane deviations are 0.25 C I t and 0.15 ('H2 groups per niolt~. Aromatics. The problem of d c t ernlining the -'I'ahle TI. Determination of CHA and CH, Groups in Saphthenes number of aromatic C-H groups \vas cornplicai ed Corrected for Per Cent Naphthene Ring by large shifts in the w a v ~number of thc absorp\veiyi,& N u n i h e r of CIIJ Groups X>i,uherof CH? (;roup* tion maximum in the 8750 em.-' aromatic ('-1i % DifferDifferregion. I t appear8 unlikely that tt single ~vavc. Sa~llitllrnt King Found T r u e enrr I'oand 'rsiir pn(.fl numbrr can be selected which will yield nrar-conCyclopentant. +14.,5 0.60 0 0.60 3.4i 5 -1.SH XIet hylcyclopentanc 83 3 1 54 1 0.54 3.14 ? - 0 86 atant values for extinction per unit aromat,ic C-H Ethylrycloyentane 72.8 1.42 1 0.42 4.11 .I -0.89 1soi)ropylcyclopentanr 63. 1 2 34 2 0.34 3.06 4 - 0 $14 group for various aromatic hydrocarbons. Inspec('yclohrxanr 103.5 - 0 73 0 -0 7 3 7.07 6 1.07 tion of the spectra of 8 aromatics indicated t h a t the Xlrthylcyrlohexanr 75.0 0.57 1 -0.43 5.48 3 0.48 average wave number for maximum absorption k:t hylr,yi-lohexan? 68..5 0.67 1 -0.33 6.30 F, 0.30 was near 8755 cm.-' and this was selected as tht. ~~-RIethylbirycloheryI 78.0 1.63 1 0,55 8 80 Y -0.20 I ,2-Dicyclohexylethane 74.2 0.46 0 0.46 11.39 12 -0.61 analytical wave number. Extinctions were oh1,3-Diryrlohexyl~ro~a~e 69.4 0.76 0 0.76 12.00 13 -1.00 tained for 15 single ring and 9 dicyclic aromatic. Decalin 103.0 -0.91 0 -0.91 9.34 8 1.34 Isopropyl Decalin 77.0 1.41 2 -0.54 8.04 7 1.04 at 8755 c m - ' and also at the 8238 and 8360 cm.-l Average error in nyrnber of g r o u p 0.56 0.86 CHZ and CH, wave numbers. The coefficients were e s t a b h h e d separately for these t,wo c l a w i
5J 0 4
Y
ANALYTICAL CHEMISTRY
490 .-
Table VII.
Determination of CH, and CH2 Groups in Blends of Kaphthenes
Ring
Number of CHa Groups DifferFound True ence
80.4 77.4 84.7 88.7 82.8
0.30 1.06 0.81 0.28 0.85
It eight Xaphthene Bird To.
s-1
N-2 N-3 x-4 S-5 Average error iri nriinbrr o f groups
7c
0.81 -0.51 1.02 0.04 0.91 -0.10 0.53 -0.25 0.76 0.09
Uunibel of ("1 Group. * LhfferFound True enrr
Table \ITI.
~
5.69 5 , 4 b 4.19 4.63 4.92 4 . Y 7 10.36 9.59 A,.% 6.40
0.20
0.23 -0.46 -0.05 0.77 0 19
0 :i;
Determination of CEI, and CH, (;roup- in Blends of Paraffins and Naphthenes Number of CH?
PS-1
PK-2
PN-3 Average error in number of groups
25.0 46.8 62.9
2.48 1.76 1.05
2.78 2.10 1.16 0.25
-0.30 -0.34 -0.11
DISCUSSIO\
____.
~-~~
-
ranges in the degree of branching are evident, as indicated by the ratioofC&toCH2.
