V O L U M E 20, NO. 2, F E B R U A R Y 1 9 4 8 5, the probability of identical partition ratios is small. I t becomes very much less if a second or third entirely different system gives a perfect distribution and especially when the more conventional criteria are also all in good agreement. The method by no means replaces or minimizes the need for the use of the older more conventional approach, even though it n-ill be found much more decisive for many cases. When the limitations of the method are properly understood, a countercurrent distribution pattern, such as those in this paper, gives an especially informative picture from the standpoint of identification. It can be used for this purpose as a physical constant, since the partition ratio for the phase pair employed can be easily calculated ( I S ) . Moreover, certain other characteristic properties can also be deduced, such as the degree of adhcrence to or deviation from a constant partition ratio a t a given concentration level. Further, the purity of the sample employed is usually a t once apparent and the characteristic partition ratio can be accurately derived, even though the sample used was not pure. LITERATURE CITED
(1) Craig, L. C.. J . Biol. Chem., 155, 519 (1944). (2) Craig, L. C., Golumbic C., Mighton, H., and Titus, E., Ibid., 161,321 (1945).
139
Craig, L. C., Hogeboom, G. H., Carpenter, F., and du Vigneaud, V.,Ibid., 168, 665 (1947). Drake, N . L., Creech, D. D., Garman, J. A., Haywood, S., Peck, R. M.,Walton, E., and Van Hook, J. O., J . Am. Chem. SOC., 68, 1214 (1946).
.
Elderfield, K. C., Craig, L. C., Lauer, W. M., Arnold, R. T., Gensler, W. J., Head, J. D., Bembry, T. H., Mighton, H. R., Tinker, J., Galbreath, J., Holley, A. D., Goldman, L., Maynard, J T., and Picus, N., Ibid., 68, 1516 (1946). Herriot, R. M., Chem. Reas., 30, 413 (1942). Hogeboom, G. H., and Craig, L. C., J . Biol. Chem., 162, 363 (1946).
Kunita, M., and Korthrop, J. H., Cold Spring Harbor S~mposia Quant. Biol., 6, 325 (1938). Moore, S., and Stein, W. H., J . Bid. Chem., 150, 113 (1943). Northrop, J. H., and Kunita, M , J . Gen. Physiol., 13 787 (1930).
Skau, E. L., and Wakeham, H., “Physical Methods of Organic Chemistry,” Vol. I, p. 13,New York, Interscience Publishers, 1945. Thorp, D., J . SOC.Chem. Ind., 65,414 (1946).
Williamson, B., and Craig, L. C., J . Bid. Chem., 168,687 (1947). RECEIVED J u n e 5 , 1947. Work done under a contract, recommended by tha Committee on Medical Research, between the Office of Scientific Research and Development and T h e Rockefeller Institute for Medical Research. Eighth in a series on identification of organic compounds. Investigators Harold Mighton and Elwood Titus were loaned t o this project from tha Columbia Office of Scientific Research and Development contract.
Mass Spect ro meter-H ydroc hIo rinat ion Analysis of Butenes F. W. RIELPOLDER AND R. A. BROWN, The Atlantic Refining Co., Philadelphia, Pa.
To improve the accuracy of determining individual butenes, isobutene is converted to tert-butyl chloride prior to mass spectrometer analysis. Two special methods of computation have been devised, modifications of the basic principles of analyses proposed by Washburn, Wiley, and Rock.
