ANALYTICAL CHEMISTRY
298 ~-
61.3- 74.474.4 82.4 32-37 38-41
Total Saninl?
..
. . . . . . . .
0.2 13.3 0 2
made when necessary by reference to an absorption band of known wave length. Satisfactory scattered light corrections (‘;in he made by use of glass and lithium fluoride shutters for the 4.1- to 9.1-micron and 9.4-to 15micron regions, respectively. Resolution requirements, of lcss importance t>hanthose of reproducibility, are met if the recorded spectrum shows a doublet structure for the carbon dioxide ahsorption peak a t 4.25 microns.
19
.
18 3
BIBLIOGRAPHY (1) Arery, W. H., J . Optical Soe. A m . , 31, 633 (1941). (2) Barnes, R. B., Gore, R . C..
Table X. Infrared Report Sheet, Total Hydrocarbon Sample Vol ‘7 cut
0-18 4
(Sample, commercial alkylate) 20.625.133.525.1 33.5 41.9 13-14 15-18 19-22
18.4 20.8
Yo
C4
Iso-cs n-Ca 2.2-DM B 2 , 3 - n~ ~B
RIP
3-RIP n-C6 2,2-DRI P 2,4-DMP 2,2,3-ThIB 2,3-DIfP 2-MH 3-MH n-Cr 2,2,4-TMP 2,2-D RI H 2,5-DRiH 2,4-DhIH 2,2,3-TM P 3,3-D.IIB 2,3,4-TMP 2,3,3-TMP 2.3-DMH 3,4- D M H 2,2,5-TM H
0-11 Gas Liquid 72 97c 27 1s 0.9 72.0 ‘0 3 27’1 0 82 .. 12 3
..
0
12
41.961.3 23-31
..
.. ..
..
N o evidenre
73
.. ..
28 2s 18 24 2
.. .. ..
..
..
..
..
.. ..
..
..
..
Apparentlv uure 2,2,4:TMP , .
N o evidence 16 17 5
2 3 2
.. ..
. ..
3.4 3.7 1.2
Liddel, U., and Williams. V. 8..“Infrared Spertrosropy”, Sew York, Reinhold Publishing Corp., 1944. (3) Barnes, R. B , McDonald. R. S.. William. V. Z..and Kinnaird. R. F., J . ,4pplied
,.
.. .. , . ..
.. .. , .
..
28 14 1
..
..
40
14 5
.
68b
..
..
..
4
8
45
11.0 9.7 1.6
0
5 4 1 8 . 3 Cu. a Leaders indicate t h a t the presence of a given compound waq not expected a n d no analysis was rarried c m r b S o t corrected t o 100%. Higher boiling material present ,
Phys., 16, 77 (1945). (4) Brattain, R. R.. P r t ~ o l e i r ~ i World, 40, 46 irehruary
1943). (5)
Table XI. Infrared -4tialysis of Synthetic Sample Containing Aromatic Hydrocarbons Synthetic Composition, Analysis Number Constituent Volume yo 1 2 3 Synthetir 55 m-Ethyltoluene 16.9 15.5 18.1 17.9 p-Ethyltoluene 17.8 16.7 17.5 17.6 1 35-Trimethylbeniene 22.0 21.4 20.8 21.1 o:Ethyltoluene 23.6 23.8 24.0 23.9 1,2,4-Trimethylbenzene 19.7 22.6 19.6 19.5 Synthetic 56 1,3,5-Trimethylbenzene 16.7 16,6 17,5 16.1 o-Ethyltoluene 3.9 . 3.0 2.6 3.7 1,2,4-Trimethylbenzene 62.9 63.9 62.7 64.6 1,2,3-Trimethylbenzene 12.2 12.4 12.3 12.5 Hydrindene 3.3 5.1 3.8 4.2
4 16.9
18.1 21.0 23.9 20.1 17.0 3.7 63.0 12.5 3.8
micron at 7.5microns. Supplrmentary corrections for zero drift can be made by use of a referencr line interpolated between zero positions determined before and after each recorded section. Corrrctions for wave-length shifting with ambient temperature are
Brattain, R. R., and Beerk. O., .l. Applied Phya.. 13, 699 (1942). ~. .,-
(6) Brattain, R. R., Rasnlussen. 11. S,,and Cravath, A . SI.. I b i d . , 14, 418 (1943). (7) Fenske. M. R., “Science of Petroleuru”, Vol. 11, p. 1629, Yew York, Oxford University Press, 1938. ( 8 ) Fry, D. L., Nusbaum, R . E., and Randall, H. SI.. b. Applied Phys., 17, 150 (1946). (9) Gordon, R . R., and Powdl, H., 6.Inst. P e t d e w m . 31, 191 11945). (10) Nielsen, J. R., Oil Gas J . , 40, No. 37.34 (1942). (11) Nielsen, J. R., and Smith, D. C., IND. ENG.CHEM., . ~ N A L . ED., 15, 609 (1943). (12) Oetjen, R. A., Randall, H. h i . , and .Indemon, W .E., Ret’. M o d . Phw., 16, 260 (1944). (13) Smith, D . C., and Miller, E. C., J . Optical SOC.Am., 34, 130 (1944). (14) Sweeney, 1%’. J., I n . ENG.CHEY.,r l x . i L . E D , 16, 723 (1944). (15) Trans. Faraday Soc., 41, 171 (1945). (16) Wright, N., Ibid., 13, 1 11941).