3.44 4.21 4.96
3 52 4.26 5.27 0.1.5
-0.08 -0.05 -0.31
I n Figures 2 and 3, showing the 8000 to 9000 om.-' spectra of a fe\v typical hydrocarbons, there is an apparent lack of resolution in that the absorption maxima aw not well separated-for example, n-heptane shows maxima a t 8250 and 8380 cm.-I but the valley betrveen these is not so deep as might be expected for bands 130 cm.-l apart. However, this spectrum obtained with the modified Perkin-Elmer spectrometer having a half-int,ensitg band vidth of 22 em.-' is as vie11 resolved as the spectrum of the same conipouud measured with an effective slit width of 5 em.-' and reported by Liddel and Kasper ( 7 ) . Evidently both instruments reveal the true shapes of the bands which are envelopes of numerous rotational lines. Therefore, increased resolution is of little value in analytical work in this region and the PerkinElmer spectrometw with the indicated modifications is adequate. The same conclusion is reached by comparing the coefficients calculated by Rose (11) from data from the larger spectrometer n.ith those resulting from the present work. I n the determination of CH, and CH? groups in paraffins, coefficients il, E , C, and D of the equation;:
assuming zero absorption by the aromatic C H group at 8235 aud 8360 cm.-l Tables IX and X show the reiulti: obtainrd for the tw-o classes. Average deviations of 0.14 CHB, 0.25 CH2, and 0.42 aromatic C H groups iwre found for the singleTahk TX. Determination of CH?, CH2, and .iromatic CH Groups in Singlering aromatics and 0.12, 0.20, and 0.48 Ring iromatics for these same groups, respect,ively, for Nuinher of .Aromatic the dicyclic compounds. S u m h e r of C H BGroups S u m b e r o f CH? Groups C H Groups __ ___ Five blends cont,aining varying DifferDifferDifferAroinu tic Found True ence Found True ence Found True ence amounts of 7 singlc-ring aromatics were Benzene 0.03 0.13 0 0.13 5.86 -0.14 0 0.03 esamined (Table X I ) . Average deviaToluene 0 08 0.08 0 1.09 5.17 0 17 1 0.09 o-Xylene 0.35 0 2 0.33 1.87 4.31 0.31 -0.13 tions were only 0.09 for all groups, m-Xylene 0.11 0 2 -0.41 0.11 2.09 3.59 0.09 7 n-Xvlr.nr 2.25 0.06 oning to the canceling out of positive 0 -0.96 0.06 3.04 0.25 E:th;lb&xne 1 -0.11 1.04 1 0.17 0.89 5.17 0.04 and negat,ive deviat,ions for this parn-Propylbenzene 1.17 0.09 2 -0.12 5.09 1 1.88 0.17 0.34 4.34 1 o-Ethyltoluene -0.35 1.88 2 0 65 -0.12 ticular group of components. 7 2.00 0 .14 3 . 8 6 1 m-Ethyltoluene 0 . 0 1 0.00 0.99 Refinery Streams. Both the paraf0 0.90 -0.63 2.37 1,2,4-Trirnethylbeneene 3.03 3 0.03 0.90 3 . 0 8 0 0 . 2 5 0 . 72 0 . 2 3 p-Isopropyltoluene 3.28 3 0.08 fin-naphthene and the aromatic fracn-Butvlbenzene 0.92 4.81 3 0.23 -0.19 3.23 1 -0.08 tert-B;tylb&zene 0 . 3 0 2 . 8 5 6.0,5 -0.1.5 0 . 