February 15, 1942
ANALYTICAL EDITION
sure data to 0.00001 gram per cc., using the saturated vapor of pure sulfur dioxide as standard. Numerical data are r e ported in another paper (4). For the precision with which these measurements were carried out, it was necessary to devise suspension points for both helix and bob so that the helix was never distorted when suspended, but hung freely. Precise indices which were out of the axis of the tube were also necessary for proper focusing. The cross point in a figure-eight loop of the helix fiber ( I , Figure 1) was used to suspend the helix from a loop of 0.1-mm. (0.004-inch) platinum wire and likewise to suspend the bob from the helix. The indices ( H , Figure 1) consisted of 0.5mm. loops in the helix fiber a t either end of the helix. They had sufficient curvature to permit the most accurate setting of the microscope cross hair a t their outer edge and were in good enough alignment so that no change in the focus of the microscope was necessary when reading the two ends. It is generally recognized that quartz helices exhibit no hysteresis after once being annealed. Quartz is also not easily corroded and has a very low thermal coefficient of volumetric per O C. For a 2-cc. quartz bob, and a expansion, 1.2 X change in temperature of 15O , the volumetric expansion amounts to 0.000036 cc. This is negligible for ordinary vapor density measurements but appreciable for precise measurements of liquid density, and will account for the small diver-
,
131
gence which the authors have found in water. However, this effect can be eliminated by making the necessary correction for the change in the volume of the bob. The volume of a hollow quartz bob of substantial wall thickness is also not greatly affected by pressures of the order of 4 atmospheres (approximate vapor pressure of sulfur dioxide a t 25’). If the bob is sealed off a t atmospheric pressure, the effect of 4 atmospheres is reduced to that of 3 atmospheres. Using Bridgman’s (1) value for the compressibility of mineral quartz, the change in the volume of a 1-cc. bob would be 0.27 X cc. per atmosphere. The error resulting from this cause can be reduced if necessary by determining the approximate compressibility of the bob before it is sealed off. It is believed that this method will be found useful for the precise measurement of densities in many systems which are not accessible to measurements by the usual methods.
Literature Cited (1) Bridgman, P. W., Am. J . Sci., 15, 287 (1928). (2) McBain, J. W., and Bakr, A. M., J. Am. Chem. SOC.,48, 690 (1926). (3) Wagner, G. H., Bailey, G. C., and Eversole, W. G., IND. ENCI. CHEM.,ANAL.ED.,13, 658 (1941). (4) Wagner, G. H., Bailey, G. C., and Eversole, W. G., J . Phys. Chem., 45, 1388 (1941).
Heats of Wetting for the Evaluation of Gas-Adsorbing Coconut Carbons HOSMER W. STONE AND RAYiMOND 0. CLINTON, University of California, Los Angeles, Calif.
H E use of heat of wetting as a means of evaluating gasadsorbing carbons has been suggested by several authors in recent years (1, Y),and restricted semianalytical methods have been developed by Macy (8) and Burstin and Winkler (4). This method has the distinct advantages, in comparison with other methods, of extreme rapidity and satisfactory accuracy, and by suitable correlation affords data as to other analytical characteristics of the carbon, as is shown below. A large number of methods for evaluating the gas-adsorbing qualities of carbons have appeared in the literature, most of which have been critically discussed by Chaney, Ray, and St. John (b), and by Berl and Herbert ( 3 ) . The majority of these methods are time-consuming and not overly indicative of the value of a carbon for a particular purpose. As has been pointed out by Chaney, Ray, and St. John ( 5 ) ,the most reliable analytical data pertaining to gas-adsorbing carbons are : 1. SERVICE TIME. Time required for the break-through of a vapor under standard conditions. VALUE. The weight of a vapor adsorbed by a 2. ADSORPTION
carbon under conditions of saturation. 3. RETENTIVITY VALUE. The weight of vapor held by a carbon through secondary valence bonds, under standard condi-
tions.
All these data are more or less arbitrary, yet under suitable conditions they afford the most reliable evidence obtainable a t present as to the gas-adsorbing characteristics of a given carbon. The chief criticism of these methods arises from the lack of definition of terms and of precise descriptions of the techniques which have appeared in the literature. It is only by a rigorous standardization of the analytical procedures that quantitative results can be achieved and duplicated. Toward this end, the authors have sought to standardize their methods as thoroughly as possible.
