Analytical Monitors Using Electrodes of Semiconductors (Germanium, Indium-Antimonide, and Silicon) J. P. McKaveney Garrett Research and Development Company, Inc., Division of Occidental Petroleum Corporation, La Verne, CaIiJ 91 750
C. J. Byrnes Crucible Materials Research Center, Colt Industries, Inc., P. 0. Box 988, Pittsburgh, Pa. 15230 A brief survey has been made of the use of semiconductor electrodes as analytical monitors. Additional illustrations of the use of germanium, indiumantimonide, and silicon by the authors are presented. In this work, the semiconductor serves as the anode of an electrochemical cell while stainless steel and platinum are used as cathodes. Germanium is shown to be a highly sensitive monitor for acid and oxidants (bromine) while indium-antimonide can be used for acid. Silicon, although highly useful for acid sensing, is presented as a monitor for oxidants (hydrogen peroxide) as well as for metals forming fluoride complexes in acid media.
UNTILRECENTLY, semiconductor electrodes have been used primarily as p H monitors and fluoride detectors. Turner (1) in 1961 appears t o have been the first t o make a practical application of n-silicon as a fluoride detector. This was followed by Steinbrecher (2,3),who modified Turner's approach of combining a n anode of n-silicon and a cathode of platinum by using illumination of the anode as well as p-silicon for fluoride analysis. M. Sparnaay and W. Wensing in 1964 (4) used a sensing electrode of germanium doped with copper and a reference calomel electrode for measuring the p H of a solution. The germanium electrode was reported to have a resistivity of 0.01 ohm-cm with n-type conduction. This aspect is puzzling since copper should give it a p-type character. Arthur and Leonard ( 5 ) also combined silicon electrodes with a reference electrode to measure hydrogen ion as well as fluoride ion concentration. I n using a wafer of n-p-n silicon, the sensor measured hydrogen ion, and with a p-n-p silicon wafer, the sensor was sensitive to fluoride ions in a p H 7 buffer solution. The patent claims application to cations as well as anions but indicates only application to hydrogen and fluoride ions and does not give any indication of crystal orientation or resistivity. S. L. Matlow (6) utilized both n and p type silicon as test and reference electrodes for measuring the p H of alkaline solutions. The patent also does not give any details on crystal orientation or resistivity. McKaveney and Byrnes (7, 8) applied semiconductors of germanium, indium-antimonide, and silicon t o the measurement of acid concentrations. The semiconductor formed the anode of an electrolytic cell while stainless steel, titanium, or platinum was used as the cathode. The n-silicon anode re~
~~
~~
(1) D. R. Turner, ANAL. CHEM., 33,959-960 (1961). (2) L. Steinbrecher et a/., U.S. Patent 3,129,148 (1964). (3) L. Steinbrecher et al., Belgian Patent 648,470 (1964). (4) M. Sparnaay and W. Wensing, U.S. Patent 3,159,783 (1964). (5) E. P. Arthur and J. E. Leonard, U S . Patent 3,219,556 (1965). (5) S. L. Matlow, U.S. Patent 3,294,662 (1966). (7) J. P. McKaveney and C. J. Byrnes, ANAL. CHEM., 42, 10231028 (1970). (8) J. P. McKaveney and C. J. Brynes, U.S. Patent 3,528,778 (1970).
