(5) Eggertsen, F. T., Groennings, S., ANAL.&EM. 3 0 , 2 0 (1958). (6) Eggertsen, F. T., Knight, H. S., Zbid.;3O, 15 (1958). (7) Eggertsen, F. T., Knight, H. S., Groennings, S., Zbid., 28, 303 (1956). (8) Forziati, A. F., Willingham, C. B., Mair, B. J., Rossini, F. D., Proc. Am. Petrol. Znst. ( Z Z Z ) 24,34 (1943). (9) Gooding, R. M., Adams, N.G., Rall, H. T., I N D . ENQ.CHEM., ANAL. ED. 18, 2 (1946). (10) Griswold, J., Van Berg, C. F., Znd. Eng. Chem. 38,170 (1946). (11) Kszanskii, B. A., et al., Zzuest. Akad. *Vaxk S.S.S.R. Otdel. Khim. Kauk 1951, 100, 1954, 286, 278, 456, 865, 1053; Bull. Acad. Sci. U.S.S.R., Diu. Chem.
Sci. S.S.R. 1954,747,917,and later publications, (12).Martin, A. J. P., James, A. T., Bzochem. J. 63, 138 (1956). (13) Nerheim, A. G., patent applied for. (14) Rossini, F. D., Mair, B. J., “The Work of API Research Project 6 on the Composition of Petroleum,” Section V, Fifth World Petroleum Congress, New York, June 3, 1959. (15) Rossini, F. D., Mair, B. J., StreilT! A. J.. “Hvdrocarbons from Petroleum, Reinhold,”Kew York, 1953. (16) Smith, H. hl., Kraemer, A. J., Thorne, H. M., “Aviation Gasolfpe and Its Component Hydrocarbons, Bur. Mines Bull. 497 (1951). (17) Smith, H. hl., Rall, H. T., Znd. Eng. Chem. 45, 1491 (1953).
(18) Urmancheev, F. A., Robinzon, E. A., Kashaev, K. G., Le, B., Zsvest. Akad. Xauk S.S.S.R., Otdel. Khim. Nauk
1957,711,1958,324. (19) Winters, J. C., patent applied for. (20) Winters, J. C., Jones, F. S., Martin, R. L., “Gas Chromatography Guides Development of a Paraffin-Isomeriaation Process;” Section V, Fifth World Petroleum Congress, New York, June 1, 1959. (21) Zlatkis, A., ANAL. CHEM.30, 332 (1958).
RECEIVEDfor review June 5, 1959. Accepted August 24, 1959. Division of Petroleum Chemistry, 136th Meeting, ACS, Atlantic City, N. J., September 1959.
Pyrohydrolysis of Cryolite and Other Fluoride-Bearing Materials H. P.
SILVERMAN’ and
F. J. BOWEN
Metals Division Reseurch laboratory, Kaiser Aluminum and Chemical Corp., Permanente, Calif. ,a-Alumina is used as an accelerator in the quantitative pyrohydrolysis of cryolite and aluminum reduction cell electrolytes in a generalized procedure which is applicable to other fluorides as well. A simple nickel reactor permits the use of standard laboratory apparatus. The results are of the same order of precision and accuracy as reported b y other investigators. Mechanisms for the pyrohydrolysis of cryolite in the presence and the absence of the accelerator are suggested.
S
authors (1, 6, 7 , 9-11) have reported on the use of pyrohydrolysis to determine the fluoride content of various salts. Warf, Cline, and Tevebaugh (If) used a n accelerator, uranium oxide (UaOs),to promote the hydrolysis of the slow-reacting alkali and alkaline earth fluorides. Haff, Butler, and Bisso (6) reported a successful procedure for aluminum fluoride but found that cryolite, sodium fluosilicate, and sodium fluoride were not completely hydrolyzed. This paper also contains a n excellent comparison of the merits of pyrohydrolysis and the Willard and Winter ( l a )distillation. The reactors described by the investigators cited above have, in general, been fabricated from quartz and platinum, requiring a large expenditure for equipment. Susano, White, and Lee (10) as well as Gahler and Porter (4) have described nickel apparatus which have given satisfactory service. The nickel apparatus described in the present paper is simple and is designed for EVERAL
