Solid state Na-β-alumina potentiometric sensor for measuring gaseous arsenic oxides

Solid State Ionics 59 ( 1993) 109-I 15 North-Holland

Solid state Na- p-alumina potentiometric for measuring gaseous arsenic oxides

sensor

J. Kirchnerova and Christopher W. Bale ’ Departmentof MetallurgicalEngineering, Ecole Polytechnique,Campus de ITJniversitkde Montrkal, P.O. Box 6079, StationA, Montreal, QC, Canada H3C 3A7 Received 16 July 1992; accepted for publication 1 September 1992

Na-palumina solid electrolyte in the form of a closed-end tube was employed as a potentiometric sensor for measuring gaseous arsenic oxides. The behaviour of the sensor was tested by measuring the EMF of two cells: Pt, air 1Na++alumina 1Pal, P&n,. Pt (cell I) as a function of PM, and PO*;and Pt, air, P- lNa+-8+thtminal P-, air, Pt (cell II) as a function of Pm, when P,, = constant, and as a function of P,,,, when PAJHs = constant. In both cells the measurements were made at 1025 and 1075 K. A linear response to Pm, and to Paa was observed over the concentration range 5 to 2000 ppm AsHs and 1 to 180 ppm As40s. The cells were found to be more sensitive to arsine than to As40s. The oxidation of As40s to pentoxide required an efficient catalyst. The EMF dependence on arsenic concentration for the two cells, as well as the EMF dependence on Pa of the ceil (I), suggest that at 1025 K the formation of pyroarsenate is the reaction which defines the electrode potential. At 1075 K the results indicate that the controlling reaction is the formation of orthoarsenate.

1. Introduction Arsenic is an impurity in certain metallurgical reserves (primarily Cu, Ni, Au and Ag orebodies) and as such it is a potential source of unwanted gaseous compounds. In pyrometallurgical processing it may form volatile oxides such as arsenic trioxide Asd06. Alternatively, in hydrometallurgical treatment or electrochemical refining it may produce amine AsH3. Arsine is a highly toxic gas with a threshold limit value (TLV) of 0.05 ppm. Clearly for environmental and security reasons, it is desirable to continuously monitor arsenic concentrations in the vicinity of metallurgical operations. In our laboratories we have been performing a systematic study on cationic conducting ceramic electrolytes for possible use as potentiometric sensors of non-metallic species. We have employed various ceramic electrolytes from the NASICON and P_alumina families in order to monitor arsenic-bearing gases [ l-3 1. A patent has been granted [ 41 on the use of Ag-P-alumina as a potentiometric sensor for Author to whom all correspondence should be addressed.

arsenic in pyrometallurgical processing. This paper presents the results where Na+alumina has been used to measure arsenic concentrations in oxygenbearing atmospheres, primarily in air.

2. Na-&alumina solid electrolyte

Severalyears ago, Gauthier et al. [ 51 proposed that alkali metal oxyanion salts could be employed as solid-state sensors for gaseous oxides. For example, NaSAs04 could be user! as a sensor for arsenic oxides although it was never tested on amine. Unfortunately, alkali salts possess several properties which make them unlikely candidates for industrial sensors. These properties include low melting points, high porosity to gases, poor mechanical integrity, and a tendency to be hygroscopic. These problems may be resolved in part by employing an alkali metal-conducting ceramic. For example, Na+ alumina could replace NaSAsO,. Unlike the sodium oxyanion salt, Na-P-alumina is a ceramic material which as a potentiometric membrane would offer a simpler design, a much higher

0167-2738/93/S 06.00 0 1993 Elsevier Science Publishers B.V. All rights reserved.

