(10) value of 2 x gram per second for normal heptane and Lovelock’s (14) value of 3 X gram per second for propane. It compares to 0.04 p.p.m. for N-pentane and 0.001 p.13.m. for ethylene as reported by Bellar et nl. ( 6 ) , based on a signal to noise ratio of 2 to 1. By use of the capillary column, the isopentane peak was sharper than shown in Figure 2 for even greater sensitivity. Total Organic Carbon Detector. The total organic carbon detector, which uses air as scavenging gas, responded to changes in oxygen and carbon dioxide contents of the sample. This is no problem in normal air samples where the oxygen content remains constant but can affect results in the analysis of combustion products. Figure 4 shows the response t o 8.5 p.p.m. of methane for varying oxygen content. By reducing the oxygen content from 20 to 12%, the response to 8.5 p.p.m. of methane is completely cancelled. Below 12% oxygen, the response is negative. When oxygen, instead of air, is used as the detector scavenging gas, no detectable change in response arises from oxygen and carbon dioxide variations. Detector response time to an increase in concentration is 2 to 3 seconds. When high concentrations are encountered, the time for returning the instrument to its zero or reference state while flushing with clean air may be as long as 1 minute, depending upon how readily the organic substance is desorbed from the flow system. The instrument is sufficiently stable in transit that sensitivity in excess of full scale response for 1 p.p.m. of
methane equivalent can be maintained. The electrometer is used without input feedback to attain the stability needed. Somewhat more sensitivity can be obtained by keeping the electrometer stationary and moving only the probe to pick up localized concentration changes. Some instrument drift arises because of variation in the air sampling stream. The primary cause is change in the flow resistance of a filter located a t the input to the pump. As the filter clogs, the sample flow rate through the detector decreases making it necessary to either clean the filter or adjust the sample bleed valve to restore sensitivity. I n addition, sample stream pressure fluctuations occur because of drifting pump battery voltage, variation in pump characteristics, and temperature. Under normal operations, occasional adjustment of the fine metering valve, F 1 in Figure 1, is required to maintain a constant pressure on gauge E. The 10-mp millipore filter is required to remove submicroscopic particulate matter from the sample gas stream. Without its use, anomolous readings may arise as dust particles are carried through the system. I n earlier tests these particles were not removed by a n inert 100- to 200-mesh C22 insulating brick packing used as a restriction to gas flow to the detector. Normal service required on the instrument consists of changing the sampling filter at a frequency dependent upon the type of service, recharging or exchanging the pump battery after approximately 6 hours of use, refilling the compressed oxygen bottle after 8 hours, and replacing the hydrogen supply after 50 hours of operation.
CONCLUSION
The flame ionization detector has proved to be rugged and readily adaptable for use in portable field instruments. Using it in a chromatograph, on-the-spot analyses have been made of C1 through CT hydrocarbons in dilute air samples. Concentrations of 0.01 p.p.m. v./v. have been determined and the practical sensitivity limit of the detector has been approached under field conditions. LITERATURE CITED
(1) Altshuller, A. P., ANAL.CHEM.35, 3R
(1963). (2) Altshuller, A. P., Bellar, T. A., Clemons, C. A., VanderZanden, E., Int. J . Air Water Poll. 8 , 29 (1964). (3) Altshuller, A. P., Clemons, C. A,, ANAL.CHEM.34, 466 (1962). (4) Andreatch, A. J., Feinland, R., Ibid., 32, 1021 (1960). (5) Bellar, T. A., Brown, M. F., Sigsby, J. E.. Ibid.. 35. 1924 11963’1. (6) Bellar, T: A.; Sigsby, J. E., Clemons, C. A., Altshuller, A. P., Ibid., 34, 763 (1962). (7) Bruderreck, H., Schneider, W., Halasz, I., Ibid., 36, 461 (1964). (8) Clemons, C. A., Altshuller, A. P., J . Air Pollution Control Assoc. 14, 407 (1964). (9) Dal Nogare, S., Juvet, R. S., ANAL. CHEM.34, 35R (1962). (10) Desty, D. H., Geach, C. J., Goldup, A., “Gas Chromatography 1960,” R. W. P. Scott, ed., pp. 46-61, Butterworths. London. 1960. (11) Dewar, R. A., J . Chromatog. 6 , 312 (1961). (12) Juvet, R. S., Dal Nogare, S., ANAL. CHEM.36, 36R (1964). (13) Kuley, C. J., Zbid., 35, 1472 (1963). (14) Lovelock, J. E., Zbid., 33, 162 (1961). (15) Reynolds, R. A., Teichman, T., Monkman, J. L., J . Air Pollutaon Control Assoc., 14, 295 (1964). RECEIVED for review September 16, 1964. Accepted January 19, 1965.