T
HE application of the mass spectrometer t o the analysis of hydrocarbon mixtures consisting essentially of CC paraffins
and olefins has been discussed in detail by Washburn, Wiley, and Rock (3). While such analyses show commendable accuracy for the paraffin constituents in the mixture, the determination of olefins is accomplished with a lesser degree of certainty. Because of the marked similarity of the butene spectra, the resolution of the butenes is greatly influenced by the stability of the instrument and the presence of small concentrations of C6 and higher olefins. In order t o improve the accuracy with which individual butenes may be determined, the authors have developed a procedure whereby isobutene is converted to tert-butyl chloride prior to mass spectrometer analysis. This reduces the number of butene mass isomers in the mixture from three to two, thus simplifying the resolution of the remaining 1- and 2-butenes. The isobutene is determined as tert-butyl chloride. Analysis of butane-butene fractions by this method has proved t o be considerably more reliable for isobutene than samples analyzed in the conventional manner. The geometrical isomers cis-2-butene and trans-2butene have nearly identical mass spectra and are usually grouped in the computation of routine analyses. In the event butadiene is present in the sample, no difficulty is encountered by the prescribed procedure, since the diolefin does not react with hydrogen chloride. The concentration of butadiene is determined accurately in the conventional manner ( 1 ) . The advantages gained by conversion of isobutene t o tert-butyl chloride become readily apparent on consideration of the mass spectra of the individual components. Thus in Figure 1 the similarity of the spectra of isobutene, 1-butene, and 2-butene is
shown, as compared with isobutme, n-butane, hydrogen chloride, and tert-butyl chloride. The over-all difference between the mass spectra of a typical butane-butene mixture before and after hydrogen chloride treatment is shown in Figure 2. Of primary importance in the hydrogen chloride-treated sample record is the noticeable reduction of the masa 56 peak due to removal of isobutene and the appearance of relatively large peaks a t masses 57 and 77, indicating the presence of tert-butyl chloride. It was determined from the analyses of a number of plant streams that no serious interference is encountered in resolving mixtures containing less than approximately 3 mole yo of pentenes. Higher concentrations of pentenes contribute significantly to the butene mass spectra, thereby increasing the probable error in the butene split. Hoviever, the majority of mixtures of practical interest are, or can be made, nearly free of pentenes; hence this limitation is not a serious factor. Repeated introduction of hydrochlorinated samples containing excess anhydrous hydrogen chloride t o the mass spectrometer had no apparent adverse effect on operation of the instrument, since no subsequent change occurred in either the spectrum or sensitivity of calibrating compounds. PROCEDURE
The method consists essentially of converting isobutene to tertbutyl chloride in the manner described by McMillan @), wherein the sample is reacted with hydrogen chloride at reduced temperature and pressure. A measured amount of the sample to be analyzed is placed in a 25-cc. reaction bulb with a known amount of hydrogen chloride
ANALYTICAL CHEMISTRY
140 50
40
30
70
60
I
-
80
90,
T1500 Units Peak Height ISOBUTENE
Ill
I
I -BUTENE
2- BUTENE I
I-
I
z W I
Y
a w a
I
* . . .
I. I
I
..
.
..,I
NORMAL BUTANE
1
I/
HYDROGEN, CHLORIDE
I
Figure 1. Mass Spectra of Calibrating C o m p o u n d s a t 40-Micron Pressure 8
gas at room temperature. The relative amounts of sample and hydrogen chloride which are blended together are a function of the isobutene content, since the partial pressure of the relatively high boiling tert-butyl chloride at room temperature must not exceed 200 mm. of mercury, its vapor pressure. For samples containing less than 50% isobutene, 300 mm. of sample and 300 mm. of hydrogen chloride gas are blended, while for samples containing more than 50% isobutene, 150 mm. of sample and 300 mm. of hydrogen chloride are used. The reaction bulb is then alternately cooled with a suitable refrigerant such as liquid nitrogen to condense the sample hydrogen chloride mixture, and warmed to vaporize partially the contents in the bulb. Usually three cycles of cooling and warming are sufficient to complete the reaction and the final warming operation brings the reaction bulb to room temperature. Under these conditions the reaction is quantitative and the tertbutyl chloride is completely vaporized. This reaction mixture containing excess hydrogen chloride is then ready to be introduced to the mass spectrometer. COMPUTATION
The first mass spectrum is obtained from the original untreated sample. I n subsequent discussion this record is referred to as record A. The second mass spectrum is then determined for the hydrogen chloride-treated sample, which is designated as record B. In order to achieve maximum accuracy in analyses of this nature, two special methods of computation have been devised according t o the type of analysis wanted. These methods, outlined below, are modifications of the basic principles of analyses as proposed by Washburn, Wiley, and Rock (3). Method 1. This procedure is recommended for CI to Cc samples where an analysis of isobutene and the sum of the nbutenes are sought.
1. The over-all composition of the sample is first computed in the conventional manner from record A with all the butenes calculated together as a group. 2. Record B is employed only to determine the concentration of isobutene relative to the n-butenes. This may be done by calculating isobutene in terms of tert-butyl chloride using mass 77, and the n-butenes by means of mass 56 after subtracting contributions due to other components in the mixture. The analysis of isobutene and n-butenes is then obtained by applying this ratio of isobutene to n-butenes to the total butene concentration calculated from record A in step 1. In order to calculate more accurately the residual peak height a t mass 56 it is advisable to apply the isobutane-n-butane concentrations as calculated in step 1 from record A to the computation of record B. The two records may be directly related by comparing the mass 43 peaks, thereby determining accurate isobutane+-butane contributions to mass 56 peak in record B. Method 2. This procedure is recommended for C1 to C, samples where an analysis of isobutene, 1-butene, and 2-butenes (cis and trans isomers being grouped) is sought. 1. Record -4is used only to calculate the analysis of propane, isobutane, and n-butane by the conventional method. 2. The concentrations of propane and the butanes are applied to record B by comparing the 43 peaks of the two records (similar to Method 1, step 2, above). Contributions of propane and the butanes to the propene, butene (and butadiene, if present) mass peaks may then be subtracted from the observed peak heights. 3. Isobutene is determined from the mass 77 peak in record
B.