3 0 0 1 .05 3 tions of 6 samples from 3 crude sources p-tert-Butyltoluene -0.40 0 -0.16 4.39 3.84 0.39 -0.40 4 -0.34 were analyzed for their content,of the 3.54 4.74 -0 34 0 m-tert-Butyltoluene 0.74 4 -0.46 .Arerage error in various carbon-hydrogen groups. The 0.14 0.25 0.42 nuinbrr of groups silica gel separation of Gooding and Hopkins (6) was used t o yield the Table X. Determination of CHI, CH2,and Aromatic CH Groups in Two-Ring saturate and aromatic fractions and Aromatics the ultraviolet, spectrophotometric test Numher of .hamatic of Cleaves ( 3 ) shelved less than 0.170 S u m b e r o f CHa Groups S u m h e r of CH? Groups CH Groups aromatics in all but one of the saturate DifferDifferDiffer.\rolnatic Found T r u e ence Found True ence Found True ence fractions. The aronlat,ic fractions from 10 -0.05 0.24 9.93 0.24 Biphenyl 0.09 0.09 the high molecular weight cuts, Tom-0.22 9 -0.31 8.69 1.30 3-llethvlbi~henvl 0.30 -0.22 -0.05 9 0.67 9.67 1.9: 0.95 -0.05 ball lB, Bradford 2B, and Midway 2-n-Pro&biphenyl 7.32 0.01 9 -1.49 1.01 0 17 Phenyl-p-tolylmethane 1.17 3B, were twice calculated using co0.36 10 -0.11 9.89 0.81 -0.19 0.36 1,l-Diphenylrthane 0 15 0 . 0 9 1 0 . 13 10 1.09 0.93 -0.07 1,l-Diphenylpropane ef€icients established for both single0.61 0.68 8 8.61 0.68 0.06 0.06 Naphthalene 0.54 0 . 1 0 7 . 3 4 7 -0.10 1.92 ring and dicyclic classes. Table XI1 -0.08 Isoprop) lnaplithalene -0.33 4 0.02 3.67 4 02 0.08 0.08 Tetralin gives the results of these analyses, Average error in which are presented for information 0.20 0.48 0.12 numbrr of groups purposes only, as no alternative method for determining these values Table XI. Determination of CH?, CH, and iiromatic CH Groups in is known. The results for the aroBlends of Single-Ring Aromatics matic CH determination are uncertain and S u m b e r of Arornatic C H Groups are probably low by 0.5 t o 1.0 group. The .irornatic Kumher of CHI Groups Suuiber of C H I Groups Blend DifferDifferDiffertotal number of C atoms per mole as calculated So. Found True ence Found True ence Found True ence from the molecular veights and the weight A-1 1.02 0.99 0.03 0.35 0.33 0.02 5.09 5.16 -0.07 0.14 4.99 4.85 0.65 -0.07 0.58 0.04 1.29 1.25 A-2 per cent naphthene ring as determined by (8) 0.06 4.76 4.70 0.88 0.98 -0.10 0.18 1.64 1.46 .%-3 4 . 2 9 -0.06 4.23 0.69 -0.18 0.12 0.51 2.12 2.00 are also given. The concentrations of CH, and A-4 4.51 0.10 4.61 0.89 -0.06 0.83 1.66 1.38 0.08 A-5 CH2 groups in four streams which were essentially aromatic free were also determined in 0.09 0.09 0.09 numbererror of groups and are shown in Table S I I I , where wide
V O L U M E 21, NO. 4, A P R I L 1 9 4 9
491
Determination of CH,, CH,, and Aromatic CH Groups i n Cuts of Crude
Table XII.
Saturates
Aromatics CH8
200
atoms per mole 8.1 14.3
CHI
ring 39 29
per mole 1.91 2.70
mole 5.19 9.30
Cryoscopic mol. wt. 101 172
8 16
I05 178
21 20
7.2 12.6
2.39 2.69
4.50 8.38
154
8
113 c
5c3
8.0 c
2.52 c
3.91 c
177
A.S.T.11, Distillation Range, F. 203-3315 401-586
Aroiiiatics, Volume
Sample ~uiiihall1 .4 Toiriball I B Bradford 2 . i Rradford 2B .\lidnay 3.4 Xlidnay 3B
'rahle XIII.
C
R
Cryoscopic mol. wt.
