270
Anal. Chem. 1980, 5 2 , 270-273
transient, the absorbance a t zero time obtained by extrapolation is directly proportional to the bulk concentration of MB and it agrees well with the calculated value. For example, the calculated absorbance of 4 x 10l6M MB at 45' is 1.28 in Table I, and the transient absorbance a t zero time is 1.27. The diffusion coefficient of MB calculated from Equation 5 is 4.7 X lo4 cm2 s-l, which corresponds reasonably well to the polarographically measured value of 4.1 X lo4 cm2 s-' (16). I n contrast to the ideal transients observed a t lower MB concentrations, the transient for 1 X M MB begins to deviate from theory after ca. 0.5 s, and the deviation increases progressively with time. We attribute this artifact to precipitation of LMB on the electrode surface. Pergola et al. (16) have shown that the solubility of LMB is approximately 7 X lo4 M, a concentration that is quickly exceeded in the reduction of 1 X M MB. Visual observation of the MSR cell during the experiment confirms that LMB precipitates. Shortly after application of the cathodic potential step, the beam a t the exit prism becomes noticeably more diffuse, indicating an increase in scattering which does not occur a t the lower MB concentrations. Transients for oxidation of LMB a t concentrations from 5X M to 1 X M generated by first reducing MB for 30 s a t 4 . 7 V and then stepping the potential to +0.1 V are plotted vs. t1Izin Figure 3. Blank experiments without MB gave no detectable absorbance transients. For MB concentrations up to and including 6 X lo4 M, the transients conform well to Equation 4. From 10 ms to ca. 3 s, the plots are linear with slopes proportional to the bulk concentrations of LMB and they extrapolate to the origin. The diffusion coefficient of LMB, calculated from Equation 4, is 4.0 X lo4 cm2 s-'. Although the transients for 8 X lo4 M and 1 X M LMB are linear, they extrapolate to finite absorbance a t zero time, an effect we attribute to the precipitation of LMB on the electrode surface. Because the precipitate causes scattering of the light beam, it is not possible to define the true absorbance a t the beginning of these transients, but it is significant that after 10 ms, both transients have slopes that correspond to a LMB concentration of ca. 7 x lo4 M, Le., a saturated solution of LMB (16). Apparently the precipitated LMB is reoxidized within the first 10 ms, after which the transient
is governed by diffusion of LMB in the usual manner. T h e electrodes in these experiments were gold, but there is no reason to believe that other conductors or semiconductors would not be suitable provided t h a t the surface is flat, nonabsorbing, and highly reflective. Preliminary experiments with both platinum foil and platinum films sputtered on glass indicate that these materials are satisfactory for MSR cells. Attempts to replace the laser with a focused xenon lamp gave greatly decreased reproducibility, presumably due to optical beam divergence within the cell. We believe that the use of a laser as the analyzing source is probably necessary for this reason. In addition to their use in monitoring reactants and products in electrochemical processes, MSR cells may prove useful for detecting short-lived intermediates. For such applications, minimizing the cell thickness and increasing the angle of the light beam relative to the electrode surface should prove advantageous in maximizing the sensitivity.
LITERATURE CITED (1) R. L. McCreery, R. Pruiksma, and R. Fagan. Anal. Chem., 51, 749 (1979). (2) D. E. Albertson, H. N. Blount, and F. M. Hawkridge, Anal. Chem., 51, 556 (1979). (3) E. A. Blubaugh, A. M. Yacynych, and W. R. Heineman, Anal. Chem.. 51, 561 (1979). (4) T. Kuwana and W. R. Heineman, Acc. Chem. Res., 9 , 241 (1976). (5) A. Prostak, H. B. Mark, Jr., and W. N. Hansen, J. Phys. Chem., 72, 2576 (1968). (6) T. Kuwana, Ber. Bunsenges. Phys. Chem., 77, 858 (1973). (7) W. R. Heineman, Anal. Chem., 5 0 , 390A (1978). (8) R. Cieslinski and N. R. Armstrong, Anal. Chem., 51, 565 (1979). (9) P. T. Kissinger and C. N. Reilley, Anal. Chem., 42, 12 (1970). (10) D. C. Walker, Anal. Chem., 39, 896 (1967). (11) M. A. Barrett and R. Parsons, Symp. Faraday Soc., 4, 72 (1970). (12) W. J. Plieth. Symp. Faraday Soc., 4 , 137 (1970). (13) B. D. Cahan, J. Horkans, and E. Yeager, Symp. Faraday SOC.,4, 36
,iiq7nl .". ",.
