ENGINEERING AND PROCESS DEVELOPMENT Subscripts 1 refers to concentrated end of lower
(5) Geankoplis, C. J., and T-Iixson, A. N.. 1s~. E m . CHFX.,42, 1 I ? L (1950). (6) H a n d , D. B., J . E'hys. C h e n ~ . 34, . 1961 (1930). (7) .Joiinson, H. F., and Bliss, H .. Tiwis. A m . I n s t . Ciirm. h ' r ~ u r s , , 42, 311 (1946). ( 8 ) Orton, K. J. P., and Jones, L). C . , J . C"hem, Soc., 115, I 104 (1919). (9) Rrhur11, IT, T., and S ~ W C IIT. . , S ~-1. , AVL.C/UL. Boc., 55,1774 ( I 933). ( I O ) I~ol-~insoxi. R. A . , and Selkit,k, R. C., J . Ciaem. Soc., 144, 1460 (1948). oc. C'hem. Ind., 67, 48 (1048). ( I 1) Smith, E. L.. and Page, ,J. E. , Ibid., 67, 110 (1948). (12) Such, d. E., and Tomlinson.
2 refers to dilute end of tower A and B refer to nondistributing components C refers to distributed component IT- refers to aqueous phase 0 refers to over-all conditions literature Cited (I) Archibald, 1%.d., J . Am. c'/I(,?)i, SOC., 54, 3178 (1932). ( 2 ) Colbuni, A. P., and Wclsh. D. G . , T W M . Ant. Inst. Cheni. Engrs., 38, 179 (1942). (3) Garwin, L.. and I I i x m i , A. S . , IND.ENG. C H m f . . , 41, 2298 (1949). (4) Chn-in, L., and Kyltmdm, It. I,.. C h e ? ? ~Eng. Progr., 47, 186 (1951).
D for review September 16, 1932. ACCLIT>;D October 20, I Presented before the Dirision of Industrial and Engineering Chemistry a.t t h e 122nd AIeetinp o f tlir h l 1 ~ 1 t 1 c . C 4 ~m v i c . u , POCIJ:TY. Atlantic City, K , .I.
ow Temperature Rates in a Flui H. NATHAN STONE', JAMES
D. BATCHELORl
AND
H. F. JOHNSTONE
Universify o f Illinois, Urbano, 111.
D
I*;.\IANU for synthesis gascs, sinokeless domestic fuels, and cheap briyuetted fuels has created new interest in low temperature carbonization pvocesscs. Simultaneously, the development of fluid-solid techniques has improvcd the economics of continuouP carbonization of powdered coals. I n this study, the rates of the initial carbonizntion wartions ha,ve been measured in a fiuidizcd bed a t temperaturcs from 770" to 1125' F. Samples of ground coals were injected int,o a preheated bed of silira, fluidized xith prrheatcd nitrogcxn, to determine the kinetics of the isothermal decomposition rcxctions in an inert atmosphere. The primary emphasis of earlier vork on low t,exnperature carbonization has been on the effect of temperature, contact timcl, rate of heating, and preoxidation upon the yields of tar or gas (9.8-IL). Sinnatt (15) and Burgcss and Wierler (3) inadc some measurements on tlic mtes of carbonization rwcations. Fuchs and Sandhoff ( 7 ) analyzed kinetics data froin several sources horn the standpoint of thermodynamics. From the freo energy changes accompanying C-C and C-~-Hbond clcavagcs, they calculated the values for t,he formation of sl)lit products and poPtulated a first-order niechanisni with activat,ion energies in the order of 32,000 calories per gram-mole. The complex nature of coal makes a conventional study of the, kinctics of decomposition difficult. The situation is one in which the exact nature of the reactants is unknown and the products earl he classified, a t beat, as groups of compounds. 'rhc thermal cracking of large polycyclic organic molecules ip generally hi:lieved to be the source of the volatile produch of carbonization. 1nvestiga.tions of the crac,king of individual hydrocarlions h a w s h o r n that such reactions are usually unimolecular. l f several reactions of diff erent rates take place simultaneously, howevcr, the order of the over-all reaction mag be variable, or fractional. If a particular r e a h o n is predominant, an apparent fixed order may result. Since there is no may to tag the molecules appearing in the products, it is not possible to disentangle the scveral simultaneous reactions vhich often yield coinmon products. Auother approach is that of grouping all the reactions into tho single process, 1 2
274
Pi,eient address, Bay State Abrasive Products Co., U'est,boro, hIaw Pi,esent addre%. Pittsbnrgh Consolidation Coal Co.. Library, 1%.
