Thermodynamics of roasting arsenopyrite

Thermodynamics of Roasting Arsenopyrite N. CHAKRABORTI and D. C. LYNCH Existing thermodynamic data for the Fe-As-S-O system were evaluted and predominance area diagrams for that system were constructed at 798 and 973 K. Isopleths for the As/S and As/O atomic ratios in the vapor phase have been added to the diagrams by solving the complex equilibria. These modified diagrams were used to evaluate the results of roasting both natural and synthetic arsenopyrite (FeAsS) in inert, reducing, and oxidizing atmospheres at 798 and 873 K. Conditions leading to the retention of As as As2Os(s) and FeAsO4(s) were also reviewed. The experimental results indicate that both reducing and oxidizing atmospheres are more effective in the removal of As than an inert atmosphere. In a reducing atmosphere arsenic sulfides are evolved and the percentage o f As removal increases with decreasing Po2. The greatest percent of As removal occurred with highly oxidizing atmospheres which generated As406 vapor. I. INTRODUCTION ARSENIC bearing minerals such as arsenopyrite (FeAsS) are often associated with copper ores. Smelting o f arsenious concentrates causes a serious emission problem since ar- senic and many of its compounds are toxic, and inorganic arsenic in the presence o f SO2 is a suspected carcinogen. In addition, any arsenic retained in metallic copper seriously affects its physical properties. This, together with the fact that recent single-step smelting processes handle arsenic less efficiently than the conventional smelting/converting operations, makes roasting studies o f arsenic-bearing minerals important.1 Arsenic and many of its oxides and sulfides are volatile and thus roasting an arseniousconcentrate should lead to the removal o f As. Unfortunately, present roasting practices result in the retention o f over 50 to 75 pct of the arsenic.2 This retention involves two processes, first the release (or lack o f release) o f arsenic-bearing vapors from the host mineral and, second, reaction o f the vapor species with other minerals, resulting in further retention o f arsenic. The process is complex, involving not only several different host minerals but also widely varying oxygen and sul- fur potentials. Multiple hearth roasters are still primarily used in the processing o f arsenious concentrates.2 The partial pressure of 02 at the surface o f the bed is approximately 0.01 to 1 atm.* This pressure decreases significantly in the interior *Throughout this paper 1 arm. = 101.325 KPa. of the bed while the activity o f sulfur increases. The impact of the variation in the Po2 and Ps2 has a significant influence on what arsenic phases are stable. Accordingly, an under- standing o fthe complex mechanism of arsenic removal from ores requires not only an evaluation o f rate data but also a thermodynamic analysis o f the pertinent As systems. In the course o f this investigation roasting experiments were conducted with arsenopyrite at temperatures ranging from 798 to 873 K in inert, reducing, and oxidizing atmo- spheres. Thermodynamic data for the Fe-As-S-O system have been compiled, and the data used in evaluating the N. CHAKRABORTI, Research Assistant, and D. C. LYNCH, Associate Professor, are both in the Department of Mining, Metallurgical and Ce- ramic Engineering, University of Washington, Seattle, W A 98195. Manuscript submittedJune 8 , 1982. experimental results. Predominance area diagrams are presented for the ternary subsets o f the quaternary system, and these diagrams have been modified to include isopleths for the As/S and As/O atomic ratios. Given the numerous arsenic-bearing gaseous species found at roasting tem- peratures, the isopleths add significant information concern- ing the composition o f the gas phase in equilibrium with the solid phases found on the predominance area diagram. II. REVIEW OF THERMODYNAMIC DATA The authors are not aware of any compilation o f thermo- dynamic data regarding As and its compounds.* The data *Since submission of this paper, a compilation, "Thermodynamic Data for Arsenic Sulfide Reactions," by A.D. Mah, USBM Report of In- vestigation 8671, 1982, has been published. This compilation does not provide information on As2S3(g). However, dataprovidedfor As4Sa(g) and As406(g) are in excellent agreement with that in Table I. The data presented for AsS(g) differ from what is presented in this paper. The data in Table I for AsS(g) have recently been found to be in agreement with new experi- mental data obtainedby K. H. Lau at SRI International, Menlo Park,CA. available in the literature for the Fe-As-S-O system have been compiled in Table I. The reliability of the data varies due to the significant differences in the results reported in separate investigations. These differences are due, in part, to the polymorphic nature of arsenic and its compounds. A r s e n i c Vapor . Arsenic vaporizes in four forms: AS4, As3, As2, and As. Thermodynamic functions for these species are included in Table I. Except for the vaporization of As4 the data were obtained from Hultgren et a l ' s com- pilation.3 These data are based on the work o f over fifteen investigations. Recently, there has been some indication that Hultgren's selected value for the standard enthalpy o f vapor- ization o f As2 at 298 K is low and that values reported by Rau,4Murray, and co-workers5 and Drowart et a l .6 are more accurate. This new value for the standard enthalpy o f vapor- ization yields a higher partial pressure of Asz than the data in Table I. The variation in P•s2is not a serious problem at the temperature o f interest in this study where As4 is the primary species.7 Sul f ides . The greatest amount o f uncertainty involves the data on arsenic sulfides. In his treatise, Mellor reported that the following solid sulfides exist: A s a S 3 , As4S4, As253, and As2Ss. 8 Subsequent investigations have found only As454, realgar in both and a and/3 structures, As2S3, orpiment and METALLURGICAL TRANSACTIONS B VOLUME 14B, JUNE 1983--239 T a b l e I. T h e r m o d y n a m i c D a t a f o r t h e Fe-As-S-O S y s t e m Equilibrium In K Temperature R a n g e K Reference Elemental Sulfides 16,116 ASh(g) = 4As(s) 18.3 -- Table I. Cont. Thermodynamic Data for the Fe-As-S-O System Equilibrium In K Temperature Range K Reference Arsenate 3As406(g) + 702 + 4Fe304(s) = 12FeAsO4(s) -188.04 + 27.6" |0-4T 105 + 3.898 " - - - 27.9 log T T 27 As4S3, dimorphite. The latter sulfide is quite unstable and is not o f interest in this study.9 Furthermore, As and S, at temperatures above the melting point o f S, are known to form a liquid solution containing as much as 70 at. pct As at 973 K.1° In the gaseous phase the following sulfide species have been identified: As4S3, AsS, As2S2, As454, and AsaS3.11'12 The presence o f AszS2 and As453 vapors in mass spec- trometric measurements may be associated with the destruc- tion of larger molecules due to electron beam bombardment. The data in the literature suggest that As2S3 and ms4S4 a r e the primary sulfide vapor species and that AsS, while it exists, is only a minor species.11This is true, as shown later, if the activities of both arsenic and sulfur are high. At tem- peratures typical o froasting operations arsenic sulfides exist in the vapor state. Accordingly, the literature review which follows deals primarily with the vapor species. Experimental data from two independent investigations reveal that vapor originating from the sublimation o f realgar is composed predominantly o f AS454 molecules. Munir, et al., using a mass spectrometer, found small quantities o f As453, As4, and As2 in the gas phase.11They noted, however, that with respect to As4Sa(g) the percentage of these species decreased with increasing temperature. Lau and co-workers, also using a mass spectrometer, found As2S2 vapor upon sublimation of realgar.12 However, these same investigators found that when precautions were taken to ensure equi- librium the vapor consisted entirely of As454 molecules. Lau and co-workers attribute the formation o f As2S2 to the effect o f the electron beam. It is apparent from these studies that realgar vaporizes a s ms4S4 molecules and that the total vapor pressure is due essentially to this species. Munir et a l . calculated the enthalpy and entropy of sublimation o f AsaS4 tO be 118---2 kJ/mole and 154 +--4 J/mole K, respectively." Johnson, Papatheodoron, and Johnson using a drop cal- orimeter found the enthalpy and entropy of fusion of realgar to be 26.67 -+0.20 kJ/mole and 45.97 + 0 . 