Macy (8) and later Bell and Philip (2) have shown the existence of a quantitative relation between the heat of wetting and the retentivity value of a specific carbon, but the possi‘t, bility of a quantitative relation between adsorption value and heat of wetting has been a matter of considerable dispute. Burstin and Winkler (4)first suggested such a relation, but, unfortunately, their data were insufficient to warrant such a conclusion. Macy (8), as the result of the analysis of a large number of different carbons, found several rather large discrepancies, and, therefore, stated that no correlation between adsorption value and heat of wetting could exist. Other authors have either been undecided ( 2 ) or have confirmed in a degree (3)the data of Burstin and Winkler. Yo correlation of heat of wetting with the service time of a carbon has been found in the literature. I n view of evidence which has been obtained by the authors, a definite relation between the heat of wetting of a carbon as determined under certain conditions as stated and the other analytical characteristics listed is postulated. This relation, however, is limited to specific types of carbon-that is, to carbons which have a common raw material source. While the present paper concerns itself with gas-adsorbing carbons derived from coconut, since this is the most efficient and most common type in use at present, nevertheless the procedure, with suitable modification, should be applicable to other common-source carbons. The question of the effect of varying the method of activation on the correlation of these characteristics of commonsource material carbons has been raised. Of the thirteen coconut carbons used in this work, five are known to have been activated by the same method. The other eight carbons were commercial carbons from various sources, whose method of activation was not known. Two of these latter carbons were prepared by high-temperature steam treatment
I
,
Vol. 14, No. 2
INDUSTRIAL AND ENGINEERING CHEMISTRY
132
Rc = retentivity value for carbon tetrachloride in mg. retained per gram of carbon Q z = heat of wetting in m-xylene, in calories evolved per gram of carbon
Throughout this article, except as otherwise indicated, "chloropicrin service time" refers to the chloropicrin service time value as calculated from the carbon tetrachloride service time through the use of an integer factor. This factor has appeared in the literature with a value of 3.0 (e), but as the authors have used a 15.0-cm. column length of carbon in the service time determinations instead of 10.0 cm., the value of 2.0 has been used in the present work.
, I
I
I
I
~1
I
I
The results of the analyses are presented graphically in Figures 1, 2, and 3, and summarized in Table I.
I
i
Experimental SERVICE TIMEAND ADSORPTION VALUE. For the determination of these values, modifications of the methods of Chaney and of Fieldner (6, 8) are used. The apparatus is shown in Figure 4. The carbon tetrachloride is dried and redistilled before use, reserving the fraction boiling at 76-77.5" C.
FIGURE1. RELATION BETWEEN ADSORPTIONVALUE FOR CARBON TETRACHLORIDE AND HEATOF IVETTING IN m-XYLENE of a spent carbon which had been used to remove hydrocarbons from carbon dioxide. The temperatures of stripping were 425" and 535" C. The stripped carbons possessed a higher degree of activation than the original source carbon. Pearce and McKinley (9) found the relation between.adsorption value and the heat of adsorption (vapor phase) to be given by the expression: Q = mX"
vhere Q is the heat evolved in calories per gram of carbon, X is the milliliters of a given vapor adsorbed, and m and n are constants, depending upon the vapor used. Burstin and Winkler (4) found the following equation: Ab
=
19 Qa
where A,, is the adsorption value in milligrams of benzene adsorbed per gram of carbon, and Qbis the heat of wetting for benzene in calories per gram of carbon. Both equations were derived from the analysis of a very limited number of carbons, and, therefore, their general applicability is slight. Previous workers (9, 4, 8) have determined the heat of wetting of carbon by benzene, and compared these data with the adsorption value for benzene. The present writers have found the use of m-xylene for the determination of heats of wetting to offer several advantages-notably, a decreased volatility, a constant specific heat over the temperature range used, and approximately 50 per cent higher values for the heat of wetting. Because of the difficulties inherent in the use of the accelerated chloropicrin service time method, it was found desirable to use the accelerated carbon tetrachloride service time method (6) with suitable modifications. The use of this latter method makes possible a saving of time, since the service time and adsorption value determinations may be made successively. Results of the analysis of thirteen samples of coconut carbons have led to the following empirical relations:
In the above equations: A:a = adsorption value for carbon tetrachloride (4) in mg. adsorbed per gram of carbon at 25' C. S,'" and 82' = service times for chloropicrin in minutes at 20" and 25' C., respctively
CARBONS TABLE I. AXALYSESOF COCONUT AD
A:'
Re
s:
Sa"
5
268 52s 38.5 33.4 589 42.6 37.0 300 252 497 36.8 31.6 41.0 291 571 35.8 605 43.2 37.4 302 736 51.0 362 44.0 930 61.2 421 53.0 562 40.4 293 35.0 214 410 26.5 30.4 42.0 29 1 36.4 586 363 52.6 45.4 758 405 58.0 50.2 872 454 64.2 55.4 1004 AD a p p a r e n t density of 8- t o 14-mesh carbon i n grams per for Qi represent average of from 5 t o 10 determinations, with deviation from t h e mean of 0.2 calorie.
