I N D U S T R I A L AND ENGINEERING CHEMISTRY
1204
u s c.14 Pd
(11)
where C is an exponential function of the expression ((u-p)d3p/pZ) and may therefore be conveniently represented by plotting on a logarithmic scale, against ((a- p ) d 3 p / p z ) as abscissas. C obviously is numerically defined, in any given set of data, as equal to U M p . I n Figure 3 the data of hlartin13 for crushed quartz grains in turbulent air streams have been plotted in this manner save that the curve has been made smooth and continuous in the transition range. Points G and C'represent the critical points between which he postulates a straight-line law connecting velocity and particle diameter, while C" evidently represents the hypothetical critical point between stream-line and turbulent conditions. Points A-A ' represent the limit cases of the present work, and show its excellent agreement with that cited by Martin. Points B-B' represent approximately the limits and equation for Audibert's experiments, which evidently are in poor agreement. Points D-D' represent the limits of tests performed by hlartin using a vertical current of air in stream-line flow, and indicate that the state of flow of the fluid stream has little effect on the size of particles supported thereby, although this, if generally true, is a rather .surprising conclusion. In actual use, Figure 3 is directly applicable to determining the fluid velocity required to support a given particle or the terminal velocity of the particle falling through the fluid. , The procedure is to evaluate the term ( ( u-p ) d 3 p / p z ) ) determine C from the curve, and solve equation (11). Unfortu-
Vol. 20, No. 11
nately, one type of problem important in practice-viz. calculation of the largest particle suspended in a fluid stream of given velocity-must be solved by indirect approximations unless it is known in advance that the case in question must lie above or below the critical region C-C'. Above this region equation (9) applies, while below point C the curve of Figure 3 is represented by the expression
u =
35.2 (u-
p)d2
P
(12)
The constant of this expression is somewhat lower than that (54.5) calculated from Stokes' law (equation 4). It must, however, be realized that there is still considerable uncertainty attached to any general expressions for the fall of particles through fluids or their support in fluid streams. As discussed in an earlier part of this paper, even for large particles falling under turbulent conditions the turbulence of the fluid may still be a factor of importance. Similarly, the factors governing the suspension in a turbulent fluid stream of small particles, which would fall through a quiescent body of the fluid under stream-line conditions, are still more uncertain. Moreover, the accurate location of the critical region (C-C', Figure 3) and the shape of the curve in this region cannot be considered to be satisfactorily established. It is hoped that these points will be clarified in time. Acknowledgment
The authors wish to express their appreciation to the Combustion Utilities Corporation for permission to publish the results of these tests.
Reduction of Tricalcium Phosphate by Carbon' K. D. Jacob and D. S. Reynolds FERTILIZER A N D FIXEDNITROGEN INVESTIGATIONS, BUREAU OF CHEMISTRY AND SOILS,WASHINGTOS, D. C
for the manufacture N T H E volatilization of phosphoric acid, the primary reaction occurring in the furnace results in the formation of elemental phosphorus and is customarily represented by the equation : Ca3(PO& 3Si02 + 5C = 3CaSiOa 5CO P2 The phosphorus is oxidized either within or outside the furnace proper and is finally recovered as orthophosphoric acid.18 From the standpoint of commercial operation of the process, one of the primary functions of the silica is to combine with the lime to form a liquid slag which can be tapped from the furnace. The investigations of Berthier, ' Hempel,'ONielsen, Mehring, and Joneslg show, however, that silica and ROSS, also plays a definite part in accelerating the reaction and reducing the temperature a t which reduction begins. The last-named investigators have further shown that it is not necessary to form a liquid slag in order to obtain practically complete volatilization of phosphorus from small mixtures of phosphate rock, silica, and coke. Definite information on the factors affecting the reduction by carbon, in the absence of silica, of tricalcium phosphate, which for most purposes may be considered as the primary constituent of phosphate rock, is essential in order to be able to draw reliable conclusions regarding the effect of silica on the reaction. Berthierl was unable to reduce phosphate of lime with carbon alone, and according to ThorpeZ2 tricalcium phosphate is not reduced by carbon in the absence of silica except a t excessively high temperatures, and even
I
+
1 Received
M a y 19,1928.
+
* Numbers in text refer t o bibliography at end of paper.