PRESENTED a t t h e Symposium on Molecular Structure and Spectroscopy, Ohio S t a t e Vniversity, J u n e 11, 1946.
Short-Cut Methods of Infrared Analysis W. D. SEYFRIED
AND
S. H. HASTINGS, HiLmble Oil & Rejining Company, Baytown, Texas
I
S MOST published work ( 2 , 5 , 7 , 13, 17) the practical application of infrared spectroscopy to the analysij of mixtures of hydrocarbons or other compounds is based upon the fact that each hydrocarbon or compound has a characteristic absorption spectrum, caused by different functional groups or different geometrical arrangements of these groups. According to Beer’s law, the amount of energy that i) absorbed at a particular wave length for a particular compound varies with the concentration of the compound in the cell, the length of the cell, and the amount of energy directed tonard the cell, according t o the following equation: 1 = Ioe-kc’
where IO = incident energy Z = transmitted energy
k = absorption coefficient c = concentration of component I = length of the cell
I n the case of gases, this! equation can be reduced to:
I ( , I/ log - = P
K
where p
= partial pressure of gas
K = absorption coefficient
In the conventional analysis of gaseous mixturw it is necessary
to obtain the infrared spectra of the pure compounds, select unique wave lengths for each compound where that compound
V O L U M E 19, NO. 5, M A Y 1 9 4 7
299
In connection with commercial-scale and pilot-unit operations for the production of butadiene from butylenes, a number of short-cut infrared procedures have been developed for the rapid, accurate determination of certain key components in samples containing other compounds wbose measurement is not necessary or which can be determined readily by other means. Although some of the procedures are applicable only for specific purposes, others are general in nature and may be applied to a w-ide variety of samples encountered in petroleum refining operations. The procedures may be divided into three categories: base-line methods (isobutylene and isobutane in admixture with other Cd hydrocarbons); procedures based upon measurements of optical density at unique wave lengths (methane, carbon
absorbs strongly and the absorption of the other compounds in the mixture is a t a minimum, calibrate a t the selected wave lengths v i t h each of the compounds likely to be present, so as to obtain absorption coefficients for each compound a t each wave length, and set 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 general, this procedure yields accurate and complet’e results, provided, of course, that there are no components present other than those calibrated for and that Beer’s law is followed for each component. In many cases, however, especially when more than about four components are present, this procedure is rather timeconsuming. In many control analyses for the operation of a plant or pilGt unit, it is necessary to know the concentrations of only one or two components in a complex gaseous mixture. In these cases it is especially desirable to develop procedures that will give the required answer in a minimum length of time rather than to rely on the more time-consuming conventional procedures. For instance, in many of the Ca hydrocarbon samples obtained in operations for the production of butadiene, it is desirable to know only the concentration of butadiene, butanes, isobutylene, and normal butylenes. These answers can be obtained by determining butadiene by ultraviolet absorption, total unsaturates by absorption in silver-mercuric nitrate solution, and isobutylene by infrared; the time required for all these determinations is considerably less than normally required for a multicomponent infrared analysis with the necessary auxiliary determinations. .$gain, in many c a m it is desirable to know the concentration of only one or two components, such as carbon monoxide or methane, in a fixed gas sample. In these cases a short-cut method for the determination of a particular component is desirable not only from the standpoint of saving time, but also to avoid errors resulting from absorption coefficient abnormalities for most of the fixed gases. In connection with plant operations and pilot-u.iit studies for the production of butadiene from butylenes, a number of short’ cut procedures have been developed for application to the anal) of certain types of samples. hlthough some of the procedures have only limited application, the principles upon which they are based may have general application to the analysis of a wide variety of samples. Other procedures are believed immediately applicable to the analysis of a number of hydrocarbon samples normally encountered in petroleum refinery operations. The short-cut procedures developed or modified in this laboratory have been divided roughly into three categories: the baseline technique originallH developed by Wright (17) techniques based upon absorption coeffirient measurements a t unique wave lengths where essentially no interfering absorption is encoun]
monoxide, and carbon dioxide in admixture M ith hydrogen and small amounts of CJ and lighter hydrocarbons); and comparison standard techniques (butadiene-1,3 product purity and butadiene1,2 in admixture with other Cq hydrocarbons). Analyses of synthetic mixtures under routine operating conditions have demonstrated that these procedures are accurate to within a few tenths of 170 and that the time requirements are in the order of 15 to 20 minutes. Data obtained during investigation on abnormalities in absorption coefficients exhibited by Farious compounds showed that, of the paraffin, mono-olefin, acetylene, and allene series, the abnormality was most pronounced for the first number of each series and that it decreaspd as the molecular weight increased.