-
RETENTIVITY VALUE
R,
Qr
28.6 31.7 27.1 30.7 32.3 38.1 45.2 30.3 22.7 31.5 38.9
43.1 47.5
ml. Values a maximum
AT 2 5 0 ~ .
2 . RELATIONBETWEEN RETENTIVITY VALUEFOR CARBONTETRACHLORIDE AND HEATOF
FIQURE
WETTING IN m-XYLENE
133
ANALYTICAL EDITION
February 15, 1942
as the bieakdoivn point, and the elapsed time recorded as "accelerated carbon tetrachloride service time". The adsorption tube-to-burner connection is then broken and the efRuent carbon tetrachloride-air mixtures vented to a hood. Passage of the carbon tetrachlorideair mixture is continued, 111th weighings of the adsorption tube each half hour, until no further gain in weight is evident (3 to 5 hours). The gain of carbon tetrachloride in milligrams per gram of carbon is recorded as the adsorption value. The "accelerated carbon tetrachloride service time" as determined above is multiplied by 2.0 ( 6 ) to convert to the approximate corresponding "chloropicrin service time". Precautions must be taken to ensure a constant 0" C. temperature of the carbon tetrachloride in the cooling bath, as well as a constant air-flow rate throughout the determination, since small variations in these factors will introduce relatively large errors.
RETEXTIVITY VALUES. These values are determined by a modification of Uacy's method (8).
SERVICE TIME FOR CHLOROPICRIN
FIGURE3. REIATIONBETWEES SERVICETIMESAT 20° 2 5 O AXD HEATOF WETTISGIIGWZ-XYLESE
ASD
A composite sample of the carbon is screened to 8- to 14-mesh on standard screens, dried 4 hours at 140' t o 150" C., and cooled in a desiccator. A volume of the dried carbon (about 25 ml.), sufficient to give a tamped column length of 15.0 cm. in the adsorption tube, is transferred to the adsorption tube and the tube and content's are weighed to 1mg. The carbon is then tamped by tapping lightly on a level wooden surface until no further diminution in volume is noted, a t which point the column of carbon in the adsorption tube should measure etactly 15.0 cm. in length. This column length of 15.0 cm. should be reproducible upon inverting the adsorption tube and retamping the carbon. A simple method of achieving this result involves the use of a flatbottomed test tube 1.3 cm. in inside diameter, and calibrated a t 15.0 cm., for the measurement of the carbon before insertion into the adsorption tube. The adsorption tube is plped in position in the thermostat, which is controlled a t 25.0 . Sufficient time is then allowed (about 30 minutes) for the carbon to reach temperature equilibrium with the thermostat. An air-flow rate of 650 ml. per minute (490 ml. per minute, per sq. em. of cross-sectional area of the carbon) is established through the apparatus, with the three-way stopcock turned so as to vent the carbon tetrachloride-saturated air to a hood. The stopcock is then turned to allow the carbon tetrachloride-air mixture to pass through the adsorption tube, the initial starting time is noted, and any necessary flow-rate adjustments are quickly made. The first appearance of a greenish tinge in the burner flame (which is adjusted to a height of about 7.5 em., 3 inches) is taken
A 3- to 4-gram sample of the saturated carbon from the adsorption value determination is transferred quickly from the adsorption tube to a glass-stoppered weighing bottle and weighed. The glass stopper is removed and the weighing bottle is placed in a vacuum desiccator lubricated with de Khotinsky cement which is in turn placed in an oven maintained at 90" to 100" C. The desiccator is connected to a vacuum pump and the carbons are evacuated a t 2-mm. pressure, with periodical weighings, until the weight loss amounts to not more than 15 mg. over a period of 0.5 hour. The retentivity value is determined by graphing the percentage weight of vapor retained against time, and extending the portion of the curve representing low rate of vapor loss, which is approximately a straight line, back to the retentivity axis, as is shown in Figure 5 ,
