R. S. Potember, S. Yamaguchi*, And C. A. Viands
Applied Physics Laboratory,
Johns Hopkins University
Laurel, Maryland 20723, U.S.A.
Editorial Preface: Dr. Richard Potember's research in data storage systems that utilize a Scanning Tunneling Microscope (STM) was mentioned in the Molecular Computing Overview that appeared in the Jan. 1, and Jan. 15, 1996, issues of 21st. This scientific paper discusses the underlying technology of this STM-based system. A background in chemistry is also helpful for a better understanding of this article. --Francis Vale.
Silver and copper tetracyanoquinodimethane (TCNQ) salts have been imaged using a scanning tunneling microscope (STM) and an atomic force microscope (AFM). Single crystal, polycrystalline thick films, and vacuum-deposited thin films on highly oriented pyrolytic graphite (HOPG), molybdenum disulfide (MoS2), silvered mica and potassium chloride (KCl) were imaged in this study. Images of the AgTCNQ thin films grown on several of these substrates showed molecular-level resolution. A previously reported field-induced switching phenomenon of silver-TCNQ has now been observed using a STM. These results may be the basis for developing a molecular-based information storage system.
1. Introduction To Research
and copper) TCNQ salts show electrical  and optical  field-induced
The resistivity of these materials switches
from a high impedance state to a low impedance state within a few nanoseconds
by applying an electric field. The optical properties of these materials
are also switched by laser irradiation. These results are generally understood
in terms of a reversible neutral-ionic transition. Additionally, it was
reported that the formation of approximately 10 mol.% neutral TCNQ (TCNQ¡) is required to achieve the low impedance state and that upon continued application of an electric field the percentage of TCNQ¡ increases
with time . The phase transition is also expected to at least locally
alter the crystal structure; however, such a change has not yet been observed
a molecular level.
Recently, the scanning tunneling microscope (STM) has been widely used for surface analysis of organic substances, such as Langmuir-Blodgett films, phthalocyanines, liquid crystals, organic conductors , superconductors, and polymers. The STM has been proposed to be used as a tool to access molecules and to manipulate them for molecular level electronic applications [5-8]. The recent invention of the atomic force microscope (AFM) allows non-conductive surfaces to be imaged with nanometer-scale resolution. The AFM is well suited for studies of the surface of these organic and polymeric materials because of their poor conductivity.
In this study, STM and AFM are used to investigate molecular level structural differences before and after a field-induced phase transition in semiconducting metal-TCNQ complexes. We explore the possibility of using the STM to switch and image individual molecules of metal-TCNQ complexes. Simple calculations predict that the field strength at the STM tip should exceed the threshold for switching. In monitoring the formation of the low impedance phase, the STM is able to distinguish the conducting metallic species and the insulating organic species. In earlier experiments, the formation of the low impedance phase has been monitored by measuring the production of TCNQ¡.
For this STM and AFM study, AgTCNQ was chosen because the crystal structure of the single crystal  as well as a high resolution electron microscopy investigation of an epitaxial film  have been reported and these results would assist in the interpretation of our STM and AFM images. We chose substrates based on their epitaxial growth potential for making AgTCNQ thin films (KCl and silvered mica) as well as atomically flat and conducting substrates typically used for STM work (HOPG and Molybdenum disulfide) which also provide a means of calibration.
2.1 STM and AFM
The STM images were taken using an STM-SU2 (designed for UHV environment, with a tripod-type piezoelectric scanner) and an SA1-800 (designed for ambient environment, with a tube-type piezoelectric scanner), both instruments were manufactured by Park Scientific Instruments (PSI), Mountain View, CA. Electrochemically etched tungsten (W) and platinum (Pt) tunneling tips (0.5 mm diameter) were used for this study. No significant differences between images obtained with W and Pt tips were observed. The constant height mode was used predominately in this work. The constant current mode was only used for topographic scans on polycrystalline AgTCNQ and CuTCNQ samples. The laser deflection sensing type AFM (BD2-800, PSI) and microfabricated Si3N4 cantilevers with a gold reflective coating (V-shape, 200 µm long, 22 µm wide and 0.6 µm thick, theoretical force constant 0.37 N/m, theoretical resonant frequency 66 Khz ) were used for the AFM work.
The AFM study was performed with a repulsive force in the range between 10-7 and 10-8 N. All STM and AFM studies were performed in air. Images were tilt corrected in some cases, then filtered using a Weiner optimal filter by IP-5 image processing software (PSI, CA). Images have not been drift corrected. Preliminary calibrations were obtained using MoS2 and HOPG. No correction for piezo nonlinearities was required. A Nanoscope II (Digital Instruments, CA) was also used to obtain STM and AFM images on some of the samples for comparison. No significant differences between images were observed from the choice of instrument.