4. Propene is calculated as a monoisotopic residual 42 peak in record B. 5. 1-Butene and 2-butenes are determined by solving two simultaneous eauations based on the residual 39 and 56 Deaks (or 41 and 56 peaks). 6 . The complete composition may be calculated from the above data in the usual manner.
Pentanes, if present in the mixture, are calculated in the same manner as the butanes. I t was verified experimentally that 2methyl-1-butene and 2-methyl-2-butene react with hydrogen chloride a t low temperature to form tert-amyl chloride, while the other pentenes do not react. No serious interference results from this reaction because mass 91 peak defines the amount of tartamyl chloride present. The unreacted pentenes are calculated by means of the residual mass 70 peak. Attempts to extend this hydrogen chloride treatment method to pentane-pentene mixtures for a more accurate resolution of the pentenes are now in progress. When it was desirable t o confirm the fact that isobutene had
Table I.
Analysis of Phillips Synthetic Mixture 4
Component
Known Composition
Ethane Propane 1,3-Butadiene Isobutene 1-Butene 2-Butene Isobutane n-Butane Isopentane n-Pentane Total
0.02 3.25 2.58 44.82 12.20 12.24 11.23 11.25 1.20 1.21 100.0
-
M.S. Analysis, HCl Treatment Mole Per Cent 0.0 3.2 2.4 45.0 11.5 13.2 11.6 10.6 1.3
1.2 100.0
Difference
-0.02 -0.1 -0.2 +0.2 -0.7 +1.0 +0.4 -0.6 f0.1
0.0 ...
Table 11. Analysis of Phillips Synthetic Mixture 8 Component Propene Propane Isobutene 1-Butene 2-Butene Isobutane n-Butane Isopentane n-Pentane Total
Known Composition 8.2 8.8 5.1 9.6
6.2 12.3 31.6 8.6 9.6 100.0
-
M.S. Analysis, HC1 Treatment Mole Per Cent 8.1 9.1 5.0 9.1 6.8 12.0 32.0 8.5 9.4 100.0
-
Difference -0.1 +0.3 -0.1 -0.5 +0.6
-0.3 +0.4 -0.1 -0.2
...
141
V O L U M E 20, NO. 2, F E B R U A R Y 1 9 4 8 Table 111. Mass Spectra of tert-Butyl Chloride and tertAmyl Chloride t ert- Bu t yl
tert-Amyl Chloride 39.4 300 41.2 216 10.9 47.0 13.6 41.0 268 39.6 491 102 465 15.1 31.5 8.80 54.4 13.0 447 28.5 ' 5.69 20.1 101 614 31.5 2.82 0.12 5.63 358 567 130 181 100 7.75 32.0 1.58 5.12
Chloride 8.38 4.78 11.2 75.6 2 64 6 03 6 43 13 3 73 0 10 8 1.99 12 6 1.0 4.90 7.03 1.20 4.17 1.18 10.2 17.3 30 1 0.06 0.07 0.07 0.09 0.54 0.07 1.09 7.75 100 5.28 32.0 0.04 0.04 0.03
hlass 26 27 28 29 35 36 37 38 39 40 41 42 43 50 51 52 53 54 55 56 57
91 92 93 106 Sensitivity 20.3 Sensitivity of n-butane a t mass 58 = 10.5.
Table IV.