1 Present address, Lockheed Missile Systems, Sunnyvale, Calif.
1960
ANALYTICAL CHEMISTRY
use in a standard globar tube furnace, so that temperatures higher than those obtained with wire-wound resistance equipment are practical. Aluminum fluoride, cryolite, and, in some cases, reduction cell electrolytes are analyzed routinely for fluoride in the aluminum industry. Devising a single, simple, rapid, and precise procedure to handle these materials necessitated a study of the use of accelerators in pyrohydrolysis. This paper describes the apparatus and procedure which were used to pyrohydrolyze quantitatively aluminum fluoride, cryolite, and reduction cell electrolytes. EXPERIMENTAL
Apparatus. T h e reactor, shown in Figure 1, consists of three basic parts. Section B, which carries t h e 0.5-inch steam inlet tube, is a cap which may be clamped t o section A to provide a n airtight seal. Section A is a 24-inch length of I-inch standard nickel tubing. One end of A is closed by a solid nickel plug welded in place and t h e other end carries a nickel collar. The collars which form the coupling between A and B are machined from 1.5-inch solid nickel rods and are welded in place. A 3/16-inch male and female radii coupling is machined on the faces of the collars (inset, Figure 1). A length of 0.5-inch nickel tubing, welded at right angles into the end of A , serves as a n exit tube for the distillate. As shown in Figure 1, a brass water jacket is silver-soldered over the exit tube so that the entire assembly serves as a condenser. The clamp connecting A and B consists of a half-round section, to which is fastened a U-shaped strap which
carries a thumbscrew. The clamp drops over the tube body and, when tightened, bears against the back of the collar. T o close the reactor the male and female radii of A and B are matched, the clamp is dropped into place, and the thumbscrew is tightened against the end cap, The entire assembly should pass a IO-p.s.i. pressure test. All parts that project from the furnace are wrapped in asbestos except the condenser and the collars over which the coupling clamp is placed. A new reactor is conditioned by passing steam through i t at operating temperature for several hours. The tube furnace must meet three conditions: The temperature of the hot zone must be controlled easily, the furnace must maintain a hot zone temperature of 1200’ C., and at least 8 inches of the reactor tube must be in the hot zone to provide adequate superheating of the steam. The long hot zone obviates the need for a steam preheater. A Lindberg Model CF-2 or a n y other two-tube globar furnace is convenient, as two reactors can be handled at the same time. The steam-generating flask used has its lead wires to the immersion heater and a water inlet tube entering near the bottom of the flask. This is convenient because the water level can be controlled readily without disconnecting the steam lead a t the top of the flask, and the heating rate can be controlled with a Variac. It is not desirable to add water during a run, however, because a pressure drop due to cooling may pull volatilized fluorides back into the steam generator. A rubber stopper carrying a glass steam exit tube is used to close the flask. Rubber tubing connects the generator to the steam inlet tube of B, which permits the end cap to be moved away from the reaction tube for insertion of samples without
disconnecting the steam lead from the generator flask. Heavy-duty nickel combustion boats are recommended. A standard boat 4 3/* inches in over-all length is convenient. This boat has a handle or lip with a hole which facilitates the insertion or withdrawal of the boat. Alundum boats are too fragile; A boat puller may be fabricated by bending a hook a t one end of an 18-inch length of a l/s-inch diameter nickel or stainless steel welding rod. A small semicircle of rod, spot-welded to the hooked end of the boat puller so that the semicircle forms a fork above the hook, is convenient for guiding the boat. T o collect the emuent, 250-ml. polyethylene beakers are used. Procedure. The samples are ground t o pass 100 mesh. About 0.25 gram of a n accurately weighed sample is mixed intimately with approximately 0.75 gram of a-alumina by rolling on a glazed paper. Approximately 0.5 gram of alumina is placed in the rear of the boat to form a layer 1 inch long. The sample mixture iS placed on this layer and covered with a final layer of about 0.5 gram of alumina. The furnace temperature is adjusted to 1200" f 50' C. and the steam generator is adjusted to deliver 8 1 ml. of condensate per minute through the reactor. A polyethylene beaker containing 50 ml. of carbon dioxide-free water is placed under the tip of the condenser so t h a t the tip of the condenser tube is immersed about 1/4 inch. The sample boat is inserted into the reactor and positioned so that the rear of the boat is a t the mid-point of the hottest zone. This normally will be the midpoint of the furnace, Steam is allowed to escape through the cap during this operation. T h e reactor is closed as soon as possible after positioning the sample by clamping the cap in place. The influx of air,