+
then calcium phosphide is formed instead of free phosphorus. Xelsen16 obtained partial reduction of tricalcium phosphate by heating with carbon a t temperatures above 1400' C. On the other hand, ROSS,Mehring, and Jones'g volatilized 90 per cent of the phosphorus present in small mixtures of equal weights of phosphate rock and coke by heating for 1 hour a t 1300" C. under close temperature control. However, conclusions as to the completeness of the reaction between tricalcium phosphate and carbon alone cannot be drawn from the results of these experiments, because of the presence of appreciable quantities of silica in the phosphate rock and coke. The present paper gives the results of an investigation of the factors affecting the reduction of tricalcium phosphate by carbon under closely controlled conditions in the absence of silica. Materials
TRICALCIUM PHOSPHATE-SWeral samples O f supposedly c. P. tricalcium phosphate were analyzed, but none of them contained the proper Pa05--Ca0 ratio and all were contaminated with varying quantities of impurities. One of these samples, designated as tricalcium phosphate KO.1, was used in some of the experiments after it was first heated at 900950' C. for 3 hours to remove free and combined water and then passed through a 200-mesh sieve. Practically pure tricalcium phosphate was prepared by slowly adding, with constant stirring, a dilute solution of pure phosphoric acid to a water suspension of pure lime and evaporating to dryness on the steam bath. The product, which contained some free lime and dicalcium phosphate, was finely ground and mixed
Sovember, 1928
INDUSTRIAL AND ENGINEERING CHEMISTRY
until uniform composition was assured. It was then heated a t 900-950" C. and passed through a 200-mesh sieve. The product, designated as tricalcium phosphate KO.2, contained 0.26 per cent uncombined calcium oxide, which was approximately equal to the excess quantity of lime originally used in neutralizing the phosphoric acid. The absence of appreciable quantities of pyrophosphate was shown by the fact that reduction by carbon did not occur below 1150" C. while reduction of pure calcium pyrophosphate started a t 1000" C. Hedvall and Heubergerg have shown that pyrophosphatss are converted into orthophosphates by he'ating with metallic oxides. The composition of the two samples was:
1205
the reaction begins. The following procedure was used in all the experiments unless stated otherwise: The boat containing the mixture of phosphate and carbon was placed in the back end of the resistor tube in the position D,and any air present in the tube was displaced by a stream of dry oxygen-free nitrogen introduced a t G. The furnace wa5 heated until the temperature of a small block of graphite at E was about 50" C. higher than the temperature a t which it was desired to run the experiment, and the boat was then pushed into the position E, the back of the boat being in the hottest part of the furnace. I n this way a gradient of temperature was obtained and danger of overheating a t points No 2 THEORETICAL between the ends of the boat was eliminated. The average h'o. 1 Per cent Per cent Per cent temperature of the boat could thus be raised to the desired 4 3 . 9 8 PlOI 45 38 point in 3 minutes, the time of heating being measured from 56.20 CaO 54 36 45 54 79 21 0.27 SOa h-one the end of this 3-minute period. After heating for the de0.42 CI None sired time, the boat was pushed into the comparatively cool PaOrCaO ratio 0.7967 0,8359 0 8447 f r o n t e n d of the resistoi C o m p a r a t i v e experitube a t about the positioii ments with these materials F , and allowed to cool in the Volatilization of phosphorus from mixtures of trishowed that practically the furnace in an atmosphere calcium phosphate and carbon begins at 1150" C., same percentages of phosof nitrogen. The temperaand under favorable conditions the reaction is comphorus were v o l a t i l i z e d ture of the boat while in the plete in 1 hour at 1325" C., or in 10 minutes at 1500" C. when they were heated with position D did not exceed The speed of the reaction at a given temperature decarbon under identical con1050" C.which was 100" C. creases rapidly as the thickness of the reaction mixditions. The higher grade lower than the minimum ture is increased beyond about 1 cm., which appears to material was used except as temperature necessary for be due largely to the low thermal conductivities of trinoted in the tables. r e d u c t i o n of tricalcium calcium phosphate and carbon. Various forms of carCARBON-Acheson phosphate. Volatile reacbon of low ash content are almost equally effective g r a p h i t e powder, lamption products were carried as reducing agents. black, sugar carbon, and a A phosphate more basic than tricalcium phosphate out of the furnace through dense form of carbon made is formed to a certain extent during some stage of the the tube H . from a lampblack base and reaction. Less than 0.2 per cent of the total phosThe temperatures of the designated as carbon flour phorus originally present is converted into calcium front and the back of the were used. The sugar carphosphide at temperatures up to 1500' C. boat were determined by bon was ground in a porceThe reduction of tricalcium phosphate in the presm e a n s of a L e e d s a n d lain-jar pebble mill and the ence of an excess of carbon follows the course of a monoNorthrup optical pyrometer other materials were ground molecular reaction in the temperature range 1250' which was standardized a t in a small iron-cylinder ball to 1400' C. frequent intervals by the mill. The 200-mesh maBureau of Standards. The terials were extracted with average of these two teni1 : 1 hvdrochloric acid on the steam bath until practically free from iron and alumina and, peratures, which did not differ by morethan 34" C. and was after washing and drying, were heated a t 800-900" C. for 45 usually about 20" C., was taken as the temperature of the mixminutes in zbn atmosphere of nitrogen. The purified ma- ture, and was maintained within 10" C. of the desired temperature in the range 1150" to 1300' C., and within *EoC. terials had the following composition: above 1300" C. Satisfactory black body conditions were VOLATILE FIXED MATTER maintained in the furnace, but a small correction7was applied CARBON AT 900' C. HYDROGENASH to the observed temperatures to compensate for absorption Per cent Per cent Per cent Per cent of light by the glass furnace window. Mixtures of uniform composition were prepared for use in the experiments by agitating the desired weights of 200-mesh materials for several hours in a bottle with a number of small The ash was almost entirely silica and the high ash content rubber-covered lead balls. A uniform flow of 1685 cc. oi of the sugar carbon was due to silica derived from the por- dry oxygen-free nitrogen per minute was maintained through celain jar and pebbles during grinding. Carbon flour was the furnace during each experiment except those in which used in the experiments except as noted in the tables. the rate of gas flow was made the variable factor. The residues from the experiments were analyzed for phosphorua Apparatus and Experimental Method pentoxide and calcium oxide, and the percentage of phosvolatilization was calculated from the change The graphite-tube resistance furnace described by Bryan, phorus lost hlehring, and Ross3 was used. A longitudinal cross section in the P205-CaO ratio of the residue as compared with the of the furnace proper, showing an improvement over the orig- ratio for the original tricalcium phosphate. The results inal design, is represented diagrammatically in Figure 1, in the accompanying tables are each the average of a t least which also shows dimension drawings of the graphite resistor two experiments in which the total phosphorus volatilized tube and graphite boat. The improvement consists of a did not vary more than 2.5 per cent. brass rod, A , having a carbon tip, B, and fitting in a stuffing Effect of Carbon from Different Sources box, C. By means of this device it was possible to obtain reliable information on the speed of reduction of tricalcium It, is known that the velocity and the temperature of rephosphate a t any desired temperature above that a t which duction of ferrous oxide16chromium trioxide,*and zinc oxide"
I S D LTSTRIALA Y D EXGINEERING CHEJIISTR Y
1206
T'ol. 20, No. 11
are influenced to a certain extent by the variety of carbon used. ' In order to determine whether this is also the case in the reduction of tricalcium phosphate, experiments were carried oilt with mixtures containing equal weights of tricalcium phosphate and graphite, lampblack, sugar carbon, and carbon flour, respectively. The results, Table I, show that in general sugar carbon and carbon flour were slightly more effective reducing agents than lampblack and graphite. GRAPH/ r E TUBE
GRRPHI 7 .