290
*
sponded t o acid normality for acids with ionization constants greater than 1 X in the presence of an NHIF electrolyte. The germanium and indium-antimonide anodes, while responding to acid, were also influenced by the other electrolyte chemicals and any metallic ions present in the sample. The purpose of this communication is to expand on the brief data in the earlier papers regarding germanium and indiumantimonide as well as to extend the scope of silicon as a n analytical monitor. EXPERIMENTAL
In this work indium-antimonide (In-Sb) was used which had been obtained from the St. Louis plant of Monsanto Chemical Co. Germanium (Ge) was obtained from Sylvania Electric Products Inc. (Towanda, Pa.). Samples were usually in the bar form prior t o removal of a suitable length with a cut-of wheel. The semiconductor was then ground to desired size (usually '/r-inch in diameter and l inch long). InSb contacts were made ohmic by abrading the surface with No. 600 silicon carbide paper followed by the use of a n acid flux and 50 :50 Pb-Sn solder for connection t o a copper wire. The ohmic connection to the G e was made after abrading the surface, tinning, and by soft soldering using a zinc chloride paste. If desired a copper plate can be made to the G e before soldering. Prior to the electrochemical process, the semiconductor surfaces were chemically polished for 30 seconds in HF (48x), HNOa ( 6 5 z ) , and acetic acid (glacial) in a ratio by volume of 2 :1 :1. Germanium. A series of voltagecurrent measurements were made to ascertain the response of the G e anode to acid and oxidizing agents. An auxiliary saturated calomel electrode (SCE) was included for reference voltage measurements. The potential control had meters for reading all voltage, cell current, and electrode potentials. For more accurate control at lower electrode potentials, a potentiometer and null meter were included in the circuit. Provisions were also made for temperature control and solution purging with hydrogen. A platinum foil was used as a n auxiliary current electrode. Figure 1 shows a schematic of a typical apparatus used for measurement of electrode current at a fixed potential. This application was first applied t o Ge and utilized a 1 'i8-inch diameter by inch thick slice of n-type with 0.8 ohm-cm resistivity and 110 crystal orientation. The G e slice was sealed to the bottom of a 11i4-inch 0.d. and 1.00-inch i.d. Lucite tube with Epoxy cement. A copper wire was soldered t o the germanium slice which had been plated with copper a t the wire contact side of the electrode. The Lucite tube was covered with masking tape and black paint t o prevent light from striking the semiconductor and producing extraneous currents. As the work with germanium proceeded, intrinsic (47 ohm-cm) and low resistivity n-type (0.04 ohm-cm) were examined as acid sensors. The germanium was in rod form (0.25-inch diameter) and of 111 crystal orientation. These types were examined as anodes (about 4.0 cm2 in area) within the cylindrical cathodes shown in Figure 4 of refer-
ANALYTICAL CHEMISTRY, VOL. 44, NO. 2, FEBRUARY 1972
Table I. Output Current for Intrinsic Ge Electrode (47 ohm-cm) at 30°C us. a SCE Volts us. 0.001N 0.001M SCE HzSO~,mA K2Cr207,mA 0.10 0.02 0.01 0.50 1.21 0.40 0.75 1.82 0.83 1 .oo 2.20 1.25 1.25 2.52 1.65 2.90 2.01 1.50 3.20 2.43 1.75 3.48 2.80 2.00 Table 11. N-Ge (0.04 ohm-cm) Anodic Current at Varied Cell Potential for H2S04Containing 0.60 Gram (NHJ2S2O8 per 200 Milliters HzSOa, 1 ,35 Volts, 2.70 Volts, 4.05 Volts! g/200 ml PA PA PA 200 210 220 0.000 260 375 450 0.010 106 0.020 339 5 26 941 0.030 391 627 434 705 1,157 0.040 466 766 1,351 0.050 488 815 1,540 0.060
025
0.50
0.75
1.00
H2SO4 (grams/200
1.25
ML)
Figure 2. N-Ge (0.8 ohm-cm) anode (6.5 cm2) response to H2S04with 0.100 gram HzO2
1600
SON
m
1
E 1200-
LLI
LL
r Figure 1. Ge electrode for acid measurement
800-
0:
ence (7). Also a series of experiments were performed in which the voltage applied t o the electrodes was increased from 1.35 to 4.0 volts. Indium-Antimonide and Gallium-Arsenide. Small anodes were prepared from gallium-arsenide and indium-antimonide (In-Sb), but all efforts failed to place a n electrical contact o n the gallium-arsenide directly with solder or after nickel plating the contact end (9). Results were more successful with the In-Sb. N-type with a resistivity of 0.0042 ohm-cm and 111 orientation was used. The anode (cylindrical) area was about 1.0 cm2 with a platinum wire (1.0 inch) cathode. Silicon. Recent work performed with silicon anodes utilized the commercial apparatus from Hach Chemical Company (Ames, Iowa). Their model 7597 acid concentration meter contains a very small n-silicon anode (0.35 cm2) surrounded by a cylindrical stainless steel cathode. The meter is line operated (115 Vac) containing a transformer, solid state rectifier, and amplifier. The meter scale covers the range from 0.001 to 0.050 normal in acid when using an NH4F electrolyte at the recommended concentration of 0.600 gram per 100 mll of measured solution. RESULTS
The data of Table I indicate the response of an intrinsic Ge electrode to varying voltage using a solution containing 0.001N&SO4 and another containing oxidizing agent (0.001M K~Cr207). When direct additions of sulfuric acid ( 5 to 30 grams HPs o 4 / l o 0 ml) were made to the cell of Figure 1 using 0.8 ohm-cm n-Ge, the higher acid concentrations appeared to (9) M. V. Sullivan and J. H. Eigler, J . Electrochem. 226 (1957).
2 0 -
Soc., 104,
r
400-
I
I
I
I
produce erratic current measurements. However, when the sulfuric solutions were diluted with water (5 to 200 ml) and in the presence of an oxidizing agent such as hydrogen peroxide, a calibration curve was obtained. Apparently, at the more dilute acid concentrations, an oxidizing agent was required to effect detectable etching with this type of germanium. The curve indicated a direct relationship between output current and acid concentration (Figure 2). T o eliminate the high curvature of the calibration beyond 0.75 gram of sulfuric acid per 200 ml, other types of germanium-i.e. intrinsic (47 ohm-cm) and n-type (0.04 ohmcm)-were examined as sulfuric acid sensors. Surprisingly, these materials were much more sensitive in acid response, requiring a dilution of 1 to 1000 of the original acid solutions containing from 5 to 30 grams of sulfuric per 100 ml. The calibration obtained, however, was linear (Figure 3). The more sensitive etching may have been due to the 111 crystal orientations compared to the 110 of the first specimen. Also to obtain linearity it was necessary to increase the applied potential to 4.0 volts (Table 11). However, when acid samples containing about 50 ppm of metal (Fe, Cr or Ni) after 1 to
ANALYTICAL CHEMISTRY, VOL. 44, NO. 2, FEBRUARY 1972
291
Table 111. N-Ge Anode (0.04 ohm-cm and 4.7 cmz)Response at 1.35 V to tL!304 Containing 0.300 Gram H 2 0 2 per 200 Milliters HzSOd, 20.0 "C, 25.0 "C, normality PA PA 0.0000
15
16
0.0002
200
0.0008
180 300 402 468
0.0010
50 1
555
0.0004 0.0006
w
a
t-
1 600c
325
432 515
1000 dilution were analyzed, the acid values were about 20% too high. The response of germanium t o a n oxidizing etchant such as bromine is shown in Figure 4. This was the material of 0.04 ohm-cm resistivity and 111 crystal orientation. In this respect, germanium is more sensitive t o bromine than hydrogen peroxide which requires the presence of acid to produce an appreciable current response (Table 111). In Figure 5, the response of In-Sb t o acid is indicated. Of the three elemental semiconductor types examined, In-Sb was the least satisfactory, both from the view point of interferences as well as applied potential response. With silicon and germanium, a curvature in calibration was usually linearized by an increase in the