3 4 J/mole K.13 These values compare favorably with those determined by Street and Munir.14 Street and Munir used differential thermal analysis and obtained values o f 25 kJ/mole and 43 J/mole K, respectively, for the enthalpy and entropy of fusion of As454. Using Munir's data for the enthalpy of sublimation, and assuming the heat of sublimation is the sum of the heats o f fusion and vaporization, the enthalpy and entropy o f vaporization are calculated to have values of 91.3 kJ/mole and 108 J/mole K, respectively. These val- ues are in agreement with Trouton's rule which states that the entropy of vaporization is approximately 92 J/mole K for all substances. 15 Ustyugov, Kudryavtsev, and Kuadzhe also measured the vapor pressure o f As4S4 in equilibrium with molten realgar using a quartz spoon gauge.16 If their data are used in con- junction with that obtained by Munir et a l . for the sub- limation o f realgar, they yield a heat of fusion approximately twice the value obtained by Street and Munir, and Johnson and co-workers. As a result of this inconsistency and the agreement between the other experimental investigations, the authors have elected to disregard the work conducted by Ustyugov et a l . Like Ustyugov et al., Strathdee and Pidgeon also used a quartz spoon to measure the vapor pressure over FeAsS(s), As4S4(/), and pyrrhotite at temperatures between 473 and 923 K.17They assumed the total pressure in the gauge to be the sum o f the partial pressure o f As4 and As454. By com- pletely vaporizing the As484(/)Strathdee and Pidgeon were able to evaluate the partial pressures o f both AS4 and As454. Their assumption that FeAsS(s), AsaS4(/), and pyrrhotite were the only condensed phases present, appears to be in error. Barton has shown that these phases are in equilibrium only at temperatures between 764 and 961 K and not be- tween 473 and 923 K, as Strathdee and Pidgeon sug- gested.~°At lower temperatures pyrite, not pyrrhotite, is in equilibrium with the other two phases. In addition, Barton has shown that the As-S liquid in equilibrium with FeAsS and pyrrhotite is not msaS4 but a liquid o f variable com- position which is rich in As. Accordingly, with knowledge o f the above information, the authors have used the data of Johnson et al., Munir and co-workers, and Street and Munir to develop equations for the vapor pressure of A S a S4 and its standard free energy o f formation. Those equations are included in Table I. Greater discrepancy exists for the vapor pressure o f AszS3 than that for AS484. Pressures for As2S3 reported by Ustyugov et al., Isakova and Nesterov, and Hsiao and Schlechten differby several orders of magnitude.16,18,19 The latter investigators used the Knudsen-Langmuir technique and assumed the condensation coefficient to have a value o f unity. That assumption may account for the unusually low values they reported. Another discrepancy in their data sug- gests that the heat of vaporization is much larger than the heat of sublimation. As a result o f this discrepancy, the authors elected to disregard these data in favor o f more recent investigations. Isakova and Nesterov used a closed vessel and a modified statistical method to obtain the vapor pressure of As2S3.18 The vapor pressures they reported for As2S3 are very similar to those obtained by other investigators for AS454. The pres- ence o f a small quantity o f realgar in the solid sample would account for this variation. METALLURGICAL TRANSACTIONS B VOLUME 14B,JUNE 1983--241 Ustyugov et a l . measured the vapor pressure of As2S3 utilizing the same procedure they used to measure the va- por pressure o f As454.16 Their values are lower than those reported by Isakova and Nesterov and yield a normal boiling point of 1000 K. Isakova and Nesterov found a much lower normal boiling point. The boiling point o f 1000 K compares favorably to values listed in other compilations.2° Lau et al., using a mass spectrometer, found appreciable quantities of As2S3, As4S4, and As2S2 vapor over orpiment.12 The presence of a significant quantity o f As2S2 vapor sug- gests that perhaps equilibrium was not obtained in these experiments. In earlier work involving realgar As2S2 vapor was also found. Although, when precautions were taken to ensure equilibrium, only As4S4 vapor was observed. _ Faure, Mitchell, and Bartlett reported that As2S3(s) de- composes on vaporization to AsS(g) and S2(g).2~ However, they used high ionization potentials of 35 and 75 eV. Such high potentials may account for the destruction of As2S3 and AsaS4 species. Lau and co-workers, using lower ionization potentials, did not observe any appreciable concentration of AsS vapor over orpiment. Molten orpiment exists in equilibrium with appreciable quantities of $2 and As4 vapor.22 If As454 vapor is also found in appreciable concentrationsover orpiment, as suggested by Lau and co-workers, then the vapor pressures reported by Ustyugov are the sum o f partial pressures o f several species. In the investigations described herein we have, for lack of better data, acceptedUstyugov's equation for the vaporpres- sure of As2S3. Although Ustyugov's equation undoubtedly yields higher vapor pressures for ms2S3 than actually exist, an estimate suggests that Ustyugov's data provide a pressure of AszS3 to within an order o f magnitude ofthe correct value. Mills and Lau and co-workers have evaluated the thermo- dynamic parameters o f AsS vapor. 23 Their data are in excel- lent agreement. Using these data an equation for the standard free energy o f formation o f AsS(g) was calculated and is presented in Table I. A comparison of the vapor pressures of AsS, As4S4, and As2S3 is presented in Figure 1. The data indicate that As2S3 and As4S4 vapors are the predominant species at high partial pressures of $2 and As4. However, at lower partial pressures for arsenic and sulfur the pattern is reversed and AsS is the major vapor species. The present discussion reveals that considerable disagree- ment exists in the physico-chemical properties o f arsenic sulfide vapors. However, sufficient data are available for calculation of thermodynamic properties for As4S4 and AsS vapors. Data for As2S3 vapor are not as reliable, and thus the data for this compound in Table I should be used cautiously. O x i d e s . There is general agreement among investigators regarding the thermodynamic properties of the arsenic oxides. This agreement is probably associated with the fact that only two solid oxides exist and only one is found in the vapor state. The trioxide, As203, exists in two crystalline polymorphic forms: arsenolite is the most common, and the other is claudetite. Transition between these two forms does not readily occur, and most investigators have examined the properties o f arsenolite as it is stable over a much broader temperature range. The other solid oxide species is the pent- oxide, AszOs. The pentoxide, a more stable solid existing at temperatures well above the normal boiling point of the tri- oxide, decomposes at approximately 1000 K forming A S 4 0 6 vapor and 02. 