The Journal of
Physical Chemistry VOLUME 98, NUMBER 16, APRIL 21,1994
Q Copyright 1994 by the American Chemical Society
LETTERS Tunneling Barriers in Electrochemical Scanning Tunneling Microscopy J. Pan, T. W. Jing, and S. M. Lindsay' Department of Physics and Astronomy, Arizona State University, Tempe, Arizona 85287- 1504 Received: November 3, 1993; In Final Form: January 21, 1994'
We have measured barriers for electron tunneling between a Pt-Ir tip and a gold substrate under potential control, obtaining values similar to those reported in ultrahigh vacuum. However, in contrast to vacuum tunneling, the data show a strong dependence on the bias applied between the tip and the substrate. They are only weakly dependent on the electrochemical potential of the substrate. The barrier changes with the direction of electron tunneling, an effect we attribute to permanent polarization in the gap. We observe a sharp dip near zero bias for tunneling in water. It is not observed for tunneling in a nonpolar solvent, and we attribute it to induced polarization in the tunnel gap.
Introduction The electrochemical scanning tunneling microscope (ECSTM) 1-4 has been applied to problems as diverseas interfacial chemistry5 and biological microscopy: yet fundamental questions about the technique remain unanswered. Vacuum tunnelingwas identified by the exponential decay of conductance, G, with distance, z,
G = Goe-2u
(1)
and the observation of decay lengths, (2~)-l,on the order of 1 A.7 Here, Go = 2ezlh = 77.52 pmh0.~1~ In a simple model 4~~= 0, the work function.1° In practice,11-144~~is found to be 1-3.5 eV less than 'P, owing to interactions between electrons at the small values of z used in STM.8J5J6 We will refer to the measured quantity, 4 ~ as~the , STM barrier. Attempts to measure the STM barrier in water have yielded a wide spread of poorly reproducible values with averages an order of magnitudelower than found in v a c ~ u m . ~Presumably, ~J~ this was a consequence of contamination in the tunnel gap.lS2l We have constructed a new microscope for ECSTM which is hermetically sealed and can drift as little as 1 A/min, operated in liquid. We are able to obtain high-quality cyclic voltammo*Abstract published in Advance ACS Abstracts. April 1, 1994.
grams in situ in the STM cell over long periods. We bave used the new STM to obtain reproducible values for STM barriers as a functionof tip bias, tunnel current, and substrate potential. We have examined different electrolytes, a nonpolar liquid and a substrate coated with an organic adlayer. Experimental Procedure Au(l11) 23 X 4 3 reconstructed substrates were grown by evaporation of gold onto heated mica22 and stored under argon prior touse. Tipswereetchedfrom0.25-mmPto,8Iro.z(Engelhard, Cateret, NJ) and insulated with Apezion wax.23 Cyclic voltammetry has been used to demonstrate that these tips do not cause appre~iablecontamination.~~ Electrochemical leakage of the tips was less than 1 PA. Our homemade ECSTM was controlled by a NanoScope I1 SPM controller (Digital Instruments, Santa Barbara, CA). We used a Pt wire as a counter electrode and an oxidized silver wire as a quasi-reference electrode. The quasireference was calibrated and checked for stability against an Ag/AgCl/KCl reference (Microelectrodes Inc., Londonderry, NH). Here, all potentials are given on the saturated calomel electrode (SCE)scale. We used degassed 18 Mi2 water from a Nanopure Bioresearch grade filtration system (Barnstead Inc., Dubuque, IA). NaC104 and HC104 were obtained from GFS (Columbus, OH), decahydronaphthalene (decalin) was from Kodak Chemicals (Rochester,NY), and cytosine was from Sigma
0022-365419412098-4205%04.50/0 0 1994 American Chemical Society
Letters
4206 The Journal of Physical Chemistry, Vol. 98, No. 16, I994
I1
TABLE 1 electrolyte water (18 MQ) HCIO4 (10 mM) HCIO4 (10 mM) NaClO4 (0.4 M) NaC104 (0.4 M) NaC104 + cytosine NaClO4 + cytosine ~~
2.O-cit
tflz z
8
1.6
0.8
potential vs SCE (V) open circuit +0.20 +0.30 +0.14 +0.34 +0.20
barrier (eV) 1.80 0.19 1.OO 0.33 1.14 0.33 1.07 f 0.29 1.05 0.12 2.15 f 0.68 3.80 0.62 1.52 0.30
* **
open circuit' n.a. a All data recorded with the tip -0.1 V with respect to the substrate starting at 1 nA current, except which was recorded at 0.1 nA starting current. decalin
0.4
*,
0.0
0
1
2 3 GAP (A)
4
5
(a)
g a
~~
2.0
2 1.5 m . -
0.5 l0 0 .