is 2.367. The mixture giving the lowest volatilization of. phosphorus contained less carbon than was theoretically required on the basis of the above reaction, and the residue was almost snow white, indicating that practically all the carbon had been used up in the reaction. No evidence of fusion was observed in any of the residues obtained in this investigation, although in a number of cases the residues from mixtures heated a t temperatures up t o 1500" C. contained a considerable quantity of BOAT undecomposed tricalcium phosphate. According to Nielsen,I6 tricalcium phosphate melts a t 1550' C., while Dieckmann and Houdremont5 give 1670" C. as the melting point. Lime and carbon, both of which have extremely high melting points, were the only other solid phases present in any of the residues in appreciable quantity.
.
Table 11-Effect of Variations in P,Oa-Carbon Ratio (Mixtures heated 1 hour a t 1400" C.) COMPOSITION OF MIXTUREP206-CARBON PHOSPHORL 5 Cas(P03z Carbon RATIO VOLATILIZED Per cent Grams Grams loa 10 0 457 99 8
14
6 4 3
If3
17 a
1.074 1.841 2 608
95 2 88 6
82 2
CadPOdz No. 1.
Effect of Rate of Preliminary Heating
Figure I-Graphite
Tube Resistance Furnace
Graphite is known t o be less reactive chemically than amorphous forms of carbon, and this is shown to be the case in the experiments carried out at 1325"C., but a t 1400" C. it was practically as good a reducing agent as carbon flour, and slightly better than sugar carbon. The slightly lower volatilization of phosphorus from the mixture containing lampblack was probably due t o the fact that the total thickness of this particular mixture was greater than that of the other mixtures of the same weight, owing to the fluffy nature of the lampblack. The effect of the thickness of the mixture on the speed of reduction is shown in the results obtained on the 10- and 20-gram mixtures heated for 1 hour at 1325" C. and this factor will be discussed further in connection with Tables I V and V. Table I-Effect of Carbon f r o m Different Sources COMPOSITION OF CARBON MIXTURE TEMPERAPHOSPHORUS TURE TIME VOLATILIZED MATERIAL Caa(PO4)z Carbon Grams Grams C. Minutes P e r cent GraphiteQ 1325 5 5 60 95.7 Lampblack 1325 5 97.7 5 60 Sugar carbon" 1325 5 5 99.0 60 Carbon flour 5 1325 99.5 5 60 Graphite 10 10 1325 68.2 60 1325 Sugar carbon 10 10 73.2 60 10 1325 74.0 Carbon flour 10 60 10 1400 Graphite 88.3 10 30 1400 Sugar carbon 10 10 30 86.7 1400 10 10 Carbon flour 90.8 30 a Caa(PO4)a No. 2 used in these experiments. Ca.,(POh No. 1 used in other experiments.
Effect of Variations i n t h e P,O,-Carbon
Ratio
Twenty-gram mixtures containing different quantities of carbon and tricalcium phosphate were heated for 1 hour a t 1400" C. The results, Table 11, show that the percentage of total phosphorus volatilized increased from 82.2, with a PzOscarbon ratio of 2.608, to 99.8 with a ratio of 0.457. The theoretical ratio for the reaction Cas(PO4)2
+ 5C
=
3Ca0
+ 5CO f
PZ
Nielsen states that when mixtures of tricalcium phosphate and carbon were rapidly heated to 1700" C. a larger percentage of phosphorus remained in the residue than when similar mixtures were slowly heated to the same temperature. Ross, Mehring, and J o n e P have calculated a theoretical equilibrium temperature of 1690" C. for the reaction Caa(PO4)z
+ 8C = Ca3P2 + 8CO
under a pressure of one atmosphere, which indicates that the lower volatilization of phosphorus obtained by Nielsen from rapidly heated mixtures was due to the formation of a larger quantity of calcium phosphide than when the mixtures were heated slowly. I n the latter case a large percentage of phosphorus would be volatilized before the temperature necessary for the formation of calcium phosphide was reached. The effect of the rate of preliminary heating was determined on mixtures of equal weights of tricalcium phosphate and carbon heated for 1 hour a t final temperatures of 1200" to 1300" C. I n the experiments on the effect of slow preliminary heating the boat containing the mixture was placed directly in the hot zone of the furnace and heated rapidly to 1000"C. Heating of the furnace was then regulated so that the mixture was brought up to the desired temperature in about 30 minutes. The procedure previously described was used in the experiments on the effectof rapid preliminary heating. Table I11 shows that a larger percentage of the total phosphorus was volatilized when the mixtures were heated rapidly to the desired temperature than when they were heated slowly. If the reaction does not proceed strictly according to the equation Cas(PO&