0 o - E l ' ' ' ' - l o " • ~ ~ s 2 S 3 ~ ~.E_AsLS/. -12 " 'r~ ~ . ~ l - ' 6 -12 -8 -4 0 Log PAsl. Fig. 1 --Arsenic sulfide isobarsat 10-5 and 10-~5atms. The isobars reflect the stability of each sulfide with respect to the others. The lower the position of the isobar on the diagram for a set pressure the more predomi- nant the sulfide vapor• Data are for a temperature of 798 K. Behrens and Rosenblatt measured the vapor pressure o f arsenolite using a Knudsen effusion technique.24 They showed that their results are in agreement with dataobtained in eleven other investigations conducted over a period of seventy years. The authors have used the results ofthis latest investigation and other data in the literature to calculate the free energy equations presented in Table I for As203(s) and A s 4 0 6 ( g). Gu6rin and Boulitrop, Vian et al., and Polukarov and co- workers examined the dissociation of the solid pentox- ide. 25'26 Vian et a l . noted several errors in the data presented by Gu6rin and Boulitrop.27 The revised data are in agreement with thoseobtained by Polukarov and co-workers who used a static dew point technique to evaluate the vapor pressure of A s 4 0 6 and O2 above the pentoxide. Since Polukarov et a l ' s free energy equation is equally reliable but expressed in a simpler form than that provided by Vian et al., the authors have selected it for inclusion in Table I. Iron Arsenides . Barton's analysis o f the iron arsenides is the only known study o f these compounds.l° Barton mea- sured the free energy o f the univariant reactions between pyrrhotite and the iron arsenides FeEAS, FeAs, and FeAs2. The sulfur content of pyrrhotite varies with the partial pres- sure of $2. Barton was thus able to use an X-ray techniquehe and Toulmin developed in an earlier investigation to deter- mine the equilibrium partial pressure o f $2 associated with the univariant reactions.28'29 Barton used these data to calcu- late equations for the standard free energies of formation o f the arsenides. His data are included in Table I. Arsenopyrite. Barton, and Strathdee and Pidgeon have evaluated the free energy for various reactions involving arsenopyrite,l°J7 Barton used the same procedure described above for the arsenides. However, for arsenopyrite he evalu- ated the vapor pressure of $2 associated with the univariant reaction between FeAs2, pyrrhotite, and arsenopyrite. With 242--VOLUME 14B, JUNE 1983 METALLURGICAL TRANSACTIONS'B that vapor pressure Barton was able to calculate the free energy of formation o f arsenopyrite. Strathdee and Pidgeon measured the vapor pressure of AS4 in equilibrium with pyr- rhotite and arsenopyrite. However, as noted earlier, their experimental results contain a source o f error. Accordingly, the authors have used Barton's data for the free energy of formation of arsenopyrite. Iron Arsenates. Vian, Iriarte, and Romero, using existing data, estimated the standard free energy o f reaction associ- ated with the dissociation of ferric arsenate (FeAsO4)to mag- netite, arsenic trioxide vapor, and oxygen. Their estimated data suggested that the arsenate will readily decompose at temperatures above 1200 K which they confirmed with ex- perimental results. The present authors have, in turn, uti- lized their data in the construction o f the predominance area diagrams. The thermodynamic information for ferrous arsenate, Fe3(AsO4)2, although provided by Vian et al., was deemed unreliable by thoseinvestigators and, thus, was not used in their study.* The present authors have reached the *Vian et al calculated AH~g8 values of the arsenates bythree methods and reported the mean of the values as the actual heat of formation. Similarly, the standard entropy was calculated by two different methods, and as before the mean value reported. For ferric arsenate the values of AH~'98 and S~98 obtained by the different methods were in close agreement. That was not the case for ferrous arsenate, and thus those data were deemed unreliable. same conclusion regarding the ferrous arsenate data. I l L P R E D O M I N A N C E A R E A D I A G R A M S The data in Table I were used to construct the predomi- nance area diagrams for the ternary subsets o fthe Fe-As-S-O system. Extrapolation of the thermodynamic data in some cases to higher temperatures was necessary. However, that extrapolation should not produce any gross error in the dia- grams. The diagrams are shown in Figure 2 as the three faces for the quaternary systemat temperatures o f 798 and 973 K. The only majordifferences between the diagrams at the two temperatures involves arsenopyrite, FeSO4, and wustite. Ar- senopyrite melts incongruently at 975 K, and thus its phase field occupies only a small portion of the diagram at 973 K. Similarly, FeSO4 decomposes at 944 K, and consequently, does not appear in the diagram at 973 K. Wustite is a stable phase at temperatures above 833 K, and thus is not present in the diagram at 798 K. The partial pressures of 02, $2, and As4 are each limited by the precipitation of a condensed phase. In the Fe-As-O and the Fe-As-S systems the As4 vapor pressure is limited by the precipitation of metallic arsenic. The oxygen potential in the Fe-As-O is limited by the reaction between As4 and 02 in the production of the solid pentoxide. This same reaction limits the oxygen potential in the Fe-S-O system, provided some As is also present.* In the Fe-As-S and Fe-S-O systems *The predominance area diagram presented for each subset of the qua- ternary system has been calculated assuming a constant potential for the fourth component. The Fe-S-O system was constructed assuming As4 to have a partial pressure of 10-~8 atmosphere, the Fe-As-O system assumes 10-18 partial pressure for $2, and the Fe-As-S system assumes an oxygen partial pressure of 10-30 atm. the sulfur potential is limited by the precipitation o f molten sulfur. The sulfur saturation line has not been added to the Fe-S-O face of the diagram, as the influence of oxygen on the activity o f molten sulfur is unknown. The solubility o fAs FeSC Fe2(SO,); / 0 Log P02 Fe2(SO,,: Log PS2 .0 ..