which imparts a smoky appearance to the effluent, may cause some mechanical loss of fluoride and should be kept to a minimum. Steam is passed for 10 minutes, after which the beaker is lowered so that the condenser tip is no longer submerged. The passage of steam is continued for 5 minutes more to flush out the reactor. The total condensate is collected and titrated with standard 0.2N sodium hydroxide solution using 4 drops of 1% phenolphthalein aa indicator. The titration is performed a t room temperature until a perceptible pink color persists for 15 seconds. The solution is then transferred to a n Erlenmeyer flask and heated until the color fades. The titration is continued until the indicator color persists for 1 minute, a t or near boiling. The entire volume of sodium hydroxide used is recorded. A blank is determined by omitting the sample and repeating the entire procedure including the proper quantities of alumina. If only aluminum fluoride is to be analyzed, all of the alumina accelerator can be omitted and the steam flow rate can be reduced to 5 ml. per minute. All other parts of the procedure including the sample weight remain the same. DEVELOPMENT OF METHOD
Standard Samples. Alcoa spectrographic standard samples, SRP-1 (aluminum fluoride) and SRP-2 (cryolite), were used as jtandards. These samples, purchased from the Aluminum Co. of America's Research Laboratories, were reported to contain 59.17 and 50.8% fluoride, respectively. Samples of reduction cell electrolyte were synthesized by melting reagent grade components t o give the desired compositions. These were in a nominal composition range of 8% calcium fluoride, 0.2y0 magnesium fluoride, and 4%
OUTLET
INLET
SUPPLY \
--..!-NICKEL
COLLAR
NLET
Figure 1.
Reactor
-I=.
Table
I. Effect of Accelerators on Pyrohydrolysis of Cryolite
Accelerator None None Si01
v*o. v;o;
u-ALO~ a-AlrOi
Table
II.
0
*
Fluoride Recovery,
28.94 29.09 39.87 50.76 48.47 50.62 50.59
56.97 57.26 78.48 99.92 95.41 99.64 99.59
%
Weight of Charge Fluoride Recovery
Combined Wt. of Charge 0.250 0.500
Fluoride Found after 15 Min., %
Fluoride
Found, % 50.6 50.7 50. 4b 49.4
vs.
Recovery, %a
99.6 99.8 1.OOO 99.2 2.000 97.2 Based upon fluoride content of 50.8%. Average of 19 determinations; a11
others are average of duplicates.
alumina with the balance of cryolite, Na&F6. Conditions to minimize volatilization were used and an account was made of all weight losses during such syntheses. Any sample for which the weight loss was significant was rejected. Preliminary Investigation. A preliminary factorial experiment was conducted to determine the significant variables, the relative effect of the variables, and the interactions between variables, if any. This led to more detailed work covering significant factors. An investigation of the pyrohydrolysis of aluminum fluoride, done prior to that of cryolitc, agreed with the results and conclusions recently reported by Haff, Dutlcr, and Bisso (6). Necessity for Accelerator. T h e need for a n accelerator in the pyrohydrolysis of cryolite is shown in Table I, where, when the same procedure is used, there is a marked difference between results obtained with and without accelerators. Although vanadium pentoxide was considered as an accelerator, there appeared to be no advantage to the use of this reagent as compared to the use of aalumina, which is readily and cheaply obtained. Uranium oxide Mas not investigated. Proportions of Accelerator to Sample. The effect of the weight ratio of alumina to cryolite was studied over a wide range. A total weight of t h e charge (sample plus alumina exclusive of the insulating layers of alumina) of 1 gram or less led to the most quantitative results (Table 11). As low total charge weights of 0.250 and 0.500 gram are undesirable for routine work because of the small sample weight involved, a total charge weight of 1 gram was used in the study of the effect of VOL. 31, NO. 12, DECEMBER 1959
1961
Table Ill. Effect of Weight Ratio, Temperature, and Steam Flow on Recovery of Fluoride from Cryolite Steam Temp., O
c.
1200 1200 1200 1200 1200 ROO 1000 1100 1200 1200 1200 1200
Flow. lIl./Min,
Condensate 8 8 8 8 8 8 8 8 8 4 6 8
\\-eight Ratio 1:l
3 :1 ti: 1 10: 1 20: 1 3 :1 3:1 3: 1 3 :1 3 :1 :3: 1 3:1
AV.
Determinations
Fluoride Found, $;
1 19 6 6 2 1
34,M 50.42 50,71 50.45 50, i 2 39.54 49.07 48.32 50.42 50.26 50.16 50.50
14 2 47 14 2 33
Standard Deviation ... 0.28 0.23 0.29 0 . 16a
...