O 10L
P r . - -vv T30 4 0
-2 0 4
--
50
TIME (min)
(b)
We have found that contaminated substrates always give small values (ca. 0.3 eV) for the tunneling barrier. This reduction is not necessarily caused by an organic adlayer. Indeed, we shall show that some organic adlayers can result in a dramatic increase in the barrier. We attribute the reduction to the presence of large insulating particles which cause mechanical distortions of the gap.20 This mechanism is illustrated by the data in Figure l b (inverted triangles). In this experiment, the tip was left over a facet which yielded particularly high values for the barrier, and data were collected over 17 min. (The electrolyte was 80 pL of 0.4 M NaC104, and the substrate was held at 0.25 V vs SCE.) After 18 min, 10 pL of a 40 pM solution of cro protein (the gift of Dr. E. Appella) was added to the cell while the tip was left in place. (We used a procedure to introduce the solution which, with clean water, did not disturb the tunnel gap.) Cro protein binds strongly to gold in these conditions (D. Lampner, unpublished). The barrier was almost instantly reduced to a small value.
Figure 1. (a) Typical conductance vs distance data for a Pt-Ir tip and
a Au( 111) substrate. Two measurements made at the same position 3 min apart are shown as open circles. The tip was translated away 600 A, and two further measurements were made (filled circles). The data are displaced vertically for clarity. The lines arc fits which yield the given barrier values. (b) Barrier values as a function of time for a series of measurements made by moving the tip from place to place over the substrate (in 18 MO water). Data were gathered at the rest potential (filled circles) and then at a nominal 0.14 (open circles) and 0.34 V vs SCE (squares). The inverted triangles show the result of deliberate introductionof large particles (cro protein, injected at 18 min) into a 0.4 M NaClO4 solution with the substrate at 0.247 V vs SCE. The tip was left over a facet which, prior to the introduction of protein, gave unusually high barrier values. The dashed line is only to guide the eye. Chemical (St. Louis, MO). Cleanliness was checked by reproducing standard ~oltammetry2~ in situ in the STM cell. The voltammogram did not change over a period of an hour with or without the STM tip introduced into the cell. Tunnel current was recorded as a function of gap distance using the NanoScope software. The z axis was calibrated using stepson the Au( 111) surface. Typical raw data in 0.4 M NaC104 are shown in Figure la. The lines are exponentials fitted with the barrier values given in the figure. Barriers obtained at any one point on the surface were very reproducible but varied considerably from point to point on the surface (Figure la). We noticed that if the tip was left in one placed for periods in excess of about 10 min, measured barriers often drifted downward, presumably because contamination accumulated in the gap. We therefore moved the tip randomly over the surface and acquired many data points over a period of time. The barrier (averaged over 10 min) usually remained stable over an hour or more, and the fluctuationswere much smaller than those reported in previous work.I7J* This is illustrated in Figure l b where we show data (circlesand squares) obtained in 18 Mfl water. (Potential control is not quantitative in this case.) Our measured barriers are the averages of data sets like this. The quoted errors are f 1 standard deviation calculated from the measured distribution.