ANALYSIS MOL. 96 PROPENE PROPANE ISOBUTANE ISOBUTENE I-BUTENE 2-BUTENE n-BUTANE
= W
Propene Propane Isobutene 1-Butene 2-Butene Isobutane n-Butane Pentenes Hexenes Heptenes Diisobutene Total
'''''' '
! ' ' ' ' ' I'4'0' 30, I ''' I
'
' ' 5,0, ',' I 'I ' I ' ,'
UNTREATED SAMPLE r T ' ' E O ' ' ' ' ' ' 90' '
6 0 ' ' ' ' ' 70' '
1500 Units Peak Height
E
*
I
HCI TREATED SAMPLE
0.1 0.02 0.1 0.0 0 0 0.0 0 0 0.0
m
'
31
0.05 0.05 0.3
0.2 0.2 97.8 0.8 0.1 0.2 0.1 0.2 0.1 0.1 0.2
'
a
Precision Limits5
hIole
38.7 17. I 14.0 14.0 10.3
100.0
c3
Analysis of Isobutene Mixture by Special Procedure
Hydrocarbon
a
completely reacted with hydrogen chloride, the sum of the mass 53 and 55 peaks due only t o butenes was calculated. Since 1-butene and 2-butene mass spectra are essentially equal for this combination of peaks, the mass 56 peak calculated for this residual was found t o approximate closely that actually found for the butenes. The data in Tables I and I1 compare the known compositions of Phillips synthetic mixtures 4 and 8 with those determined by the present method. Reference to these tables shows that the calculated and known compositions of isobutene, 1-butene, and 2butene agree t p within 0.1 t o 1.0 mole %. The mass spectra of tert-butyl and tert-amyl chlorides are tabulated in Table 111. The over-all time requirements for the analysis of C, fractions were found to be 2 man-hours by Method 1, and 2.5 man-hours by Method 2. The hydrochlorination procedure consumed about 0.5 hour, while the remainder of the time was spent on niassspectrometer analysis.
20
30
40
50
MASS
60
NUMBER
70
80
90
(m/e)
Figure 2. Mass Spectra of Typical BB Feed at 40Micron Pressure
Precision limits based on two aeperate analyses.
Table V.
Uncertainties in Determination of Each Butene Isomer Due to Presence of Other Butenes (Conventional h1.S. analysis a n d HC1 treatment method) Uncertainty, microns pressure
-
Untreated Butene RIixture Butene Computed
Rlass
Isobutene
1-Butene
2-Butene
2m
Root error squared
u. I D 0.17 0.22
1-Butene
2-Butene
41 55 56 57 2s
41 55 56 57 2s
0.0382 0.0153 0.0535
0.0391 0.0176 0.0466
0.1070
0.1032 0,1098 Sensitivity fluctuations Peak fluctuations Total fluctuations 0.0390 0.0349 0.0246 0.0277 0.0145 0.0171
....
0.0381 0.0214 0.0167
....
0.0762
....
....
0.0380 0.0198 0.0549
....
....
0.1123 0.0527 0,1550
....
0.3200
0.1120 0.0737 0.0484
....
0.0781 0.0798 0.2341 Sensitivity fluctuations Peak fluctuations Total fluctuations Butene mixtures contained 1 micron of each butene,
.... .. .. o:i9 0.20 0.27
tert-Butyl chloride
... ... ... ... 0.0240
:
0 0033 0,0206 0.0480
....
....
....
HC1-Treated Butene hlixture 1-Butene
2-Butene
Zm
Root error squared
.. .. .. .......
....
.... ....
u.u1
....
.. ..
0.0417
....
0.0317 0 0.0734
.... ....
.. .. .. ..
'
....
0.0314
0.1031
0.0374 0 0.0748
0.0724 0,0206 0,1962
.... ....
.. .. .. ..
....
o:iz
....
..
.... ....
o'ii
....
....
. I . .
..
0.0240
0.0354
0.0317
0.0874
..
0.0037 0 0166
0.0353 0 0.0707
0.0417 0.0734
0.0808 0.0166 0 1848
.... ....
....
..
.. o:i3 0.14 0.20
....
0.0407
.... .... ....
....
.... .... ....
0.01 0.01 ,
....
0
....
.. ..
0.13 0.17
.. ..