The retentivity values may also be determined by evacuation a t lower temperatures if wished, with achievement of comparable results. Lowering the temperature greatly extends the amount of time required to complete a determination, however-for example, evacuation at 25 requires 90 to 100 hours for the attainment of retentivity values, while a t 100" 3 to 5 hours are required, as shown in Figure 5 . O
HEATSOF WETTING. A cylindrical silvered Dewar flask, 10 em. high by 3.0 cm. in inside diameter, is used as a calorimeter. A 15' to 45" thermometer, graduated in tenths or twentieths of a degree is supported above the Dewar flask in such a manner that the thermometer bulb is about 5 mm. above the bottom of the flask. The thermometer is calibrated for 40-mm. immersion. The heat capacity of the assembly is determined by the heat of neutralization of sodium hydroxide and hydrochloric acid using the value H N e 13,895 calories at 20" (IO). The authors' calorimeter, of the type outlined, had a heat capacity of 5.0 calories per de.f?i.OO-ml. portion of dry m-xylene (boiling point 137-9") is pipetted into the calorimeter, and the temperature of the cal-
n
FILLING TUBE
TO VENT
k
AIR
CARBON COPPER COIL
IO' LONG 1/4'
TUBING
J THERM0 STAT
FIGURE 4. APPARATUS FOR DETERMIXATION OF ADSORPTIONVALUES AND SERVICE TIMES
WIRE
134
INDUSTRIAL AND ENGINEERING CHEMISTRY 0 0
0
z n
W
E
sw
a c
u
W
s
m> a
2
s c? IW
FIGGRE5. GRAPHICAL hlETHOD F O R DETERRIINATION O F RETENTIVITIES orimeter and contents is adjusted t o a predetermined value, identical with that of the carbon sample t o be tested, by means of small test tubes containing hot or cold water. This temperature is then recorded to within 0.02' as "initial xylene temperature". A sample of the carbon t o be tested (previously screened t o 8- to lbmesh and dried 8 hours at 140') is weighed t o the nearest milligram and added rapidly t o the calorimeter. The contents of the calorimeter are mixed by a gentle rotation of the thermometer until a maximum temperature is reached. This temperature is read t o within 0.02" and recorded as "final temperature". Unused m-xylene may be recovered by transferring the contents of the calorimeter, at the conclusion of the run, t o a No. 5 Whatman filter. The calorimeter should be wiped dry with a clean, lintless cloth between each run. The temperature of the carbon should be known with a n accuracy of 0.3" C. prior to its addition to the calorimeter. If it differs from the initial xylene temperature by more than 0.3", the calculation of the initial mixing temperature of the carbon-xylene mixture from the known initial xylene temperature and the temperature of the carbon is then made, and this value subtracted from the final temperature and recorded as T,, the temperature rise. From the known heat capacity of the calorimeter, H , and the weight of carbon used, m, the corresponding heat of wetting, Q5,of the carbon for m-xylene in calories per gram of carbon is given by means of the formula: Qz =
2
(H
+ 8.32 + 0.17 m) calories per gram of carbon
The figure 8.32 is derived from the weight and specific heat of the xylene used, and 0.17 is the specific heat of carbon. This equation holds true within the temperature range of 18" to 35" C., since within this range variations in the specific heat and density of xylene are neglible. The weight of dry carbon used in the above determinations is adjusted to give a temperature rise of from 3" to 4". The room used for conducting the tests should be as free from drafts and rapid temperature changes as is possible.