volatile matter (s)
4
volatile matter (g)
(L)
:ind tr:triiig the progreep with time. The yield of gas can be IUCBSurcd as a function of time, or saniples of the solid can he withd r a m st frequent intervals and analyzed t o find the rate of di:crease of volatile matter. Pinre the fluid bed is ideally adtiptod lor solid sampling, the latter method was employed lor 1 hip studl,. The rate can bo represented by
where I' is the volatile matter content at' any time, in pounds per. pound of fixed carbon, and V e is the constant volatile matter content that is u1timat)ely reached after heating a t a given temperature. For the over-all process, n is a function of (V - V,) a t coiistaiii temperature that canbe determined from the experimental resulk. Several proposed theories attribute the control of carbonizatioii reactions to t,he p h y s i d structure of coal. Berkowit,a ( 8 ) maintains that the pore diameters determine the coking and carlionizing properties. The carbonizat'ion rate should thus be R function of the particle diameter. Others have emphasized the differences in the rates of decomposition of the various banded ingredients. These physical effects should be more rcadily observed than the chemical cleavages. Rapid Injection of Powdered Coal into Preheated, Fluidized Bed Is Experimental Procedure
Reactor. The reactor, showi in Ngure 1, was constructed of No. 310 stainless steel welded tubing, 3 inches in diameter. All parts and fittings in the high temperat,ure zone were also made of stainless steel. The reactor was heated externally with six 500rr-att, strip heaters. Four of these were equally spaced about the l o m r section and were controlled automatically. The other tivo were equally spaecd about tho upper section and were nianually controlled. The ends of the thermocouple well and of the solids sampling tube were in the vertical axis of the column. The bed was supported by a stainless st'eel porous plate. The entire column was heavily i n d a t ' e d t o prevent heat, losses.
INDUSTRIAL AND ENGINEERING CHEMISTRY
Vol. 46, No. 2
,
ENGINEERING AND PROCESS DEVELOPMENT
'PRESSURE TAP
.CYCLONE DUST COLLECTOR
'a
TO SOLIO-. SAMPLING TUBE
TO
coal was introduced, the fluidizing gas was cooled and recompressed to 3 pounds per square inch pressure and returned to the cycle. After the coal was introduced, the recycle valve wm closed and pure nitrogen was fed to the system. The off-gam were then sent to a tar tra ping network consisting of three tra s immersed in ice baths. $he first was packed with glass woo1 an{ the other two were simple U-tubes, as shown in Figure 3. These traps collected most of the condensable vapors. The gas samples were collected in evacuated 1-liter flasks fitted with three-way stopcocks a t the top and two-way stopcocks a t the bottom. The solid sampling tube was attached through a three-wv plug cock to a water jacketed pipe to which a 500-ml. filtering flask could be fastened. By ulling a vacuum on the flask, a partially cooled sample could e! withdrawn when the plug cock was turned to the proper position. Before a sample was drawn, nitrogen was passed through the sample line to ensure that the solids withdrawn were representative of the bed a t the fiampling time. Procedure. Figure 3 shows the gas, solid, and liquid flow diagram of the entire operating unit used in the first set of esperiments. The reactor was charged from the top with approximately 13 pounds of silica and then closed. Sampled coal (2 pounds) was sealed in the feed hopper and the cooling water wm adjusted. The entire system was then purged of air with nitro-
I STACK
D E T A I L OF SOLID SAMPLING
TUBE
.
113V
Figure 1 .
NITROGEN INLET
Fluid Bed Reactor
In the first series of experiments (I@, the coal was introduced by a screw, also constructed of stainless steel, and designed for high speed operation. The screw had a pitch of 1 inch and clearance of 1/16 inch. It was operated a t a speed of 210 r.p.m. and the time required to deliver 2 pounds of coal was 17 seconds. The coal hopper and part of the screw housing were water jacketed to prevent premature heating of the coal. The hopper was provided with a nitrogen by-pass line so that the pressure could be maintained slightly above that inside the reactor. In the second series of experiments ( I ) , a nearly instantaneous coal injector was connected to the column in place of the screwfeeder. The injector consisted of a steel tube 2 inches in diameter and 30 inches long, equipped with a piston machined for a loose sliding fit and with a flange to hold a thin frangible disk of aluminum foil between the injector and the fluidized bed. X t r o gen gas pressure, applied behind the piston and frangible disk, was used to rupture the disk and blow the coal charge from the tube into the bed. Figure 2 shows the injector details. BOLT FLANGE
Figure 2.