+
5C = 3Ca0
+
5CO
+
PZ
but involves the formation of a more basic calcium phosphate, which is more difficult to reduce than tricalcium phosphate, then it might be expected that better results would be obtained by rapid than by slow preliminary heating. On the other hand, if the reaction does not involve the formation of a compound more basic than tricalcium phosphate, then the percentage of total calcium oxide present as free calcium
0
November, 1928
IA'D UXTRIAL A N D ENGINEERING CHEMISTRY
oxide in the residue should be the same as the percentage of the total phosphorus volatilized. Table 111-Effect of R a t e of Preliminary Heating (Mixtures of 5 grams Caa(P0a)n and 5 grams carbon, heated 1 hour) RATIOOF FREE RATEOF CaO FORMED TO TEMPRELIMINARY PHOSPHORUSTOTAL CaO PHOSPHORUS PERATURE HEATING VOLATILIZED A S FREE CaO VOLATILIZED 0 c. Per cent Per cent 0.245 3.9 Slow 15.9 1200 11.8 0,509 23.2 Rapid 1200 0.818 4 4 . 4 3 6 . 3 Slow 1250 0.863 43.6 Rapid 50.5 1260 0,976 8 2 . 6 Slow 8 4 . 6 1300 0.975 84.7 86.9 Rapid 1300
The residues from these experiments were ignited at 900" C. t o burn off excess carbon and were analyzed for free calcium oxide by the method developed by Lerch and Bogue" for the determination of uncombined lime in Portland cement. The results given in Table I11 have been corrected for the small quantity of uncombined calcium oxide originally present in the calcium phosphate. In all the residues the percentages of total lime present in the form of uncombined lime were less than would be present on the basis of the non-formation of a compound more basic than tricalcium phosphate, and the residues from the mixtures subjected to slow preliminary heating with final temperatures of 1200" and 1250" C. contained proportionately less free lime than those from mixtures subjected to rapid preliminary heating a t the same final temperatures. These results indicate that a compound more basic than tricalcium phosphate is formed to a certain extent during some stage of the reduction. The work of Dieckmann and Houdremont5 indicates that this basic compound is oxyapatite, 3Ca3(P04)2.Ca0,and it is apparently formed by the interaction of tricalcium phosphate and lime rather than by a partial reduction of the tricalcium phosphate itself. Lassieurll has observed the formation of a compound more basic than Ca3(P04)2during the reduction of tricalcium phosphate by hydrogen a t 1300' C. Effect of Thickness of Mixture
The effect of the thickness of the mixture on the percentage of phosphorus volatilized in a given time a t temperatures of 1250" to 1400" C. was determined in a series of experiments. The boats in which the mixtures were heated were all of the same dimensions, but as a result of their circular shape the maximum thickness of the mixtures was not directly proportional to their weight. The results in Table 15' show that the percentage of total phosphorus volatilized in a given time a t a given temperature decreased considerably with increase in thickness of the mixture beyond 1.1em. Table IV-Effect of Thickness of Mixture COMPOSITION OF MIXTURE TEMTHICKNESS OF PHOSPHORUS CaJ(POd2 Carbon PERATURE TIME MIXTURE VOLATILIZED Grams Grams Per cent Grams O c. Minutes Cm. 60 53.0 0.263 2.5 1.5 1250 0.6 1250 60 1.1 50.5 0.501 5 3 60 1250 1.6 25.1 0.498 10 60 99.5 0.956 1325 1.1 5.9 60 74.0 1.422 1325 100 10 1.6 69.5 1.969 1325 2.2 15' 15 60 2.011 1325 3.0 20 20 60 50.7 0,959 1400 1.1 50 5b 30 99.8 1,697 1400 1.6 88.3 10Q 106 30 2.2 15 l5h 1400 64.0 30 1 .904 e Caa(PO4)z No. 1. b Graphite.