33 32
93-337 340-542
108-391 372-634
38
Weight
%
114
per
Determination of CH? and CH2 Groups in Xonaromatic Refinery Streams
Wellght Sample Coiiiiiiercial mixed hexanel. Commercial octanes Alkylate Hot acid octane
\loleLulai V eight
89l/? 100 107 116
C Atoms per Mole 6.28 7.04 7.48 8.12
SaphPhene Ring IO 28 -9
- 11
CHI
(xroups per Mole
2.21 2.19 4.3.5 4.59
... 99
=
€2 =
d (moles CH3/mole) C (moles CHlImole)
CHI Groups per Ratio Mole CHa/CH2 n ;a " 3.78 " _ 4.03 0 54 1.81 2.40 0.81 6 67
E~~~~
W
d-3 I a W a
0---
+ -X
0----
CYCLOPENTANES CYC LOH EX A NE S 2 RING CYCLOHEXANES /' DECALINS /
8 3 0
[I
c3
-2
N
Lost 11.4
7.5 13.0
CHn per mole 0.34O 3.0P 3.31b
.Irornatic C H per mole 3.86'3 3.535 2.7Ob
2.4.ja 2.07b
3.50b
3.1Sa
1.88" 1.03b
0.62L" 3.5% 3.86b
3.34a 2.33a 1.27b
per
1.83" 2.89a
2.47b
=
A (moles CHa/mole)
+
B (moles ('Hzjmole)
+ B (moles CH?/mole) + D (moles CHZ/mole)
bjere evaluated using €1 and first measured a t the wave numbers of maximum absorption in the 8250 and 8400 cm.-' regions, and -econd, a t the preselected wave numbers of 8235 and 8360 cm. -1 The numerical values so obtained are shown in Table XIV along with those reported by Rose (Table V, 11) for measurement& made a t 8264 and 8400 cm. -1 The agreement between the coefficients found in this and in the wwlirr work is indeed good, considering that the latter were
'
CH3 mole 1.64a 2.3Za 1.96b
based on a considerably different group of paraffins and measured with a very different spectrometer a t slightly different wave numbers. In the case of aromatics the coefficients for the equations
+D (moles CH?'mole) = F (moles CHJmole) + moles aromatic CH
c~~~~ =
€1
C atoms per mole 7.6 12.7
E~~~~
C (moles CHr/niole)
G (moles CH?/mole)
+ H '(
mole
are different from those proposed by Rose, as shown in Table
xv.
Rose's values for coefficients A , R, C, and D were established from paraffin data. In this work fair agreement with the earlier work is found in the single-ring aromatic data but the two-ring aromatics show considerably greater variations. I n the latter case it is believed that a n insufficient number of CH, and CH2 groups were present in the limited number of compounds examined to establish these coefficients !vi th confidence. Inspection of Tables IX and X shows that the deviation. in the determination of the aromatic CH group are not random and that there are a t least two systematic trends. In 6 cases out of 5 where para substitution exists, the determined number of aromatic CH groups is low, while the presence of a tertiary butyl radical yields high results. In p-tert-butyltoluene these opposing effects largely cancel out. & i s it is probable that petroleum fractions contain more para-substituted aromatic3 than tertiary butyl compounds, the coefficients .horn above are likely to yield low results in the determination of the aromatic C H group. Greatly improved results would be expectrd in the analysiq of petroleum fractions if the coefficient3 were established
I
u u.
Table XlV.