.. 0.12 0.16
ANALYTICAL CHEMISTRY
142 n-BUTENE ANALYSIS IN ISOBUTENE SAMPLES
I n special cases it is desired to determine a 1-butene and 2butene split for samples containing approximately 95% isobutene. Although the hydrogen chloride treatment described eliminates isobutene as such, the resulting tert-butyl chloride interferes seriously in the resolution of the small concentration of 1-butene and 2-butene. T o increase the relative partial pressure of 1-butene and 2-butene compared to the partial pressure of tert-butyl chloride in the hydrogen chloride-treated sample, a simple partial condensation is made bv cooling the reaction bulb to -40' C. prior tointroduction to the mass spectrometer. .4t the reduced temperature and pressure existing in the reaction bulb practically all of the tertbutyl chloiide is condensed while only a relatively small amount of the butenes is dissolved by the liquid phase. I t was found that this process increased the butene concentration in the vapor phase by a factor of about 10. A sample may then be removed from the reaction bulb a t -40" C. and introduced directly into the mass spectrometer. The mass spectrum vhich is obtained from the 'vapor in the reaction bulb a t -40" C. (record C) is used in conjunction with the two mass spectra already described from the untreated sample and the hydrogen chloride-treated gas sample for computation of the analysis Records A and B are computed in the manner outlined in Method 2, with the exception that the more stable 57 peak is used as the base peak for tert-butyl chloride instead of the 77 peak. Record C is used only to indicate the relative amounts
of I-butene and 2-butenes by solving two simultaneous equations based on the 39 and 56 peaks. This ratio may then be applied to the results calculated from records A and B to complete the analysis. A typical analysis of a sample containing more than 95% isobutene which was treated by this latter method is shown in Table
IV. ACCURACY OF METHOD
To determine what advantage the method might have in reducing errors due to changes in sensitivity caused by fluctuations of the instrument, a mathematical analysis was made of a represrntatire mixture. Error breakdovns, calculated by an inverse method (1) for an assumed sensitivity fluctuation of 1% are shown in Table V. I t is apparent that the uncertainties for a mixture containing tert-butyl chloride are smaller than for the mixture in n-hich isobutene is present; hence greater accuracy should result .from the use of hydrogen chloride treatment than would be possible from the conventional mass spectrometer analysis. LITERATURE CITED
(1) Consolidated Engineering Corp., Pasadena, Calif., private com-
munication. (2) McMillan, ISD.ENG.CHEM.,AKAL.ED.,9, 511 (1937) (3) Washburn, Wiley, and Rock, Ibid., 15, 541 (1943). RECEITED J u n e 9. 1947. Presented before the Division of Petroleum Chemistry at the 112th Meeting of the AMERICAN CHEMICAL SOCIETY, New York
PI'. Y .
Determination of Sodium, Potassium, Aluminum, and Zinc In Derivatives of Carboxymethylcellulose E. P. SAR'ISEL, SPENCER H. BUSH', ROBERTA L. WARREN, AND A. F. GORDON* The Dow Chemical Company, Midland, Mich. &lethodsare described for the determination of sodium, potassium, aluminum, and zinc in derivatives of carboxymethylcellulose. They were developed on the basis of accuracy and rapidity of analysis for the application to research and control purposes.
T
HE rapidly increasing commercial importance of the metallic salts of carboxymethylcellulose makes desirable the development of analytical methods for the accurate determination of the specific metal ions contained therein. Procedures have been devised for the determination of sodium, potassium, aluminum, and zinc in their respective salts of carboxymethylcellulose. Rlost of the methods presented in this paper are satisfactory for the analysis of varying amounts of one element in the presence of the other elements mentioned. For example, small amounts of sodium are likely t o be found in other salts of carboxymethylcellulose and it would be useful to analyze for sodium and aluminum in B mixture of the sodium and aluminum salts of carboxymethylcellulose. I n so far as possible, rapid methods of analysis have been developed, utilizing colorimetric and volumetric techniques. Considerable investigation was required t o find a suitable rapid method for the digestion of the salts of carboxymethylcellulose, in order that the cellulosic residue might be destroyed and leave the metallic constituents in a soluble state. Wet-ashing with a sulfuric acid-nitric acid mixture is commonly used for the digestion of organic material and is satisfactory for this type of 1 ?
Present address, University of Michigan, Ann Arbor, hlioh. Present address, Purdue University, Lafayette, Ind.
product, but it is time-consuming. Lindner (8, 9) and others (3) have reported that 3oY0hydrogen peroxide aids in the digestion of plant tissue. The use of 30% hydrogen peroxide subsequent t o a preliminary treatment with sulfuric acid greatly speeds up the digestion process. Dry-ashing for the destruction of cellulose residues is usually more rapid than n-et-digestion but suffers from the fact that at elevated temperatures insoluble oxides may be formed with some metals, which gives low analytical results. This difficulty may be overcome and most of the benefits of dry-ashing retained bv ashing a t some moderate temperature such as 400" C. The ashing of the sample should be only partially completed and at a moderate temperature it should be sufficient to remove all of the volatile constituents and to reduce the bulk of the original sample to a carbonaceous residue. Subsequent to dry-ashing, the residue is taken up in concentrated sulfuric acid with continued additions of 30% hydrogen peroxide until the solution becomes clear. When digestion is complete, the clear liquid remaining in the calibrated 50-ml. Kjeldahl digestion flask (Figure 1) is diluted to the mark with distilled water and suitable aliquots are removed for metal analysis by the various procedures described belon. Khen it is knovin that sodium is present alone in an organic compound, a convenient method of analysis (7) is t o