Detail of Coal injector
The nitrogen used for fluidizing the bed was supplied from a bank of cylinders through a reducing valve, pressure regulator, and flowmeter. It was reheated in an electrical tube heater of 1500-watt capacity. 8uring the warm-up period before the February 1954
'
J Figure 3.
Flow Diagram of Carbonization Apparatus
A = Fluid bed reactor B = Cyclone separator C = Tor traps D = Gas samplers E = Screw conveyor F = Solid sampler G = Thermocouple weU H = Manometer
I = Pressure regulotor J = Nitrogen supply K = Recycle blower 1 = Rotameter M = Gas cooler N = Nitrogen preheater 0 = Coal hopper P = Porous plate
gen. All heaters were set at, full capacity and the recycle system was regulated to the minimum fluidizing velocity. This required a flow of 0.6 standard cubic foot per minute snd gave sufficient turbulence for good heat transfer and uniform teniperat,uw distribution. While the fluidized bed of silica was approaching the control temperature, the gas sampling bottles were evacuated to less than 10 mm. of mercury absolut,e pressure. The tar trap coolers were packed with a mixture of ice and water. When the temperature of the bed reached the initial control temperature, about 50' to 70" F. above the desired reaction temperature, tho recycle flow was switched to a fresh nitrogen flow of 1.3 cubic feet per minute. After about 2 hours, when the temperat,ure controller was functioning smoothly, the screw conveyor was
INDUSTRIAL AND ENGINEERING CHEMISTRY
275
ENGINEERING AND PROCESS DEVELOPMENT soluble in fuming sulfuric acid, oxygen, hydrogen, carbon monoxide, methane, and nitrogen. An est,iniate was made of the total sulfur content of the gases b y burning samples of 500 ml. n i t h 0.1 cubic foot of natural gas in air and absorbing the acidic constituents, other than carbon dioxide, in dilute standard sodium carlwnatc solution containing hi-drogeri peroxidc, which W:LS then titrated to the methyl red end point. Blank Tuns were made for correction for the sulfur content, of the natural gas. Two noncaking roals from Illinois and Cblorado m r c carbonized a t temperatures from 800" to 1125' F., using thc screwfeedcr. A caking coal from the Pitkburgh seam was carbonizrd from 770" to 950" F.,using the gas injector method. The proximate malyses of the coals arc shown in Table I. At the completion of the run, t,he fluidizing gas and heaters were jhut off and the reactor was opened to the stack. Cooling was so slow, howevcr, that the t,emperaturc drop was less than 50" F. in t,he first hour. \\-hen the bed as empt'ied, samples of t#hcremaining material rwre analyzed for the "stcady state" volatilc riiattcr content.
Table
I.
Proximate Analyses of Coals Carbonized -49
Moisture Free,
Receired,
70
Illinois Y o . 2 Seam, Wilmington Mine lioisture 9.9 ... Volatile matter 38.0 '12.2 Fixed carbon 46.7 51.8 Ash 5.4 6.0
__
Total
3Ioiature Ash F'rce, and
70
%
100.0
100.0
i
.
.
44 . B 55.1
... ___ 100.0
Colorado Upper Mesa Verde Seam, Crested Butte Mine hIoisture 3.5 ... ... I'olatile matter 38.0 39.4 41 5 Fixed carbon 53.6 55.5 58 5 A41 4.9 5.1 ,.. Total
__
__
100 0
100.0
100.0
Pittsburgh S e a m , hIontour-10 bline hioi3ture 1.53 ... Volatile matter 40.08 40.7 Fixed carbon 54.96 58.8 3.46 3.5 Ash
Total
-_
__
100 00
100 0
42.2 57.8 .
.
(
.
100 0
Analr-ses made by Analytical Division of Illinois State Geological Survey and lakyoratories of Pittsburgh Consolidation Coal Co.