Results and Discussion We summarize data for a number of samples in Table 1. The average barrier is greater than 1 eV (i.e., 2 K 1 1 A-*), even reaching3.8 eVinonecase,values that mightbetakenasindicative of "vacuum tunneling". However, these data were all obtained at -0.1 V tip bias, and the barriers changed as this bias was changed. Figure 2 shows the dependence of the barrier on tip bias for the substrate at 0.14 and 0.34 V (SCE) in 0.4 M NaClO,. (The potential for zero charge is at about 0.25 V in this electrolyte.) Because the barrier also depends upon the tip-to-substrate separation, we recorded data as a function of the initial set-point current, IO. The barrier falls a little with increasing 10when the tip is negative, but the effect is within 1 standard deviation. The work function of an isolated electrode is given by @ 4.8eY+ ~ N H E ,where ~ N H Eis the electrode potential on the normal hydrogen scale.2s (This result should be corrected for polarization at the interface,I6but this effect is small for gold.) Here, we see that the STM barrier (at negative tip bias) behaves in just the opposite manner, falling as the substrate potential is increased. However, the most noticeable features of these plots are the asymmetry between positive and negative tip bias and the sharp dip at a small positive tip bias. In Figure 2a, we see that about 0.5 eV more energy is required for an electron to tunnel from the tip to the substrate than vice versa. This is similar to the dipole energy that has been measured for water on a noble metal substrate26so the asymmetry may be a consequenceof the alignment of water molecules in the gap (see the inset in Figure 3a). Such ordering will change with potential, and this might account for the changes that are observed as the substrate potential is altered. The sharp dip near zero bias is unaffected by changes in the substrate potential that are much larger (200 mV) than the width
-
The Journal of Physical Chemistry, Vol. 98, No.16, 1994 42Q7
Letters
3.0
3.0 n
%
U
a Y
2.0
2.0
0.1nA o 0.2nA CI 0.5nA 1nA 1.0 x 2nA
a a U
*
ml.O
0.0
0.0 0 +0.05 +0.1 TIP-SUBSTRATE BIAS (V)
-0.1 -0.05
(a)
-0.1 -0.05
0 +0.05 +0.1 BIAS (V)
TIP-SUBSTRATE
(b)
Figure 2. Barriers as a function of tip to substrate bias for various initial currents, IO (as marked in the figure), for (a) the substrate at 0.14 and (b) 0.34 V vs SCE. Data were obtained in 0.4 M NaC104, and each point is the average of many experiments with the error bars representing *l standard deviation. Inset (right of (b)) shows (top) induced dipole (arrow) for tunneling from tip to substrate, for zero bias (middle), and for tunneling from substrate to tip (bottom). The 1 standarddeviationbars overlap, givinga misleading impressionof the spreads (the bars are symmetrical). The zero-bias dip is about 2.5 times a typical standard deviation in both cases.
1
I
OS0-! -0:1 -0.k
I '
d
I
+ O h 5 +d.l
'
TIP-SUBSTRATE BIAS (V)
(a)
1-0 . 0 -0.1 -0.05 0 +0.05 +0.1 TIP-SUBSTRATE BIAS (V)
(b) Figure 3. Barriers for tunneling as a function of tip to substrate bias (initial current = 1 nA) for (a) decalin and (b) a cytosine adlayer on Au(ll1) at +0.2 V vs SCE in 0.4 M NaCIO4. Each point is the average of many experiments with the error bars representing *1 standard deviation. Fluctuations near zero bias are about 0.3 times a typical standard deviation in (a) and 0.1 times a typical standard deviation in (b). The inset in (a) shows permanent dipoles opposing tunneling (top) from tip to substrate and enhancing tunneling from substrate to tip (bottom).