(14) T. Takamura, K. Takamura, W. Nippe, and E. Yeager. J . Electrochem. Soc.. 117, 626 (1970). (15) R. Kayser, Ph.D. Dissertation, Georgetown University, Washington, D.C., 1975. (16) F. Pergola, G. Piccardi, and R. Guidelli, J . Electroanal. Chem., 83, 33 (1977).
RECEIVED for review July 25, 1979. Accepted November 8, 1979. This work was supported in part by the Office of Naval Research. C.E.B. thanks Corning Glass Works for fellowship support.
Palladium-Palladium Oxide pH Electrodes W. T. Grubb" and L. H. King General Electric Company, Research and Development Center, P. 0. Box 8, Schenectady, New York 1230 1
Wire electrodes for sensing pH are produced by coating palladium wires with sodium hydroxide and then oxidizing In oxygen at 800 OC. After suitably lnsulatlng to define the active area and cover up bare palladium, the wires are useful for pH measurement approximately from pH 3 to pH 11. The electrodes are stable for more than 6 years In distilled water storage with no essential change in pH response. They are most useful under mildly oxidizing conditions, e.g., aerated aqueous solutions. Strong reducing agents such as ascorbic acid and active redox couples such as ferro-ferricyanide destroy the pH response, but the electrodes eventually recover when stored in neutral aqueous media. The voltage vs. pH behavior Is close to that predicted for the reaction 0003-2700/80/0352-0270$0 1.OO/O
Pd
+ H 2 0 = PdO + 2H' + 2 e
For pH measurements, the glass electrode is not quite universal. When glass is chemically attacked, or a glass electrode cannot be made small enough or rugged enough, other pH electrodes find application. A pH electrode based on a polymer-carrier system permeable to hydrogen ions has recently been described ( I ) . More commonly, electrodes based on pH-dependent electrochemical redox reactions have been used. The reversible hydrogen-platinum electrode, for example, forms the basis for an operational definition of the pH scale (2). The use of a palladium-palladium hydride electrode to measure the p H of HF-containing solutions has recently been described (3). A number of metal-metal oxide p H 0 1980 American Chemical Society
ANALYTICAL CHEMISTRY, VOL. 52, NO. 2, FEBRUARY 1980
electrodes, most notably those based on antimony, have been employed as well as those based on quinone-hydroquinone systems ( 4 ) . Reference 4 reviews the general subject of alternate p H electrodes. In general the electrodes based upon pH-oxidation-reduction systems are not applicable over the entire range of oxidation potentials found in aqueous solutions. For example, hydrogen platinoid metal systems have limited utility in aerated solutions, although by careful design Jasinski has extended the life of palladium-palladium hydride electrodes to a week or so in aerated solutions (3). Antimony-antimony oxide electrodes lose their response characteristics when the oxide layer becomes too thick, and structures permitting periodic removal of excess oxide by sanding ( 5 ) or brushing have been devised (6). T h e use of a metal more noble than antimony for such electrodes is indicated. However, metals that are too noble will not form oxides of sufficient stability. The metal should be noble, b u t not too noble. This paper describes the use of palladium which can be oxidized in a special manner t o form p H sensing wire electrodes possessing favorable characteristics for use in mildly oxidizing, e.g., aerated aqueous solutions. They have excellent lifetime and response characteristics. However, in contrast to antimony and hydrogen-platinoid metal systems, they are sensitive to reducing conditions. Electrodes for p H based on iridium-iridium oxide were reported by Perley (7); however they were found susceptible t o voltage drift problems (8). R. A. Macur (9) discovered that improved iridium electrodes were produced by high temperature oxidation of Ir wires coated with KOH. This lead was extended t o other group 8 noble metals and palladium was discovered t o be the optimal case ( I O ) . T h e following describes the procedure developed for preparing palladium oxide p H electrodes.