Evidence, that the decrease in the percentages of total phosphorus volatilized from the thicker mixtures was a t least partly due to the fact that temperatures in the interior of the charge were lower than were actually measured on the exterior of the boat, was obtained in rough experiments in which mixtures of phosphate and carbon were heated in a mufl.e furnace and the temperatures a t various depths
1207
determined by means of thermocouples placed parallel to the surface of the mixture. At 1000° C. the temperature gradient a t a depth of 2.5 cm. was 20" C. per centimeter. However, there was undoubtedly an appreciable conduction of heat along the thermocouple wires into the interior of the charge, and consequently the actual temperature gradient v a s higher than that observed experimentally. Furthermore, the observed temperature gradient represents equilibrium conditions as far as heat transfer is concerned, so that 15 ith rapidly heated mixtures the gradient is necessarily higher before equilibrium is established. Further evidence on the effect of the thickness of the charge on the completeness of the reaction was obtained in a series of experiments in which the percentages of the total phosphorus volatilized from different levels of the same mixture were determined. I n order to obtain accurate samples for analysis thin slots were made a t the proper intervals in the back of the boats used to hold the mixtures. The residues could then be sharply divided into horizontal sections by the insertion of metal slides after the boat was removed from the furnace. The results in Table V show that in all cases the greatest volatilization of phosphorus occurred in the top layers of the mixtures and the smallest in the center layers. The top layers were heated by direct radiation from the furnace walls and the bottom layers were heated by conduction through two layers of graphite forming the true and false bottoms of the boat. In one experiment the mixture was purposely heated very slowly to the desired temperature in order to obtain as uniform heating as possible throughout the charge. This r e sulted in more uniform volatilization of the phosphorus but there was still an appreciable difference in the percentages of total phosphorus volatilized from the different layers. Table V-Volatilization of Phosphorus a t Different D e p t h s of Mixtures of Tricalcium P h o s p h a t e a n d Carbon
DEPTH
COMPOSITION OF TEM- RATEOF AT WHICH PHOSPHORUS MIXTURE PERA- PRELIMINARY RESIDUEWAS VOLACaa(PO& Carbon TIME TURE HEATING SAMPLED TILIZED Grams Grams Mtnutes C. Cm. Per cent 10 10 15 1400 Rapid 0 to0.73 71.2 0 . 7 3 to 1 . 4 6 38.9 1.46 to2.19 55.4 10 10 60 1300 Rapid 0 to0.73 67.9 10
10
60
1300
Slow
20
20
60
1325
Rapid
01 . 47 63 t o 21 . 14 96 0 to0.73 0 . 7 3 to 1.46 1.46 to 2 . 1 9 0 to0.50 0.50to 1.00 1 . 0 0 to 1 . 5 0 1.50 t o 2 . 0 0 2.00 to2.50 2.50 to 3 . 0 0
45 67 . 15 85.7 72.3 77.7 68.6 50.1 37.2 37.7 39.6 61.6
Effect of Pressure and Rate of Flow of Nitrogen
I t was necessary to maintain a minimum flow of 24.5 cc. of nitrogen per minute through the furnace during the experiments in order to prevent condensation of phosphorus on the furnace window and to permit accurate temperature observations with the optical pyrometer. A series of experiments was carried out to determine the effect of pressure and rate of flow of nitrogen on the percentages of total phosphorus volatilized from 20-gram mixtures of equal weights of phosphate and carbon heated a t 1400" C. for 1.5minutes. Constant pressures greater than atmospheric were obt'ained by passing nitrogen through the furnace against a column of water of the proper height contained in a series of bottles. A small vacuum pump was used for the experiments a t reduced pressures. The results are given in Table VI. In all cases the gas flows were measured directly a t the pressures indicated. With a nitrogen flow of 245 cc. per minute, the percentage of total phosphorus volatilized dropped from 56.7 under a pressure of 366 mm. of mercury to 34.8 under 1240 mm. pres-
INDUSTRIAL A N D ENGINEERING CHEMISTRY
1208
sure. On the other hand, variations in pressure from 656 mm.. to 1122 mm. did not have any considerable effect when the flow of nitrogen was increased to 1685 cc. per minute. The effect of the rate of gas flow would undoubtedly be more pronounced if the gas were passed directly through the mixture rather than over the surface only. Table VI-Effect
of Pressure a n d R a t e of Flow of Nitrogen over
Mixtures (Mixtures of 10 grams Cas(P0a)z and 10 grams carbon heated 15 minutes a t 1400' C ) NITROGBN PHOSPHORUS NITROGEN PHOSPHORUS PRESSURE FLOW VOLATILIZEDPRESSURE FLOW VOLATILIZED M m H g Cc.per mtn. Per cent M m Hg Cc 9er mrn. Per cent 56 7 1240 366 245 245 34 8 514 245 49 9 656 1685 a6 7 45 8 765 762 245 1685 53 2 39 6 1122 899 245 1685 52 6
Effect of Temperature and Time of Heating
Tables VI1 and VI11 show that reduction of tricalcium phosphate by carbon, in the absence of silica, begins a t about 1150' C. but the reaction is comparatively slow a t temperatures below 1300' C. Practically complete volatilization of phosphorus was obtained from mixtures of 5 grams of phosphate and 5 grams of carbon in 1 hour a t 1325' C., and 98.4 per cent of the total phosphorus was volatilized from similar mixtures in 10 minutes a t 1500" C. With mixtures of 10 grams of phosphate and 10 grams of carbon it was necessary to increase the temperature to 1375' C. in order to obtain practically complete volatilization of the phosphorus in 1 hour. The results of these experiments are shown graphically in Figures 2 and 3. Table VII-Effect of Temperature (Mixtures heated for 1 hour) 10 GRAMS Cao(P0a)z 10 GRAMS CARBON 5 GRAXSCaS(POd2 5 GRAMS CARBON Phosphorus TernTemPhosphorus Volatilized perature perature Volatilized Per cent OC O c. Per cent 1150 2.3 11505 4.2 1200 23.2 1200 9.9 1250 50.5 1250 25.1 1300 86.9 1300 53.7 13250 99.5 1350 85.5 137Sa 99.8 C a ~ ( P 0 3 zNo. 1.
+
+
.