tered] and comparison standard techniques. The specific apphcation of these principles, together with modifications effected in the apparatus and procedures employed and fundamental information obtained on absorption coefficient abnormalitie., ia disrwsed in subsequent paragraphs. APPARATUS AND EXPERIMENTAL TECHNIQUES
The studies discussed in this report. were performed using B Beckman infrared spectrophotometer (Model IR 107), manufactured by Sational Technical Laboratories, equipped wit’h a rock salt prism and housed in a constant-temperature, constanthumidity room. The instrument, has been des,cribed in the literature ( 7 ) . Since the installation of this instrument, a number of modifications and addit,ions have been found desirable or necessary. I n order to minimize contact of samples with stopcock grease, the original glass gas-handling system, equipped wit,h stopcocks, \\-as replacrd with a copper tubing manifold equipped with Hoke valves. This installation has been in use for some time and has been found to operate satisfactorily : considerable reduction in doir’-ntime has been achieved through elimination of glass breakage. To avoid the necessity for taking saniple bombs containing liquid hydrocarbons into the room housing the spectrometer, a system has been established whereby liquid samples are vaporized into glass bulbs over saturated magnesium sulfate solution. In order to prevent the inadvertent admission of this solution into the ahsorpt,ion cell while transferring samples, a protective device consisting of a set of electrodes and a relay-operated solenoid valve was installed in the sample intake line. ll7henever magnesium sulfat,e solution comes in contact, with the electrodes and thus closes the circuit, the. solenoid valve claw$, thus stopping the flow of solution through the linr. h Pirani gage was installed i n tlw thermopile system to make possible the determination of pressure in this desired; the gage has indicated that the pressure in the system has been less than 2 microns for over 2.5 years of continuous operation. I n order to eliminate fluctuations in current through the Sernst glower, i t n-as necessary to install a constant-voltage regulator in the main power supply line. To improve the accuracy of the pressure measurements i n the gas handling system, especially in the low pressure ranges, a Kallacc-Ticrnan manometer was installed. Because the original turret mechanism huppliad with the inhtrument was not adequate for covering all the wave lengths desired, this mechanism was replaced with a Lyave-length drive consisting of a micrometer barrel attached to the instrument case an5 a case-hardened shaft hetween the micrometer anti the prism turntable. In general, conventional methods \yere einployed in the operation of the infrared spectrometer used in this investigation. and no attempt is made to describe these operations in detail in the present report. Special techniques or procedures that were developed or employed during the course of tliese investigation? are diicussed in some detail, ho~\-erei,.
A N A L Y T I C A L CHEMISTRY
300 Table I.
Materials Employed Purity, Mole 7& M o s t Probable Impurity
Compound
99.3 99.8
Butanes Butylenes
99.9 99 8 99 9 96 3
nC4 Iso-Ca
...
...
Isobutane n-Butane B u t lene-1
cis-Jut ylene-2
...
trans-Ca-2 cis-C4-'
99 u 99.0
trans-Butvlene-2 Methane " Carbon monoxide Carbon dioxide
HZ
Hz, COz, C H I
93.0 99.9
..
:hame
6.8 ..