applied voltage. However, with In-Sb a decrease in sensitivity was noted in going from 1.35 t o 2.7 volts (which was more linear). An additional increase t o 4.0 volts caused further decreased sensitivity as well as heavy discoloration of the electrode's surface. Figure 6 indicates the response of the n-silicon anode of the Hach Model of the acid concentration meter to hydrogen peroxide in the presence of fixed amounts of NH4F and acid. As discussed earlier, silicon gives a significant response to oxidizing agents only in the presence of acid and a fluoride electrolyte. The potential across the electrodes in the commercial line operated apparatus is equivalent t o 1.0 volt dc. The curvature of the calibration could be made more linear for this application if a higher voltage were available. The titration of H F with ferric iron (Fe*+) is illustrated in Figure 7. A solution of ferric nitrate and 0.84 molar in iron was used. The equivalence point of 5.70 ml or 0.0048 mole of Fe3+ for 0.0144 mole of H F indicates formation of a n FeF, complex. DISCUSSION
Acid Response for Ge and In-Sb. In addition l o the data in Table I for intrinsic Ge, current us. voltage measurements were also made for the low resistivity (0.04 ohm-cm) Ge. In both cases, increased voltage resulted in a linear increase of current with no indication of reaching a limiting current plateau. Cell voltages (anode t o cathode) were also increased t o 12 volts and again no limiting current was obtained. Also the presence of ionized salts appeared t o add to the current in a linear fashion with voltage increase. The data of Table I1 indicate that it was necessary to increase the cell voltage to about 4.0 volts on the G e anode t o obtain linearity beyond 0.020 gram of H2SO4 per 200 ml or approximately 0.002N. The data for 4.05 volts are approximately the same as that shown in Figure 3 except a further dilution was made from 200 ml to 1 liter. There is also a slight difference between the two current responses a t 4.05 volts for the 200 ml and 1 liter data at a given acid concentration since the (NH&SzOg was fixed at 0.600 gram total for both solutions. 292
800 v)
dI
I
I
I
I
0.02
0.06 0.10 0.14 0.18 BROMINE (GRAMS/200ML)
Figure 4. Response of N-Ge anode (4.7 cm2) to bromine at 1.35 V
0
0.04 0.06 0.08 H2 S O 4 (grarns/500ML)
0.02
Figure 5. N-In/Sb (0.004 ohm-cm) anode (1.0 cm2) response to &So4 with 0.600 gram (NH& &Os at 2.7 V
The response of the 0.04 ohm-cm G e to acid in the presence of H 2 0 2is indicated in Table 111. The advantage of using H 2 0 2over K2Cr207 or (NH4)2SzOgis that little or n o residual current is present in the absence of acid. The data indicate linearity to almost 0.0010N H2S04at 1.35 volts when using peroxide. Like silicon, G e is also affected by temperature, and precise measurements would require temperature compensation. In general it was felt that metal or salt ions had too strong a n effect t o make use of germanium as a n acid sensor for metal containing baths. However, the high sensitivity for germanium to acid should make it applicable as a monitor for extremely dilute acid solutions which d o not contain appreciable amounts of metal ions. Such applications might involve systems where acid gases such as SO2 or SO, are collected from air in aqueous solutions of oxidizing agents. It should also be recalled (7) that germanium gives such high sensitivity to acids only when using an oxidizing agent such as (NH4)2S208or H202. In respect t o making germanium more specific to acid as a detector in the anodic form, a possible solution may be inferred from the work of Sparnaay ( 4 ) who indicated that a copper dopant effectively accomplished this for germanium as a pH sensing electrode. H e noted that the potential between the germanium sensor and calomel reference electrode was
ANALYTICAL CHEMISTRY, VOL. 44, NO. 2, FEBRUARY 1972
9 1
0.030 (3
z
I
0.040 (3
f D Q.
