Anal. Chem. 1991, 63,2203-2206 (3) Dawson, J. H. J. Lecture Notes /n Chem/sfry;Springer: Berlin, 1982; Vol. 31, p 331. (4) Doyle, R. J., Jr.; Buckley, T. J.; Eyler, J. R. Lecture Notes ln Chemkfry; Sprlnger: Berlin, 1982; Vol. 31, p 365. (5) &to, Y.; Chlkara, C.; Inoue, M. Shlhuvyo Bunsekl 1983, 37, 115. (6)Hays, J. D.; Dunbar, R. C. Rev. Scl. Insfrum. 1984, 55, 1116. (7) Alford, J. M.; Wllllams, P. E.; Trevor, D. J.; Smalley, R. E. Int. J . Mess Spectrom. Ion Processes 1988, 72, 33. (8) Meek, J. T.; Stockton, G. W. Fourier Transform Ion Cyclotron Resonance Mass Spectrometer with Spatially Separated Sources and Detector. U S . Patent 4,686,365, 1987. (9) Guan, S.; Jones, P. R. Rev. Scl. Insfrum. 1988, 59, 2573. (10) Beu, S. C.; Laude, D. A,, Jr. Znt. J . Mess Specfrom. Ion Processes 1991, 704, 109-127. (11) Comlsarow, M. 6. J . Chem. Phys. 1978, 69, 4097-4140. (12) Comisarow, M. E.; Marshall, A. G. Chem. Phys. Lett. 1974, 25, 282. (13) Comisarow, M. E.; Marshall, A. G. Chem. Phys. Lett. 1974, 26, 489. (14) Chin, L.; Lin-Wang, T.-C.; Ricca, T. L.; Marshall, A. G. Anal. Chem. 1987, 59, 449. (15) Marshall, A. 0.; Lln-Wang, T.C.; Rlcca, T. L. Chem. phys. Lett. 1984, 105, 233. (16) Kerley, E. L.; Russell, D. H. Anal. Chem. 1989, 67, 53. (17) McIver, R. T., Jr.; Hunter, R. L.; Baykut, G. Anal. Chem. 1989, 67, 489. (18) Kofel, P.; Allemann, M.; Kellerhals, Hp.; Wanczek, K.-P. Int. J. Mess Spectrom. Ion Processes 1988, 72, 53. (19) Laude, D. A., Jr.; Beu, S. C. Anal. Chem. 1989, 67, 2422. (20) Hogan, J. D.; Laude. D. A., Jr. Anal. Chem. 1990. 62, 530.
2203
(21) Honovich, J. P.; Markey, S. P. Int. J . Mess Specfrom. Ion Processes 1990. 98. 51. (22) Hofstadler, S. A.; Lade, D. A., Jr. Int. J . Mess Specfrom. Ion Processes 1990, 707, 65. (23) Rempel, D. L.; Huang, S. K.; Gross, M. L. Int. J . Mass Specfrom. Ion Processes 1988, 70, 163. (24) Huang, S. K.; Rempel, D. L.; Gross, M. L. Int. J . Mess Specfrom. Ion Processes 1988, 72, 15. (25) Beauchamp, J. L.; Armstrong, J. T. Rev. Scl. Insfrum. 1989, 40, 123. (26) Van De Guchte, W. J.; Van Der Hart, W. J. Int. J . Mess Specfrom. Ion Processes 1990. 95. 317. (27) Scheikhard, L.; Blundeschling, M.; Jertz, R.; Kluge, H.J. Int. J . Mess Specfrom. Ion Processes 1989, 89, R7. (28) Guan, S. J. Chem. phvs. 1989, 97, 775-777. (29) Goodman. S. Proceedings of the 37th ASMA Conference on Mass Spectrometry and Allied Topics, Miami Beach, FL, May 21-26, 1989; p 1218. (30) Comisarow, M. 6. A&. Mess Specfrom. 1980, 8 , 1698-1708.
RECEIVED for review February 15, 1991. Accepted June 14, 1991. This work is supported by the Welch Foundation (F-11381, the Texas Advanced Technology and Research Program (No. 4515), and the National Science Foundation (CHE9013384 and CHE9057097).
