02 1
Ind. Eng. Chem. Prod. Res. Dev. lg82, 21, 621-626
Prediction of Solvent Effects in Aqueous-Organic Solvent Delignification Steven M. Hansen' and Gary C. April' Chemical and Metallurglcal Engineering Department, The University of Alabama, Un/versity, Alabama 35486
Bulk deiignification results for organic solvent delignification were shown to occur In two steps which can be described by first-order kinetics. Isokinetic relationships determined from these rate constants show that the mechanisms of bulk delignification do not change in various solvent systems. Based on these isokinetic relationships, first step bulk deligniflcation results in new solvent systems were predicted within 6 % of experimental results and second step predicted results were within 12% of those experimentally determined.
Organosolv Delignification The recent world petroleum shortage has led to a renewed interest in alternative sources of chemical feedstocks and fuel. The biomass conversion of lignocellulosic material by organosolv treatment is currently receiving considerable attention. In this work, aqueous-organic solvent treatment is used to remove hemicellulose and lignin from Southern Yellow Pine to produce a cellulosic product suitable for pulpstock or as a source of chemical feedstock. The hemicellulose hydrolysis products and lignin can be separated and converted into feedstock chemicals or fuels by various schemes. In the early 1930's Aronovsky and Gortner (1936) compared the effects of different alcohols on delignification of Aspen wood. They found that residual pulp decreased with increasing solvent molecular weight and at the same molecular weight, residual pulp increased as the straighbchain carbon length decreased. It was determined that the alcohol should contain at least four carbons in order to produce a well-pulped residue and that 1-butanol was the most effective alcohol in removing lignin from wood. Kleinert (1975) investigated the organosolv pulping of spruce and poplar with ethanol and found that delignification occurred in two distinct stages which could be described by first-order kinetics. It was determined that delignification was considerably faster in the first portion of the cook and similar results have been widely reported in the literature (April et al., 1979; Lora and Wayman, 1973; Springer and Zoch, 1966). Bulk delignification was determined to be a composite phenomenon involving the breakdown of the lignin macromolecule as well as solubilization of the breakdown products. Previous work (Hansen and April, 1980; Lora and Wayman, 1978; Springer and Zock, 1966) has shown delignification to be a complex problem involving the breakdown of the lignin and carbohydrate molecules, solubilization of the breakdown products, and repolymerization of the breakdown products. Although it has been shown that the nature of the solvent is an important delignification parameter, adequate modeling to determine the effect of different solvents has not been performed. Aqueous solutions of ethanol, 1-butanol, phenol, and butyl Cellosolve (Zethoxybutane) were used as a baseline solvent set. Analysis of the apparent kinetics of the E. I. du Pont de Nemours & Company, Inc., Experimental Station, Wilmington, DE. 0196-4321 1821122 1-0621$01.25/0
baseline set shows that an isokinetic relationship for the removal of lignin and residual pulp exists. Through the use of this relationship, the delignification of Southern Yellow Pine in new solvent systems can be predicted. T h e Effect of Solvent on Reaction Kinetics The study of the nature of an organic solvent medium on the kinetics of a given reaction (or series of reactions) has been investigated by Leffler (1955) and others (Kulkarni et al., 1980; Leffler and Grunwald, 1963; Petersen et ai., 1961) to determine the existence of isokinetic relationships. An isokinetic relationship relating the kinetics in various solvent systems has been proven for over 80 widely diversified organic reactions (19551, and an analysis of the errors involved in calculating and proving the existence of an isokinetic temperature is given by Petersen et al. (1961). In the isokinetic analysis, the kinetic rate constants are described according to transition state theory as k= exp( exp( -M* (1)
T) m)
h The enthalpy (m*) and entropy (AS*)of activation are considered to be constant over the temperatures used. It was found that moderate changes in solvent functionality often alter the enthalpy and entropy of activation and in a series of related reactions this variation is usually not independent. If the solvent performs closely similar functions in a reaction series it is often found that the activation enthalpy is a linear function of the activation entropy and the relationship is characteristic of that reaction series. Defining the linear relationship as
m* = &*,
+ PAS*
(2)
where Ai%,* is the intercept and /3 (having temperature units) is the slope. The free energy of activation (which determines the rate constant at any temperature) is given as @=
m* - T&*
(3)
and combining eq 2 and 3 gives
AF* = moo* - (T-P)AS*
(4)
At T = P, all the reaction rate constants become the same (AF = constant = m0*) and ,6 is therefore termed the isokinetic temperature. One of the limitations and yet useful features of the isokinetic relationship is that it can only be expected to 0 1982 American Chemical Society
622
Ind. Eng. Chem. Prod. Res. Dev., Vol. 21, No. 4, 1982
Table I.