Analysis of Alkyltin Bromides by Gas Liquid Chromatography R. D. STEINMEYER,’ A. F. FENTIMAN, and E. JACK KAHLER Battelle Memorial Institute, 505 King Ave., Columbus, Ohio
b N o method has been reported previously for the quantitative determination of mixtures of butyltin bromides. The butyltin bromides may be quantitatively converted to their more volatile butylmethyltin analog with methyl Grignard reagent and then determined b y gas liquid chromatography with an estimated accuracy of about *2%. The method appears applicable to a number of tetralkyl and/or tetraryltins and alkyl or aryltin halides. 1 Present address, Dow Corning Corp., ;\lidland, bIich.
520
e
ANALYTICAL CHEMISTRY
D
the investigation of the nature of radiation-induced reactions of alkyl bromides and tin, it became apparent that no suitable method for the analysis of butyltin bromide mixtures was available. Initial attempts to analyze these materials by gas liquid chromatography (GLC) produced curves suggesting severe decomposition of the sample. Since the boiling points of these bromides are quite high, efforts were directed toward preparing more volatile and more thermally stable derivatives which could be analyzed chromatographically. The replacement of the bromine URING
atoms of alkyltin bromides by methyl groups drastically reduces their boiling point. I n addition, the butylmethyltins, unlike the butyltin bromides, are stable in the presence of one another even at elevated temperatures. The butylmethyltins are easily prepared from the corresponding butyltin bromides by treatment with methyl Grignard reagent. Proesch and Zoepfi (6) have reported the GLC analysis of ethylmethyltins using a column of high-vacuum grease operated isothermally a t 70’ to 90OC. With the butylmethyltins boiling points are so widely spread (78’ to 31OOC.)
Table I. Instrumental Conditions Injection port temperature, 300" C. Detector temperature, 300' C. Carrier gas, helium dried over 5-A., molecular sieves, 65 p.s.i.g. pressure at tank Recorder chart speed, 0.5 inch/min. Attenuator setting, XI then as required to keep peaks on scale Bridge current, 200 ma. Sample size, 10 ~ 1of. ethereal solution
Selection of Analytical Conditions. Samples containing the butylmethyltin compounds in diethyl ether were used to select instrument conditions for subsequent work. The helium carrier gas flow rate was arbitrarily set a t 100 ml./min. and other conditions as in Table I. In order to effect complete separation of tetramethyltin and the large quantities of diethyl ether present, it was necessary to start the temperature program at 50' C. A t a program rate of 10' C./min., all sample components were well resolved with the least volatile, tetrabutyltin, eluting a t 270' C. The tin compounds elute from the column a t temperatures which may be related directly to their boiling points, Figure 1. Internal Standard. Requirements of an internal standard are discussed by Harvey and Chalkley ( 2 ) . The internal standard should have a retention time a t about the midpoint of the chromatogram or in this instance an elution temperature of about 220' to 230' C. Figure 1 indicated that the boiling point would be 235' to 240' C. To nfinimize errors due to mechanical losses, the internal standard was added to the crude sample before preparation steps were begun. Thus, materials that react with Grignard reagents could not be considered as standard. Phenylcyclohexane (Eastman, White Label, b.p. 237.5' C.) seemed to possess the necessary characteristics and hence was selected as the internal standard.