Preparatlon of Solid Membrane Chloride Ion Selective Electrodes by Ion Implantation Robert S. Glass,* Ronald G. Musket, and Keith C. Hong Chemistry and Materials Science Department, Lawrence Livermore National Laboratory, Livermore, California 94550 The development of ion-selective electrodes (ISEs)has had a major impact in analytical measurement science, and they will continue to enjoy widespread application. Currently, there are a wide range of electrodes that are commercially available to the experimenter. ISEs come in many shapes and sizes, and use various chemical and material designs. Several all solid-state ISEs are available. Within recent years, a sizeable effort has also been devoted to the development of ion-sensitive field effect transistors (ISFETS), which, like ISE’s, are also potentiometric sensors. There are several excellent recent monographs dealing with the current state of ion-selective devices available to the interested reader (1-6). As an ISE for chloride ion, the solid-state membrane electrode has been available for some time. Common membrane materials include single crystals, disks cast from melts, sintered materials, and pressed polycrystalline pellets. Often, AgCl is dispersed in a matrix of Ag,S as this has been found to decrease photosensitivity and increase sensitivity toward the chloride ion (3). The theory of operation for ISEs based upon solid-state membranes has been extensively discussed ( 3 , 4 , 7). For the chloride ISE, a Nernstian response is expected. We have found that if chloride ions are implanated into a Ag substrate, a solid-state membrane ISE can be produced, which shows (approximately)the expected Nemstian response in solutions containing chloride ion over several decades of concentration. In addition, very good reproducibility of the potential-concentration response between different implanted specimens was obtained. To our knowledge, the only previous work using ion implantation to prepare ion-selective devices was creation of membranes operating on the principle of cation-exchange equilibria similar to those relevant to the glass electrode. In one study (8),using low fluences (1013-1016ions/cm2),Si+ and Li+ were implanted into alumina wafers in order to create a Na+ ISE. ISFETS sensitive to Na+ and K+ have been created by implantation into the gate region of a FET using moderate ion fluences (1016-1017ions/cm2). For example, Li+ and Al+ 0003-2700/9 110383-2203$02.50/0
have been implanted into the gate SiN layer covering a layer of Si02to create a Na+ ISFET (8);Na+ has been implanted into an oxidized gate Si3N4layer (initially covered by an A1 buffer layer, which was removed following implantation) to create a Na+ ISFET (9);and Na+ and Al+ or K+ and Al+ have been implanted into a gate SiOz layer to create ISFETS sensitive to Na+ and K+, respectively, (10). In these previous studies, the sensitivity depends upon the number of surface-active ion-exchange sites (e.g., (AlOSi)-), created by the implantation process. We believe our study is the first where ion implantation has been used to create solid-state membrane ISEs of the AgX type, with solid metal contact, which respond to components that the solution and the membrane have in common. As is well-known, the mechanism of anion response of these ISEs can be explained in terms of a buffering action on the free Ag ions near the membrane surface (3, 4 , 7). Our ultimate motivation in undertaking this work was to develop a convenient and reproducible method for fabricating arrays of ISE’s and “micro-ISE’s”,with individual elements selective to different chemical species. The study reported here is a first step in that direction. Using ion implantation as a fabrication method allows precise control over the structure and composition of the ion-sensing membrane layer. Therefore, special properties can be realized (i.e., rapid response). In addition, the mass production and low cost features of this method of fabricating ISEs implies that they may be disposed of following use.