3.3
0'52 0 84 0.01a 0.29
Recovery, C'
68 99 99 99
8 3 8 3 8 8 6 1 3 9
99 77 96 95 09 98 08 7 99 4
Range. the alumina-cryolite ratio on fluoride recovery. Table I11 shows that as long as the neight ratio of alumina to cryolite is a t least 3 t o I, maximum recovery is attained and there is no deleterious effect if the ratio is higher than 3 to 1. Thus, if a 3 to 1 weight ratio of alumina to sample is used, an adequate quantity of alumina will be present for a pure sample and no harm will result if the sample is contaminated with alumina or inert material. Figure 2 illustrates the effect of thrb alumina-cryolite ratio on the rate of pyrohydrolysis. The data were obtained b y titrating the condensates eontinuously a t room temperature as they w r e collected. Because the hot titration described in the procedure n as not used, the results are probably low. Although a n increase in rate is observed nith an increase in the weight ratio from 3 : 1 to 6 : 1, the recovery is essentially complete after 8 or 9 niinutep rrgartlless of the weight ratio. The data suggest, therefore. a total charge of 1 gram (sample plus alumina exclusive of insulating layer n ere chloride and sulfate.
The catioiis did not o f f u aiiy special difficiilt'y. ISecausc cations were cxpected to interfere by the formation of stable fluorides. the addition of an acccltmtor ivould onlj- decrt>aer aii interference n.hicli \vas already insignificant ; thus. it sc~enit~l unniwssarj. to inrcsstigat'e the h h a v i o r of cations upoil adtlition of mi accvlerator. The prcwnce of an acct,ltbrator might incrcase the influencc of anionic impuritirs on thc. apparent fluoritli. rccowry. Tlic cffcct of som(' of tlw mow conimon anions 011 the dctrrniination of fluoride in SRP-2 cryolite is shon.n iii T a b k VI, the o n l ~ largc . crror lwing introduced by chloride. Application
to
Other
Materials.
Various materials were pyrohydrolyzed by the described prowtlure. T h e materials and the results obtained are listed in Table YII. -4gain difficulty was encountered in evaluating t h e accuracy of the pyrohydrolytic procedure berause of the lack of an accurate referee procedure. However, t h e results obtained oil :t VOL. 3 1 , NO. 12, DECEMBER 1959
1963
Table VI.
Effect of Various Anions on Pyrohydrolysis of Cryolite
Material Added Ammonium sulfate Sodium chloride Potassium nitrate Potassium carbonate Table VII.
% of Total Sample
Determinations
10 10 10 10
3. 3 3 3
Fluoride Found,
%
Std. Dev.
45.4 45.4 45.4 45.4
45.5 46.2 45.6 45.6
0.3 0.1 0.7 0.0
%
Application of Pyrohydrolysis to Miscellaneous Materials
Description of Material Sublimed .41Fa SRP-1 A 1Fa Hand-Dicked natural cryolite Particulate separated from fumes evolved by reduction cells
Fluoride Content of Material, Yo Willard-\Ninter P-hydrolysis AverDetermi- AverCalcd.0 age Rangeb nations age Rangeb 67.8 59.17
66.9
...
54.3
... ...
... Synthetic cryolite
Fluoride Present,
... ...
...
Recovery by Pyrohydrolysis,
yo
9 47
66.8 58.6
0.12 0.27
98.5 99.0
4
53.9
0.09
99.2
0.08 0.01
3.1
0.7 0.2 0.3 0.0 0.4 0.2
2 2 2 2 2
43.61 30.75 43.31 34.56 44.3 46.8
+ 2H10
...
...
2
32
2
...
...