110

J. Kirchnerova, C. W. Bale/Solid state Na-P-alumina potentiometric sensor

mechanical stability, and a lower sensitivity to humidity. It has already been demonstrated that Na-b-alumina is suitable not only for the direct measurement of sodium activities in various environments [ 6-81, but also for monitoring indirectly several different gases, including oxygen [9-l 21. The function of NaB-alumina as a non-sodium bearing gas sensor depends on the reaction of the given gas and sodium at the electrolyte interface to form a stable compound. Through this reaction the activity of sodium at the interface and thereby the potential of the sensor are defined. Itoh et al. [ lo] have obtained good results on the performance of this electrolyte as a sensor for SO2 in air, using a gold-sodium (Au-Na) alloy as a new reference system. We have recently identified Na-P-alumina as potentially suitable for monitoring arsenic oxides [ 11. The intention was to develop a potentiometric cell for the determination of thermodynamic constants of various base-metal arsenates. The performance of Na-B-alumina as an arsenic oxide sensor was evaluated by measuring at 1025 and 1075 K the EMF values of cell (I), in which arsine gas (ASH,) was used as the source of arsenic oxide: Pt, air) Na-b-alumina IPO,, PAsH, ,Pt .

(1)

Similar experiments were carried out on cell (II ) , in which both arsine gas and arsenolite vapor ( AsdO ) were the sources of arsenic oxide: air, Pt Pt, air, PAsris(Na-B-alumina 1PAwo6,

(II)

Sodium oxide reacts with arsenic pentoxide to form several stable arsenates [ 13,141. Although no reliable thermodynamic data for their stability are available, from the melting points the formation of sodium pyroarsenate (Na4As207) may be expected to control the electrode reaction up to its own melting point _ 1118 K. At higher temperatures and/or very low arsenic pentoxide activities, the formation of sodium orthoarsenate (Na3As0,) should be the controlling reaction. Based on the literature information about gaseous arsenic trioxide [ 15 ] and on our earlier work on arsenic oxide sensors [ l-3 1, we can assume that at high temperatures ( > 1020 K) Asz05 is the predominant gaseous arsenic oxide species participating in the

electrode reaction. In this case the two possible electrode reactions can be written as: 4Na++AszOS(g)+02(g)+4e-=Na4As207(s), (1)

(2)

Arsine is unstable and at high temperatures it rapidly reacts with oxygen to form arsenic oxides. In principle both arsine gas and arsenolite vapor could be employed to produce the same final arsenic oxide composition. The total arsenic oxide concentration, expressed for simplicity in terms of species As,O,, where P&,=0.5 PAsH,(initial) or Pj_,,,,=2 PAs406 (initial), can be related to the partial pressure of arsenic pentoxide by means of the equilibrium constants KD and K,+i. These constants are for the dissociation of AsdOe to Asz03 and for the oxidation of the monomer to higher oxides respectively [ 2 ] :

X

(

l+ i KJ+iPg: .

i=l

>

(3)

Eq. (3) is based on the assumption that the concentration of the diarsenic oxide species is preponderant over tetraarsenic and monoarsenic oxides. Regardless of the values of the oxidation constants and the total number of different oxides, for a constant PO,the concentration of arsenic pentoxide can simply be expressed as: PLO5 = i3P&OX>

(4)

where j? is a constant. Consequently the potential at the working electrode can be written as: E w= aAs

+ bAs lot? &ox

,

(5)

where the constants aASand bAs depend on the reaction at the electrode and are respectively related to the Gibbs free energy of reaction and its stoichiometry. For reaction ( 1) bAs should equal to 2.3RTl 4F,whereas for reaction (2) bA,should equal to

2.3RT/6F. On the other hand, for constant P&,, the electrode potential E" is a complex function of PO,which only becomes linear for limiting cases such as K,+2P02 >> 1 or Z:=r K3+iPg:-X1:

J. Kirchnerova, C. W. Bale/Solid state Na-/l-alumina potentiometric sensor

EL,. = a02 + bo, logPO,.

(6)

In such a particular case, the value of the constant b,, would indicate the stoichiometry with respect to oxygen. While the expression for E” should be the same for the two cells, the equations for the potential E’ at the reference electrode are different. For the cell (I) the reference potential Ef is given by: E,‘= AG”(Na20)

+ 2.3RT

2F

2F

log dL0

2.3RT

-

~l%vO,.