EXPERIMENTAL SECTION Specimens for implantation consisted of Ag cylinders approximately 0.64 cm in diameter and 1.27 cm in height. Prior to implantation, the end to be implanted was polished to a mirror finish by using a slurry of silica powder with average particle size of 0.5 pm. Along with the implantation samples,two unimplanted control specimens were also prepared. The samples for implantation were then mounted in a fixture that was emplaced into a 200-kV ion implanter with the sample surfaces perpendicular to the ion beam. Hydrogen chloride gas was fed into the hot cathode ion source. The %C1+ions were mass spectrometrically selected 0 1991 American Chemical Society
2204
ANALYTICAL CHEMISTRY, VOL. 63, NO. 19, OCTOBER 1, 1991
for implantation. In the present study, we used three sequential ion beams of energies 200,70, and 33 keV. The fluences for these energies were 15, 5.9, and 4.4 X 1OI6 Cl/cm2, respectively. The Monte Carlo code TRIM89 (11) was used to calculate the lowfluence profiles for these energies and relative fluences. The half-maximum chloride concentration of the highest energy implant gave the nominal depth of the implants as about 160 nm. However, sputtering during the implantations probably reduced both the surface concentration and the depth of the chlorine. Electron spectroscopy for chemical analyais (ESCA)was applied to the determination of the composition of the surface layers and the chemical states of the silver and chlorine. A Mg Ka X-ray source provided the primary X-rays. For sputter profiling, a beam of 3-keV argon ions was used. Reagent grade AgCl in pelletized form and a nonimplanted silver specimen were the standards. The ion-implantedISEs were used without further treatment, although it is usually recommended by commercial manufacturers that the electrodes undergo prior conditioning in dilute chloride solution. The Ag cylinders were sealed into a tube of heat shrink Teflon tubing so that only the implanted face (or unimplanted polished face in the case of the control) of the cylinders was exposed. To make contact to the cylinders, a drop of mercury was placed in the tube to contact the opposite (unimplanted) end of the cylinders. Electrical contact was made to a copper wire, which was inserted into the tube and immersed into the mercury drop. All solutions were prepared by using high-puritydeionized water (18 MQcm resistivity, Millipore, with activated charcoal organic filter) and ultrahigh purity NaCl (Ultrex, Baker). Prior to making measurement,2 mL of ionic strength adjuster (5 M NaNO,, Orion) was added per 98 mL of solution. All potential measurements were made by using an Orion Model EA 940 expandable ion analyzer. The reference electrode was a saturated calomel (Fisher). In order to compare the behavior of the implanted Ag cylinders to a commercial ISE, we used an Orion Model 94-17A chloride ion selective electrode. All solutions were continuously stirred when measurements were made. Unless otherwise stated, potential measurementswere recorded after 2-min immersion in the chloride solution, although sufficient stability was often achieved in much shorter time periods. A strip chart recorder (Omega) was used to record the response-time transients and reproducibility data for alternate immersions in 0.28 and 2.8 mM C1- solutions. All measurements were made at 23 "C.
RESULTS AND DISCUSSION ESCA showed that the as-received surface of the AgCl standard contained approximately 19 at % C and 8 at % 0 contaminants. In situ argon ion sputtering was used for cleaning. After 5 min of sputtering, the carbon impurity was removed, and application of the standard sensitivity values (12) yielded 9 at % 0 , 5 2 at % Ag, and 39 at % C1. Extending the sputtering time for another 15 min resulted in a surface with 5 a t 70 0, 58 a t % Ag, and 37 at % C1. These results indicate that oxygen extends to some greater depth in the standard. The apparent discrepancy between the calculated and expected concentrations for Ag and C1 may be attributed to either preferential sputtering of the C1 or to application of incorrect sensitivity factors. In any case, if we take the Ag/Cl concentration ratio of 58/37 = 1.6 to be that for sputtered AgC1, then we can scale the measured relative concentrations for the 3SC1+-implantedsilver. Figure 1 shows scaled C1 concentration as a function of sputtering time. The near-surface region (i.e., within about 3 nm) of the as-implanted specimen (and subsequently exposed to the general atmosphere) was dominated by impurities. Only 15 sec of sputtering were required to remove the impurities, leaving only silver and chlorine. The -14 at % C1 for the first minute compares favorably with the deeper plateau of -17 at 70C1 calculated by using TRIM89, indicating that significant sputtering occurred during the implantation. Under our sputter-cleaning conditions, the sputtering rate for 100 nm of Taz05on T a was -5 nm/min. Considering the relative sputtering rates for Ag and Taz05, we estimate that the
-
n 1
',
0.0
I
4.0
2.0
8.0
SputMng time (mln)
Figure 1. ESCA spectrum showing the CI concentration as a function of sputtering time. The sputtering rate is estimated to be 14 nm/min.