EXPERIMENTAL Simple air or oxygen oxidation of palladium wire produced electrodes of erratic, drifting voltage characteristics. Limited attempts at anodic oxidation were likewise unsuccessful. Success was attained by coating one end of a palladium wire with potassium hydroxide or sodium hydroxide and then exposing the coated end to high temperature, e.g., 800 "C in air or oxygen. The temperature range of approximately 350-850 "C in oxygen produced pH-responsive electrodes. At or above 900 "C, the palladium oxide decomposed as thermodynamics for PdO would predict (11). The above method was investigated over a broad range of preparation variables, and the following describes an optimized procedure. Palladium wire 0.75 mm (0.03 inches) in diameter was annealed, straightened, cut to desired length, and then rounded on one end by rotating against a fine file. About a 0.5-cm length of the rounded end was cleaned either by dipping into aqua regia for about 20 s or by grit blasting for a comparable length of time using 90-mesh alundum while rotating the wire. The wire was then dipped into a 50% aqueous solution of sodium hydroxide (reagent grade) to a depth of about 0.5 cm and dried for about 10 min in a stream of dry nitrogen (free from COJ while rotating the wire to even the coating of NaOH. These operations were carried out at room temperature. The coated end of the wire was then placed in an oxygen-swept furnace at 800 & 20 OC for 20 min. This procedure resulted in a black oxide coating on the portion of the wire that had been covered with NaOH. The oxided end of the wire was then protected by a snug fitting piece of Teflon tubing over a length of about 0.3 cm and grit-blasted above the Teflon covered portion to remove excess oxide and define a sharp boundary for the active end of the electrode. The bare portion of the wire was then electrically insulated with Alkanex (General Electric Co. No. 9504 resin, which is a thermosetting polyester resin in aromatic solvents). It was applied by dipping, dried at 100 "C for 15 min and cured at 210 "C for 4 min. Other methods of insulation such as heat-shrinkable Teflon tubing would serve as well or better. The important consideration was to cover all
Flgure 1.
271
Photograph of a palladium-palladium oxide pH electrode
the bare wire area with insulation, just meeting or very slightly overlapping the oxide. When examined in cross section in a scanning electron microscope at 1000 x, the oxide layer was quite uniform with a thickness of about 6 pm and appeared to be uniform and nonporous. Procedures leading to adherent, nonporous oxide layers in this thickness range gave excellent pH response characteristics. Incomplete coverage of the palladium with oxide, oxide spalling, excessive overlapping of the oxide with insulation, or leaving bare Pd a t the oxide boundary are effects to be avoided. Other electrode geometries, e.g, finer wires, may require modification of the procedure to achieve these goals. The palladium wire employed here was Rivet Grade (99.9% Pd from Engelhard Industries). Palladium oxide formed as above was scraped from the wires and Debye Scherrer patterns were obtained. These showed, besides the normal PdO structure (tetragonal with a,, = 3.04 A, cg = 5.34 A), extra lines that could be indexed on a tetragonal structure, a. = 9.05 A, co = 5.5 A. It is inferred that pH-sensing oxide is somewhat different from normal PdO. However its structure or composition was not further investigated. The electrodes were evaluated for their pH response in a series of pHdrion buffer solutions (Micro Essential Laboratory, Brooklyn, N.Y). Voltages were measured using an Instrumentation Laboratory Model 245 high impedance Deltamatic pHMillivolt Meter. The actual pH of the buffers was determined with a calibrated glass electrode. The voltages were measured against a standard Ag/AgCl/saturated KC1 fiber junction reference electrode.