Q
Vol. 20, No. 11
to that for a monomolecular reaction because the speed of the reaction is determined by the dissociation of the solid tricalcium phosphate into one solid phase, calcium oxide, and one vapor phase, phosphorus pentoxide. This dissociation is similar to the dissociation of calcium carbonate into calcium oxide and carbon dioxide, a reaction which is known to be monomolecular. Table VIII-Effect
+
of T i m e of Heating at Various Temperatures 10 GRAMSCia(POi)r 10 GRAXS
+
5 GRAMSCaa(P0a)z 5 GRAMSCARBON CARBON TemPhosphorus TemPhosphorus perature Time Volatilized perature Time Volatilized C. Minutes Per cent C. Minutes Per cent 1200 30 8.7 1400 5 29.3 10 43.0 60 19.1 15 55.2 90 21.1 20 66.2 120 29.8 30 86.7 180 47.7 1250 30 24.4 40 96.5 60 50.5 50 89.3 90 67.1 60 99.8 120 76.1 180 86.7 1500 2.5 60.1 5.0 82.7 7.5 93.4 10.0 98.4
Using the results given in Table VIII, the velocity coefficients for the reaction a t 1250"and 1400" C. were calculated from the equation for a monomolecular reaction. The uniformity of the results, Table IX, shows that reduction of tricalcium phosphate by carbon takes place in such a way that the velocity of the reduction corresponds to that for a monomolecular reaction. The velocity of the reaction may be affected to some extent by the rate of diffusion of phosphorus from the mixtures, but the experimental results do not give any definite indications that this is the case. I n each of the experiments the mixture of phosphate and carbon was heated to the desired temperature in 3 minutes under uniformly regulated conditions, and the time of heating a t the particular temperature was then measured from the end of this initial heating period. A number of experiments showed that with final temperatures of 1250" and 1400" C., respectively, a negligible quantity of phosphorus was volatilized from the mixtures during the 3-minute initial period
The minimum temperature (about 1150" C.) a t which volatilization of
phosphorus was obtained in these experiments was 250' C. lower than the temperature (1400' C.) a t which Nielsen16was able to obtain reduction of tricalcium phosphate, and 150" C. lower than that (1300 " C.) specified by Blome.2 We may conceive of the reaction between tricalcium phosphate and carbon as taking place in two stages:
++
ca3(Po4)*$3Ca0 Pdh P206 $- 5C = 5CO Pz
(1) (2)
A l t h o u g h tricalcium phosphate may have very low dissociation pressures a t the temperatures of the present experiments, its dissociation would be accelerated through reaction (2) which undoubtedly is very rapid a t temperatures above 1100" C. If this is true, the velocity of the reaction between solid tricalcium phosphate and carbon will be determined by the velocity of the dissociation of tricalcium phosphate, which is relatively slow, and not by the velocity of the reduction of phosphorus pentoxide by carbon, which is relatively rapid. Therefore, it would be expected that the velocity of the reduction of tricalcium phosphate by carbon would correspond
of heating. No appreciable error was therefore introduced into the calculations by measuring the time of reaction from the end of this initial heating period. Such a procedure was not justified, however, in the experiments a t 1500" C., because the reaction was so rapid that as much as 30 per cent of the total phosphorus was volatilized during the 3-minute period required to heat the mixture to 1500" C. Consequently, it-was not possible to calculate the velocity coefficient for the reaction a t this temperature.
INDUSTRIAL A N D ENGINEERING CHEMISTRY
November, 1928
Table IX-Velocity Coefficients for t h e R e d u c t i o n of Tricalcium P h o s p h a t e by Carbon a t Various Temperatures 10 GRAMS Caa(PO4)n" 4- 10 GRAMS 5 GRAMSCas(PO4)n 5 GRAMS CARBON, HEATEDAT 1400' C. CARBON. HEATEDAT 1250' C.
+
K =
Time Minutes 0 30 60 90
120 180 a
A-A' Grams 2 269 (A) 1 715 1 123 0 747 0 542 0 302 Av.
Sample No.
K =
A
PZOS
I/tlog,A--X
Time A-linules
0' 0 0 0 0 0
0 5 10 15 20
0093 0117 0123 0119 0112 0113
30 40
1.
A PnOa A-X l/tloge C X Grams 4 398 (A) 3.109 0.0687 2.507 0.0560 1.970 0.0535 1.487 0.0643 0.585 0.0673 0.154 0.0838 AV. 0.0639
....
1209
between free lime, liberated in the reduction of tricalcium phosphate, and undecomposed tricalcium phosphate, rather than by direct partial reduction of the phosphate itself. This basic compound is probably oxyapatite. 3 Cas(P04)2.CaO. and is somewhat more difficult to reduce at lower temperatures than is tricalcium phosphate. The total reaction would thus involve not only equation (3), but also, to a certain extent, a reaction that may be represented by the equation 3Cas(POa)z.CaO
+ 15C = lOCaO + 15CO + 3Pz
(4)
I
Iw
Formation of Calcium Phosphide at Temperatures up to 1500" C.
Several of the residues from experiments carried out a t temperatures ranging frorn 1250" to 1500" C. were analyzed for phosphorus combined as calcium phosphide. The method used was a modification of the procedure proposed by Moser and Brukl .I4 A sample of 3 grams of the residue mas weighed into a dry 200-cc. flask, which was then connected through a dropping funnel to a cylinder of nitrogen on the one side and to an absorption tube containing 50 cc. of 0.2 N mercuric chloride solution on the other. Air was displaced from the apparatus by a stream of oxygen-free nitrogen, and the sample mas then decomposed with 30 cc. of saturated sodium chloride solution added from the dropping funnel. The phosphine evolved w&sremoved by a slow stream of oxygen-free nitrogen and absorbed in the mercuric chloride solution, which was then transferred to a beaker and treated with concentrated nitric acid on the steam bath to convert the lower acids of phosphorus into orthophosphoric acid. The phosphorus content of the solution was then determined by the volumetric molybdate method. The results in Table X show that calcium phosphide was not formed in appreciable quantities a t temperatures up to 1500" C. The maximum conversion into calcium phosphide, of the total phosphorus originally present, was 0.17 per cent during 10 minutes heating a t 1500" C., 98.4 per cent of the total phosphorus content of the mixture being volatilized a t the same time. Calcium phosphide was not present in residues from experiments a t temperatures u p to 1250" C. Ross, Mehring, and JonesIg have calculated a theoretical temperature of 1690" C. for the formation of calcium phosphide by direct reduction of tricalcium phosphate with carbon, and the results of the present investigation show that experimentally temperatures in excess of a t least 1500" C. are required for production of appreciable quantities of this compound. of C a l c i u m Phosphide PHOSPHORUS Ca3P2 I N CONVERTED PHOSPHORUS RESIDUE INTO CaSPn VOLATILIZED Per cent Per cent Per cent None h'one 19.1 None None 50.5
Table X-Formation TIME OF
HEATING Af znulcs 60
TEMPERATURE
c.