sputtering rate in Ag was -14 nm/min; thus, the depth corresponding to a sputtering time of 4.75 min (i.e., the depth of half-maximum C1 concentration) is -66 nm. The binding energies of the Ag 3d and C12p electrons were compared for all three specimens after sputtering to minimize contaminants. The nonimplanted silver, the AgCl standard, and C1-implantedsilver (both in the highest C1 region and after reduction of the chlorine to -4 at %) had the same binding energy of 367.5 eV for the Ag 3d51zelectrons. For the sputtered AgCl standard, the binding energy of C1 2p5I2was found to be 197.8 eV. Both binding energies have values intermediate to those in the literature for AgCl (13, 14). However, the 169.7-eVdifference between these Ag and C1 binding energies agrees with that of one previous study (13) and is within 0.3 eV of the other (14). In the region of highest C1 concentration, the C1-implanted silver had a binding energy difference of 170.6 eV, because the C1 2pbI2electrons were a t a binding energy about 0.9 eV less than that for the sputtered AgCl standard. The binding energy of the C1 2psIz electrons was similarly reduced. There was both (a) an overlap of the C1 2p3I2peak for the C1-implanted silver and the C1 2p5 peak for the sputtered AgCl standard and (b) evidence for 61 2p31z electrons with a binding energy corresponding to that of the sputtered AgCl standard. Our preliminary interpretation of these measured binding energies is that the chlorine in the as-implanted silver was present mainly as Ag2Cl with some evidence for AgC1. More detailed studies are needed to evaluate this conclusion. In order to investigate reproducibility, we compared the response of three separate ion-implanted ISEs to chloride ion in the concentration range 0.028-28 mM. These data are shown in Figure 2. The reproducibility is observed to be quite good. The slope (dE/d log [Cl-1) for each electrode was -51 mV with correlation coefficients of 0.989 or better. For clarity, only the least-squares line for the total data set from all three ion-implanted ISE's is indicated (slope of -54 mV). When data for only the range 0.28-28 mM are considered, the slopes for the individual electrodes increase, ranging from -55 to -58 mV for the three ISE's shown in Figure 2, with correlation coefficients of 0.997 or better. Also shown in the inset to Figure 2 is a comparison of the response of a commercial C1- ISE to that for one of the ionimplanted ISEs (no. 3 in the major figure). The slope which we measured for the commercial ISE was -50 mV (we have no explanation for the anomalously low slope for this particular electrode; previous experience with commercial ISEs, especially less well-used electrodes, showed that they yielded more ideal slopes of -54 to -60 mV). As is shown in the figure, the ion-implanted ISE compares quite favorably to the commercial electrode. This indicates that the interfacial processes
ANALYTICAL CHEMISTRY, VOL. 63,NO. 19, OCTOBER 1. 1991 300 EI
P
300
3w1
I
h
0
v)
m 200
>
c 0
lonlmplint#3 WIlA
200
1\ 1
u)
i >
. 1[Cl-1,1 m M 1 0
100
2205
Ionlmplrnt#l
Ion lmplrnl #l: 2 hi8 lator
Ei
0
v)
200
d >
> E
Y
100
-> E
.-A
L
C
4)
CI
8
100 017-
.o 1
0
-01
.1
1
10
100
[CI-1, mM Figure 2. Electrode-to-electrode reproducibility examined by potential response as a function of chloride ion concentration for three indivklual ion-implanted ISE’s. For clarfty, the least-squares curve is drawn only for the total data set from aH three lon-implantedISE’s. For 0.028mM, the values for ISE 1 and 2 are indistingulshable. The Inset shows a comparison of the response of a commercial chloride ISE ( e )to that for one of the ion-Implanted ISE’s (8).The least-squares line for the inset is drawn only for the commercial electrode. For 28 mM, the points for the two electrodes are indistinguishable.