RESULTS An electrode prepared as above is pictured in the photograph of Figure 1. It has the visual characteristics for good performance, uniform black oxide coating on the active end, and good insulation above. When a series of 34 electrodes was prepared in a consistent manner, the voltage vs. p H slopes varied from the ideal value of 0.059 V/pH to generally slightly lower values, and the voltage a t a single p H was always somewhat variable. For the 34 electrode group, the standard deviation in slope was f 1 . 2 m V / p H and the standard deviation in voltage at p H 7.0 was f 1 7 mV. The reasons for such variation are at this point speculative. However, the voltages
272
ANALYTICAL CHEMISTRY, VOL. 52, NO. 2, FEBRUARY 1980 8:
r
"\
-
:". .". 23: I!I-
2
i
\
13: -
\
\
\
1 k : k I b I z
VH
Flgure 2. New PdPdO electrodes. E - E AgAgCI, satd KCI, room temperature. 0 Electrode No. 416, A electrode No. 439, 0 Electrode No. 422. (---) Pd
+ H,
0
-
PdO
+ 2H' + 2e
Electrode No. 416, f Electrode No. 419. (---) Pd 2Hf 2e
+
of the electrodes indicate that they behave approximately according to the equation
E = E"
- (2.3RT/F) p H
(1)
where
E" = -(AGO HzO
-
AGO P d 0 ) / 2 F
(2)
and the pH-determining reaction is Pd
+ H20
-
PdO
+ 2H + + 2 e-
Flgure 3. Old PdPdO electrodes. E- EAgAgCI, satd KCI, 20 f 2 OC. 0 Electrode No. 421, 0 Electrode No. 427, A Electrode No. 426, X
(3)
A few electrodes showed exactly ideal slope but generally varied from the ideal potential of Equation 3 above. There could be an asymmetry in the oxide layer leading to an asymmetry potential. Nonideal voltage behavior in glass electrodes is rationalized in this way. An intended application of the present electrode was in the fabrication of COz sensors in which the p H of a bicarbonate solution is measured and the bicarbonate solution is in equilibrium with an ambient partial pressure of COz through a barrier permeable to COS but not to electrolytes. This has been accomplished and results have been reported (12). The particular application involved in-vivo COz measurements. This required further miniaturization of the electrode of Figure 1. T h e lifetime of the electrodes was investigated by storing a group of them in distilled water. Initial behavior of three electrodes shortly after preparation is shown in Figure 2. The slopes range from 54 to 57 mV/pH. They gave a 98% voltage change in 1 min to a p H change of 1 unit (usually from 7 to 8). A reevaluation after 20 months in distilled water showed no degradation in performance. After an additional 72 months in distilled water, a group of electrodes gave the performance shown in Figure 3. These data were obtained a t 20 f 1 "C, and one electrode showed exactly ideal slope. Response times were unchanged, and the electrodes appear unaffected by aging for nearly eight years. Distilled water storage is obviously satisfactory, and the presence of indifferent electrolytes or buffers also is not detrimental nor is dry storage, based on less extensive data. The dashed lines in Figures 2 and 3 are plots of Equation 3 using thermodynamic data from Latimer for the free energies of H 2 0 and PdO (13). The PdO value in Latimer differs about 2 kcal/mol from that reported by Coughlin ( I I ) , who sets f 2 kcal/mol uncertainty on the values for PdO. Coughlin gives AGO of PdO as a function of temperature. The data of Coughlin (11)for AGO of PdO and the data of Ives and Janz ( 4 ) for the Ag/AgCl electrode yield a temperature coefficent of about -1 mV/OC a t pH 7 for the P d / P d O vs. Ag/AgCl combination. This number was approximately confirmed with
+ H,O
-
PdO
+