ofthedip (about 10mV). Therefore, thedipisnotelectrochemical in origin. In vacuum tunneling, a sharp dip in barrier can occur as a consequence of mechanical interactions between tip and sub~trate.~'However, it occurs at a fixed value of conductance where attractive forces pull the tip into the surface. This is not the case here where the changes in l o correspond to an order of magnitude change in conductance,yet the dip occurs at the same value of bias. This change in conductance correspondsto about 1 A change in an 8 A gap (estimated from eq 1). Thus, the only variable that changes significantly in these experiments is the electric field in the gap. One reasonable explanation would be induced polarization of a cluster of molecules in the tunneling gap. Modeling this process as a two level system, we obtain
Here AE is the measured barrier, A E o the intrinsic barrier, and 6E the dipole energy of the oriented cluster (total moment Np) in an electric field E. The plus sign arises because such polarization would always add to the electrostatic energy for tunneling (see the inset in Figure 2b). We have made no attempt to correct the electric field for the 'dielectric constant" in the gap. The polarization process is, in itself, the dielectric response of the gap, and a microscopic model is needed for a self-consistent description. Our goal here is to see if a crude model for E gives a remotely plausible value for Np. The solid lines in Figure 2 are fits using eq 2 which allow for the intrinsic asymmetry of the barrier and take E to be zero at some small positive value of bias, VO(in order to compensate for the contact potential). With an average gap, ZOwe have NpEl keT = Np(V- Vo)/kBTZo = a ( V - VO)(ignoring local field effects and treating the gap as planar). The fits shown in Figure 2 were obtained with a = 0.072mV-1 and VO= 13 mV (a) and a = 0.084 mV-' and VO= 7 mV (b). We have carried out a WKB calculation of the current as a function of the gap (Lindsay and Pan, unpublished) using the distance dependence of the bamer calculated from a jellium model.16 This shows that the average gap, ZO,varies from 7 to 8 A, little different from a naive estimate using eq 1. Using ZO= 7 A to estimate E and taking I.C (for water) to be 6.3 X 10-30 C m, we find N = 33 ( a = 0.072) and N- 38 ( a =0.084). Thesenumbersaretoolargetobeinterpreted literally as the number of water molecules in the gap. We have tested this proposal qualitatively by measuring barriers in decalin, a nonpolar liquid of low polarizability. The results are shown in Figure 3a. Sampleswere prepared in a low-humidity environment, but we did not treat the tip and substrate to remove any residual bound water; the asymmetry may reflect this. The dip, however, is absent. We therefore attribute the zero-bias dip that is observed in water to the presence of liquid water in the tunnel gap at low electric fields. We have also measured barriers in the presence of cytosine adlayers, prepared as described elsewhere.28 We added cytosine to a final concentration of 3 mM in a 0.4 M NaC104 supporting electrolyte. These adlayers were imaged in all the conditions used for measuring barriers, so we believe that the adlayers remained intact during measurements of the barrier height. We measured some unusually high values in certain conditions (see Table 1). Barriers as a function of tip bias are shown in Figure 3b. The absence of a dip implies that liquid water is not present in the gap in this case. The large anisotropy implies that fixed polarization accounts for much of the large barrier. However,
4208 The Journal of Physical Chemistry, Vol. 98, No. 16, 1994
although the dipole moment of cytosine (2.5 X C m) is bigger than that of water, theSTM images imply that it is aligned in the planeof the substrate.28 It is possible that water is excluded by a cytosine molecule which is stuck to the tip and aligned along the z axis.
Conclusions Fixed and induced polarization in the medium appears to dominate the STM barrier for ECSTM. Thus, although values are comparable to vacuum STM, the tunneling mechanism is quite different, depending upon the microscopic structure of the liquid. This implies that a microscopictheory (probably including thermal fluctuations) is required for a detailed understanding of ECSTM contrast. However, we have obtainedcrude, qualitative, agreement by assuming that the microscope roughly follows the contour along which quantum point contact (C = Go) occurs.6 Acknowledgment. We thank Norton Lang, Dieter Kolb, Wolfgang Schmickler, Mike Weaver, Otto Sankey, and Kevin Schmidt for useful discussions and Jack Larsen, David Lampner, and Andy Vaught, who made substrates for this work. Financial support was received from the ONR (N00014-90-J-1455),NSF (Dir 89-20053), and NIH (1 R21 HG0081801Al).