(7)

The literature values for AG”(Na20) [ 161 and for log u&o (the activity of sodium oxide in S-alumina) [17] leadtoE;=-3241.6+0.6986TmVfor Po2 co.21 atm. It follows that the EMF of cell (I) can be expressed in terms of P.&,, as: EMFI = ( C, )~s - b*s log &ox where (CI),,=Ef-a,_+,,

2

(8)

or in terms of Paz:

EM% = ( G )oz - bo, log Paz ,

(9)

where (Ci)o,=Ei-6,. In the case of cell (II) the expression for the reference potential EfI is similar to that for the working potential EP, namely: EiI =UA~ + bAs1% (PX20x) r .

(10)

For the cell (II), when at both electrodes the same values of PO, are used, EMFii is given by: EMFu=b,,log

t&o,)‘-b,sb

When (PLO, ) r= (f%,~~ ) ‘“,

(f’isz~x)~.

(11)

EMFii should be zero.

3. Experimental The experimental procedure and the sensor design were similar to those successfully employed in an earlier study on a Ag-B-alumina potentiometric sensor [1,3]. The sensor (fig. 1) was constructed from a Na+ alumina closed end round-bottom tube, 3 cm long,

111

0.35 cm outside diameter with 0.1 cm wall thickness. The tube was connected to a four-bore alumina tube 0.625 cm diameter and 40 cm long. Both tubes were sealed together by COTRONIX 940 fast cure ceramic adhesive. Both sides of the electrolyte surface were platinized by using 0.1% hexachloroplatinic acid in an aqueous solution, and lined with platinum mesh held in contact with platinum wires. The Pt/Pt-10% Rh thermocouple was placed in contact with the inner surface (which served as a reference) with the wires passing through two of the holes of the alumina tube. The remaining two holes were furnished with a piece of stainless steel tubing 0.156 cm ( 1 / 16” ) in diameter connected to 3 16 SS SWAGELOCK fittings. An alumina tube (not shown in fig. 1) 0.625 cm diameter and 40 cm long was used to direct the gases to the exterior electrode surface. The assembled sensor was positioned in a closed alumina tube 60 cm long and 3.1 cm in diameter with an opening for the entry and exit of gas mixtures. The exterior alumina tube was shielded against electrical field effects by means of a grounded stainless steel screen. The assembled galvanic cells were heated in a Lindberg Heavy Duty horizontal furnace where the temperature was controlled to within 1“C by a Hewlett-Packard 240 temperature programmer. Temperatures of the galvanic cells were measured with type S (Pt/Pt-10% Rh) thermocouples, read to within 1“C with a calibrated Doric Trendicator 402 digital thermocouple meter. EMF values were measured to within 0.1 mV by a TACUSSEL Aries 20000 millivoltmeter (impedance 1012fi) or a Keithley high impedance ( lOi a) 6 15 Electrometer. Both the temperature and the EMF values were recorded on a SOLTEC 1242 dual pen recorder. Arsine-oxygen-nitrogen mixtures were prepared by mixing arsine diluted in nitrogen (98 and 5020 ppm - analyzed by the supplier, Union Carbide) with air and oxygen. Arsenic trioxide-air mixtures were prepared by passing air through solid arsenolite held in a stainless steel tube 15 cm long and 2.5 cm diameter maintained at a constant elevated temperature by means of heating tape. Approximately 1 cm thick bed of arsenolite powder mixed with alumina beads (0.1 to 0.2 cm in diameter) was placed at the exit of the res-

J. Kirchnerova, C. W. Bale/Solid state Na-/l-alumina potentiometric sensor

112

COTRONIX SEALANT

Pt

(4

/ PI - 10% Rh

I

Ai203

TUBE

Pt

&UMINA AND

FLACK Pt MESH

Fig. 1. Schematic representation of the (a) sensor; (b ) complete assembly.