course, the analysis becomes considerably more involved and time-consuming, or recourse must be had to less exact methods involving the use of empirical charts. Possible causes of the observed curvature of the plot of optical density versus partial pressure include insufficient resolving power, deviations from the perfect gas laws, deviations from Beer's law, or detection by the thermopile of stray or' extraneous light that has been reflected or refracted from the optical surfaces inside the monochromator (false energy), Various methods have been described for eliminating or reducing false energy, principally by employing shutters or appropriate filters, such as magnesium oxide ( 7 ) . I t has been the experience of this laboratory that, although the use of a magnesium oxide filter is effective in reducing stray light, changes in the composition and/or physical characteristics of the filter during use result in changes in the amount of false energy absorbed; hence, the use of such filters results in the need for frequent recalibration. In general, it has been found that the effect of stray light is best minimized by mathematical treatment, as suggested by Brattain et al. (8) and according to principles suggested by States and Anderson (15). This procedure involves the determination of the approximate false energy a t a given wave length by observing the galvanometer deflection with the sample cell filled to atmospheric pressure with a compound that is known will give total absorption a t the wave length employed. In some cases, this procedure gives a satisfactory value for stray light; however, in many instances, the value obtained is not sufficiently large to give a linear relationship in the density versus pressure curve. The procedure does give an approximate value for false energy which may then be applied to the calibration data by means of the following equations: d = log
(Io - FE) ( I - FE) ~
tvhere d
=
optical density
Io = incident light, mm. of galvanometer deflection 10.5 10.9 IO.8
0 ' 20
40
60
I
I
1
1 88
17
I
1
I
Piyre 3
6
100
Concentration, Mole Per Cent ( i n Hydrogen)
In order to utilize absorption spectra of various compounds obtained in other laboratories, it was necessary to calibrate the microdrive employed on this instrument in terms of wave length. This calibration was obtained by the usual procedure of determining the absorption maxima of several pure hydrocarbons over wave-length regions where strong absorption was known to occur and plotting the microdrive reading a t the point of maximum absorption versus the corresponding known wave length in microns. The calibration curves so obtained were then used t o determine the approximate microdrive interval corresponding t o any desired nave-length region. Absorption coefficients ( K values) were determined for each component a t the desired wave length and slit widths by filling %heabsorption cell to the desired pressure with air-free portions .of the pure components and observing the optical density for each pressure (expressed in millimeters of mercury). Absorp tion coefficient data were obtained over a pressure range to cover a range of optical density from 0.2 to 0.7. In calibrating the spectrophotometer for most analyses, it was mecesmry t o determine the absorption coefficient for each component a t one or more wave lengths. Frequently it was found t h a t a plot of optical density versus partial pressure for the pure compound would not give the straight-line relationship predicted by Beer's law. When linear relationships are not obtained, of
4
i
I
x 2 h
Y
Y
Y
8
i
d
l
"
"
"
"
I Finure
4
10
5
0
20
40
68
80
Concentration, Mole Per Cent (in Hydrogen)
108
V O L U M E 19, NO. 5, M A Y 1 9 4 7
301 acetylene through dimethylacetylene, allenes (propadiene and butadiene-1,2), and oxides of carbon are shown in Figures 1 through 5. It is evident that in each series, the variation of absorption coefficient with concentration was greatest for the compound of lowest molecular weight and that the effect decreased as the molecular weight increased. The effect of increasing molecular weight is shown more clearly in Figure 6, in which the ratio of the absorption coefficient for the pure compound to the absorption coefficient for the same compound in a 5% blend a t the same partial pressure is plotted against the molecular weight of the compound for each homologous series. It was found that the abnormality becomes practically negligible for the third and fourth members of each series and that the effect is less for the first members of series having higher molecular weights. It was found also that the degree of abnormality depended not only upon the concentration but also upon the diluent employed. This phenomenon has been discussed extensively in the literature (1-4, 8-12, 14, 16). Data published recently by Coggeshall and Saier (9) for methane are in close agreement with the data shown
35
30
5
x 25
G v
c)
‘gI 20
5
g 15 *s B
-3
10
5
0 20 40 60 80 Concentration, Mole Per Cent
100
,
Figure 5. Absorption Coefficient Abnormalities for Carbon Monoxide and Carbon Dioxide
I = transmitted light, mm. of galvanometer deflection FE = false energy, mm. of galvanometer deflection K = -Sd
PI
where
K = absorption coefficient d = opticaldensity P’ = partial pressure of pure component, mm. of mercury a t 15“ C. S standard pressure, mm. of mercury
-
If the initial value for false energy employed is correct, the R value obtained will be constant (to 11 to 2%) regardless of pressure over the optimum pressure range; however, if the assumed false energy is low, the K values will shou. an increase with decreasing pressure. If the assumed false energy is high, the K values will decrease with decreasing pressure. By successive approximations, a value for false energy will usually be found which will yield an essentially constant K value. The value thus found may then be applied for all compounds and absorptions a t the wave lengths investigated. Methods employed for dealing with certain deviations from Beer’s law are discussed below. The materials employed in the development of the procedures discussed in this report, together with their purities and most probable impurities, are shown in Table I.