Results Show Carbonization Is ThreeStage, Partially First-Order Process
Figure -1 s h o m the release of the volatile matter from the 11linois coal in the fluidized bed. Figures 5 arid 6 show thc results for thv Colorado and Pittsburgh coals, respectively. I n all cases, there is a rapid evolution of gas initially, and then the rate drops off sharply to a long period o f nrarly constant evolution. The
c
, o
400
mo
0;
iooo'r. 1200
1800
2000
2400
2800
32%
7 I M E , SECONDS
Figure
276
4,
Devolatilization of Illinois Coal
Figure
5. Devolatilization of Colorado Coal
INDUSTRIAL AND ENGINEERING CHEMISTRY
Vol. 46, No. 2
ENGINEERING AND PROCESS DEVELOPMENT early part of the reaction is of great importance. When the screw feeder was used, the total charging time for the coal sample was 17 seconds and the first solid samale was removed a t 15 seconds. Consequently, coal was still entering the bed and the temperature was falling when the first sample was withdrawn. This left an uncertainty concerning the actual temperature and contact time of the first sample for the Illinois and Colorado coals. The temperature and contact time of the later samples are progressively more certain. Characteristic temperature-time relationships are shown in Figure 7 . To improve the precision of the data in the early period the instantaneous injector was developed and used with the Pittsburgh coal. A recent paper by Shapatina and coworkers (14), which appeared in the Russian literature after the first series of experiments were completed, describes experiments on the devolatilization of powdered coal samples using a technique somewhat similar to that used here. The authors conclude that the evolution of gas a t any temperature, in the range 300" to 600' consists of three stages. The first stage corresponds to the rapid evolution of the loosely bound water, carbon dioxide, and some carbon monoxide. The second stage, corresponding to the evolution of valuable volatile matter, is considerably slower and lasts for 3 to 4 minutes; its rate is first order with respect to the reacting organic mass. The third stage consists of slow residual outgaesing, in which 15 to 20% of the vola1,ile matter is evolved. The same observation can be made on the results obtained for each of the coals used in the tests described in this paper. The rate data do not follow Equation 2 for the entire period for any constant value of 71, even when V , is set equal to zero; but the results do follow a firsborder process for a period of time, using an equilibrium volatile content given by the analysis of tho final residue. At lower temperatures, the first-order reaction extends over a greater time interval, but the fractional approach to completion a t which the first-order reaction no longer holds is roughly the same for all temperatures. For the Colorado coal, first-order equations fit the time ranges between 15 and 75 seconds a t 1000° F. and between 15 and 600 seconds a t 800' F. For the Pittsburgh coal, three stages of first-order processes in the devolatilization are discernible. These are characterized by the three ranges observed by Shapatina for Russian coals, with remarkable agreement in the limits of the volatile content for each temperature. The time intervals, however, are quite different, evidently because of the different methods used for heating. The activation energy calculated for the range from 800' to 1000" F. for the Colorado coal is 34.6 kcal. per gram-mole. This is in good agreement with the estimate of Fuchs and Sandhoff. Values of the activation energy for the range 770" to 950" F. for the Pittsburgh coal are initially lower and decrease from 26.2 to 6.7 kcal. per gram-mole from the first to the third stage of devolatilization, as shown in Table 11. Shapatina found an activation energy of 5.3 kcal. per gram-mole.
c.,
5
I
I
I
I
I
I
I
I
Table 11.
First-Order Reaction Constants for Devolatilization Range Completion,
k,
Temp.,
F.
Time,
%
Sec.-i
see.