Table I. Hysteresis Effects in PotentialConcentration Curves for an Ion-Implanted ISE
forward pot.,
reverse pot.,
[Cl-1, m M
mV vs SCE
mV vs SCE
0.028 0.28 2.8 28
239.1 194.5 138.8
231.1 193.6 138.8
83.3
are similar, as may be intuitively expected. However, concentrations of defects are no doubt different in the two cases. The defect structure of solid membranes is known to affect the value of Eo (7). At the time of measurement (2 min after immersion)neither the commercial ISE nor the ion-implanted ISE exhibit significant potential drift; rather, erratic spikes of a few tenths of a millivolt are observed, both positive and negative. As expected, the behavior for the commercial and the ion-implanted ISE’s contrast with the behavior for unimplanted Ag, which cannot function as an ISE for C1-. The slope for this material is on the order of -46 mV, and the drift at the time of measurement is much larger, being typically several (Le., 4-5) mV/min and monotonic. It is desirable that ISE’s exhibit small or no hysteresis effects. The fulfillment of this criterion for the ion-implanted ISE’s is demonstrated by the data in Table I. As mentioned above, we first made measurements in the 0.028 mM solution,
.1
1
10
100
[CI-1, mM Figure 3. -Jnditioning effects for an ion-implanted ISE. Comparison is made for the initial response ( e )and the response after 2-h immersion in deionized water (a). In both cases the experimental measurements started with 0.028mM and proceeded to higher concentrations. then at successively greater concentrations,and then reversed. Virtually no hysteresis effect existed at C1- concentrations of 0.28 mM or above (0.9 mV for 0.28 mM and 0 mV for 2.8 mM). A 7-mV potential difference existed for measurements made at 0.028 mM. This effect is somewhat larger than desired, with reproducibility of 1 mV or less being sought after for ISE’s. It may be possible to realize greater stability if the specimens are annealed, or are preconditioned in a manner suggested for commercial electrodes, or if we create a less photosensitive matrix. Also, in this initial study we made no attempt to optimize the surface concentrations of chlorine. As mentioned above, another consideration for practical ISE’s is the necessity for prior conditioning. In order to investigate this for the ion-implanted ISE’s, we compared the initial response of one of the ion-implanted ISE’s to ita response following immersion for 2 h in deionized water. As Figure 3 demonstrates, conditioning effects are fairly small, ranging from a maximum of 7 mV for 0.028 mM C1- and around 3-5 mV for all higher concentrations. However, it is evident that there is some curvature in the initial data, which disappears following conditioning. Again, the relatively stable behavior for the ion-implanted ISEs contrasts with the much less stable behavior of “pure” Ag (which often contains sufficient impurities such as S, which make them partially responsive to halides), where the “initial“ and “aftern immersion values deviate by 45 mV for 0.028 mM and 21-38 mV for the higher concentrations. Finally, a fourth criterion used in the evaluation of the of ISE’s is the response time, the reproducibility of the final potential reached, and the rate of attainment of equilibrium. In order to evaluate these parameters for the ion-implanted ISE’s, we alternately exposed them to chloride ion solutions of 0.28 and 2.8 mM concentration and recorded the potential transients. The result of this study for one ion-implanted ISE is shown in Figure 4. The response time, T ~ is ,defined as the time required for the potential to change by 90% of the difference between the initial and final potentials. For transitions in either direction T~ is approximately 1.5 s, and usually less. Good reproducibility of behavior is observed. As
2206
ANALYTICAL CHEMISTRY, VOL. 03, NO. 19, OCTOBER 1, 1991 220
0.28mM 2.8 mM
I
200
c
0 v)
ui >
180
> E
Y
=.-
160
U
C P)
c
0
n 140
120
ACKNOWLEDGMENT We would like to acknowledge the support of Robert G. Patterson for the implantations and Cheryl L. Evans for the ESCA spectra and useful discussions with Jackson E. Harrar. I
0
1
2
3
4
5
6
7
8
one can create modified layers of precise thickness and stoichiometry. Arrays of ISE’s, each selective to a different chemical species, can be conveniently patterned. Micro-ISE’s, and arrays of micro-ISE’s can be fabricated for specialized purposes, such as for small analyte solution volumes or for in vivo or bedside monitors of body fluids. Ion implantation offers the possibility for mass production of ISE’s; therefore, they may be disposed of following use similar to what has previously been proposed for photolithographically produced voltammetric arrays (15). Finally, more than one species can be implanted. For example, for the chloride ISE we can implant both 35Cl+and 32S+.This is useful for creating the desirable effect of dispersed AgCl in an Ag2S matrix, ideally at 50:50 mol 70,which, as mentioned above, should have reduced photosensitivity compared to AgCl in Ag. In fact, in these initial studies we attempted such a dual implant; however, we found that SFs, which was used as a source gas for 32S+, caused the ion source to corrode and short out electrically. In future work we will attempt to obtain these more desirable matrices.