methyltin compounds
that temperature programmed gas chromatography was projected. Further, since the completeness of the reaction of the butyltin bromides with methyl Grignard reagents was unknown and it was not known if the crude bromide samples contained nonvolatile residue, it was felt advisable to use an internal standard for quantizing the chromatograms. This paper describes the analysis of butylmethyltin mixtures via a GLC method that appears to be generally applicable for the analysis of tetraalkyl and/or tetraryltins and alkyl or aryl tin halides. EXPERIMENTAL
The instrument used for the analyses was an F&N Model 720 dual-column, linear-temperature programmed gas chromatograph equipped with a fourfilament hot wire thermal conductivity cell detector and a Minneapolis-Honeywell 1-mv. recorder. Samples were delivered to the instrument with a Hamilton 50-~1.syringe. The columns were 10 feet long 1/4-inch 0.d. 304 stainless steel tubing coiled in a 4- to 5-inch diameter with the ends bent parallel to the axis of the coil. The coils were packed with 15.5 =t 0.2 grams of 20 wt. % G.E. SE-30 silicone gum on 60- to 80-mesh Chromosorb-W, regular. The columns were conditioned in the instrument a t 250' C. overnight with a carrier gas flow rate of 100 ml./min. and the exit ends of the columns disconnected from the detector. Standards. Tetramethyltin and tin tetrabromide were purchased from K & K Laboratories, Inc., and tetra-
butyltin from Eastman Organic Chemicals. Tributyltin bromide, dibutylfin dibromide, and butyltin tribromide were prepared by methods suggested by Jones ( 4 ) , Gruttner ( I ) , and Ingham (S), respectively. The corresponding methyl derivatives were prepared by treating the butyltin bromides with methylmagnesium bromide, and purified by distilling twice a t reduced pressure. Purity of the methyl compounds was checked by GLC, whereas the bromides were converted to methyl derivatives prior to GLC assay. With the exception of the tetrabutyltin which contained 7.2% of a lower boiling unknown and tributyltin bromide which contained 7.9% tetrabutyltin the purity of the standards was greater than 99%
Table II.
Compounda Me4Sn MerSnBu MeqSnBuz MeSnBu8SnBu4 a Me = methyl Table 111.
Compound Me4Sn
Me nR 11 ._."_
Composition of Butylmethyltin Standards Concentration, grams/100 ml. No. 2 No. 3 No. 4 1.584 1.012 0.516 0.470 0.229 0.983 1.038 2,019 1.483 0.481 1 ,444 0,984 1.388 0,268 1.837
No. 1 0.328 0,081 2.515 0.205 0.470 group, Bu
=
No. 5 2.081 0.072 0.482 1.939 0.938
butyl group.
Peak Height Calibration Factors for Butylmethyltin Compounds
Std. 1 1.35 1.27 1 30 1 39 1 39
MelSnBuz MeSnBul SnBu4 a Average of three runs.