First Step Rate Constants and Activation Parameters ~~
solvent ethanol 1-hutanol ( 5 0 / 5 0 ) 1-butanol ( 3 0 / 7 0 ) 1-butanol ( 7 0 / 3 0 ) 1-butanol w / H 2 S 0 , butyl Cellosolve phenol
temp, K 448 478 448 478 448 478 448 478 388 418 44 8 44 8 478 44 8 178
residual pulp ________ k , , min ai7*
lignin remaining
-AT*
Southern Yellow Pine 0.00223 1 5 224 37.49 0.00697 0.00283 13 1 4 7 41.28 0.00922 0.00246 1 3 688 40.73 0.00689 0.00404 1 2 132 43.17 0.01039 0.00522 53.55 0.01208 6 031 0.021 0 5 0.00412 11 849 43.81 0.01012 0.00894 8467 49.49 0.02039 ~~
h , . min-' 0.00435 0.01046 0.00479 0.01 1 7 5 0.00 562 0.01311 0.00495 0.01280 0.01 251 0.01951 0.02870 0.00615 0.01 500 0.01749 0.02703
AFT*
-AS*
11 525
44.42
11 8 0 8
43.60
11 134
44.79
1 2 556
41.84
3 976
57.52
11 726
43.29
5 231
55.71
Table 11. Second-Step Rate Constants and Activation Parameters residual pulp solvent ethanol 1-butanol butyl Cellosolve phenol
lignin remaining
temp, K
h,, min-'
AR*
-AS*
h , , min-'
44 8 478 44 8 478 448 478 448 478
0.00020 0.00079 0.00026 0.00063 0.00043 0.00131 0.00036 0.00141
1 6 550
38.99
11 633
49.78
1 7 400
36.03
18 743
33.29
0.00108 0.00221 0.00018 0.00052 0.0004 1 0.00171 0.00311 0.00 54 1
apply to a series of reactions in which the solvent does not change the reaction mechanism. The analysis does not require prior knowledge of the microscopic mechanism (Kulkarni et al., 1980) and any rate constant which accurately describes the data can be used. The existence of an isokinetic relationship is strong evidence favoring a common mechanism for a related series of reactions. Although a variety of reactions are occurring in organic solvent delignification, previous results (Kleinert, 1975; April et al., 1979; Lora and Wayman, 1978) have shown that bulk removal rates for residual pulp and lignin (in various solvents) can be described by first-order rates. Based on these results, it is postulated that the steps in the complex removal mechanisms are not changed by the solvent system. Although the exact reaction mechanism cannot be determined, the isokinetic analysis can be applied using the apparent first-order rates which describe the bulk removal results. Isokinetic relationships have been found for the removal of residual pulp and lignin. The isokinetic relationships determined for the baseline solvents were found to hold for new solvent systems and could, therefore, be used to predict bulk delignification results in new solvent systems.
Experimental Procedures Organosolv delignification experiments were conducted in a 350-mL stirred batch reactor equipped with a mechanical agitator, an electrically heated jacket, and an automatic temperature controller. The reactor was loaded with 10 g of wood meal and 150 mL (50/50 by volume) of solvent mixture. The electrical heater and agitator were turned on and after some warmup period (30-45 min) the system reached equilibrium and was held at the desired temperature for the required treatment time period. After this period the heater was turned off and removed and the bomb was allowed to cool to room temperature. The residual pulp was separated from the solvent by filtration and the volume of recovered solvent was deter-
AI7*
-AS*
9 281
52.21
14 127
44.94
1 3 079
46.65
6 920
55.37
mined. The wood meal was washed in ethanol, dried, and weighed, and the residual pulp was analyzed for lignin remaining by a standard TAPPI method (T3220s-74). The volumetric swelling was determined using blocks of Southern Yellow Pine with dimensions corresponding to the axial, radial, and tangential dimensions of the wood. The volumetric swelling was determined by measuring the block dimensions before and after exposure to the solvent systems (well mixed). A more detailed account of the processes in this work have previously been described (April et al., 1979,1980; Hansen, 1981; Hansen and April, 1980).