0 [I
W
m
3 2
5
Coefficient .4
-1
B
m
+
0
/
z
+
./'
2>
, . IC D
+ a---
W
0
IO0
WEIGHT PERCENT CYCLOPARAFFIN
RING
Figure 5. Deviations in Sumber of CH2 Groups as Function of Weight Per Cent h-aphthene Ring
0.0070 0.0195 0,0270 0.0102
Table XV. Coefficient
Comparative \-slues
This Work .It cm.-l of maximum At 8235 and absorption 8360 cin. -1
Table V ( 1 1 )
0.0066 0.0194 0,0260 0.0098
0.00629 0,01989 0.02703 0.00736
Comparative Yalues
This Work Single Ring T x o Ring
Table V ( 1 1 ) 0.00629 0.01989 0.02306 0,0073.5 0.00327 0.00126 0.01121
ANALYTICAL CHEMISTRY
492 using data from the aromatics most likely to be present. The dicyclic coefficientslisted cannot be recommended for the analysis of this class in pet’roleum fractions, as they were derived from a group of compounds which are probahly not typical of the distribution found in petroleum. Data on substitutea Tetralins would be especially desirable, as this class is a probable component of crudes. Unfortunat,cly,no source of supply is known for pure members of this series. The data in Table X I I , showing the concentrations of tht. various types of CH,, CHI, and aromatic C H groups in cuts of crude, cannot be used to check the accuracy of the near-infrared method. However, the values determined for t,he satiirate fractions are a t least, possible, in t,hat combinations of purr compounds ran be set up which will satisfy all the observed data. For example, the values determined for molecular weight, prr cent napht,hene ring, and concentrations of CH, and CH, groups found in the saturate fraction of Tomball 1.A corresponds (wit,hin the limits of experimental accuracy) to a blend of 0.5 mole of ethylcyclohexane, 0.4 mole of methylheptane, and 0.1 mole of n-nonane. Similarly, a blend of 0.7 mole of dirthylhutylcyclohexane plus 0.3 mole of n-pentadecane would match the experimentally determined data for the Tomball 1B saturatra. Obviously thesc t n o streams are very much more complex than the examples shown and an almost infinite number of blends could be suggested which would also match the esperimcmtal data. Severtheless, these examples show that reasonahlr results are obtained in the ana,lysis of saturate fractions of crudes. Hoaever, in t,he case of the aromatic fractions, no combination of compounds can be postulated which will match the cxperimentally determined data. When calculated as single-ring aromatics, in all cases the sums of the CHs and aromatic C‘H groups are less than 6, the minimum value possible. A s previously suggested, the determined number of aromatic CH groups is likely to be low, owing t o a greater concentration of para-substitut,ed compounds in the fractions of crude than were in the calibrating hydrocarbons. The use of the two-ring aromatic coefficients for t,he three heavier fractions also yields data that cannot be matched by hypothetical blends. In thew latter cases the closest approach to a fit with the determined values for the functional groups are with blends of substituted Tetralinp. Therefore, this near-infrared method cannot be used t,o describe the “average” aromatic molecule, presumably because of neakncsses in t,he determination of the aromatic C H group. The analysis for CH, and CH2 groups in aromatics appear..: to be satisfactory and may be of occasional interest. The results from several paraffinic refinery streams shown iri Table X I suggest that near-infrared spectroscopy may br of value in predicting fuel performance in reciprocating engines. .Is is well kuown, the octane numbers of paraffins increase rather regularly with increases in the amount of branching, although this relationship is not quantitatively precise as indicated by F-1 octane numbers of 21.7, 26.8, 26.7, and 33.5 for thr three monomethylheptanes and ethylhrxane, all with the same numher of CHBand CH, groups. - h d . although errors of the ordcxr of 5 to 10 units might br obtained in the prediction of octane numbers of pure hydrocarbons from near-infrared determined concentrations of CH3and CH, groups, it is probable that a much closer correlation could be established between octane numbvr and near-infrared dRta for a stream from a given refinery unit or type of procctss. This type of oct,ane number prediction niight be of especial usefulness in evaluating small amounts of protiurtP obtained from laboratory scale experiments. SLW.MARY
With a slightly modified, commercially available, small prism spectrometer, absorpt;on data can be obtained in the 8000 t,o 9000 cm.-’ (1.10 to 1.25 H) infrared which, by using average molecular extinction values, allow estimat,es of the numhrr of CH:, CH2, and aromatic C H groups in hydrocarbons.