9 1 0 1 1 1 2 1 3 1 4
Time (min) Figure 4. Response time and reproducibility recorded for alternate immersion in 0.28 (El) and 2.8 (*) mM C t solution for an Wmplanted ISE. The electrode was rinsed with deionized water between immersions.
the membrane layer thickness affects response time, and ion implantation can generate membrane ISE’s of well-defined thicknesses, implantation should prove to be a useful method for adjusting the response time in a reproducible way. There are several other criteria used in the evaluation of ISEs. These include selectivity of response, detection limits, long-term drift, etc., which we have not evaluated. A full evaluation of these parameters will be the subject of a more complete paper. In these preliminary studies, we have not attempted to optimize parameters for fabrication of ISE’s based upon AgX membranes. For instance, we have not determined the optimum beam parameters, such as fluence and energies, for implantation of 36Cl+. We may be able to use a much smaller dose and thereby diminish the implantation time required. We would like to understand sources of small deviations in reproducibility, which are perhaps related to conditioning effects, and improve stability. In summary, ion implantation appears to be a useful method for creating ISE’s of the AgX type. Using the implantation method, it should be possible to produce novel materials that can function as ISE’s and show selective responses to a variety of anions and cations. Because it is “line-of-sight”and one can easily control the fluence and energy of the ion beam,
LITERATURE CITED (1) Janata, J. Principles of Chemical Sensors; Plenum Press: New York, 1989. (2) Madou, M. J.; Morrison, S. E. Chemical Sensing wlth SolM State Devices; Academic Press: Boston, 1989. (3) Koryta, J.; Stuiik, K. Ion-selective Electrodes, 2nd ed.; Cambridge University Press: New York, 1983. (4) Morf, W. E. Studies in Analytical Chemistty 2 . The Principles of Ion Selective Electrodes and of Membrane Transport; Elsevier Scientific Publishing Co.: New York. 1981. (5) Pungor, E., Ed. Ion-selective electrodes, 3, Analytical Chemistry Symposk, Series, Elsevier SclenMc Publishing Co.: New York, 1981; Vol. 8. (6) Freiser, H., ed. Ion-Selective Electrodes h Analytical Chemlstry; P l a num Press: New York, 1978 and 1980; Vols. 1 and 2. (7) Buck, R. P. In Ion-&lecttVe Elecbpdes in Analytical Chemistry; Freiser, H., Ed.; Plenum Press: New York, 1978; Vol. 1, pp 1-137. (8) Sanada, Y.; Aklyama, T.; Uilhira. Y.; Niki, E. Fresenius’ Z . Anal. Chem. 1982, 3 i 2 , 528-529.(9) [to, T.; Inagaki, H.; Igarashi, I I€€€ Trans. Eiectron D ~ V .1988, ED35. . ~56-64. .~ , . (10) Hoffmann, W.; Huller, J.; Pham. M. T. 5th Symposium on Ion-Selecthe Eiectrodes, Mutfafured; Pergamon Press: Oxford, U. K. 1988; pp 387-395. (1 1) Ziegier, J. F.; Biersack, J. P.; Littmark, V. The Stopping and Range of Ions in Solids; Pergamon Press: New York, 1985. (12) Wagner, C. D.; Riggs, W. M.; Davis, L. E.; Moulder, J. F.; Muilenberg, G. E. Hendbcek of X-Ray photoelectron Spectroscopy; Physical Electronics Division, Perkin-Elmer Corp.: Eden Prairie, MN, 1979. (13) Strydom, C. A.; van Staden, J. F.; Strydom, H. J. J . Electroanai. Ct”. Interfacial Electrochem. 1990, 277, 165-177. (14) Simon, D.; Perrin, C. C . R . Acad. Sci. Paris, Ser. II 1983, 297, 239-242. (15) Glass, R. S.; Perone, S. P.; Ciarlo, D. R. Anal. Chem. 1990, 62, 1914- 1918.
RECEIVEDfor review April 1, 1991. Accepted July 8, 1991. Work performed under the auspices of the U.S. Department of Energy by the Lawrence Livermore National Laboratory under Contract No. W7405-ENG-48.