10
20
30
40
50
60
Molecu1ar;Weight
Figure 6. Coefficient Abnormality PS a Function of lMolecular Weight for Various Hydrocarbon Series
ABSORPTION COEFFICIENT ABNORMALITIES
During the course of calibrating the infrared spectrophotometer for the determination of carbon dioxide in this laboratory, it was observed that the absorption coefficient for this compound varied with concentration, even a t the same partial pressure of carbon dioxide. Further investigation revealed that this phenomenon was exhibited by a number of other compounds, particularly by the first members of each homologous series (except cyclopropane). Data obtained on paraffins from methane through butane, mono-olefins from ethylene through butylenes, acetylenes from
in Figure 1. Unfortunately, data obtained on two other compounds (carbon monoxide and carbon dioxide) by Coggeshall and! Saier cannot be compared with the data for these compounds shown in Figure 5 because the absorption measurements were made a t different wave lengths. It is generally agreed that t h e phenomenon of increasing absorption coefficient with deereasing concentration a t the same partial pressure is attributable to the “pressure broadening” of spectral lines, caused by collision damping. In general, the consensus of the literature appears
ANALYTICAL CHEMISTRY
302 ~
.. -~
~
Table 11. Blend S o . Synthesis, mole '7( Isobutane n-Butane Butylene-1 cis-Butylene-> trans-Butylene-? Butadiene-1 ,S Isobutylene Analysis, mole % ' Isobutylene (date
is used, it is necessary to take into account the variation of absorption coefficient with concentration for butadiene1,3in the isobutylene wave-length region, and an empirical correction curve must be constructed. S o correction need be applied if the concentration of butadiene is less than about 1%. If butadiene-1,2 is present, corrections must be applied based upon the concentration of butadiene-l,2 as determined by infrared ahsorption at about 3.1 microns.
Isobutylene in Synthetic Samples
HS-l 0 8 30 2
15.8
'44 0 0 10.3
10.3 10.6(8-11-441 10.3 10 6(8-11-441 10,0(8-15-44) 10.1(8-25-441 9.6(9-6-44)
10.5(9-30-44) 10 4(10-:3-441 10.5(10-20-44)
10 6(10-31-44)
HS-2 0 0
HS-3
HF-i
0 1
9 4
0.0 0.3
13 9
10
1'5 3 4 i 3
4
30.1
47.2
le.?
4 ,
38 8
4.7 5 O(10-10-44) 4.7
39.2(9-1-45)
13 1 10 6
13.2 0.2 4 , 4.5
4.6(10-10-441 4.5 4 6(10-10-44) 4.5 4 7(10-31-44)
a.a
5.4
5.0(10-10-44)
Data obtained by this procedure on a number of synthetic samples are shown in Table 11. I t will be seen that the procedure is accurate, on the average, . . to within less than *0.3%, based on the total sample, and that the calihration is very stable over prolonged operating periods. (The data on blend HS-1 were obtained over a period of 2.5 months of routine operation using the same calibration.) In Table 111 are shown data comparing the infrared procedure with conventional chemical procedures employing absorption in 65% sulfuric acid or reaction with anhydrous hydrochloric acid. In general, the agreement b e h e e n the various methods is considered good. The time requirement for an analysis by the spectrometric procedure is in the order of 15 minutes. Deteimination of Isobutane in Admixture with CI Hydrocarbons. The baseline procedure for the determination of isobutane is similar to that described for isobutylene. In this procedure, absorption measurements are made in the range of
4,8(10-31-44)
... ...
...
to he that the analysis of sample> containing componente whose absorption coefficient ic influenced by changes in concentration is not' feasible. However, such analyses are possible by empirical methods, provided all factors influencing the absorption coefficient are taken into accouht in the analysis; procedures have been developed for the determination of carbon dioxide, carbon monoxide, methane, ethylene, butadiene-1,3, and ot,her compounds exhibiting absorption coefficient abnormalities. The procedures developed in this laboratory for methane, carbon dioxide, and carbon monoxide are discussed below. BASE-LINE SIETHODS O F ANALYSIS
Accurate analyses for specific componente can often be obtained in a minimum of time in the presence of other absorbing compounds by the application of base-line techniques. In general, such procedures are based upon absorption measurements over a wave-length interval where the compound in question exhibits a unique absorption peak and the other compounds in the mixture possess relatively small absorptions. By obtaining readings at the peak and in adjacent valleys on either side of the peak, it is possible to construct an empirical curve relating the partial pressure of the key component to a mathemat'ical term incorporating the observed relationships a t each pressure. Specific applications of this technique for the analysis of isobutylene and isobutane in admixture with other C,hydrocarbons are discussed below. Determination of Isobutylene in Admixture with C, Hydrocarbons. .4n accurate, rapid method of analysis for isobutylene in admixture with other C, hydrocarbons is very desirable in many refinery operations, particularly in operations for the production of butadiene by the dehydrogenation of normal butylenes. .4 base-line procedure, based upon absorption measurements a t only two wave lengths (at about 11.5 and 11.8 microns) on the slope of the isobutylene peak (see Figure 7) has been developed; in this procedure, it was found possible to select the points of maximum and minimum isobutylene absorption in such a manner that the mathematical relationship employed resulted i'n the elimination of background absorption of all the other Ca hydrocarbons except butadiene-l,3 and butadiene-1,2.