Results Delignification Results/Isokinetic Analysis. Delignification experiments were performed in the baseline solvent systems (aqueous ethanol, 1-butanol, phenol, and butyl Cellulose) and experimental data were found to fit two first-order reaction curves as described by Kleinert (1975) and others (April et al., 1979, 1980). Results of the analysis of the first and second step rate constants and activation parameters are shown in Tables I and 11, respectively. A 95% confidence interval of 0.5% was determined for repeated analysis of percent residual pulp and lignin remaining and correlation coefficients of 0.95-1.0 were obtained for the determination of the apparent rate constants. The activation parameters determined were assumed to be constant over the temperature range investigated. A plot of the enthalpy vs. entropy of activation for the first step bulk delignification processes is shown in Figure 1. The figure shows a single isokinetic relationship for removal of residual pulp and lignin. The first-step isokinetic temperature was determined (by linear regression) as 561 f 18 K and was found to be linear over a range of 11500 cal/g-mol. For the uncertainty levels obtained in this work, a linear range of 5200 cal/g-mol was sufficient to show that the observed relationship was not due to
Ind. Eng. Chem. Prod. Res. Dev., Vol. 21, No. 4, 1982 623
-
?roceoo RP
Ai?;,
lam01 363182825
Table 111. Results of Rate Constant Predictions with the Isokinetic Relationship
K
561.Si17.9 561.8117.9
36318t825
LR
B,
175 "C
-
12000
8000
-
-
0
_______
-
m
Residual Pulp (R')
0
Lignin Remaining ( L R )
9 5 % conlidence inlerval
I -50
-
0
I
I
-45
-40
ENTROPY of ACTIVATION, cal,/gmo/e OK
Figure 1. Isokinetic relationship for first-stepdelignification processes.
C
c
,/'.a
.
6OOqF.
0 Residual Pulp
----
, /
I
I -48
(RZI
Lignin Remaining (, R ) 95%conirdence interval
I
I
-40
ENTROPY of ACTIVATION, d , g m o i e
I
I
.32
OK
Figure 2. Isokinetic relationships for second-stepdelignification. errors in determination of the first-step rate constants. The existence of the isokinetic relationship is evidence that the mechanisms of first-step bulk delignification do not change in different solvent systems. These results were expected because of the similarity in the shape of the bulk delignification curves. The single first-step isokinetic relationship shows that lignin and residual pulp are removed by a common mechanism. These results confirm findings that removal of lignin corresponds to the loss of residual Pulp. Figure 2 is a plot of the activation parameters for the second-step removal processes. The figure shows that the second-step results do not fall on a single line and separate isokinetic temperatures were determined for the two removal processes. The isokinetic temperature for the second step removal of residual pulp was determined to be 428 f 77 K and a value of 690 f 52 K was found for lignin removed. Due to large uncertainty levels, the observed ranges for second-step isokinetic relationships were not sufficient to prove that the linearity was independent of experimental error (7100 cal/g-mol observed vs. 12 000 cal/g-mol required). The second-step isokinetic parameters should, therefore, be viewed only as estimates. A summary of the isokinetic parameters is shown as an insert in Figures 1 and 2. With the knowledge of the isokinetic relationships, rate constants at any temperature (in the range of interest) can be predicted from a single rate constant at a known temperature. Thus, batch delignification results could be predicted if the rate constants at a single temperature could be predicted or related to a particular solvent parameter. The isokinetics model can be verified by using isokinetics parameters determined from the baseline solvents to predict rate constants in new solvent systems.
205 "C
process delign k"lrt,l, k'i"q3 ktxqr step process min min min % diff Solvent: THFA/Water . 1 RPa 0.00265 0.00777 0.00770 1.0 1 LR' 0.00425 0.01077 0,01100 2.0 2 RP 0.00051 0.00335 0.00311 7.6 2 LR 0.00099 0.00209 0.00231 9.5 Solvent: Ethylene Glycol/Water 1 RP 0.00333 0.00910 0.00893 1.9 1 LR 0.00493 0.01192 0.01171 1.8 RP 0.00054 0.00394 0.00361 9.1 2 2 LR 0.00066 0.00150 0.00164 8.5 a ?6 RP = percent residual pulp. % LR = percent lignin remaining. % difference = 100 x (lpredicted experimental I)/( experimental).