1200 60 1250 0.026 60 1300 0.064 86.9 6OP 1400 0.053 0.061 88.6 2.5 0,019 0.051 1500 60.1 5.0 1500 0.023 0 057 82.7 0.046 7.5 1500 0.105 93.4 10.0 0.076 0.173 9.8 . 4 1500 . 0 ,Mixture composed of 16 grams Caa(P04)n and 4 grams carbon; other mixture2 composed of 5 grams Cas(P04)n and 5 grams carbon.
,Mechanism of Reaction between Tricalcium Phosphate and Carbon
The reaction between tricalcium phosphate and carbon to give free phosphorus is usually represented by the equation Ca3(P01)2 5C = 3Ca0 5CO PZ (3) but it has already been s h o r n in the present paper that a compound more basic than tricalcium phosphate is formed to a certain extent during some stage of the reaction, and also that this basic compound is apparently formed by reaction
+
+
+
EFFECT OF HEATING FDR ONE HOUR AT
VRRiOUS TEMPEAHTURES
Equations (3) and (4) represent the reduction of tr)dciuni phosphate as occurring entirely in the solid phase a t temperatures up to a t least 1500" C., since fusion of the non-volatile components of the system does not occur in this temperature range. The investigations of Hedvall and HeubergerQm dicate that reactions between solid phases may occur under certain conditions, but it is evident that the completenesq of such reactions is entirely dependent on the fineness and t i l e intimacy of contact of the reacting materials, provided intermediate gas or liquid phases are not formed. If it is assumed that the reaction between tricalcium phosphate and carbon occurs entirely in the solid phase then it mould be necessary to have the materials ground practically to molecular dimensions and very intimately mixed in order to obtain complete reduction. On the basis of our present knowledge of the reaction, we are by no means justified in excluding the possibility of gas phases also entering into the reduction. It is possible that the carbon monoxide formed in reactions (3) and (4) may itself act as a reducing agent towards tri' calcium phosphate according to the equ'a t ion Ca$(PO& 5CO = 3Ca0 5COz PZ (5) A thorough investigation of this possible reaction has not been made, but Lassieur" reports that he was unable to obtain any reduction of pure calcium phosphate a t 1300" C. Negative results mere also obtained by Meyer13 in attempts to reduce Thomas slag by carbon monoxide a t temperatures up to 1200" C. According to Schloesing.20calcium and aluminum phosphates when mixed with silica are reduced by carbon monoxide a t a white heat. In view of the fact that silica is known to have a pronounced effect in accelerating the reduction of tricalcium phosphate by carbon, it is possible that it may have a similar effect on the reduction by carbon monoxide. However, in the absence of further experimental data, it seems very probable that in the range of temperatures, 1150" to 1500" C., covered in the present investigation, reduction of tricalcium phosphate by carbon monoxide does not occur to any considerable extent in the absence of silica. As a possible explanation for the completeness of the reduction of solid calcium phosphate by solid carbon, it may he assumed that tricalcium phosphate has certain dissociation
+
+
+
1210
INDUSTRIAL A N D ENGINEERING CHEMISTRY
pressures a t high temperatures. The reaction may then be represented as occurring in two stages-equations (1) and (2)-and in a series of qualitative experiments it was found that reduction of pure phosphorus pentoxide by carbon begins a t about 800” C., which indicates that a t 1200” C. reduction would probably be complete and practically instantaneous. At 1200” C., for instance, the dissociation pressure of tricalcium phosphate may be so small as to be practically unmeasurable, but in the presence of carbon the phosphorus pentoxide liberated would be immediately reduced to elemental phosphorus and dissociation of the tricalcium phosphate would consequently be accelerated to a considerable extent. Furthermore, extremely fine grinding and intimate mixing would not be necessary in order to obtain practically complete reduction of the phosphate because of the presence of a vapor phase in the form of phosphorus pentoxide. Tricalcium phosphate undoubtedly has definite dissociation pressures a t high temperatures, and experimental evidence on this point would be valuable in explaining the mechanism of the reaction with carbon in the solid state. Bibliography 1-Berthier, Ann. chim. phys., [2], 83, 178 (1826). 2-Blome, Metallurgie, 7, 659, 698 (1910).