a few measurements up to 50 "C. The behavior of the P d / PdO electrode a t higher temperatures has not been investigated beyond this. As previously mentioned, the P d / P d O electrodes are expected to be sensitive to reducing agents. A particular electrode was affected by ascorbic acid a t a level of only by weight, falling 100 mV in potential in a few minutes. Its original behavior recovered after about 15 h in aerated buffer. Strong redox couples also affected the potential. A Pd/PdO electrode in pH 6 phosphate buffer responded to the addition of 0.05 M potassium ferri- and ferrocyanide by losing all p H response and assuming the potential of the ferro-ferricyanide couple. Nitrogen deaeration of the buffer solutions causes some downward drift in potential. In 20 h, a group of seven electrodes in pH 7.4 phosphate buffer drifted downward by as little as 8 mV and as much as 38 mV in 20 h. However, all electrodes retained their normal response to changes in pH. A wide variety of indifferent electrolytes, such as phosphates, borates, carbonates, nitrates, chlorides, sulfates, and amines such as Tris buffer, (HOCHZj3NHz,were not detrimetnal to pH measurement with these electrodes. The application of the P d / P d O electrodes has not been exhaustively explored. In general, they are best adapted to mildly oxidizing conditions as distinct from other types of less noble metal-metal oxide electrodes. As such they fill a gap in the repertoire of nonglass electrodes for pH measurement. A major application of this electrode in a sensor for in-vivo COz measurements has been described in detail elsewhere (12, 15-17). This device incorporated the P d / P d O wire p H electrode surrounded by but electrically insulated from a tubular Ag/AgCl electrode, both in contact with a gelled electrolyte containing bicarbonate and chloride. The entire arrangement was surrounded by a COz-permeable, electrolyte-impermeable silicone copolymer layer. Internal p H changes reflect the ambient COz partial pressure in this device which was minaturized for insertion through a catheter to accomplish in-vivo measurements. Both human and animal tests have been carried out. A very interesting recent application is in a combination sensor electrode for in-vivo applications. The CO2-permeable membrane in a COP sensor is replaced by a pH sensitive polymer membrane (14). An internal P d / P d O electrode senses COz by pH change against an internal reference electrode. The p H is sensed against an external reference electrode. This works because the pH sensing membrane and the Pd/PdO electrode cancel out internal pH changes with respect to the external reference electrode and only external pH affects potential with respect to the external reference electrode. The
Anal. Chem. 1980, 5 2 , 273-276
fabrication (18)and behavior (19) of this dual sensor have been described elsewhere.
ACKNOWLEDGMENT T h e authors thank L. W. Niedrach and 0. H. LeBlanc for helpful discussions and S. F. Bartram for X-ray determinations. LITERATURE CITED (1) 0. H. LeBianc, Jr., J. F. Brown, Jr., J. F. Klebe, L. W. Niedrach, G. M. J. Siusarczuk, and W. H. Stoddard, Jr., J. Appl. ptrysioi., 40, 644 (1976). (2) R. G. Bates, J . Nectroanai. Chem., 2, 93 (1961). (3) R. Jasinksi, J . Electrochem. SOC.,121, 1579 (1974). (4) D. J. G. Ives and G. J. Janz, "Reference Electrodes, Theory and Practice", Academic Press, New York, 1961, (5) I.Levin, Chem. Anal., 41, 89 (1952). (6) P., Wulff, W. Kordatzki, and W. Ehrneburg, Z . Electrochem., 41, 542 (1935). (7) Ga. A. Perley and J. B. Godshalk, US. Patent 2 416 949 (1947). (8) L. W. Neidrach, unpublished data.
(9) (10) (11) (12) (13) (14) (15) (16) (17) (18)
(19)
273
R. A . Macur, U S . Patent 3 726 777 (1973). W. T. Grubb and L. H. King, U.S. Patent 3 7 0 9 8 1 0 (1973). J. P. Coughlin. U . S . Bur. Mines Bull., 542, 35 (1954). L. W. Niedrach, Report 73CRD194, June 1973, General Electric Co., Corporate Research 8 Development, P.O. Box 43, Schenectady, N.Y. 12301. W. M. Latimer, "Oxidation Potentials", 2nd ed.,Prentice-Hall, New York. 1953, p 203. J. F. Brown, G. M. J. Scusarczuk, and 0.H. LeBlanc. U.S. Patent 3 743 588 (1973). L. W. Niedrach and W. T. Grubb, U S . Patent 3 7 0 5 0 8 8 (1970). 0. H. LeBlanc. L. W. Niedrach, J. F. Brown, and R. W. Lawton, Report 77CRD078, General Electric Co., Schenectady, N.Y. J. Neurnark, A. Bardeen, and J. P. Kampine, J. Neurosurg., 43, 172-179 (1975). R. A. Macur, 0. H. LeBlanc. and W. T Grubb, U S . Patent 3 9 5 0 8 8 9 (1975). R. L. Coon, N. C. J. Lai, and J. P. Kampine, J . Appl Physioi., 40, 625 (1976).
RECEIVEDfor review June 21, 1979. Accepted November 5 , 1979.