'
Verification of the Isokinetic Model In order to verify the isokinetic model, additional Southern Yellow Pine batch delignification data were obtained in aqueous ethylene glycol and tetrahydrofurfural alcohol (THFA). Using experimental rate constants at 175 "C (shown in Table 111)and isokinetic parameters determined from the baseline solvent systems (Figures 1 and 2), the activation parameters were determined by simultaneous solution of eq 1 and 2. The values of the activation parameters were then used (in eq 1) to determine the rate constants at 205 "C (see Table 111). Table I11 also shows the experimental delignification results at 205 "C and the percent difference between the predicted and experimental results. For the first-step delignification processes, the predicted difference between the experimental and predicted results was 1-2%. For the second step, the difference was approximately 7-9%. Thus the predicted results are more accurate for the first-step delignification processes (where the isokinetic parameters are accurately known). These results serve to verify the isokinetic model by showing that the isokinetic parameters obtained from the baseline data can be used to predict rate constants in new solvent systems with reasonable accuracy. Relationships between Swelling and Bulk Delignification Results An important parameter for organosolv delignification is the ability of the solvent to swell the wood structure (Stamm, 1935; Schuerch, 1965; Hansen, 1981). When a polar solvent is initially contacted with wood, swelling is seen to take place. One explanation of swelling is that the molecules of the polar solvent are attracted to the dry solid matrix and held by hydrogen bonding forces. The repulsion of adjacent adsorbed polar groups causes straightening of the cellulose chain and swelling of the matrix. Nonpolar ends of adsorbed polar solvents would also result in repulsion of adjacent cellulose chains. Additional swelling of the cellulose matrix could be caused by diffusion of the organic solvent into bound water in the cellulose matrix (causing swelling equal to the volume uptake) or by repulsion forces between nonpolar ends of adsorbed solvents and the cellulose matrix. The swelling of wood can, therefore, be viewed as the driving force for opening of the wood structure and fiber liberation. Figures 3-7 show photomicrographs of raw Southern Yellow Pine and residual pulp after treatment in the baseline solvent systems. The photomicrographs show increasing fiber liberation for the solvents as: ethanol,
024
Ind. Eng. Chm. Rod. Res. Dev.. Vd. 21, No. 4. 1982 ~
II i
-Figure 3. Photomicrograph (X7.5) of raw Southern Yellow Pine.
.
pi(we 6. Photomicrograph (x7.5) of residual pulp after treatment in 1-butanol/water (run 5-81,
I "
1
i
.. " . ., .
. .
. .. .. . . .(i:
., . . Figwe 4. Photomicrograph (x7.5) of residual pulp after treatment in ethauol/water (run S-371.
I-butanol, butyl Celloeolve, and phenol. Tbis order ia the anme for swelling, removal of lignin, and residual wood (Hansen, 1981). These results were expected since a relationship between lignin removal and fiber liberation is basic to all chemical pulping systems. Since bulk delignification was shown to be a fiber liberation process (Hansen, 1981;Rydholm, 1965)and the repulsive forces (which cause swelling) can be viewed as a driving force for fiber liberation, a relationship between bulk deligniication and solvent swelling was anticipated. A plot of bulk delignification results (lignin remaining and residual pulp) vs. volumetric swelling of the wood is shown in Figure 8. This figure shows that residual pulp decreases linearly with volumetric swelling and a curvilinear relationship between lignin remaining and volumetric swelling. These results are consistent with previous work showing that lignin is preferentially removed in solvents which cause extensive fiber liberation. The relationships between volumetric swelling and hulk delignification re-
I
.?.~'' +'>%
-.
A+I
pi(we 6. Photomicrograph (x7.5) of residual pulp after treatment in butyl Cellosolve/water (run S-60).