52: ii
51.48
51.78 48.7
number of materials b y a WillardWinter distillation followed b y a thorium nitrate titration are compared in Table VII. Although calcium fluoride was n o t completely hydrolyzed by t h e recommended procedure, its presence in t h e reduction cell electrolyte apparently did not cause a n y large error in t h e total fluorine analysis of t h a t material (Table VII). This is probably because t h e reduction cell electrolyte melts below t h e temperature of pyrohydrolysis and t h u s allows a more intimate contact between accelerator and sample than can be obtained between two solids. Calcium fluoride itself melts at a much higher temperature. Behavior of Nickel Reactors. Under normal conditions t h e nickel reactors served satisfactorily for 3 to 6 months of constant use. Cryolite, when not mixed with alumina, and chloride, n hen present in large amounts--lO% or more of the sample expressed as sodium chloride-tended to shorten the life of the tube, The condition of the reactor was determined by running a standard, such as SRP-1 aluminum fluoride or SRP-2 cryolite. Low recoveries and high blanks were characteristic of reactors in unsatisfactory condition. ANALYTICAL CHEMISTRY
+
2SaF 4HF NaAI02, (1)
-+
+
of the total fluoride would be found in the condensate. I n practice, only about 7/12 (57.6%) of the fluoride was recovered. An x-ray diffraction study of the solid residue indicated that in addition to sodium fluoride as the major constituent, small amounts of sodium aluminate (NaAIOFz),and cryolite, there was a fairly large amount of a compound found which could not be identified by reference to known x-ray patterns. The presence of unreacted cryolite would account for some of the missing fluoride. If the unidentified compound were of the type NaAlOF, or NazO zAlZO3,it would account in part for the low recovery as shown b y the following reactions: 2/3
66
a Theoretical for pure compounds. Calculated for synthesized materials; reported value for SRP-1 A1F3. * Difference between duplicate determinations except where four or more determinations, then figure is standard deviation. c Nominal composition consists of approximately 8% CaFz, 0.270 hlgFt, 4% AI203 with balance NaaAlFo.
1964
DISCUSSION
Mechanism. If cryolite in t h e absence of a n accelerator reacted as expected (5): ‘ga3AIF6
42.9 31.2 42.7 35.3 45.6 47.8
...
Synthetic aluminum reduction cell electrolytec Calcium fluoride
steam over these area6 and consequently the rate of rehydrolysis are reduced. This results in low recovery. I n subsequent blank determinations this fluoride is slowly pyrohydrolyzed and causes high blanks. The apparatus as designed obviates the need for a separate steam preheater. By placing the sample boat in the rear half of the hot zone of the furnace, the front half of the hot zone serves as a preheater. The results obtained are in good agreement with those obtained using a separate preheater.
Standards should be run a t the start of each operation, particularly if the tubes have been allowed to stand idle over a long period of time. Reactors which have become unsatisfactory can be restored to usefulness by cutting out approximately 20 inches of the mid-section and welding a new piece of nickel tubing in place. Occasionally i t was possible to restore a reactor to usefulness by polishing the inside of the tube with a wire brush powered by a ‘/r-inch electric drill. The brush was made by welding a small wire brush to the end of a welding rod which could be chucked into the drill. The tube, after such polishing, was reconditioned by passing steam through i t for several hours. Nickel is attacked only slightly by hydrofluoric acid bemuse of a chemically inert nickel oxide layer which is formed at high temperatures on the surfaces of the reactor. The inert form of nickel oxide is gray-black and has a metallic luster. The normal reactive green nickel oxide is formed below this inert layer. The low rcsults are probably caused by flaws developed in the inert layer which permit the hydrofluoric acid to penetrate and react with the reactive layer to form nickel fluoride. As the nickel fluoride is formed in more or less protected pockets, the flow of
Na341F6
+ HzO XaAIOFt + 2 S a F + 2HF -+
(2)
or (22) Na3A1Fn (32 1 ) HzO -+ 2(3z 1 ) HF 2(3z - 1 ) N a F NaZO. ( 2 )Ah03 ( 3 )
+
+
+
+
+
Thus, for every mole of cryolite that reacted as postulated in Equation 2, only of the fluoride would form hydrofluoric acid. If the cryolite reacted as indicated in Equation 3, the percentage of fluoride converted to hydrofluoric acid would depend on the value of x-e.g., if z were 2, 7/12 of the fluoride would be converted to the acid:
+
+
+
4Na8A1FB 7Hz0 + 14HF lONaF Na20.2.41~03 ( 4 ) When cryolite was pyrohydrolyzed in the presence of an excess of a-alumina, the fluoride n a s quantitatively converted to hydrofluoric acid. The x-ray pattern of this solid residue revealcd the presence of p-alumina (P\’a20.11Al2Os) and a new unidentified compound with a pattern different from that d i s cussed above as the major components. Sodium aluminate and a-alumina in small quantities mere also found. If the over-all reaction had been, as sus-
pected, the sum of the following reactions : Na3AlFa
+ 2H20 2NaF
2NaF
+
+ 4HF + NaA102
+ l l A 1 ~ 0 3+ HzO
-+
Na20.11Al203
+ 2HF
(5) (6)
to yield:
+ 3H20 + 11Al~O3 + Naz0.11A120~+ NaA102 (7)
NaaAIFs 6HF
sodium fluoride reacted to form @alumina, i t is suspected t h a t other reactions similar to Equation 6 took place to form compounds of the type, Na20.zA1203, where z was intermediate between 1 and 11 and the value of z was dependent upon the ratio of a-alumina to cryolite in the original mix.