-S-As liquid saturation ;i ~ /FeAsS 92 .44 F ' e a A s eAs2 ' -46 ~e-" , l ' / , I . . . . :,~ " / _ / : , ~ :,o : 8 / ' ~ -', "2 'o Lo~P~, As saturationAs20~ saturation (a ) o / \As~O~ sa tu ra t ion Log P02 (b) Fig. 2 - - T h e 3 faces of the Fe-As-O-S predominance area diagram at (a) 798 K and (b) 973 K. The composition of the phases FeO, FeS, and FeAsS varies with Po2, Ps:, and P A s 4 . For simplicity the phases are repre- sented by the i r stoichiometric formulas. in the liquid increases with PAs4, thereby decreasing both the vapor pressure of $2 in equilibrium with the liquid and the sulfur content of the liquid. The S-As liquid saturation line was added to the diagrams using Barton's data~° and the Gibbs-Duhem equation: d log 7A8 = - Xs/XAs d log % [1] log 7As at X~s METALLURGICAL TRANSACTIONS B VOLUME 14B, JUNE 1983--243 where XAs is the mole fraction of As in the S-As liquid, and X),~ represents a specific liquid composition at which the activity of As is known. The analysis has been presented in detail elsewhere.22 Predominance area diagrams in their conventional form show the stable solid phases as a function o f the partial pressures o f two gaseous species in the system. However, such diagrams are not adequate to explain volatilization when a complex gas phase is present. For the Fe-As-S-O system only the latter three elements are found in the gas phase, and if the three faces o f the quaternary system shown in Figure 2 are examined separately,only two elemental spe- cies are found in the gas phase for any one o f the ternary systems. The authors have evaluated the composition o f the gas in equilibrium with the solid phases shown in the Fe-As-S and Fe-As-O diagram. This information has been added to the predominance area diagrams in the form of isopleths of constant As/S and As/O atomic ratios. The isopleths for the Fe-As-S system were evaluated by examining the complex equilibrium between As4 and $2 and the following gaseous species: As3, As2, AS, AsS, ms4S4, As2S3, 54, 56, and $8. These equilibria provide eleven un- knowns and nine equations. The As/S atomic ratio defined by the following equation: As S 4PAs4 + 4PAs4S, + 2PAszS~ + PAsS + 3PAs~ + 2PAs2 + PAs 8Ps8 + 6es6 + 4Ps4 + 2Psz + 4PAs4S, + 3PAsES3 + Pass [21 provides another unknown, As/S and another equation. The total pressure ofthe arsenic and sulfur species is given by the following equation: pAs-S = PAs4 + PAs3 + eAs2 + PAs + PAs4S4 + PAs2S3 + PAsS + Ps2 + Ps4 + es6 + Ps8 [3] When the values ofPAs-sand As/S are set there exist eleven equations containing eleven variables. That set of eleven equations was solved using a digital computer and an itera- tion technique. Similarcalculations were performed for the Fe-As-O system where the vapor phase was assumed to con- sist o fASh, As3, As2, As, ms406, and 02. In that analysis the total arsenic-oxygen pressure (pAs-O) and the As/O atomic ratio were expressed as: pAs-O = eAs4 + PAs3 + Pas2 + PAs + PAs406 + Po2 [4] A__~s = 4PAs4 + 3PAs~ + 2PAs~ + PAs + 4eAs40~ [5] O 2Po2 + 6PAs406 The concept o fusing atomic ratios, total pressure, etc. as constraints for multiequilibria problems has been known to the scientific community for many years. Denbigh has re- viewed these principles in his discussion o f the Gibbs phase rule. 33 Other authors have included similar discussions in v a r i o u s textbooks. 34'35 Tevebaugh and Cairns utilized these basic concepts in 1962 to examine the C-H-O system. 36'37'38 Since this initial work others have performed similar calcula- tions39-44 and reiterated the basic principles.44 Initially a rou- tine similar to that in Reference 44 was utilized to solve the simultaneous equations. Later that technique was abandoned in favor of a more efficient successive substitution routine. Both the isobars and the isopleths for the atomic ratios were added to the predominance area diagrams in Figure 3. The isobars for pAs-O in Figures 3(c) and 3(d) are horizontal at low partial pressures of 02. At these low values o fPo2 the vapor phase is composed primarily o f As4. This, of course, accounts for the nearly vertical As/O isopleths. At Po2values above 10-21 atm at 798 K, or at Po2 values greater than 10-15 atm at 973 K, As406 is the major vapor species and pA~-O can be approximated as eAs406. Values o f PrAs-° increase rapidly with increasing values of Po2. This rapid increase is a reflection o fthe stability of the AS406 molecule. The As/O isopleths cannot be added to the diagram at the higher 02 partial pressures as the As/O atomic ratio has a value of approximately 4/6 over the entire region where A S 4 06 is the primary vapor species. In the Fe-As-S system $2, As2, and As4 are the primary vapor species. This is evident from the isobars for pA~-S in Figures 3(a) and 3(b), which are either horizontal o r vertical. If a sulfide vapor species were predominant the isobars would lie diagonally on the diagrams and have a negative slope. The isobars and the As/S isopleths indicate that arsenopyrite exists in equilibrium with a vapor phase composed essentially of arsenic. IV. EXPERIMENTAL Materials. Natural and synthetic arsenopyrites were used in this investigation. The synthetic material was prepared in the laboratory from -200 mesh iron, arsenic, and sulfur powders. All the powders had a purity o f 99.99 pct or better. A vycor tube was sealed at one end and filled with equimolar amounts of iron, arsenic, and sulfur. A vycor rod was in- serted into the tube to minimize free space and any loss of the reactants to the vapor state. The tubes were sealed under a vacuum of approximately 10-6 atm and subsequently heated at 873 K for seven days. At the end of this period the tubes were quenched in water, and the reacted products were analyzed by X-ray diffraction. That technique indi- cated the presence of arsenopyrite only. The ore used in this study came from the Gold Hill prop- erty in Utah. The naturally occurring arsenopyrite was found to be associated with large amounts o f silica and pyrite. The natural material was crushed to + 1/8 inch pieces and hand- picked for pyrite. After removing the pyrite the ore was crushed again. A heavy-media separation technique was subsequently used to remove silica. The silica was found to be so intimately mixed with the arsenopyrite that a consid- erable fraction o f the silica remained with the arsenic- bearing mineral. Chemical analysis revealed a final arsenic content o f approximately 20 wt pct while the formula FeAsS corresponds to an arsenic content o f 45 wt pct. X-ray analysis revealed that the final concentrate contained, in addition to arsenopyrite, some pyrite and a considerable fraction o f silica. Experimental Technique. Natural and synthetic arseno- pyrite samples ( < 0 . 3 g and prepared as noted above) were roasted in various atmospheres using a two-zone labora- tory furnace with a preheating arrangement for the inlet gas (see Figure 4). A chromel-alumel thermocouple was used to monitor the roasting temperature while an Fe- constantan thermocouple was used in the gas preheating region of the furnace. The thermocouples were connected to two three-mode controllers. The hot zone of the roasting 244--VOLUME 14B, JUNE 1983 METALLURGICAL TRANSACTIONS B - 8 o~ " -12 -1( 0 ~ / S - A s liquid saturation -~--I - - - __,~ --,... , C l ~--,-± --~ --I',-,-!~ " " / F e S / / I / I . , A " , . S I I ,..-Uo- .-/._,,///,~'.~-I" AI /F.+~ , ~ . . v i A I II-~ " / I I " I ~ ; A s / / IF~.'As.2' ii : .II I / , i / - /119 , / 1 1 / I , / , I , / II, -15 4 2 -8 -4 0 Log PAs4 (a) 0 -2 -4 -6 - 8 a" -1oa< J -12 -I~ -16 -18 -30 ~ f A s sa tu ra t ion L . I, I Z__\- . ~ s , - - - ~ --X,, = 47'i- . ..o.\ ~:~ - - ~ , - ~ - . , \ \ \ io~o' \ / J , / , \ , \ X , -28 -26 -24 -22 -20 -18 --16 -1/+ -12 Log P02 (¢) ( S - A s l iqu id saturation 0 "-- ~ ' - ~ FeS2 __7 - - ..__o~ ,, - - , , , - I ;_ ' / m 1>-10 ,, ~ z ' ° ~ , ¢ ~ , / ; AI , / i O " - 8/ " , , 2 / / ~ - i ~ ; I A " J / I 1 / I - q I / r 4 A I , l b . , ' ; 46 / -16 -12 -8 -/. 0 Log PAs~ FeAsS *2[ As sa tu ra t i on - . , 01181/.106/ I_ I / 11 l / / -'Is ~ I-Z\,_< i , ~ - - ~ , \ 7 " " ~ - - - " ' T ' Fe/.