sults (at a known time and temperature) can be used to predict bulk delignification rate constants based on volumetric swelling. Prediction of Bulk Delignification Results The previously formulated model was used to predict bulk delignification results in aqueous solutions of THFA and ethylene glycol as well as blended solvent systems (Hansen, 1981). Results were predicted at 175,195,and 205 OC and compared to experimental results a t 175 and 205 "C. The predicted and experimentalresidual pulp and lignin remaining results in THFA (volumetric swelling of 8.0%) are compared in Table IV. In aqueous THFA residual pulp results were predicted within 5% of the experimentally determined values. The predicted values for lignin removal were within 2% for the first-step process and 3-12% for the second-step results. The predicted values for the second-step curves reflect the high degree of un-
I d . E%. chem.Rod. Rea. Dev.. Vol. 21. No. 4, 1982 625
Table IV. Comparison of Redicted and Experimental Results exptl results pred results temp, "C time, h %RP' 9iLR 9i RP 9i LR
9i difference 9i LR
% RP
Ethylene Glycol/H,O 175 175 175 205 205
0.50 1.00 2.50 0.60 1.60
95.6 81.9 18.0 72.5 58.4
86.3 74.4 70.1 65.4 59.1
92.04 84.11 82.13 74.95 62.32
86.82 15.31 68.90 65.95 52.10
3.72 3.44 5.30 3.38 6.72
0.60 1.31 1.71 0.85 11.77
THFA/H,O
a 5% RP = Percent residual pulp. (experimental).
- ~
,
9i LR = percent lignin remaining.
.~' y
certainty in the determination of the second-stepisokinetic parameters and are similar to results obtained using experimental data (model verification results). Residual pulp and lignin remaining predictions and experimentalresults in aqueous ethylene glycol (volumetric swelling of 8.4%) are also compared in Table IV. These results are similar to those found for THFA/water. The residual pulp predicted values were within 2% and the lignin remaining results were predicted within 2% for the first-step and between 2 and 12% for the second-step predictions. These results show that thia model can be used to predict bulk delignification reaults in aqueous organic solvent systems. S iresults have also been obtained for blends of several organic solvents (Hansen, 1981).
I I
i
9i difference = 100 x (1,predicted- experimentall)/
I
Figure ?. Photomicrograph ( x 7 . 5 ) of residual pulp after treatment in phenol/water (run 5-44].
Conclusions From this work it was concluded that solvents which cause increased fiber liberation will show an increased removal of lignin and residual pulp. This conclusion has previously been shown for chemical pulping processes. From the existence of isokinetic relationships it is coneluded that changing the solvent systems d m - n o t change the mechanisms for removal of pulp or lignin. The existence of a single isokinetic relationship for the first-step processes shows that the same mechanistic steps control the removal of both lignin and residual pulp components. It w a ~found that the isokinetic relationships determined from the baseline data can be used (with one rate constant a t a known temperature) to predict bulk delignification results in new solvent systems. A relationship beween bulk delignification results and volumetric s w e U i was found. This relationship was used with the isokinetic model and was shown to accurately predict bulk delignification results in new solvent systems at temperatures in the range of interest. Literature Cited
Am. G.
I
11
I
u
I 13
I
I)
SWELLING, v0l.S
Figure 8. Relationship between bulk delignifieation results and volumetric swelling.
m.
C.: lim8" S. M.; 8hamats. R.: J.; B r a a d . W. 0. "Aql*las CTganlc sdwnl DapnlRcatbn Of s"woo4s"; V e m l t d at 89m Nstbml A I M Menling. PMland. OR. 1980. Am. 0. C.;KamI. M. M.: Reddy. J. A,: Bowas. G. H.: H a m . S. M. TAPPI 1979. 62. 83. A m n W y . S. I.: Gortna. R. A. I d . Eng. Umn. j933. 28. 1370. wansen. s. M. m.0. ~)ssatatlan. TM u n l ~ ~ of n Av ~ L U ~wm. . M. 1981. Mnsen. S. M.; Am. G. C. "8IFewJslc&s h a Wood: *quean Organic Ak&?d Treamant": prwsnld at the 2nd Chmlcal Campss of me Nom, A I " Contkmnt. Las V w s . NV. 1980. Kbinert. T. N. TAPPI 1975. 58. 170. K t h r n i . M. G.; & M a r . R. A,; DWaswamy. L. K. Uam. Eng. . % 1980, I. 35. 823. Lanler. J. E.; GNnwald. E. '"Ratesand EqulRwm of Or-nlc R e w " " ; Wilay: New Y a k . 1983. Latter. J. E. J. &g. Uam. 1955. 20. 1202.
Ind. Eng. Chem. Prod. Res. Dev. 1882, 21,626-629
626
Lora, J. H.; Wayman, M. TAPPI 1978, 61, 47. Petersen, R. C.; Markgraf, J. H.:Ross, S.D. J . Am. Chem. SOC.1961, 83, 3819. Rydholm, S. A. "Pulping Process"; Intrscience: New York, 1965. Schuerch, C. Ind. Eng. chem. Prod. Res. Dev. W85, 11, 61. Springer, E. L.; Zoch, L. L. Sven. PapperstMn. 1966, 69, 513.