Vol. 20, N o . 11
3-Bryan, Mehring, and Ross, IND. END. CHEM.,16, 821 (1924). 4-Carothers, Ibid.. 10, 35 (1918). 5-Dieckmann and Houdremont, 2. anorg. allgem. Chem., 120, 129 (1921). 6--Flacke, Z.Elektrochem., 21, 37 (1915). 7-Foote, Fairchild, and Harrison, Bur. Standards, Tech. Paper 170, 117 (1921). &Greenwood, J . Chem. SOL.(Londgn), 98, 1483 (1908). 9-Hedvall and Heuberger, Z.anorg. allgem. Chem., 130, 49 (1924). 10-Hempel, 2. angew. Chem., 18, 132 (1905). 11-Lassieur, Orig. Communications, 8th Intern. Cong. A p p l i e d Chem., 2 , 171 (1912). 12-Lerch and Bogue, IND.ENG.CHEM.,18, 739 (1926). Mitt. Kaiser-Wilhelm Inst. Eisenforsch. Diisseldorf, 9, 2 i 3 13-Meyer, (1927). 14-Moser and Brukl, Z.anovg. allgem. Chem., 121, 73 (1921). E-Nernst, “Theoretical Chemistry,” p. 758, Macmillan & Co., London, ’ 1916. 16--h’ielsen, Ferrum, 10, 97 (1912). 17-Preuner and Brockmoller, 2. physik. Chem., 81, 129 (1912-13). lS-Ross, Carothers, and Merz, J. IND. ENC. CHEM.,9, 26 (1917). 19--Ross, Mehring, and Jones, Ibid., 16, 563 (1924). 2O-Schloesing, Comet. rend., 69, 384 (1864). 21-Swann, J. IND.ENC.CHEM.,14, 630 (1922). 22-Thorpe, “Outlines of Industrial Chemistry.” p. 256, Macmillan Co., New York, 1916. 23-Waggaman, Easterwood, and Turley, U. S. Dept. Agr., Bull. 1179 (1923). 24-Zeller and O’Harra, School Mines Met., Univ. Missouri, Tech. Bull. 6, 3 (1925).
Chemistry of the Cellulose Determination’ Clifford E. Peterson and Mark W. Bray U. S. FOREST PRODUCTS LABORATORY, MADISON,WIS.
ROSS and Bevan2 in 1880 reported that the cellulose content of plant materials may be estimated by subjecting the moistened material to the action of chlorine gas and then removing the reaction products of the lignin and other bodies with a hot dilute solution of sodium sulfite. H a ~ gin , ~his experiments with wheat straw, found that the lignin could not be entirely removed by the use of sodium sulfite solution after chlorination. When, however, he substituted for the sulfite a 1per cent solution of sodium hydroxide a t 70” C. for 10 minutes, he obtained practically the same yield of cellulose as by the original method and was unable to detect lignin in the residue with zinc iodide stain. To obtain from wood by the Cross and Bevan method a uniformly white residue of cellulose that develops no pink coloration on treatment with sodium sulfite solution, four to five chlorinations are generally r e q ~ i r e d . ~Commercial pulps produced by either the sulfite or the alkaline processes are found always to contain appreciable amounts of lignin, even after bleaching with calcium hyp~chlorite.~This lignin seems to be more resistant to chemical action than the major part of the lignin present in wood; three chlorinations are usually required to complete its removal. Miller, Swanson, and Soderquist6have shown that a hydrolysis of wood with dilute acid renders lignin largely insoluble by the ordinary sulfite cooking process. The lignin is still removable, however, by chlorination. Similarly, Michel-Jaffard’ reports that he was unable to pulp Bordeaux pine by the sulfite process after a preliminary extraction with sodium hydroxide solution. I n addition,
C
Received May 16, 1928. J . Chem. SOC.(London). 88, 666A (1880); Chem. News, 42, 77 (1880). “Uber die Natur der Cellulose aus Getreidestroh,” Berlin (1916). 4 Schorger, J. IND. ENC.CHEM.,9, 556 (1917). 6 Bray and Andrews, Paper Trade J . , 76, No. 3, 49; No. l 9 , 4 9 (1923). 6 Ibid., 81,No. 9, 58 (1926). 7 Papier, 27, 213 (1924): Paper I n d . , 6, 869 (1924); Paper Trade J., 79, No. 18, 159TS (1924). 1 2
Schafer and Peterson8have observed that the rate of delignification of flax straw with sodium sulfite is retarded by a previous digestion with caustic soda. Klasong has shown by chemical means that two forms of lignin exist in plant substances. He estimates that in spruce wood 63 per cent of the total lignin is alpha- or acrolein-lignin, and that 37 per cent is beta- or carboxyl-lignin. Ritter,Io working on red alder and western white pine, has made a mechanical separation of two forms of lignin, which he has characterized physically. One form occurs in the middle lamella or partition wall between cells, is light brown in color, shows structural form, and has a comparatively high methoxy1 content (13.6 and 10.8 per cent in red alder and western white pine, respectively); the other form, found in the cell wall proper, is darker, is amorphous, and has a low methoxyl content (4.8 and 4.3 per cent, respectively, in the two species). From the foregoing considerations it is conceivable that one or more of the chemicals used in the cellulose determination exert a hydrolytic effect, which renders a part of the lignin soluble with difficulty, or that one of the types of lignin just described is more resistant to chemical action than the other. -4third and more likely view is that the lignin in the middle lamella is removed readily because of the separation of the fibers that occurs during chemical treatment, while the cellwall lignin is afforded a mechanical protection that retards its reduction by the chemical reagents. Wood contains also pentosans, or furfural-yielding substances, part of which are removed during the pulping processes and also during the cellulose determination and part of which remain in the cellulosic residue. I n view of the various points of uncertainty attaching to the Cross and Bevan chlorination process, it was thought that a study of the changes in chemical composition of spruce wood Pafier Trade J . , 8 6 , No. 3, 51 (1928). Paper I n d . , 4, 262 (1922). 10 IND. ENG.CHBM.,17, 1194 (1925). 8
9