Calibration factorsa Std. 2 Std. 3 Std. 4
Std. 5
1.32 1.28 1 30 1 38 1 34
1.29 1.32 1 24 1 39 1 37
1.33 1.31 1 31 1 38 1 40
1.36 1.34 1 29 1 38 1 39
VOL. 37, NO. A, APRIL 1965
Av . 1.33 1.30 1 29 1 38 1 38
s
521
Determination of Calibration Factors. Five standard solutions con-
taining approximately equal weights (about 2 grams) of phenylcyclohexane and varying amounts (about 0.5 t o 2 grams) of the standard butylmethyltin compounds in 100 ml. of diethyl ether were prepared (Table 11). Triplicate analyses were made of each standard mixture a n d calibration factors were calculated. The average calibration factors determined for each component in each standard solution are given in Table 111. For the sake of simplicity, peak height was measured rather than peak area. A 12-inch ruler graduated in tenths of an inch was adequate to make measurements. Since the peaks were recorded on different instrument attenuator settings, the recorded peak heights were multiplied by the attenuator factor (Xl, X 2 , X4, etc.) to normalize the height values. Sample Preparation. Approximately 4.0 grams of the reaction product or known mixture a n d 2.0 grams of internal standard, phenylcyclohexane, accurately weighed, were dissolved in 25 ml. of anhydrous ethyl ether. This solution was added from a dropping funnel to 20.0 ml. of 3.OM methylmagnesium bromide in ethyl ether (Columbia Organic Chemicals Co., Inc.), contained in a 100-ml. three-necked flask equipped with a reflux condenser and a Truebore stirrer with a Hershberg wire paddle. The addition took 15 to 20 minutes, during which time the ether came to a mild reflux. The mixture was refluxed for an additional hour. At the end of this time 8.0 ml. of a saturated aqueous solution of ammonium chloride was added a t such a rate that the mild reflux was maintained; 45 to 50 minutes was required for the addition. The granular precipitate of the
Table IV.
70
90
110
1%
150
0
ANALYTICAL CHEMISTRY
210
2 3
250
270
290
Chromatogram of butylmethyltin compounds
magnesium salts which separated were removed by filtration through a sintered glass funnel and washed three times with 15 ml. of ether; mild suction was used to draw the ether solution through. The clear ether solution of butylmethyltins, essentially dry a t this point, was placed in a 100-ml. volumetric flask and brought to volume with more ether. A 10-p1. aliquot of this solution was then analyzed by gas liquid chromatography. Smaller samples may be analyzed by reducing all weights and volumes proportionately.
Good results were obtained with as little as 1 gram of crude sample. Analysis of Butyltin Bromide M i x tures. Six synthetic mixtures of tin
tetrabromide, butyltin bromides, and tetrabutyltin were processed and analyzed as described above. A representative chromatogram is given as Figure 2. The weight per cent compositions of the mixtures were determined by the following formula: (Wt. of int. std.) (peak ht. of component) (Factor 1) (Factor 2) (100) peak ht. of int. std.) X (wt. of bromide mixture) yo of component Factor 1 is the response calibration factor mentioned earlier, Factor 2 is a wt./wt. correction to determine the weight of the bromide compound from the weight of the methyl compound (Table IV), and the other terms are self-evident, The results of these analyses are given in Table V.
Factor 2 2.45 1.88 1.49 1.21 1.00
438.3/178.7 415.4/220.7 392 51262.7 369 6j304.7 346.7/346 7
ComStd. 15 ponent Prepd. Obsd. SnBr4 12.7 3.0 23.4 BuSnBr3 14.1 40.1 39.9 BuzSnBrz 29.1 Bu8SnBr 19.9 BuaSn 13.2 5.9 Recovery 101.3 Duplicate determinations. b Single determinations.
190
Peaks are: ( 1 ) tetramethyltin, (2) butyltrimethyltin, (3) dlbutyldimethyltin, ( A ) phenylcyclohexane, (5) tributylmethyltin, (6) unknown from (7)tetrabutyltin
Mol. wt. Bu,SnBr4-,/ Mol. wt. Bu,SnMe4-
Table V.
170
Column Temperature, C Figure 2.
Factors to Convert Weight of Butylmethyltins to Weight of Butyltin Bromides
To convert from/to Sn?vIer/SnBrr BuSnMe3/BuSnBr3 Bu~SnMez/BuzSnBrz Bu3SnMe/Bu3SnBr Bu4Sn/Bu4Sn
522
50
Prepared and Observed Compositions of Butyltin Bromide Standards
Std. 2a Prepd. Obsd. 20.9 11.3 18.3 27.2 22.3
7.6 23.1 18.3 38.9 __ 12.7 100.6
Composition, wt. 70 Std. 3b Std. 4b Prepd. Obsd. Prepd. Obsd. 13.4 11.7 ... ... 14.3 14.6 8.0 8,1 70.7 7.2 0.7
70.2 7.4 0.5 97.9
51.8 19.5 14.4
51.3 20.4 --14.5 100.8
Std. 5b Prepd. Obsd. 9.8 18.3 66.5 5.0 0.4
9.2 19.1 66.0 5.2 0.5 100.0
Std. 6b Prepd. Obsd. 11.6 26.9 50.2 10.3 1.0
10.3 26.8 49.7 10.7
1.0 98.5
( 2 ) Harvey, D., Chalkley, I)., Fuel 34,
Table VI.