Judged by the results from blends of pure components, averagr deviations of less than 0.15 CH, and CH? groups ail1 be found in the analysis of paraffins and aromatics, and less than 0.4 of these same groups in naphthene or paraffin-naphthene hlends. Although the average deviation in the determination of the aromatic CH group was less than 0.1, it is believed that this accuracy should not be expected in the analysis of petroleum fractions. The concentrations of these groups can be determined in roniplex mixtures over wide molecular weight ranges. These data cannot he derived by other methods. LITERATURE CITED
(1) Barnes, It. B., McDonald, K. S.,\Villiams, 5’. Z., and Kinnaird. R. F., J . Applied Phus., 16, 77 (1945). (2) Brackett, F. S., and Liddel, U., Smithsonian Misc. Collection.. 85, No. 5 (1931). (3) Cleaves, A. P., “Ultraviolet Spectrochemical Analys:s for Aromatics in Aircraft Fuels,” NACA Wartime Report, ARR No. E5B14 (1945). (4) Doss, M.P., “Physical Constants of Principal Hydrocarbons.” 4th ed., New York, Texas Co., 1943. ( 5 ) Gooding, I%.hl., and Hopkins, R. L., “Determination of Aroniatics in Petroleum Distillates,” Division of Pctro!eurn Cheiii. SOC.,Chicago, Ill., 1946. istry, 110th Meeting of - 4 ~CHEM. (6) Hogness, T. R., Zscheile, F. P., and Sidwell, A . E., J . PhJ1.r Chem., 41, 379 (1937). (7) Liddell, U., and Kasper, C Research Satl. Bur. Standanls, 11, 599 (1933). (8) Lipkin, M . R . , Martin. C. C., and Kurtz, S. S.,J r . , Isn.E;vc CHEM., ~ A L ED., . 18, 376 (1946). (9) McAlister, E. D., Phus. Reo., 34, 1142 (1929). (10) Mchlister, E. D., Matheson. G. L., and Sweeney, 15’. .I., Rev. Sci. Instruments, 12, 314 (1941). (11) Rose, F. W., J . Research Satl. Bur. Standards, 20, 129 (1938). (12) Sachanen, A . N . , “Chemical Constituents of Petroleum,” p. 114, S e w York, Keinhold Publishing Corp., 1945. RECEIVED October 29, 1948. Presented before the Division of Analytical and hlicro Chemistry at t h r 114th Meeting of the ERICAN AN C H E \ n r . A I S O ~ I E TSt. Y , Louip, 110.
Corrections Louis Lykkrn has called attention to a serious mistake ill t h e paper entitled “Accumulation of Traces of hrsen2te by Coprrcipitation Tvith Magnesium .Imnioniuni Phosphate” [Kolthoff, I. M., arid Carr, C. \I7.,ANAL.CHEII.,20, 728 (1948)l. On paye 720, first rolumn, sixth line under heading “Coprecipitatiori of Arsenate with Magnesium Ammonium Phosphate,” “an amount of 500 mg.” should read ”100 my.” The same correction should be made three lines ahove ‘.Itecornmended Procedure” and four lines under “Recommended Procedurr.” The following misprints should be corrected: page 730, third line before end of first column, 0.014% instead of 0.14%; literature reference (@, page 384 instead of 881. In addition, the following clarificat,ion should be made. 0 1 1 page 728 in the second column, third line under the heading “Procedurr and Analysis,” thc. thrre follolving sentences should read : .lfter the solution had b w n made distinctly acid with hydr:+ chloric acid, an excess of magnesia mixture \vas added. (The magnesia misturc was prepared hy dissolving 50 grams of magncsium chloride hexahydrate and 100 grams of ammonium el-iloridc in 500 ml. of water. .I slight excess of ammonia was added, and the solution was allowed to stand overnight. ilny precipitate which formed was removed by filtration. The solution was slightly acidified with hydrochloric acid and diluted to 1 1it e r .) I. 11, I\OLTIIOFY C. \\-. C . m R I11 the article on “Ferrous 11etallurgy” by H. F. Beeghly [ANAL.CHEM.,21, 241 (1949)], the third line from the bottom of thr firqt column on pagr 243 Yhould read: “0.05 to 30% nickrl.”