-oF I \F '-10V \ / _ l _ _ i ~ _ , ~ \ ' ; , , I \ ~I ;1;Ie k - -22 -20 -18 -16 -14 -12 -10 -8 -6 Log P02 (b) (d) Fig. 3--Predominance area diagrams with isobars (p~s-s and p~s-o) and isopleths (As/S and As/O) added. The isobars are represented by broken lines ( - - - - - ) while the isopleths are depicted as light solid lines ( ). All values listed are base l0 logarithms. Diagrams (a) and (b) represent the Fe-As-S system at 798 and 973 K, respectively, while diagrams (c) and (d) represent the Fe-As-O system at the same temperatures. furnace was approximately 6 cm long and showed a maxi- mum temperature variation of only ---3 K. The temperature at any particular point within the hot zone fluctuated by only ---1 K when steady state conditions were achieved. Vycor crucibles containing arsenopyrite samples were placed in the hot zone o f the furnace. The bed depth was kept less than 3 mm to minimize any diffusion or mass transfer problems within the bed. During an experiment the furnace was purged con- tinuously with the reactant gas. The gas flow was controlled by needle valves and measured by a series o f capillary flowmeters. Before metering, all the gases used in the ex- periments were passed through drying columns containing Drierite. Residual O2 in the CO2 and He gas streams was removed by passing these gases over Cu-chips heated to approximately 673 K. Ascarite was used for the removal o f residual CO2 in the CO gas stream. The experiments were conducted for periods ranging from 15 minutes to two hours under oxidizing, reducing, and inert atmospheres using different combinations o f CO, METALLURGICAL TRANSACTIONS B VOLUME 14B, JUNE 1983--245 He quench \gaScolumnCleaning CU furnace two zone furnace gas exit I ' / ~' ) ~ / specimen tc i J port CO mixer pro heater $~ ~stopcock volvo f Iowme er Fig. 4--Experimental apparatus. CO2, He, and 02 gases at 798 K and 873 K. The air present in the furnace was driven off by purging the furnace with the reactant gas before any arsenopyrite samples were intro- duced. Any arsenic-bearing vapors formed during roasting were immediately transported by the flowing gas stream to cooler portions of the furnace and there condensed. Roasted materials were collected at regular intervals for chemical analysis. To prevent oxidation, hot samples were quenched in helium before the furnace was actually opened to the atmosphere. The roasted material was digested in a HNO3-HCI-HF- HClO4 acid mixture. Before treating any arsenic-bearing ore with HF for digestion, it was necessary to establish oxidizing conditions to prevent the arsenic precipitating out in an insoluble form. The material was initially treated with a HNO3-HC1 mixture before any HF was added. Perchloric acid was added only at the last stage o f digestion to dis- solve any remaining material. Atomic absorption spectro- photometry was used to determine the amount of arsenic. Some of the determinations were made on a Perkin-Elmer 305B atomic absorption (AA) spectrophotometer, while an IL 551 AA unit was used for the rest of the analyses. V. RESULTS The experiments were conducted to investigate the effect o f particle size and temperature, and to investigate the influ- ence of oxidizing and reducing atmospheres on the evolution o f arsenic from arsenopyrite. The results are shown in Figures 5 through 9. Figures 5 and 6 show the effect of first temperature and then particle size. The effect o f a reducing atmosphere is shown in Figures 7 and 8, while the effect o f an oxidizing atmosphere is illustrated in Figure 9. A summary of the products obtained on roasting is presented in Table II. I I I I 30 • ._a /OC ._u ~ • 0 ~ • < ,o Pco Pco ° 5 ° f / • 873 K 0 , I , I ~ I , I 0 30 60 90 120 Time (min.) Fig. 5--Influence of temperature on roasting of -150 +200 mesh natural arsenopyrite. 70 > 0 60 1,5 temperature 798K Pco/Pco = 20 part•col size • -115 +150 • -200 +270 E rv, u 3C - - , < ~ 15 i ~ i ~ i • I I I 0 30 60 90 120 Time (min.) Fig. 6 - - E f f e c t of panicle size on the roasting of natural arsenopyrite. VI. DISCUSSION The Fe-As-S-O system is inherently complicated due to the presence o f numerous volatile species. Arsenic in arsenopyrite can be removed to the vapor phase as any of its volatile compounds: As454, As2S3, AsS, As406, As4, As2, etc. The nature of these emissions depends on the oxygen, sulfur, and arsenic activities in the roasting environ- ment. Analysis o f the process is further complicated by the uncertainty in the available thermodynamic data. The information in predominance area diagrams, although repre- senting equilibrium conditions, can be used to evaluate roasting phenomena. As with most other gas solid reactions, the rate o f arsenic removal from arsenopyrite is enhanced by increasing tem- perature and available surface area as shown in Figures 5 and 6. It appears from Figure 5 that an increase in tem- perature from 798 K to 873 K increases the initial rate of arsenic release 1.4 times, for -150 +200 mesh FeAsS particles, roasted in an atmosphere where the CO2/CO ratio is 50. The influence o f temperature on the kinetics of the chemical process is actually greater than that shown in 246--VOLUME 14B, JUNE 1983 METALLURGICAL TRANSACTIONS B 1 0 0 1 w ' u ' J ~ ' I t t / t02 = ."p__o~ ~ . Pc0,zl PC0= 2 0 E 40 ~60~ / • Pc0/Pc0= 50 /"-J < 2 0 ~ [ - ] / A / ~ " A --1 / // O a ~ 7 l N O • a l J 201 I / a ~ • / • ' ° ~ / . , 1 Fig. 7 - The additionof CO tothe inert atmosphere significantly increases 0 0 ~ 3 06 0 9 0 1 2 0 the rate of arsenic removal• The experimental results were obtainedfrom synthetic arsenopyrite at 798 K andfor a particle size of -170 +200 mesh. 8 0 t r I temperature 798K • PCO = 1 atm. • PHe=1 arm., • PCO/PCO= 20 u - - n 60 • Pco/Pco ~ - g • • •/ / o f , , , / - _ . . . . • _ - o ~ ~o.,,,-. o ~ O • 0 0 , I , t , I , 30 60 90 120 Time (min.) Fig. 8 --Increasing the CO2/CO ratio decreases therateof arsenic removal toward that which occurs with an inert atmosphere. The results are for natural arsenopyrite with a particle size of -150 +200 mesh, except for the mineral roastedin an inert atmosphere. The latter results were obtained from synthetic arsenopyrite of -170 +200 mesh size. Figure 5. Increasing the temperature also increases the par- tial pressure of 02 (see Table III) which, as shown later, decreases the rate of As removal. Figure 6 shows that METALLURGICAL TRANSACTIONS B Time (min.) Fig. 9 - - A large increase in Po2 volatilizes As as As~O6 vapor and thereby increases the rate of As removal. IncreasingPo2 in a reducing atmosphere decreases the rate. The experimental results were obtained from natural mineral (-200 +270 mesh size). arsenopyrite particles of - 2 0 0 +270 mesh roast about twice as fast as those particles of -115 +150 mesh size. This variation in reaction rate is in reasonable agreement with the expected variation in surface area when the two sizes of particles are taken into consideration. Industrial practice indicates that the percentages o f As evolved from an arsenious concentrate varies significantly with the roasting environment.2Arsenopyrite was roasted in inert, reducing, and oxidizing atmospheres. The results are reviewed separately. Inert Roast. Under inert conditions thermal dissociation o f arsenopyrite is responsible for arsenic removal. Arsenic is evolved a s As4 vapor according to the reaction 4FeAsS(s) ~ 4FeS(s) + As4(g) [6] During the roasting process elemental As is precipitated in the cooler portions