Prepared and Observed Mole Per Cent Compositions of Butyltin Bromide Standards 1 and 2
Composition, Mole Per Cent Compound SnBra BuSnBrs Bu&Brz BuPSnBr BuaSn
Prepd. 11 13 40 21
15
Std. 1 Obsd. 3.6
22 41 30 6.5
f9.O
Prepd. 18 11
+9: 0
18 28
Diff. -8.4
25
-8.5
RESULTS AND DISCUSSION
I n the early phases of the work, poor reproducibility of both the retention time and relative peak height of the first peaks in the chromatograms was evident. Reproducibility was improved considerably by consistently programming the column oven to 280" C., shutting off the controller, opening the oven and permitting it to cool for exactly 15 minutes (during which time the programmer was reset to 50' C,), closing the oven, injecting the next sample when the column oven temperature indicator reached 50' C., and then turning on the programmer. Through this technique, the elution temperatures of sample components were reproduced to within two degrees. The reproducibility of the relative peak heights is reflected by the reproducibility of the peak height calibration factors given in Table 111, where the greatest variation is *1.87,. I t is also to be noted from Table E11 that even though the concentration of the sample components varied from less than 0.1 gram to more than 2 granis per 100 ml. of solution, the calibration factors remained constant. Apparently, with a 3-gram sample diluted to 100 ml., the detector response is linear from as low as 3y0 to as high as 70% sample component concentration. Difficulties were encountered in working with the standard alkyltin bromide mixtures. Recovery data (Table V) show that the Grignard reaction is indeed quantitative and that no loss of sample occurred during other steps in the sample preparation procedure. Xevertheless, drastic differences between the added and observed quantities of all the components excluding the dibutyltin dibromide were noted in Standards 1 and 2 . By converting the weight per cent values to mole per-
Std. 2 Obsd. 7.0 21
18 40 14
Diff.
+-1110 +'12 -11
centages, as in Table VI, the reason for these discrepencies becomes apparent, The molar loss of tin tetrabromide and tetrabutyltin is equaled by the gain of butyltin tribromide and tributyltin bromide. What has occurred is the room temperature reaction of equal molar amounts of tin tetrabromide and tetrabutyltin to form butyltin tribromide and tributyltin bromide prior to reaction with the Grignard reagent. The same redistribution reactions were encountered by Matsuda and Matsuda (5) during attempts to prepare standard solutions of other organotin halides. Standards 3 through 6 xere made to contain minimal amounts of either tin tetrabromide or tetrabutyltin and the agreement between observed and added concentrations is much better, Table V, but even here the observed values for tin tetrabromide and tetrabutyltin are consistently low and the observed values for butgltin tribromide and tributyltin bromide consistently high. Dibutyltin dibromide is the only material that does not take part in this redistribution reaction. The data obtained for this material are than more indicative of the accuracy of the method and the observed values are within lYO of the actual percentage in every case. Apparently, accuracy of the method is a t least as good as the precision with which the calibration factors for the butylmethyltin compounds can be determined. Such factors varied only d ~ 1 . 8 7over ~ a concentration range of 3 to 70%. ACKNOWLEDGMENT
The authors thank R . E. Wyant and J. F. Kircher for their helpful suggestions. LITERATURE CITED
Gruttner, G., Krause, E., Wurnik, M., Chem. Ber. 50, 1549 (1917).
(1)
191 (1955). (3) Ingham, I