+
stoichiometry would require that the mole ratio of a-alumina t o cryolite in the charge be at least 11 t o 1. I n practice, a mole ratio of approximately 6 to 1 (weight ratio 3 to 1) gave a n essentially complete recovery of fluoride as hydrofluoric acid. Thus, while some of the
LITERATURE CITED
(1) Banks, C. V., Burke, K. E., 0’ Laughlin, J. W., Anal. Chim. Acta 19, 239 (1958). (2) Brownlee, K. A., “Industrial Experimentation,” 4th ed., Chap. I11 and IV, Chemical Publishing Co., New York, 1952. (3) Chilton, J. M., Horton, A . D., AXAL. CHEM.27, 842 (1955).
(4) Gahler, A. R., Porter, G., Ibid., 2 9 ,
296 (1957). (5) Griotheim. Kai. “Contribution to the The&y of the Aluminium Electrolysis,” 22, I Kommisjon Hos F. Bruns Bokandel, Trondheim, 1956. (6) HafS, L. V., Butler, C. P., Bisso, J. D., ANAL.CHEM.30, 984 (1958). (7) Hibbits, J. O., Ibid., 29, 1760 (1957). (8) Marstilles, C. M., Research Laboratory, Aluminum Co. of America, private communication. (9) Menis, O., Powell, R. H., ANAL. CHEM.30, 1546 (1958). (10) Susano, D. D., White, J. C., Lee, J. E., Ibid., 27, 453 (1955). (11) Warf, J.C., Cline, W. D., Tevebaugh, R. D., Zbid.,26, 342 (1954). (12) Willard, H. H., Winter, 0. B., IND. ENG.CHEM.,ANAL.ED. 5 , 7 (1933).
.,
E.
RECEIVED for review March 17, 1959. Accepted August 17, 1959.
Evaluation of a Commercial Alkyl Aryl Sulfonate Detergent as a Column Packing for Gas Chromatography D. H. DESTY and C. L.
A. HARBOURN
Petroleum Division, Research Centre, The British Petroleum Co., ltd., Sunbury-on-Thames, Middlesex, England
,A solid anionic household detergent containing about 17% of an alkyl aryl sulfonate has been recommended as a general-purpose packing for gas chromatography. The effects of flow rate and particle size on column efficiency have been studied at an operating temperature of 134” C. with samples having a wide range of retention volumes. Comparative data are also given at other operating temperatures. The optimum conditions for specific separations are discussed. The separation behavior of the packing has been studied and retention volume data for a large number of volatile materials, mainly hydrocarbons and sulfur compounds, are given for an operating temperature of 245” C. This packing has a wide field of application especially for analytical purposes at high operating temperatures where the ability to construct long columns of high efficiency having a low pressure drop is particularly valuable.
A
a variety of stationary phase liquids have been employed in gas-liquid chromatography, kieselguhr or powders obtained from diatomaceous earth firebrick have been used almost exclusiyely.as supports. Gohlke and McLafferty ( 2 ) demonstrated that a commercial, household, solid detergent made a good general-purpose column LTHOUGH
packing combining the functions of both stationary phase and support. Little detailed information has, however, been published on its performance. The spray-drying technique used in the preparation of this type of detergent appears t o have a wide application in the preparation of column packings for gas chromatography. The characteristics of a commercial alkyl aryl sulfonate detergent similar to that employed by Gohlke have, therefore, been investigated particularly with respect to the factors determining column efficiency. I n addition, the nature of the separation produced with hydrocarbons and sulfur compounds has been examined. APPARATUS
The chromatographic instruments employed in this work are of two different types. The first, with which most of the work was carried out, has as detector the gas density balance devised b y A. J. P. Martin. Two models were used, one similar to that originally described b y James and Martin ( 6 ) , and the other a modified design for use a t high temperatures. An improved recording system comprising a Leeds & Northrup DC microvolt amplifier feeding a 10-mv. high speed (1 second full scale) potentiometric recorder has been incorporated in both instruments. The second, which was used only for the determination of the relative reten-
Table I. Sieve Analysis and Water Content
Mesh Size (BSS) > 18 18-30 44-72‘ 72-100