Colorado Coal hctivation energy, E (800' t o 1000" F.)= 3 4 . 6 kcal./gram-mole 0.0081 0.0165 0.0187
800
900 1000
15-600 15-120
1545
30-70 45-75
15-75
Pittsburgh Coal
Activation energy, kcitl./gram-mole Zone 1 (770" to 950° F.) = 26 2 Zone 2 (7705 to 950" F.) = 2 3 . 4 Zone 3 (770' t o 950' F.) = 6 . 7 770
kl ka k3 ki ks ka
850 950
= 0.090 = 0.008 0.0012
= = = =
0-50
0-47 47-68 +68 0-61
ki = 0.012 kr, = 0.002 ka = 0.0008
50-300 +300
0-12 12-120
61-84 84
+0-59 59-82 +82
0.14
0.018 0.0015
f120 0-5
5-60
A 60
After the initial stages of devolatilization, there is a long period in which the rate is practically independent of the amount of volatile matter remaining, and some physical chsracteristic of the char, such as pore diameter, controls the rate of gas evolution. The decrease in activation energy with extent of devolatiliaation is consistent with this idea. The point a t which no further devolatilization occurs was reached only with the Colorado coal a t 1000" F., but previous work has indicated that the devolatilization reaches a definite state of completion for each temperature ( 5 , 6 ) . After Evolution of Oxygen and Sulfur Compounds, Hydrocarbons Are Volatilized in Second Stage of Process
The yields and composition of the gases evolved agree with those obtained when other methods of carbonization are used. The composition of the gas changes sharplv in a temperature range characteristic of the coal-e.g., a t 1000" t,o 1100" F. for the Illinois coal, the predominant component of the evolved gas changes from methane to hydrogen. Below this temperature, alkyl free radicals are hydrogenated to saturated paraffins. Typical gas analyses are shown in Table 111. il comparison of the relative rates a t which the several gaseous products were volatilized from the Pittsburgh coal was made by integration of the gas composition curves to find the total yield during the 10-minute period of fluidization. Typical data are shown in Table IV for 900' F. The carbon monoxide and dioxide are formed rapidly near the beginning of the devolatilization,
1
V -770.F. P 850 *F. 0 950%
-
I
I loo I
I
I
I
I
I
I
I
Figure 7. Figure 6.
February 1954
Devolatilization of Pittsburgh Coal
I
200
I 300
I
I
400 x . 3 TIME, SECONDS
I Eo0
I
700
em
Temperatures Recorded in Fluidized Beds for Colorado Coal
INDUSTRIAL AND ENGINEERING CHEMISTRY
277
ENGINEERING AND PROCESS DEVELOPMENT Table 111.
-
Analyses of Gases from Carbonization of Coal in Fluidized Bed
(Colorado Crested Butte coal at BOOo F.) Time in Fluidized Bed. SOC. 20 50 110 230 Oxygen- and nitrogen-free basis. YC Acid gases 28 19 13 11 Unsaturated compounds 12 16 13 tl Carbon inonoxide 15 11 IIydrogen 0 0 0 Methane 45 64 64 74 Sulflira 1.06b 1.05b 0.6412 O;~IVJ
-
:
Includes ai1 acid gases except carbon dioxido. ~ ~it. i ~ ~ ~ / ~ ~
a ~
Table IV.
~
.
Rates of Formation of Gaseous Products
(Coai from Pitt5burgIi s e m i , Montour-10 mine 15 -.
Carbon
3
dioside
Carbon
0
0
2R
88
100”
Hydrogen Methane
G
10 i-i
1.5 25
26 38
45 65
1
10
7
0
CTn3aturatei
IC.!
Itt ROOo
Carbonizing _______ Time, Sec. 30 60 120 210 3G0 480 600 Cumillative of total yield after 10 ruin.----10 91 100a .. .. . 9
52
70
.
..
64 72 69
83 8s 86
,
,
.
1011
ion
100
Reuction complete at. a h o u t 180 see.
while methane, hydrogen, and the unsaturated hydrocarbons a i e produced more uniformly over the entire period. This tigrws with the observations of Shap:ttin:L. The small amount of data obtained on the sulfur content of the gases indicate that clcvolatilizatioii of the sulfur compounds also takes placc more rnpidl!, than the formation of methnne and other hgduocnrbons. Effect of Particle Size on Carbonization Rates. In ordcr l o study the effect of particle sizc on carbonization r a t w i n the, fluidized bed, the Colorado coal Ras recrushed to a finer sizc. The Ratch method was used to determine thc nwqs nietlittn diametcr from screen analyses ( 4 ) . The original coal h i d a nicdian diameter of 309 microns, and the recrushed coal had a medinn diameter of 222 microns. Figure 8 shows t,hc rel:ttivci rates of devolatilization for the two si The initiid rate is rnucli inorc rapid for the sinall parsicles, hiit thr:w is littlc differento i n the rate. :ift,cr thc initial period.