Discussion The authors have found that the proposed modified methods for the analysis of a gas-adsorbing carbon give more accurate and reproducible results than do the standard procedures. I n the determination of adsorption values and service
Vol. 14, No. 2
times, the use of fritted-glass bubbler tubes ensures saturation equilibrium between the air and the carbon tetrachloride. Similarly, by circulating the air-carbon tetrachloride mixture through a copper coil immersed in the thermostat prior to carbon exposure, and by the use of a thermostated adsorption tube, discrepancies due to heat effects are eliminated. The modified adsorption tube permits thermostating during a determination, and by means of the stopcocks prevents vapor loss during a weighing. Further, by making use of a longer carbon column for service time determinations (15.0 cm. instead of 10.0 cm.), the observed service time is lengthened and the effect of experimental error thus decreased. The method of determining retentivity values has been modified by the use of weighing bottles and a vacuum desiccator in the procedure, to permit more ease of manipulation and the simultaneous analysis of a large number of samples. Especially notable is the large increase in accuracy and reproducibility in the heat of wetting determinations, through the use of an improved calorimeter and the substitution of m-xylene for benzene as the wetting liquid. Heats of wetting obtained by the proposed method average about one and one-half times as large as those obtained by the standard Chemical Warfare Service method, on the same carbon. The comparison of these m-xylene heats of wetting with the corresponding service times, retentivity values, and adsorption values, as derived from the proposed procedures, leads to straight-line or smooth-curve functions. This statement, however, does not hold true for the standard methods of analysis when compared with the heats of wetting as determined according to C. TV. S. specifications. Using the modified procedures, duplication of results can be achieved in the determination of retentivity and adsorption values to within 1 0 . 5 per cent. Constancy of service time values is dependent not only upon the temperature of the carbon tetrachloride saturation tower, and of the thermostat, but upon gas-flow rate per unit cross-sectional area of carbon and other factors such as the packing of the carbon and the observance of the first trace of carbon tetrachloride break-through. For this reason, some slight skill is required to duplicate results to within *0.25 minute. The heats of wetting are readily duplicable to within 0.3 calorie. I n accord with the proposed method of analysis for calculating service times, adsorption values, and retentivity values from the corresponding heats of wetting in m-xylene, the average mean deviation in the heats of wetting-namely, 0.2 caloriewill lead to a maximum error in the calculated values of the other analytical characteristics of *0.25-minute service time, and less than *5 mg. in adsorption and retentivity values. It is therefore apparent that the heat of wetting not only is more rapid and requires less skill than other analytical procedures for the evaluation of a carbon, but also yields results of equal accuracy.
Summary An improved method has been developed for evaluating gas-adsorbing coconut carbons by means of the relation between the heat of wetting and other analytical factors. Equations enable calculation of service time, retentivity values, and adsorption values from the heats of wetting of coconut carbons. Modified methods for the determination of service times, adsorption values, retentivity values, and heats of wetting are given. Advantages of the heat of wetting method are its rapidity, its small requirements of equipment, and its quantitative characterization of a coconut carbon.
Acknowledgment The authors are indebted to A. B. Ray of the Carbide and Carbon Chemicals Corporation for supplying six of the car-
February 15, 1942
ANALYTICAL EDITION
bons used in this work, and for supplying certain data regarding these carbons.
Literature Cited K*, and Berl* E*, z' physik' Chem', 122, 81
( l ) Andress, (Igz6); Z . angew. Chem., 35, 722 (1922). (2) Bell, S. H., and Philip, J. C., J . Chem. SOC.,1934, 1164. (3) Berl, E., and Herbert, W., Z . angew. Chem., 43, 904 (1930). (4) Burstin, H., and Winkler, J., Przemysl Chem., 13, 114 (1929).
135
(5)
Chaney, N. K., Ray, A. B., and St. John, d.,IND.EXG.CHEM.,
(6)
Fieldner, A. C., Oberfell, G. G., Teague, 12. C., and Lawrence,
15, 1244 (1923).