Stamm, A. J. Ind. Eng. Chem. 1935, 27, 402.
Receiued for review December 21, 1981 Revised manuscript received May 24, 1982 Accepted August 23, 1982
Factors Affecting the Properties of Nickel Sintered Structures for
Vlctor A. Tracey INCO Europe Limited, Birmingham 616 OAJ, United Kingdom
INCO filamentary Nlckel Powder Types 255 and 287 have been sintered into porous structures with porosities of the order of 80%. Sintering time, temperature, and atmosphere have been varied and the porosities and strengths of the structures have been measured. The results have been analyzed by multiple regression and the effects of the parameters on the strength/porosity trends have been indicated. The trends have been related to the structural changes taking place during sinterlng, and they indicate that the optimum conditions are in the region of 12 min at 980 OC (1800 O F ) in wet atmospheres with dew points of 20 O C (68 O F ) and with the hydrogen content as high as possible but compatible with safety of operation.
Introduction The INCO nickel carbonyl process (Mond et al., 1890) has been developed to produce pure nickel powders with a variety of special shapes. Two of these powders, INCO Nickel Powder Type 255 (1981) and INCO Nickel Powder Type 287 (1981) are filamentary in nature, and when sintered they yield structures with porosities of the order of 80%. One of the principal applications of these structures has been for the electrodes of nickel-cadmium batteries where the pores are impregnated with "active mass" (Falk and Salkind, 1969). Greater potential usage is being actively pursued for the positive electrodes of the nickelzinc and nickel-iron batteries for traction applications (Carr et al., 1976; Yao et al., 1981). Although the porosities required can be obtained by sintering using a variety of times and temperatures (above 700 O C ) sufficient strength must be developed to allow handling of the porous material through processing, and the sintered bonds must be adequate to withstand any corrosion that might occur on impregnation and through cyclic stresses developing in service. This paper summarizes the effects of sintering time, temperature, and atmosphere on the porosity and strength developed when sintering the two powder types over their bulk density range, Table I. Materials and Procedure Thirty-seven batches of INCO Nickel Powder Type 255 and fourteen batches of INCO Nickel Powder Type 287 were employed in the investigation. The powders were made into slurries using methyl cellulose as the slurry former and coated onto nickel mesh using a doctor blade unit to control thickness and provide a final dried coating of about 1 mm (Tracey, 1979). The coated meshes were sintered for times up to 15 min, at temperatures between 850 and 1050 "C (1562 to 1922 OF) in atmospheres of nitrogen varying in hydrogen content
Table I. Bulk Density and Equivalent Porosity Ranges of INCO Nickel Powder Types 255 and 287
a
powder type
bulk density, range, g/cm3
calcd theoret equiv porosity range, %
255 287
0.50-0.65 0.75-0.95
94.4-92.7 91.6-89.3
(ASTM Standard B 3 2 9 4 1 ) (Inco, 1975).
up to 75% and with dew points from -40 to +20 O C (-40 to 68 O F ) . The porosity achieved in the porous nickel was calculated and the strength of the sinter was measured in three point bending (ASTM B528-70). For the two series of tests 140 samples were evaluated. The effect of the independent variables on the strength and porosity developed was assessed by multiple regression analysis. The trends in the results have been followed by feeding in basic process data which might be considered as typical from the aspects of process operation and economics, i.e., time of 5 min, temperature of 950 "C (1742 O F ) , and an atmosphere of nitrogen -5% hydrogen at a dew point of 20 "C (68 OF). Results (i) INCO Nickel Powder Type 255, Bulk Density 0.5 to 0.65 g/cm3. The porosity change over the bulk density range was about 3%. The mean porosity was 81-8270 (Figure 1). Strength was improved by about 60% on raising temperature from 850 to 1050 OC (1562 to 1952 O F ) , with the majority of that improvement, about 40%, being achieved between 850 and 950 "C (1562 and 1742 OF) (Figure 2). The strength improvement observed with increasing time fell of above 10 min (Figure 3). Increase in hydrogen content of the atmosphere raised the strength of the sinter by 10-15% (Figure 4). Flammability tests indicated that above 14% hydrogem, the gas mixture would burn freely in air.
0196-4321/82/1221-0626$01.25/00 1982 American Chemical Society