.
A curve was constructed of partial pressure of isobutylene versus the expression d.M - fldr,, where d.w = density at point of maximum absorption, dL = density a t point of minimum absorption, and jl = a constant, selected so that the product of fidL T o d d essentially equal d.w for all the Cn hydrocarbons other than isobut'ylene, butadiene-1,3, and butadiene-1,2. (In this laboratory the value of f1 was taken as 1.5.) If but'adiene-l,3 is present, the analysis can be performed on the residue obtained after absorption of the butadiene in maleic anhydride, or corrections can be applied for butadiene absorption in the isobutylcne Lmve-length region after the determination of the butadienrt concentration by ultraviolet ahsorption. If the latter proceduri.
-8 11.3
El
*
11.4
11.5 11.6 W a v e Length, Microns
11.7
11.8
Figure i. Spectra Employed to Select Ware Lengths for Short-Cut Isobutylene Analysis
V O L U M E 19, NO. 5, M A Y 1 9 4 7
303
Table 111. Comparison of Procedures Sample
Synthesis
HS-2 T - 3 feed T-5 overhead
4.7
...
...
Table IV.
Infrared 65% H2S04 M o l e yo isobutylene 4.6 5.0 2.3,Z.l 2.5 1.7,1.7 1.8
Table VI.
Anhydrous
HC
1-1
Cosorbent
...
5.5 10.5 20.3 4.5 9.9 5.7 3.8 4.3
...
10.0 20.0 4.3 9.1
4 .7;4 . 7 9.3,9.3 5.1 3.7 3.9
... ...
...
1-2
1-3 a
8.2 61.6 0.0 14 7
Infrared .Vole per cent carbon monoxzde
5.0
4.6,5.0 2.4 1.9
Isobutane in Synthetic Jlixtures
Blend KO. Synthesis, mole % n-Butane Butylene-1 Butylenes-2 Butadiene-1.3
Determination of Carbon >Ionoxide
Synthesis"
22.9 15.4 25.4
Blended with hydrogen
'70.6 4.8 0 0
o n
n n
Table VII.
Determination of Carbon Dioxide
Rurrell
Infrared V o l e p e r cent"
Table V. Synthesisa 73.9 49.3 34.5 24.5 14.8 6.0 1.5
...
...
5,1 3.4 2.0 4.0 2.5 2.2 4.7
Determination of Methane
Infrared M o l e per cent methnne 73.0 49.4 35.2 25.0 14.2 6.0 1.5 8 9 9 9
Podbielniak Analj .I.
...
...
9.0 9.9
Blended wit.h hydrogeii
8.2 to 8.8 niirronc: (a shitrp absorption peak for isobutanc occurs a t about 8.46 microns); points of maximum and minimum isobutane absorption can be selected so that the optical densities for the other C, hydrocarbons, except n-butane, are approximately the same. n-Butane, if present in the sample, can be corrected for by employing the first answer for isobutane in conjunction with the values obtained for total unsaturates in the sample to arrive a t an exact value for isobutane by a succession of simple approximations. Data obtained using synthetic mixtures with this procedure are shown in Table IV. This method is usually accurate and reproducible t o within about * 1.0%, based on the total sample, and requires about 15 minutes per determination. S l Y G L E P O I l T BIETHODS OF ANALYSIS
In many types of samples, the number of components is so limited and absorption spectra are so characteristic that it is possible to obtain accurate analyses by determining the absorption coefficient a t a single wave length. In most samples of this nature, the components are of such low molecular weight that they exhibit the phenomenon of changing absorption coefficient with concentration and with different diluents, as discussed above, so that most procedures of this nature are empirical and are applicable only to the particular types of samples for which calibration data are obtained. Determination of Methane in Admixture with Other Gases. As developed in this laboratory, this method is based upon the unique absorption peak of methane a t 7.67 microns and is limited to the determination of methane in samples consisting predominantly of hydrogen and containing small quantities of carbon monoxide and hydrocarbons containing three or less carbon atoms. (Such samples are obtained from sources such as absorber tail gas streams and from C1 fractions from Podbielniak distillations.) Because the absorption coefficient for methane varies with concentration as well as with partial pressure, it is necessary to calibrate the spectrophotometer at a standard pressure and using
a
5.0 3.2 2.6 3.7 2.7 2.2 4.3
In admixture with hydrocarbon gases through C I .