Nitrosamine Air Sampling Sorbents Compared for Quantitative Collection and Artifact Formation D. P. Rounbehler, J. W. Reisch, J. R. Coombs, and D. H. Fine* New England Institute for Life Sciences, 125 Second Avenue, Waltham. Massachusetts 02154
Eight wet and dry sorbents were evaluated both for their abiliy to retaln added nitrosamine standards and for artifactual formation of nitrosamines in the presence of added precursor amines and air containing oxides of nitrogen. The dry solid sorbents included activated charcoal, activated alumina, silica gel, Florisil, Tenax, and Thermosorb cartridges. The wet sorbents conslsted of impinger traps containing 1 N KOH and pH 4.5 phosphate-citrate buffer/20 mM ascorbic acid. The air concentrations of nitrogen oxides, preloaded amines, airflow rates, and air sample volumes were chosen to simulate actual field sampling in industrial atmospheres where nitrosamines and their precursors could be present. ThermoSorb/N was found to be the only sorbent which was both free of artifact formation and capable of retalnlng 100% of the preloaded nltrosamines.
Several investigators have recently reported finding airborne nitrosamines in samples taken in urban and industrial atmospheres (1-8). Since many nitrosamines are known to be potent animal carcinogens (9), these findings of airborne nitrosamines are of concern. However, in sampling air for nitrosamines, it is important t o avoid false analytical results due either to in-situ artifactual nitrosamine formation from airborne precursors or t o the inability of the chosen sorbent t o quantitatively retain the nitrosamines. Others have examined the problems of artifactual nitrosamine formation due t o analysis and sample preparation (10-13). In this study, we examined several types of sorbents for their ability t o collect and quantitatively retain a variety of volatile nitrosamines under simulated air sampling conditions. We also examined each sorbent for artifactual formation of nitrosamines from trapped amines and air containing nitrogen oxides. 0003-2700/80/0352-0273$01.00/0
EXPERIMENTAL Reagents and Materials. Activated charcoal 40/60 mesh (Alltech Associates), activated alumina F-1 80/90 mesh (Analabs), Florisil60/80 mesh (magnesium silicate, Floridin Co.), Tenax-GC oxide, 35/60 mesh (a porous polymer of 2,6-diphenyl-p-phenylene Applied Science Laboratories), silica gel 60/80 mesh (Applied Science Laboratories), and ThermoSorb/N (a mixture of metal silicates which have been activated in a reducing atmosphere together with amine trapping agents and a co-eluting anti-oxidant, Thermo Electron Corporation) were all tested as received. All solvents were distilled in glass (Burdick and Jackson). Potassium hydroxide (ACS grade pellets), phosphate-citrate buffer solution, and ascorbic acid were purchased from Fisher Scientific. Morpholine, piperidine, pyrrolidine, and diisopropylamine were purchased from Eastman Chemical. Dimethylamine (40% solution) was purchased from Aldrich Chemical, and the nitrogen oxide gas mixtures were purchased from Scientific Gas Products. All reagents were used as received. Nitrosamine standards were supplied by the Analytical Services Laboratory of Thermo Electron Corporation. All GC column materials were purchased from Supelco. Analytical. The nitrosamine determinations were made using a Gas Chromatograph (GC) (Thermo Electron, model 661), interfaced to a TEA analyzer (Thermo Electron, model 502) as described by Fine et al. (14). The GC was fitted with a 0.32 cm i.d. X 500 cm stainless steel column packed with 5% Carbowax 20M on Chromosorb W 100/120 mesh and was operated isothermally a t 160 "C with argon as the carrier gas a t 15 mL/min (15). All analytical determinations were made by comparing chromatographic responses to that of known standards by interfacing the TEA to an integrating recorder (Hewlett-Packard, model 3380A). Apparatus. The apparatus for generating known mixtures of nitrogen oxides in air and for maintaining specific flow rates is depicted in Figure 1. A constant flow of gas (2 L/min) was maintained by pumping air through a gas mixing apparatus controlled by a rotameter and monitored by a mass flow meter (Hastings, model ALL 10K). Oxides of nitrogen gas were intro1980 American Chemical Society