References and Notes (1) Sonnenfeld, R.; Hansma, P. K. Science 1986,232,211. (2) Liu, H.-Y.; Fan, F. R. F.; Lin, C. W.; Bard, A. J. J. Am. Chem. Soc. 1986,108,3838. (3) Scigenthaler, H. STM in Electrochemistry. In Scanning Tunneling Microscopy II; Wiesendanger, R., GUntherodt, H. J., Eds.; Springer-Verlag: Berlin, 1992;p 7. (4) Bard, A.;Fan, F.R. F. Applicationsin Electrochemistry. In Scanning Tunneling Microscopy: Theory, Techniques and Applications; Bonnell, D. A,, Ed.; VCH: New York, 1993;p 287. (5) Gao, X.;Weaver, M. J. J. Am. Chem. Soc. 1992,114,8544. (6) Jing, T.; Jeffrey, A. M.; DcRose, J. A.; Lyubchenko, Y. L.; Shlyakhtenko, L. S.;Harrington, R. E.; Appella, E.; Larsen, J.; Vaught, A.;
Letters Rckesh, D.; Lu, F. X.;Lindsay, S. M.Proc. Narl. Acad. Sci. US.A. 1993, 90,8934. (7) Binnig. G.; Rohrer, H.; Gerber, C.; Weibel, E . Physica 1982,109/ llOB,2075. (8) Lang, N. D. Phys. Rev. B 1987,36,8173. (9) Kalmeyer, V.; Laughlin, R. B. Phys. Rcv. 1987,835,9805. (1 0) Chcn, C. J. Introductionto Scanning TunnelingMicroscopy;Oxford University Prcss: New York, 1993. (1 1) Binnig, G.;Garcia, N.; Rohrer, H.; Soler,J. M.; Flores, F. Phys. Rev. B 1984,30,4816. (12) Schuster, R.; Barth, V.; Wintterlin, J.; Behm, R. J.; Ertl, G. Ultramicroscopy 1992,42-14,533. (13) Gimzewski, J. K.; Mbller, R. Phys. Rev. B 1987,36,1284. (14) Winttedin, J.; Wiechers, J.; Brune, H.; Gritsch, T.; Hofer, H.;Behm, R. J. Phys. Rev. Lett. 1989,62,59. (15) DeAndres, P.;Flores,F.; Echenique, P. M.; Ritchie, R. H. Europhys. k i t . 1987,3, 101. (16) Schmickler, W.; Hcndemn, D. J. Elecrroanal. Chem. 1990,290, 283. (17) Lindsay, S.M.;Barris J. Vac. Sci. Technol. 1988,Ab, 544. (18) Binggeli, M.;Carnal, D.; Nyffenegger, R.; Scigenthaler, H. J. Vac. Sci. Technol. B 1991, 9, 1985. (19) Coombs, J. H.;Pcthica, J. B. I B M J . Res. Dro. 1986,30,455. (20) Lindsay, S. M.;Thundat, T.; Nagahara, L. Imaging Biopolymers Under Water by ScanningTunnelingMicroscopy. In BfologicalondArtiflcia/ IntelligenceSystems;Clementi, E., Chin, H. S.,Ed&;ESCOM: Lcidcn, 1988; p 124. (21) Lindsay,S. M.;Thundat, T.;Nagahara, L. A. J. Microsc. 1988,152, 213. (22) DeRose, J. A.; Lampner, D. B.; Lindsay, S . M. J. Vac.Sci. Technol. 1993,All, 776. (23) Nagahara, L. A.; Thundat, T.; Lindsay, S . M. Rev. Sci. Instrum. 1989,60,3128. (24) Angerstein-Kolowsh,H.;Conway, B. E.; Hamelin, A.; Stoicoviciu, L. J. ElectroanaI. Chem. Interfacial Electrochem. 1987,228,429. (25) Kbtz, E.R.; Neff, H.;Muller, K. J. Electrwnal. Chem. 1986,215, 331. (26) Bange, K.; Straehler, B.; Sass, J. K.; Parsons, R. J. Elecrroanal. Chem. 1987,229,87. (27) Gwo. S.; Shih, C. K. Phys. Reu. 1993,847, 13059. (28) Tao, N.J.; DeRose, J. A,; Lindsay, S . M.J. Phys. Chem. 1993,97, 910.