ervoir between two layers of a purified glass wool. A glass shielded chrome&alumel thermocouple with the tip positioned in the center of the arsenolite bed was used to measure the temperature to within 0.1 ‘C. The arsenolite bath was connected to the gas entrance of the cell assembly by means of a stainless steel tubing 0.3 17 cm ( 1/ 8 II) in diameter coupled with SWAGELOCK fittings. This part of the gas line was heated by means of heating tape held at 10°C above the temperature of the arsenolite reservoir. Gas flowrates were adjusted from 0.5 mQ/s to 1 mQ/s. The partial pressure of PAs406was calculated by using the equation by Behrens and Rosenblatt [ 18 ] and the measured temperature of the arsenolite reservoir.

4. Results and discussion Before each experimental run with arsine or arsenic trioxide, the sensor was heated for several hours first at 923 K and then to 1025 K in a flow of dry air at each electrode in order to obtain a steady EMF ( + 10 mV) . A relatively high sensitivity to the flowrate of air at the external electrode was observed. It was then necessary to “As activate” the electrodes, i.e. to form in situ a layer of the arsenate by an exposure to an arsenic oxide-bearing atmosphere. Initially no additional catalyst was used to promote the arsenic pentoxide formation. Under these conditions an attempt to measure the response to

113

J. Kirchnerova, C. W. Bale/Solid state Na-b-alumina potentiometric sensor

PAs,os = 31.1 ppm in the cell (Ia), cell (I) with As,O, as the source of AslOX, failed. This agreed with the results obtained earlier in experiments with Ag-P-alumina electrolyte [ 21. On the other hand, when arsine was used as a source of arsenic oxide the sensor responded im= 501 ppm and PO,= 0.9atm mediately. With PAsH, a steady EMF value of cell (I) was obtained in 5.5 h. This value remained reproducible ( + 10 mV) during experimental run which lasted several days. The response time to temperature and concentration changes was longer than with Ag-P-alumina [ 21, but within 30 mitt, and steady EMFs were obtained. This response time compares well with that observed by Itoh et al. [ 10 ] who used Na-P-alumina as a sensor for SO2 gas. Although the cell used in the present study was much less sensitive to flowrates than the one consisting of two air electrodes, the effect was not negligible, particularly at higher arsine and/or oxygen concentrations. The upper flowrate limit in our experiments was set to 1 mQ/s. The dependence of EMFi on arsine concentration in synthetic air (Po, co.21 atm) at 1025 and 1075 K is represented in fig. 2. It can be seen that at both temperatures a linear response was obtained over a wide range of concentrations although the slopes are different. At 1025 K the slope of 54 mV corresponds within the experimental error to 2.3RT/4F,suggesting reaction (1) as the controlling step. At 1075 K the observed slope was 70.5 mV corresponding to 2.3RT/3F, i.e. a value twice as large as the theoretical value for reaction (2). This agrees with our ear-

.

1025

0

1075K

3Na + AsOz + O2 = NaSAsO, ( s)

(12)

or alternatively, the existence of AsO~.~species. In fact, a monoarsenic oxide species has been observed by mass spectrometry above CaS(AsO,)* [ 191 and it has also been proposed as taking part in the mechanism of the dissociation of some orthoarsenates [ 201. Furthermore, at both temperatures the electrode reaction appears to be kinetically controlled. Similar behavior was observed in the case of the AgP-alumina sensor [ 2 1. In a second series of experiments carried out with cell (II), a catalyst consisting of a layer of purified asbestos fibers impregnated with platinum was placed at the end in the gas-inlet tube. In this experiment