0.030
W E
d 0.24
0
0.32
H 2 0 2 (grams/lOOML)
2.0
4.0 6.0 8.0 M L Fe3+ ( 0 . 8 4 M )
Figure 7. Titration of 0.2885 gram (0.0144 mole) of HF in 100 ml with Fe3+ using N-Si electrode
Figure 6. N-Si anode (0.35 cm2) response to HzOZwith 1.10 grams NH,F and 0.0730 gram HCI at 1.0 V and amplification
found to be greatly dependent o n the varying kinds and quantities of foreign ions in the solution. However, by doping copper, the germanium the germanium with up to 1 mole became selective for pH. The residual current for In-Sb at zero H 2 S 0 4concentration shown in Figure 5 was caused by the (NH4)&08 present in the solution. The earlier study had indicated that NH4F could not be used at the usual concentration because at a potential of 1.35 volts, 1.60 grams of N H 4 Fper 500 ml caused a current of 4300 PA. Therefore in the presence of hydrolyzable metal ions, In-Sb would not be the most effective acid sensor. The one aspect of the In-Sb work which appeared intriguing related t o its high sensitivity for fluoride salts. This implies a possible application as a fluoride monitor. Perhaps by decreasing the applied potential on the In-Sb below 1.35 volts (the lowest examined), more optimum response might occur for fluoride ions as well as hydrogen ions. It has come to our attention (10) that satisfactory ohmic contacts are made to Ga-As by tinning electroless nickel contacts with a lead-tin solder. Sharpless ( I / ) has shown that tin is important in producing ohmic contacts to Ga-As. In this work, he also indicated Ga-As to be superior to either silicon o r germanium in relation t o temperature effects. Temperature sensitivity has been a minor disadvantage with silicon and germanium sensors. The use of a conductive epoxy may also be suitable for making electrical contact to the various semiconductors. Straub (12) has made extensive use of this material in fabricating silicon monitors for acid applied t o H2S04, HCI, and chromate metal treating baths. Response to Oxidants and Other Species. Figure 6 indicates that silicon can be used as a sensitive indicator for hydrogen peroxide. Figure 4 had indicated the data for germanium response to bromine. The limited studies t o date indicate that both types of semiconductors are sensitive t o a wide range of oxidizing agents with germanium appearing to be the most sensitive. The data of Figure 7 utilize the response of silicon to HF as a sensor for establishing the composition of a n iron fluoride complex. This intelligence was informative as earlier work ~~
(10) D. R. Turner, Bell Telephone Laboratories, Murray Hill, N.J., private communication, April 9, 1970. (11) W. H. Sharpless, BellSystem Tech. J., 38, 259 (1958). (12) W. A. Straub, United States Steel Corp., Applied Research
Laboratory, Monroeville, Pa., private communication, September 8, 1971.
(13, 14) had indicated the dominant fluoride complex t o be FeF2+ in dilute acid. The earlier work was all indirect in that total fluoride ion was measured as well as free fluoride ion and the difference attributed t o formation of FeF2+. The n-silicon electrode is a selective monitor for HF and appears t o offer a direct method of establishing the stoichiometry of fluoride complexes in acid media. Another application of the n-silicon electrode apparent from Figure 7 is the determination of metal ion content for oxidation states of metals forming fluoride complexes. Thus the determination of the amount of Fe3+, Ti4+, Alas, etc. present singly in a solution could be performed. It is assumed that application would not be made for mixtures of different metals forming fluoride complexes. A specific example could be the determination of the amount of Fea+ present in a mixture with Fez+. Such a need presently exists for the measurement of the amount of Fe3+ contained in FeC13 etchants used by the electronics industry. As the etchant becomes used, the ferrous iron content builds up and the etching or pickling ability decreases. The determination of Fe3+ in the metal bearing sample could be performed in a n acid medium using a soluble fluoride salt as titrant. The authors had indicated in their earlier studies (7) on the anodic behavior of n-silicon t o HF that the resistivity should be less than 0.05 ohm-cm for satisfactory current response. A recent publication of Theunissen et a/. (15) in Holland examined the anodic dissolution behavior of n- and p-type silicon. F o r n-silicon complete dissolution occurred in 5 aqueous H F for