J. N., Ihid.. 11. 524 11919). (7) Herbst, H.. kolloid 2.,'38, 314 (1926). (8) Macy, R., J . P h y s . Chem., 35, 1397 (1931). (9) Pearce, J. N., and McKinley, L., Ihid.,32, 360 (1925). and Rowe,-4.W., J , 4 1 n , Chem, so,., 44, 684 (10) Richards,T.w-., (1922). PRESENTED before the Division of d n a l v t i c a l a n d Micro Chemistrv a t t h e 102nd Meeting of t h e AXERICAN CHEMICAL SOCIETY, .itlantic City, N J.
Turbidimetric Determination of Polymerized Hydrocarbons in Solution BASSETT FERGUSON, JR., AND MARK D. SNYDER, Ugite Sales Corporation, Chester, Penna.
I
X DEALING with unsaturated hydrocarbons subject to polymerization it is frequently necessary to determine the polymer content in a solution of the monomer. As the polymer is usually relatively high boiling, the monomer may be flash-distilled under vacuum and the residue weighed and considered polymer. This method is more or less inaccurate, especially for small quantities of polymer. Some users of these materials have empirical specifications for maximum polymer content based on the insolubility of the polymer in methanol or petroleum naphtha. If a given dilution of the sample with one of these solvents does not produce a distinct cloud of precipitate, the polymer content may be assumed to be below the predetermined value. No mention has been found in the literature, however, of extending this principle to give an accurate analytical method. To obtain a rapid and accurate analytical method for the analysis of polystyrene in solution of styrene and other aromatics, the dilutions for maximum sensitivity of the sample with methanol and petroleum naphtha were determined. It was found that two parts of absolute methanol added to one part of sample will give a distinct cloud when 0.0005 per
cent of polymer is present in the original sample, and nine parts of Atlantic Refining Co.'s S o . 44 naphtha added to one part of sample will give a distinct cloud when 0.005per cent of polymer is present. Furthermore, at these dilutions the effect on the polymer solubility of small changes in temperature is small. Analyses are normally performed at a room temperature of 20" to 25" C., but it' was found that differences in results are negligible betn-een 15" and 30" C. Using these optimum dilutions, a set of standards \vas made up covering t'he range of 0.0005 to 0.3 per cent polymer direct'ly. Higher polymer contents could be measured by first diluting the sample 10 to 1 or 100 to 1 with xylene. As had been feared, hon-ever, the standards containing the higher polymer concentrations deteriorated in a fen- weeks' time, owing to the settling of the precipit'ated polymer. Some means of measuring the turbidity vias then sought. Measurement of the height of a column of the liquid through which fine print could be read (as used in the kauri-butanol test) was tried, but great differences were found between different operators' ability to distinguish the fine print through the turbid liquid. A Burgess-Parr sulfur turbidimeter (Arthur H. Thomas Co. catalog No. 9334) proved to be a satisfactory instrument for the measurement. A column of liquid is placed above an incandescent filament of standard (adjustable) intensity and a movable tube dipping into the liquid indicates the point of extinction of the filament by the turbid liquid. Figure 1 shows per cent of polymer corresponding t o turbidimeter readings for 2 t o 1 methanol dilution and 9 to 1 S o . 44 naphtha dilution. For low polymer concentrations, 200 ml. of absolute methanol are added t o 100 ml. of sample in a stoppered cylinder and mixed by inverting several times. After 15 minutes the mixture is poured into the turbidimeter and a reading is taken using 0.5 or 1 volt on the filament, depending on the turbidity. For somewhat higher polymer concentrations, 180 ml. of S o . 44 naphtha are added to 20 ml. of sample and the same procedure is followed. For polymer concentrations above the scale, 10 to 1 or 100 to 1 dilutions of the original sample with toluene or xylene may be used.
The turbidity increases during the first few minutes after adding the methanol or naphtha, but approaches a maximum after approximately 10 minutes. The 15-minute period was chosen so that time differences of a minute or two in making readings make no appreciable difference on the result. A number of other polymers are known to be precipitated from a solution of the monomer by methanol and certain other solvents and it is believed that this method could be readily adapted to polymer analysis in such cases.