o f the furnace assembly. Further evidence o f that reaction was provided by X-ray analy- sis, which showed FeS to be the primary solid product. Theoretically, all the arsenic can be removed by reac- tion [6]. However, as a practical matter the process is slow (see Figure 7). The generation o f arsenic sulfide vapor, while possible, should not occur in significant quantities. Clark has shown that arsenopyrite can range in composition from FeAs0.9Sl.1 to FeAsHS0.9 .3° With this information and assuming that FeS is the only solid formed, then from a mass balance the maximum percent of arsenic, in sulfur rich arsenopyrite, evolved as arsenic sulfides is 11 pct. Since the predom- inance area diagrams in Figures 3(a) and 3(b) indicate that the vapor in equilibrium with arsenopyrite is rich in arsenic, the excess sulfur in FeAs0.9Sl.1 is most likely retained in the Fe~_xS phase than evolved as arsenic sulfide vapor• VOLUME 14B, JUNE 1983--247 Table II. Summary of Experimental Results Experimental Condition Solid Roasted Product Color* of Distilled Product Remarks inert atmosphere; 798 K Pco = 1 and 0.5 atm; 798 K PcoJPco = 20; 798 and 873 K Pco2/Pco = 50; 798 and 873 K Po2 = 0.1; 873 K pyrrhotite black Fe304 yellow and black Fe304; wus t i t e orange to red Fe304; FeAs orange to red Fe203 white elemental As evolved blackcolorof distilled product associated with carbon deposition *The color of the distilled product can be used to identify the form of As evolved: Black--elemental arsenic; yellow, orange, and red--arsenic sulfides; white-- arsenic oxides. Table I lL Partial Pressure of 02 Associated with Reducing Environments Temperature (K) CO2/CO Po2(atm) 798 50 3" 10-2s 20 5" 10-26 -10 (Pco = 1 atm)* 1 • 10-26 6 (Pco = 0.5 atm)* 5 ° l0 27 873 50 5" 10-22 20 8" 10-23 *Partial pressures listed are the initial values for C O entering the reactor. The COz/COratios listed are calculated from the CO2, CO, C equilibrium. In a dynamic system, such as that used in this inves- tigation, the vapors evolved from the mineral specimen are continually purged from the reactor, and since only a thin layer of powdered mineral was used, the specimen is in continuous contact with a fresh inert gas stream. The partial pressures of arsenic and sulfur above the mineral specimen are decreased by the purging action; thus, equilibrium be- tween the vapor phaseand arsenopyrite no longer exists, and reaction [6] proceeds. An inert roast, however, does not guarantee the thermal decomposition of arsenopyrite by reaction [6]. The presence o f sulfur-bearing species such as pyrite in a static bed may restrict the evolution, in an inert roast, o f arsenic from arsenopyrite. Pyrite can provide a high value of Ps2 upon decomposition to pyrrhotite, and as indicated by the pre- dominance area diagrams, the high value of Ps: can trap arsenic in a S-As liquid, thereby preventing escape o f ar- senic to the vapor phase. A static bed, such as that found in a hearth roaster, allows for build-up o f both the S and As content of the gas phase due to the inability o f gaseous species to diffuse through the bed rapidly. Hence, in hearth roasting under inert conditions S-As liquid formation may occur. Phase diagrams for the S-As system developed by Barton show that approximately 50 at. pct As can be re- tained in the liquid at 798 K. Sulfur-arsenic liquid formation did not occur in this in- vestigation. This effect is attributed to the shallow beds and the purging action o f the gas stream. Reducing Atmosphere. Roasting under a reducing atmo- sphere increased the rate and percent o f arsenic removal in comparison with an inert atmosphere. The influence, on the roasting process, o f the addition o f CO and CO2 to the inert atmosphere is shown in Figures 7 and 8. The equilibrium partial pressure o f 02 associated with each gas mixture is listed in Table III. While the addition of CO to the inert atmosphere increased the rate o f arsenic removal, sub- sequent addition of CO2 decreased the rate. The data in Table II indicate that with the addition of CO and CO2 to the roasting environment, two major changes occur. First, the evolution o f elemental arsenic with an inert atmosphere is replaced by the volatilization of arsenic sul- fides. The sulfides formed a red to yellow condensate in the cooler portions of the furnace and were easily identifiable. Condensation of elemental arsenic no longer occurred. The second change involves the solid reaction product. The addi- tion o f CO and CO2 to the roasting environment resulted in the formation of wustite and Fe304. In an inert atmosphere the solid product was pyrrhotite. The use of CO and CO-He gas mixtures resulted in carbon deposition and CO2 formation. These gas mixtures, as shown in Figures 7 and 8, resulted in the removal of more than 60 percent of the arsenic in less than two hours. The percentage o f As removal may have been less because the amount o f carbon deposited on the roasted product is un- known, yet X-ray analysis did not reveal the presence of carbon in the roasted product. Thus, any error associated with carbon deposition and the percent of arsenic evolved should be insignificant. The addition o f CO2 to the reactant gas raises the partial pressure of 02. The increase in Po2, as shown in Table III, is slight. However, this increase in the value o f Po:, however slight, significantly reduced the rate of arsenic removal as shown in Figure 8. The experimental results in Figure 8 reveal that the amount o f As evolved after two hours de- creased from 60 percent with pure CO (Po2 = 1 • 10-26 atm) to 23 pct with a reactant gas having a CO2/CO ratio o f 20 (Po2 = 5 " 10 -26 atm). The amount of As evolved was decreased further to 15 pct by increasing the CO2/CO ratio to 50 (Po: = 3 • 10-25 atm). The presence o f oxygen in the roasting environment re- sults in the release of sulfur to the vapor phase. This release occurs as As4S4, AszS3, and AsS. Furthermore, it is apparent from X-ray data that oxygen replaces sulfur in the roasted product or reacts with sulfur, forming SO2. The latter reac- tion leaves As, in the form of iron arsenides, in the solid product. The following reactions are thought to occur: FeAsS(s) + CO2(g) ~ FeO(s) + CO(g) + AsS(g) [7] 4FeAsS(s) + 4CO2(g) ~ 4FeO(s) + 4CO(g) + As4S4(g) [8] 3FeAsS(s) + 2CO2(g)~ 2FeO(s) + FeAs(s) + 2CO(g) + As2S3(g) [9] FeAsS(s) + 2CO2(g) ~ FeAs(s) + SO2(g) + 2CO(g) [lO] 248--VOLUME 14B, JUNE 1983 METALLURGICAL TRANSACTIONS B At temperatures below 833 K and/or at higher values o fPo2 wustite in reactions [7] through [9] is replaced by Fe304. The iron arsenide FeAs in reactions [9] and [10] is replaced by FeAs2 at higher values of PAs,, or by Fe2As at lower values of PAw Reactions [7] and [8] release As and are desirable if the loss of sulfur from a concentrate can be tolerated. Reaction [10] results in the retention o f the arsenic, while reaction [9] involves both retention and release of As. Arsenic is retained in the solid product by increasing the partial pressure of 02. Raising the value ofPo2by increasing the CO2/CO ratio decreases the partial pressure o f $2. The equilibrium for this process is represented by the follow- ing reaction: 4CO2(g) + S2(g) ~ 2SO2(g) + 4CO(g), / p \4p2 Keq = ( ~ CO ) r SO2 [11] \Pco2/ Ps~ Rearranging the equation for the equilibrium constant yields: Ps2 = P~oJ[Keq(Pco2/Pco)4] [12] At 798 Keq has a value o f 1.9 • 10-16. Increasing the COs/CO ratio from 20 to 50 decreased