drastically changed. The particles from the dull and bright bands reacted in a different manner. Those from the dull bands w r c porous and spongy, while those from the bright bands first cshibited small bubbles over the entire surface of the particles. At longer cont’act times these bubbles grew and burst, leaving craters in the surface. At highcr temperatures, the disruptions of the part’icles were more pronounced. Some particles appeared if the!- had exploded, leaving hollow ahelh, and others showed Iissurclike cracks. Single particles containing both dull and lxight bands shoived the describcd differences in adjacent bands. -1s noted by Sinnatt (I@, thc p:trticlea from the bright bands smlled more than the dull particles. I n accordance with 13erkoxitz’ idea, the char froni the bright particles should have finer pores. This is consistent with the results obtained, n-hich shon-ed fine bubbles appearing in the bright particles, while thr dull particles v m e porous and spongy in appearance. Conclusions
Coal carbonization reactions in a fluidized bed do not follow a vimplc, fixed order, but the order depends upon the degree of tievolatilization and upon the temperature. I n the initial stage, the oxygen and sulfur compounds are evolved. The reactions are rapid and folloy approximately first-order mechanisms. In the final stage, the devolat~ilizationapproaches a zero order reaction. The initial rate of devolwtiliaation increases as the particle size decreases, but size is riot important after the first 15 seconds of exposure to the high temperature fluidized bed. The banded ingredients of coal carbonize a t different rates. Acknowledgment
A part of this work was supp31 ted by a fellowship of the Pittsbm gh Coilsolidation Coal Co. The Illinois State Geological Jurvey furnished the facilities of its pilot plant laboratory. ‘Phe authors wish to t h m k 0. JY. Ilres and Paul Henlinr for their help nith the analyaes and with the operation of the equipment ufied in the study. Liferafure Cited Ratdielor, J. D., h2.P. thesk in chemical engineerinp. I’niversity of Illinois, 1951. Berkowitz, S . , Fuel, 28, 97 (1949). Burgess, M ,J., and Wheeler, R. V., Ibid., 4, 208 (1926). Da,lla Valle, J. hl., “~’Iicronie~itics,” p. 96, New York, Pitman Publishing Gorp., 1943. Fieldner, A. C., and Davis, J. D.. 1;. S . B w . M i n t s , A%Iloi~ogTapik 5 (1934). Fieldner, A. C., el al., Ibid., Tech. Paper 524 (1932). Fuchs, W., and Sandhoff, A. G., IND. ENG. CHEM.,34, dB7 (1942).
IIolroyd, R., and FTheeler, 11. V., Fuel, 9, 40, 76, 104 (1930). Lowry, €1. H., “Chernist,ry of Coal Gtilieation.” pp. 459-77, New York, John Wiloy S: Sons, 1946. IIorgan, J. J., and Soule, R. P.. C h e m & M e t . h g . , 26, 923, 977, 1025 (1922).
I OQ
I 40
,
80
Parr, 3. W., “Constitution of Coal,” New York, Columbia Gniversity Press, 1926. Parr, S. W,, Proc. diad Intern. Conf. Biticmimus Coal, 1928, I
I20
I
I
I60
200
240
280
340
TIME. SECONDS
Figure 8.
Effect of Particle Size on Devolatilization of Colorado Coal at 1000” F.
Vol. I, p. 54.
Selvig, W.A , Ode, W.H., and Davis, J. D., U. S.Bur. ilIi?les, Tech. Paper 668 (1944).
Shapatina, E. A., Kalyuzhnyi, V. V., and Chukhanov, Z. IC., Doklady A k a d . Nauk S.S.S.R., 72, 869-72 (1950); Chem. A b s t ~ .44, , 10294 (1950).
Sinnatt, F. S., Proc, 2nd Intern. Conf. Bituntinozis Coal, 1928, Rates of Carbonization of Banded Ingredients. Samples of char obtained at various contact times and temperatures were examined microscopically for information regarding the relative rates of decomposition of the banded ingredients. After 15 seconds in the fluidized bed, a t the lower temperatures, the particle,. iiere nonuniform. At the highcr temperatures they n-ere 278
Vol. I,p. 560. Stone, H. S . , Ph.D. thesis in chemical engineering, Univcmity of Illinois, 1950. ACCEPTED October 28, 1053. REOXWEDfor review June 6 , 1953. Presented at the meeting of the Division of Gas and Fuel Chemistry of the AMERICANCHEMICAL SOCIETT-, Pittsburgh, Pa., April 1953.
INDUSTRIAL AND ENGINEERING CHEMISTRY
Vol. 46, No. 2