a series of blends containing from 0 to 100% of methane in admixture with the other components expected to be present in the samples to be analyzed. I n this laboratory, a standard total pressure of 500 mm. has been selected, and a calibration curve of 10,000 log zo/z mole per cent methane versus the expression: total sample pressure is employed in analytical work; hydrogen is used as the diluent gas in obtaining calibration data. In Table V are shonrn results obtained on synthetic and regular samples when using this procedure. The results indicate that the method is accurate on the average to within *0.5%, based on the total sample; the time requirement is about 15 minutes. Determination of Carbon Monoxide in Admixture with Other Gases. This procedure is applicable to the determination of carbon monoxide in concentrations ranging from 0 to 20% and in admixture with hydrogen and small quantities of CI to Ct hydrocarbons. The procedure is based upon the absorption of carbon monoxide at 4.3 microns; as in the case of the methane determination, the absorption coefficient varies with concentration and pressure, so that an empirical relationship must be employed and the calibration must be obtained a t a constant uressure. Calibration data are dotted as mole per cent carbon 10,000 log 10,T Data monoxide versus the expression: total sample pressure' obtained on synthetic and routine plant sampies by this procedure and by absorption in Cosorbent are tabulated in Table VI. The procedure is apparently accurate to within about =tO.5%, based on the total sample; values obtained by Cosorbent were generally somewhat higher than values obtained by infrared. The time requirement for this procedure is about 15 minutes. Determination of Carbon Dioxide in Admixture with Other Gases. This procedure may be applied to the determination of carbon dioxide in any concentration in admixture \%ithany other gases. The procedure is based upon obtaining absorption measurements a t 4.2 microns, in the region where carbon dioxide possesses a unique absorption peak. Interference rioni other compounds present is eliminated by obtaining absorption measurements a t the key wave length before and after removal of the carbon dioxide from the mixture by absorption on Sscarite. The calibration data are plotted as mole per cent carbon dioxide -
io'ooo log I O i z corrected for the total sample pressure absorption Obtained on the carbon dioxide-free .,ample. Com-
versus the expression:
ANALYTICAL CHEMISTRY
304 Table VIII. Aniil p
Determination of Butadiene-1,P
Butadiene-1,2 Synthesis
i p
Butadiene-1,3, Synthesis
M o l e p e t cent 0.5 $0 I
,
:3
Butylenes, Synthesis
.Ifole p e r c e n t 0 6
1.0 7.5
99.4 99.0 70.0
0.0 0.0
22.5
parative rehultr on rout,ine aaniples obtained by this procedure and by absorption in potassium hydroxide solution are shown in Table VII. In general, the agreement betiyeen infrared and absorption procedures is in the order of il fev tenths of lY0. Although there is no advantage from the standpoint of time requirement (20 minutes) over the standard Burrell procedure, the accuracy of the infrared method is believed to be somewhat better for some types of samples and the infrared procedure is applicable t o the determination of carbon dioxide in relatively small quantities of sample. The method is sensitive to about 100 parts per million of carbon dioxide. COMPARlSON STAKDARD METHODS OF ANALYSIS
In some infrared analyses it is advantageous to apply a comparison standard technique whereby absorption measurements are obtained at key wave lengths on a similar sample of known composition immediately prior to taking the same readings on a sample of unknown composition, the difference between the readings being a direct measure of the amount of a particular absorbing substance present. This procedure is particularly valuable for analyses where small changes in calibration values or in wave length or slit width settings could cause large errors in analysis. Application of this principle t o the determination of product butadiene purity and butadiene-1,2 concentrations is discussed below. Determination of Butadiene Product Purity by Infrared Absorption. I n operations for the production of butadiene, it is highly desirable to have available a rapid, accurate control tebt, SO th:it operating conditions can be quickly adjusted when the purity of the final product is below the specificatiori minimum (98.0 weight % ' butadiene) or when the purity rises so high that appreciable quantities of butadiene are lost from the raffinate from the extraction process. d4procedure for the determination of the purity of specification grade butadiene, by determining the impurities by means of a single absorption measurement at 6.9 microns, has been. reported by Brady ( 6 ) ; a modification of this method has been adopted for use in this laboratory. I n the modified method, spectra are obtained on propylene and on a blend of all C, hydrocarbons known to be present in specification grade butadiene, in the region of 6.9 microns. (Propylene has frequently been found