Ps2 by a factor of 39 assuming Pso2 is constant. The predominance area dia- grams in Figure 2 reveal that a decrease in Ps2 provides an environment in which iron arsenide is stable. Thus, any FeAs produced remains in the solid roasted product until either the partial pressure of As4is decreased or Po2 is raised. X-ray analysis of the roasted product revealed the presence of FeAs in specimens roasted in an environment with a CO2/CO ratio of 50. The arsenide was not found in appre- ciable quantities at lower CO2/CO ratios. Oxidizing Atmospheres. The Fe-As-O predominance area diagram in Figure 2 suggests that with high values o f Po2 arsenic can be retained in the solid product by formation of FeAsO4 or AssOs. Yet the experimental data in Figure 9 shows that arsenopyrite roasted in an ozidizing atmosphere (Po2 = 0.1 atm) resulted in the greatest arsenic removal. The predominance area diagrams for the Fe-As-O system in Figure 3 include the As/O isoplethsand the PrAs-° isobars. At 798 K and O2 partial pressures below 10-sl atm the isobars are horizontal and the arsenious vapor is essentially composed o fAs4. At values ofPo~ greater than 10TM atmthe isobars have a negative slope and the vapor consists essen- tially of As406. The near vertical nature o f the isobars re- flects this stability. The total vapor pressure which can be approximated as eAs406 increases rapidly with increasing Po2. The isobars in Figures 3(c) and 3(d) show that at a total pressure less than or equal to one atmosphere FeAsOa(s) and As2Os(s) are only stable at low partial pressures of As4. At 798 K and at an As4 partial pressure of 10-9 atm the total vapor pressure o f arsenic species rises rapidly at P02 values above 10-21 atm. At this partial pressure of 02 and at greater values of P02, the arsenic vapor is primarily As406. The partial pressure o f As406, given PAs4 ---- 10-9 atm, is 0.1 atmosphere at the Fe3Oa-Fe203 equilibrium. At the Fe203-FeAsO4 equilibrium, the vapor pressure o f As406 is well in excess of one atmosphere. The formation of FeAsO4 under the stated conditions is not possible unless the system is pressurized. Arsenopyrite at roasting temperatures, as indicated earlier, has a high arsenic activity which can promote As removal in an oxidizing atmosphere. Figure 3(a) shows that FeAsS exists at PAs4 values ranging from 10-3 to 1 atm at 798 K. At 973 K the range of partial pressure is less but the values are greater. Although the equilibrium values o f PAs4 noted above will not exist in a nonequilibrium process such as roasting, relatively highPAs4values can be expected. Such values of PAs4, as noted earlier, promote the volatilization o f arsenic a s A s 4 0 6 vapor at high pressures. The use in this investigation of a shallow bed and speci- mens rich in As prevented retention of As in the roasted product. The use o f a highly oxidizing atmosphere at a total pressure of one atmosphere prevented equilibrium from being established. The arsenic reacted with the O2 forming AS406 vapor. Because the equilibrium value o f PAs406 is above one atmosphere, As406 was rapidly volatilized and removed in the flowing gas stream. The partial pressure o f As4 in the gas stream should effectively approach zero due to reaction between As4(g) and excess O2(g)producing addi- tional AS406 vapor. The predominance area diagrams for the Fe-As-O system show that reduction o f the partial pressure of As4 leads to oxidation o f AsaO6(g) to As205(s). Oxidation of AS406 can also occur within the porous solid bed, if the bed is sufficiently thick. In that situation As is depleted from the upper portion o f the bed before it is totally evolved from the lower bed. The high value of Po2 and low value o f PAs4 in the upper bed favor As2Os(s) formation. These same conditions also favor reaction between As406(g) and Fe203(s) in the production of FeAsO4(s). Arsenate will form at higher partial pressures o f As4 than necessary for the formation o f AssOs(s). Assuming the kinetic processes are fast, arsenate in lieu of As2Os(s) should preferentially be formed in the bed. Once e i the r the solid oxide or arsenate is formed, the highly oxidizing condition existing in the roaster (Po2 = 0.01 to 0.1 atm) prevents their decomposition. The high partial pressure o f 02 is accompanied by low equi- librium partial pressures o f As406 and As4. Even if the chemical steps involved in the decomposition o f the solid oxide and the arsenate are fast, the rate o f mass transfer of both As406 and As4 from the solid phase is diminished by the low equilibrium concentrations o f both As406 and As4. In- creasing the roasting temperature increases the equilibrium concentrations o f AS406 and ASh and thus increases the rate of decomposition o f both the oxide and arsenate. VII. CONCLUSIONS A high percentage of As removal from an arsenious con- centrate during roasting appears possible. There are several economic advantages for early removal o f As in the smelting process and, as such, the authors direct their concluding remarks toward industrial practice. The reader is cautioned that some o f the following conclusionsapply only to arseno- pyrite and may not be valid for concentrates containing appreciable amounts o f other arsenic-bearing minerals. Temperature. Increasing temperature increases the rate o f As removal from arsenopyrite. The predominance area dia- grams also reveal that As2Os(s) and FeAsO4(s)become more unstable as the temperature is increased. While As2Os(s) decomposes at approximately 1000 K, decomposition of FeAsOa(s) requires a temperature in excess of 1200 K, a temperature well above the normal roasting temperatures o f 773 to 973 K. METALLURGICAL TRANSACTIONS B VOLUME 14B,JUNE 1983--249 While high temperatures may promote rapid As removal they also lead to larger reductions in the sulfurcontent ofthe concentrate. Such a loss o f fuel may not be tolerated in the converter. Roasting Environment. Inert, reducing, and oxidizing at- mospheres can be used to eliminate As from the concentrate. The last appears to be the most promising, but only when the bed thickness is small. The modified predominance area diagrams in Figures 3(c) and 3(d) reveal that As406 vapor can be readily volatilized without As2Os(s) or FeAsO4(s) formation if the arsenic potential is high. Accordingly, the blending of an arsenious concentrate with non-arsenic- bearing concentrates may actually hinder As removal by promoting FeAsO4(s) and As2Os(s) formation. Thermodynamic calculations indicate that in an inert a t m o s p h e r e A s 4 , A s 2 , and $2 are the primary vapor species in the Fe-As-S system. Experimental data confirm those calculations. When a reducing atmosphere is used, sulfur is also re- leased and arsenic sulfides are the major form of As emis- sion. The oxygen present in the reducing atmosphere reacts with Fe forming iron oxides. Increasing the partial pressure of 02 in the reducing atmosphere decreases the rate o f As removal. The added oxygen decreases the partial pressure of Ps2, which, in turn, leads to iron arsenide formation. Accordingly, the mixing o f an arsenous concentrate with excess powdered coal and the igniting o f the mixture in air may prove to be a practical means of As removal. Although increasing the partial pressure of 02 in a reduc- ing atmosphere was found to decrease the rate o f As re- moval, the predominance area diagrams in Figures 3(c) and 3(d) indicate that a further increase in the value of Po2 