K

Regression

-

-5.6

lier results obtained with NaCaAsO, electrolyte [ 11. Fig. 3. represents the dependence of EMFr on log PO, for two different arsine concentrations each at 1025 K and 1075 K. At 5 ppm A&IS, the behavior is linear with the slope of 55 mV at 1025 K suggesting again the reaction ( 1) as the controlling step, with K,,, PO,x= 1. At higher arsine concentration the EMFi versus log PO, starts to become nonlinear, indicating that K3+2PO,> 1 only for PO2> - 0.4atm. At 1075 K the limiting slope of the above function is 105.8 mV corresponding to 2.3RT/2Finstead of the expected value 2.3RTl3F. The unexpectedly high slopes bA, and b,, observed at 1075K suggest that the electrode potential is controlled by the following reaction to form sodium orthoarsenate:

-5.2

-4.6

-4.4 log

-4.0

-3.6

PAsH3

Latrnl

-3.2

-2.6

-2.4

Fig. 2. EMF, of the cell (I) as a function of log PAti) in synthetic air at 1025 and 1075 K.

600 -1.4

’ -1.2

-1.0

-0.8 log

PO,

-0.6

-0.4

-0.2

1 0.0

[atml

Fig. 3. EMF, of the cell (I) as a function of log PO, at 1075 K for 301 ppm and 502 ppm AsHI, and at 1025 K for 5 ppm and 9.8 ppm AsI%

J. Kirchnerova, C. W. Bale/Solid state Na$-alumina potentiometric sensor

114

the sensor assembly was first cooled to room temperature, and then again heated up to equilibrate at 1025 K for several hours under a flow of dry air at each electrode. A stable value of 520? 10 mV was obtained. Thereafter, the working electrode was allowed to equilibrate at the flowrate of 0.5 mQ/s with dry air containing 7.7 ppm As,O,. In this second series of experiments the sensor responded immediately and after about three hours a steady value of 599.1 mV was observed (cell (Ia) ). This value remained constant within 2 mV over a period of approximately eight hours. Arsine was then introduced at the internal (reference) electrode. When a new steady EMF was established the EMF of cell (II ) was measured at 1025 K as a function of PAsH, for two different values of PAs40s(expressed as log (Pi,,) r), and as a function of PA-o6 for PAdis = const. Similar measurements were repeated at 1075 K for three different values of log (PLO, ) r . The results are represented in figs. 4 and 5 for 1025 K, and in fig. 6 for 1075 K. The results indicate a linear dependence and the slopes at both temperatures are the same as those observed for EMFr, i.e. 5 1 mV and 7 1 mV for 1025 K and 107 5 K respectively. However, non-zero values of EMF were recorded at the points of equal (&o, ) r and ( PLz~x ) w or log (P~so,)l. The deviations seem to be dependent on (P14820x)r,indicating again that the electrode reactions are kinetically controlled with different mechanism for arsine and arsenolite. A knowledge of E; permits one to evaluate AGR

20 0 -20

e

-40 -60

-6.0

-5.2

-5.0

-4.0

-4.6

-4.4

-4.2

-4.0

log PbH,[atml Fig. 5. EMFIl of the cell (II) at 1025 K as a function of log PAs406for log (PASH,/4),= -5.81.

00

60 40 r=

20

L

0

k lAJ -20 -40 -60

-5.6

-5.4

-5.2

-5.0 log

-4.6 P,,,

-4.6

-4.4

-4.2

-4.0

[atml

Fig. 6. EMFIl of the cell (II) at 1075 K as a function of log PAs”,.

for the two deduced electrode reactions from both EMF, versus log PAsH3and EMFi versus log PO,. This leads within 5 kJ to identical values of - 1350 kJ for reaction ( 1) at 1025 K, and - 1040 kJ for reaction (2) at 1075 K. Both values of AGR are reasonable, but further studies are required in order to isolate the actual electrode reactions and to determine the thermodynamic constants of all equilibria involved. Only then can one confidently employ the cell in thermodynamic measurements.

40

!A

-5.4

-5.6

-5.2 k3

-4.0 Pb,p,

-4.4

-4.0

-3.6

[atml

Fig. 4. EMFIr of the cell (II) at 1025 K as a function of log Pm, forlog (PA,),=-5.84, -5.25, and -4.97.