t o be present as a contaminant in specification grade butadiene; hence, this hydrocarbon has been taken into account in the method along with the C4 hydrocarbons.) From these spectra a point is selected a t or near 6.9 microns, a t which the C, hydrocarbons and propylene have essentially the same absorption coefficient; by using this wave length for analysis, possible errors caused by changes in the ratio of propylene t o C4impurities are eliminated. T o avoid possible errors resulting from calibration fluctuations or from irreproducibilities in slit width and wave-length settings, a comparison standard procedure was adopted for this determination. This technique consisted simply of determining the optical density a t the specified wave length of a sample of known purity (preferably a synthetic) immediately prior to running the unknown samples. The difference in optical density obtained (delta D)can then be applied to a curve of delta D versus mole per cent but'adiene-l,3; the mole per cent of butadiene-1,3, in the unknown is then read from this chart. Comparative data have been obtained on a large number of regular plant samples by the modified infrared procedure and by the official Rubber Reserve specification procedure (absorption of
butadiene in maleic anhydride, followed by scrubbing of the unabsorbed impurities from the maleic anhydride with carbon dioxide, absorption of the carbon dioxide in caustic, and final measurement of the residue gas). Typical data obtained on 36 samples analyzed by both methods during the period from Sugust 11 to 27, 1945, showed an average discrepancy of about, *0.2y0 butadiene between the infrared and the chemical methods. This value is sorneir-hat higher than expected and is probably attributable primarily to operating errors in the chemical procedure. The time requirement for the infrared procedure is about 20 minutes. Determination of Butadiene-1,2 in Admixture with Cq Hydrocarbons. Butadiene-1,2 is produced in small quantities in thcx butplents dehydrogenation reaction and, if not removed in the c~straction and purification operations, tc3n.k to accumulate as itn impurity in recycle butadic.nv. In order t o control operations for the removal of this hydrocarhon, it is necessary to have available an accui'ate, rapid :inaiyticrl procedure for its determination in admixture nith other hydrocarbons, including large proportions of butadiene-1,3. ,4procedure has been developed in this laboratory based upon the unique absorption peak of butadiene-1,2 a t about 5.1 microns; a t this wave length butadiene-1,3, butanes, and butylenes have low atxorptions of about the same order of magnitude. By comparing t'he optical density of an unknown sample with the optical density a t the same wave length of a sample containing a known concentration of butadiene-1,2, it is possible to determine the concentration of butadiene-1,2 in the unknown sample. Csing this technique, it has been found possible to obtain accurate values for butadiene-1,2 in samples containing butadiene-1,3 in concentrations from 0 to 99.5y0. Data obtained on synthetic mixtures using this method are shown in Table VIII. This method is apparently accurate to within about +0.2%. The time requirement for the procedure is in the order of 20 minute$.. ACKhOWLEDGMENTS
The work discussed in this report was conducted under the auspices of Rubber Reserve Company and Humble Oil & Refining Company, to which the authors express thanks for permission to report the data obtained. The contributions of various members of the company's organization, particularly that of J. -4.Anderson in supplying the procedure for correcting for false energy, are acknon-ledged alio. LITERATURE CITED
(1) -%ngstrom,K., A n n . P h y s i k . 6, 163 (1901). (2) hvery, W. H., J . Optical SOC..4m., 31, 633 (1944). (3) Bahr, E.V., Physik. Z . , 12, 1167 (1911). (4) Bahr, E. V., Verhandl. deut. p h y s i k . Ges., 15, 673 (1913). (5) Barnes, R. B., Liddel, U., and Williams, 1'. Z., ISD. ESG.CHEM.,
4x.k~. ED.,1 5 , 8 3 (1943). (6) Brady, L. J.,I b i d . , 1 6 , 4 2 2 (1944). (7) Brattain, R. R., C a l i f . Oil W o r l d Petroleum Ind., 36 ( 2 ) , 9 (1943).
(8) Brattain, R. R., Rasrnussen, R. S'., and Cravath, 4.RI., J . A p p l i e d P h y s . , 1 4 , 4 1 8 (1943). (9) Coggeshall, N. D., and Saier, E. L., Ibid., 17, 450 (1946). (10) Dennison, D. M.,P h y s . Rev.,3 1 , 5 0 3 (1928). (11) Hertz, G . , V e r h a n d l . deut. p h y s i k . Ges., 13, 617 (1911). (12) Lorentz, H. A., PTOC. Amst. Acad. SOC.,8, 591 (1906). (13) Nielson, J. R., Oil Gas J., 40 (37), 34 (1942). (14) Nielson, J. R., Thornton, V., and Dale, E. B., Rev. M o d e r n Phys., 16, 3 and 4 , 3 0 7 (1944). (15) States, M. N., and Anderson, J. C., J . Optical Soe. Ann., 32, 659 (1942).
(16) Stephens, F. M.,"Symposium on Mass, Infrared, and Ultraviolet Spectrometry", R.F.C., O5ce of Rubber Reserve (Feb. 2124, 1944). (17)
Wright, N., IND. ENG.CHEM.,ANAL.ED.,13, 1 (1941).
PREBENTED before the Division of Analytical and Micro Chemistry at the 110th Meeting of the AMERICAN CHEMICAL~OCIETY, Chicago, Ill.