beyond 10-21 atmospheres (at 798 K) should lead to the volatilization of As as As406 vapor. The same diagrams also indicate that magnetite will also form in a reducing atmo- sphere. Extensive magnetite formation is undesirable as it leads to entrapment of concentrate and matte in slag in the reverberatory fumace. Bed Dimensions. The experimental and theoretical results indicate that bed thickness may be the most important parameter in the rapid and efficient removal o f As from a concentrate. Bed thickness should be minimized to decrease the residence time o f vapors in the bed. By so doing, the potential for reaction between the arsenic-bearing vapor and other compounds in the concentrate is decreased. Furthermore, the results of this investigation indicate that As depletion in the upper portion o f the bed can lead to oxidation of AS406 vapor to As2Os(s) or to reaction of the vapor with iron oxide in the production of FeAsO4(s). The experimental data indicate that formation o f those As retaining solids can be prevented by decreasing the bed thickness. Vapor Species. The various molecular forms for arsenic emissions may result in different operational problems for smelters processing an arsenious concentrate. Of all the volatile arsenic species elemental arsenic has the lowest vapor pressure, and it should be the easiest to precipitate and collect if its oxidation can be prevented. Given the large negative standard free energy for the formation of As203(s) and As20~(s), prevention o f oxidation does not appear possible. A similar situation exists with the arsenic sulfides. The collection of arsenic a s As203(s) in a baghouse or electrostatic precipitators may not be advisable depending on future environmental and work place emission standards. The vapor pressure of arsenic trioxide is 7.0.10 -6 atm at 400 K, the upper operating temperature for baghouses. Although that vapor pressure may seem small, the pressure corresponds to an arsenic concentration o f 6.4" 10 + 4 / z g o f As per m3 of gas, a value 6400 times the present work place environment limit in the United States. Conversion o f the trioxide to the pentoxide produces a stable solid compound not readily decomposed in oxidizing atmospheres. Arsenic pentoxide reacts with moisture from air forming a paste-like substance containing H3AsO4 and 3As205" 5H20.8 Both compounds are stable at temperatures up to 450 K. Generation o f large quantities of the pentoxide during roasting may lead to fouling o f flue ducts, fans, baghouses, and electrostatic precipitators. ACKNOWLEDGMENTS The financial support for this investigation given by the United States Bureau of Mines (contract number J0205066) is gratefully acknowledged. The authors would also like to express their appreciation to Mr. Kevin Lund and Mr. Donald Lewis for their technical assistance. REFERENCES 1. A.K. Biswas and W. G. Davenport: Extractive Metallurgy o fCopper, Pergamon Press, Toronto, 1976, vol. 20 , p . 224. 2 . I.J. Weisenberg, P.S. Bakshi, and A. E. Vervaert: J. Met., October 1979, vol. 31 , pp . 38-44. 3 . R.H. Hultgren, P. D. Desai, D. T. Hawkins, M . Gleiser, K. K. Kelley, and D. D. Wagman: Selected Values o f Thermodynamic Properties o f the Elements, ASM, Metals Park. OH, 1973, pp . 39-46. 4 . H. Rau: J. Chem. Thermodyn., 1975, vol. 7, pp . 27-32. 5 . J.J. Murray, C. Pupp, andR. F. Pottle: J. Chem.Phys., 1973, vol. 58 , pp. 2569-78. 6 . J. Drowart, S. Smoes, and A. Vanderauwera-Mahieu: J. Chem. Thermodyn., 1978, vol. 10, pp . 453-64. 7 . D.M. Dabbs: University of Washington, Seattle, WA, unpublished research, 1982. 8 . J .W. Mellor: A Comprehensive Treatise on Inorganic and Theoretical Chemistry, Longmans, Green and Co., London, 1929, vol. IX, pp . 139-41 and p . 265. 9 . S. Maske and B. J. Skinner: Econ. Geol., 1971, vol. 66 , pp. 901-18. 10. P.B. Barton, Jr.: Geochim. Cosmochim. Acta, 1969, vol. 33 , pp . 841-57. 11. Z.A. Munir, G. B. Street, and H. E Winters: J. Chem.Phys., 1971, vol. 55 , pp. 4520-27. 12. K.H. Lau: SRI International, Menlo Park, CA, unpublished research, 1982. 13. G.K. Johnson, G.N. Papatheodorou, and C.E. Johnson: J. Chem. Thermodyn., 1980, vol. 12, pp. 545-57. 14. G.B. Street and Z. A. Munir: J. Inorg. Nucl. Chem., 1970, vol. 32 , pp. 3769-74. 15. O. Kubaschewski and C. B. Alcock: Metallurgical Thermochemistry, 5th ed., Pergamon Press, New York, NY, 1979, vol. 24 , p . 185. 16. G.P. Utsyugov, A.A. Kudryavtser, and B.M. Kuadzhe: Neorg. Matr., Izr. Akad. Nauk. SSR, 1968, vol. 4 , pp . 1338-39. 17. B.A. Strathdee and L.M. Pidgeon: Can. Min. Metall. Bull., 1961, vol. 54 , pp . 883-87. 18. R.A. Isakova and V. N. Nesterov: Tr. Inst. Met. i Obogashch., Akad. Nauk. Kaz. SSR, 1962, vol. 5 , pp . 29-33. 19. C.M. Hsiao and A. W . Schlechten: J. Met., 1952, vol. 4 , pp. 65-69; also, Trans. AIME, 1952, vol. 194, pp . 65-69. 20 . R.C. Weast and M . J. Astle, eds.: CRC Handbook o f Chemistry and Physics, 60th ed., CRC Press, Boca Raton, FL, 1979, p . B-58. 21 . F.M. Faure, M.J. Mitchell, and R.W. Bartlett: High Temp. Sci., 1973, vol. 5 , pp . 128-37. 22 . D.C. Lynch: Metall. Trans. B, 1982, vol. 13B, pp . 285-88. 250--VOLUME 14B, JUNE 1983 METALLURGICAL TRANSACTIONS B 23 . K.C. Mills: Thermodynamic Data for lnorganic Sulphides, Selenides and Tellurides, Butterworth, London, 1974, p . 739. 24. R.G. Behrens and G.M. Rosenblatt: J. Chem. Thermodyn., 1972, vol. 4 , pp . 175-90. 25 . H. Gu6rin and R. Boulitrap: Compt. Rend., 1953, vol. 236, p . 83 . 26 . A.N. Polukarov, M . Zh. Makhmetor, and E.A. Buketov: Russ. J. Phys. Chem. (Eng. Trans.), 1971, vol. 45 , p . 1527. 27 . A. Vian, C. Iriarte, and A. Romero: Ind . Eng . Chem.Process Des. Dev., 1963, vol. 2 , pp . 214-23. 28. P.B. Barton, Jr. and P. Toulmin, III: Geochim. Cosmochim. Acta, 1964, vol. 28 , pp . 619-40. 29 . P. Toulmin, III and P.B. Barton, Jr.: Geochim. Cosmochim. Acta, 1964, vol. 28 , pp . 641-71. 30 . L.A. Clark: Econ. Geol., 1960, vol. 55 , pp . 1345-81 and 1631-52. 31 . D.C. Lynch: Metall. Trans. B, 1980, vol. liB, pp . 623-29. 32 . R.C. Weast and M . J. Astle, eds.: CRC Handbook o f Chemistry and Physics, 60th ed., CRC Press, Boca Raton, FL, 1979, p . D-45. 33. K. Denbigh: The Principles o f Chemical Equilibrium, Cambridge Univ. Press, 1955, pp . 180-93. 34. T. Rosenqvist: Principles o f Extractive Metallurgy, McGraw-Hill, New York, NY, 1979, pp . 63-70. 35 . S.I. Sandler: Chemical and Engineering Thermodynamics, John Wiley and Sons, New York, NY, 1977, pp . 310-14. 36 . A.D. Tevebaugh and E. J. Cairns: Final Technical Report; Section II Contract No. DA-44-009-ENG-4853, ARPA Order No. 247-61, Ft . Belvoir, VA, December 1962. 37. E.J. Cairns and A. D. Tevebaugh: J. Chem. Eng. Data, 1964, vol. 9 , pp. 453-62. 38. A.D. Tevebaugh andE. J. Cairns: J. Chem.Eng.Data, 1965, vol. 10, pp . 359-62. 39 . H.A. Liebhafsky andE. J. Cairns: Fuel Cells andFuel Batteries, John Wiley and Sons, New York, NY, 1968, pp . 463-67. 40 . K. Natesan: MFPG Conference Proceedings on Prevention o fFailures in Coal Conversion Systems, U.S. Dept . of Commerce, N. B. S. Special Publication 468, April 1976, pp . 159-67. 41 . O.K. Chopra and K. Natesan: High Temp. Sci., 1977, vol. 9 , pp . 243-56. 42 . S. Mohnot andB. G. Kyle: Ind .Eng .Chem.Process Des. Dev., 1978, vol. 17, pp . 270-72. 43 . S. S chechter andH. Wise: J . Phys. Chem., 1979, vol. 83 ,pp . 2107-11. 44 . M . Donley and Y.K. Rao: J. Cryst. Growth, 1982, vol. 58 , pp . 273-78. METALLURGICAL TRANSACTIONS B VOLUME 14B, JUNE 1983--251