5. Conclusions

The experiments carried out in this study confirmed that Na-P-alumina would be suitable as po-

J. Kirchnerova, C. W. Bale/Solid state Na-j&alumina potentiometric sensor

tentiometric sensor for gaseous arsenic oxides. A linear EMF response was observed over a wide range of arsenic (oxide) concentrations, but with different slopes for 1025 K and 1075 K, suggesting different electrode reactions at the two temperatures. The results indicate that the electrode reaction defining the potential is kinetically controlled and that different mechanisms of arsine (ASH,) and of arsenic trioxide (As,O,) oxidation to pentoxide ( AszO,) are responsible for the apparently different sensor sensitivity to these gases. To permit the use of the system for the thermodynamic measurements, further studies are required in order to determine the equilibrium conditions and to evaluate the thermodynamic data for gaseous arsenic trioxide oxidation. Acknowledgements

This work was in part performed for CANMET, Department of Energy, Mines and Resources, Ottawa, Canada, under DSS contract no. 2 1ST. 234404-92 13. Partial financial assistance from the Natural Science and Engineering Council of Canada is also gratefully acknowledged. The authors are indebted to Dr. J. Skeaff (CANMET) for his many discussions. References [ 1] J. Kirchnerova and C.W. Bale, Development of a Solid State Arsenic Probe for Use in Pyrometall. Processes, Final report contract 23440-4-9213 (Dept. of Energy, Mines and Resources, Ottawa, 1986 ) .

115

[2] J. Kirchnerova, C.W. Bale and J.M. Skeaff, J. Electrochem. sot. 137 (1990) 3505. [ 31 J. Kirchnerova, C.W. Bale and J.M. Skeaff, Sensors Actuators 2 ( 1990) 7. J. Kirchnerova, C.W. Bale and J.M. Skeaff, Solid State Arsenic Probe for Use in Metallurgical Processes, U.S. Patent no. 4,842,698 (June 1989). 1M. Gauthier and A. Chamberland, J. Electrochem. Sot. 124 (1977) 1579. D.J. Fray, Metall. Trans. 8B (1977) 153. A.DubreuilandA.D. Pelton, in: Light Metals 1985,ed. H.M. Bohner (AIME, Warrendale, Pa, 1985) p. 1197. [ 8 ] 0. Takikawa, A. Imai and M. Horata, Solid State Ionics 7 (1982) 101. [9] D.E. Williams, Brit. UK Pat. Appl. G.B. 2,119,933 (1983). [ lo] M. Itoh, E. Sugimoto and Z. Kozuka, Trans. Jap. Inst. Met. 25 (1984) 504. [ 111 T. Ogata, S. Fijitsu, M. Miyayama, K. Kuomoto and H. Tanagita, J. Mater. Sci. Lett. 5 (1986) 285. [ 12 ] K.T. Jacob, M. Iwase and Y. Waseda, Adv. Ceram. Mater. 1 (1986) 264. [ 131 B.K. Kasenov, CM. Isabaev and A.N. Polukarov, Zh. Fiz. Khim. 53 (1979) 2173. [ 141 C.M. Isabaev, E.A. Buketov and B.K. Kasenov, Zh. Neorg. Khim. 27 (1982) 3163. [15]H.Blitz,Z.Physik.Chem.l9(1986)417. [ 161 C.B. Alcock and G.P. Stavropulos, Can. Met. Quart. 10 (1971) 257. [ 171 A. Dubreuil, M. Malenfant and A.D. Pelton, J. Electrochem. Sot. 128 (1981) 2006. [ 181 R.G. Behrens and G.M. Rosenblatt, J. Chem. Thermodyn. 4 (1972) 175. [ 191 J. Drowart, in: Int. J. Mass Spectrom. Ion Phys. 14 (1992) 243. [20] J. Kubo, K. Shigematsu and T. Ishihara, Nippon Kagaku Kaishi 119 (1981) 1691.