2.2 Sample preparation
Single crystalline and polycrystalline AgTCNQ were prepared by a slow diffusion method in an H-tube. TCNQ (0.2 g) was placed on one side of the H-tube and a cleaned silver plate (10 by 25 by 0.1 mm) was placed on the other side separated by a fine glass frit. Dried acetonitrile was used as the solvent. Large flat grains (over 100 by 100 µm size) of metallic-red-wine colored polycrystalline AgTCNQ grew over a 3 week period at room temperature covering about half of area of the silver plate. Single crystals (metallic-red-wine color, rectangular needle shape, typically 0.1 by 0.1 mm thick and 0.3-3 mm long) grew out from the edge of the silver plate and were collected separately. X-ray diffraction analysis (Molecular Structure Corp.,TX) showed an orthorhombic cell (a=6.965 _, b=16.666 _, c=17.431 _, V=2023.4 _3), in good agreement with the reported crystal structure .
Thin films of AgTCNQ were prepared on several substrates, including highly oriented pyrolytic graphite (HOPG, ZYA grade, Union Carbide, OH), single crystal molybdenum disulfide (MoS2, mined in Quebec, CANADA) single crystal potassium chloride (KCl, Optovac, MA), and epitaxially grown (111) Ag on mica (ASTM V2 grade, Asheville-Schoonmaker Mica Co., VA).
A unique dual-source vacuum deposition chamber, developed at Johns Hopkins, was used for the preparation of AgTCNQ films . Background pressure was kept under 10-6 Torr. The temperatures of the substrate and the metal and organic sources were monitored by thermocouples. The film thickness was determined from a quartz crystal monitor (Veeco QM-311). The density of TCNQ used for the thickness monitor was 1.34 as measured by pyconometry. HOPG, MoS2, KCl and mica were cleaved in air prior to use and were heated to 350 - 400 ¡C under vacuum for 2 hours before deposition. Silver films (1500 _) were deposited on cleaved mica surfaces which were kept at about 300 ¡C. TCNQ (typically 2000 _) was deposited on the substrate at 40 ¡C, followed by the deposition of a stoichiometric amount of silver metal. For the silver on mica film, TCNQ (300 _) was deposited directly onto the silver. In some cases (HOPG, molybdenum disulfide, and silvered mica) the films were post-heated (100 ¡C, under nitrogen flow) to complete the reaction. Topotactic formation of a AgTCNQ film has been reported from an epitaxially grown TCNQ film on KCl . A film was prepared by this method and the KCl substrate was dissolved away in order to image the first layer of AgTCNQ.
3. Results And Discussion
3.1 Single crystalline and polycrystalline AgTCNQ
We observed topographic images on a number of samples prepared by various methods. Relatively high tunneling voltage (Vt = 2.5-5.0 V) and low tunneling current (It, typically less than 0.5 nA) were required to obtained stable tunneling conditions. For this work, both positive and negative tunneling bias was used. We observed that stable images only appeared after several scans at which time a lower tunneling voltage could be used.
On one sample, a 250 _ periodic groove on the polycrystalline AgTCNQ appeared after several scans. These lines may have been created previously by larger area scans ( about 3.2 x 3.2 µ scan, 128 line scans along X axis for 1 image). The conductivity of AgTCNQ (2 x 10-5 S/cm ) is probably too low to initially allow stable tunneling. When the tip first contacts the surface, the applied electric field of the tip most likely switches the AgTCNQ from its high impedance state to its low impedance state, and, as a result stable tunneling is achieved. The switching threshold of metal TCNQ complexes is reported to be in the range of 103 - 104 V/cm . Field-induced switching could occur on AgTCNQ under the scanning conditions of the STM.
3.2 Thin AgTCNQ film on MoS2.
The film of AgTCNQ we prepared on MoS2 showed an ordered grain structure with grains of the order of 50 _ wide by 100-200 _ long. Molecular-scale scans, however, only showed noise distorted atomic images of MoS2. It was noted in an earlier STM study  that an ordered structure of cyanobiphenyl liquid crystal layer formed on MoS2. A van der Waals heteroepitaxial growth mechanism is common on MoS2 substrates as reported for inorganic compounds  and an organic compound . We believe that a similar growth mechanism is probably responsible for the highly oriented structure that we observed on silver-TCNQ charge transfer complex grown on MoS2.
3.3 Thin AgTCNQ film on silver/mica.
We observed molecular resolution of AgTCNQ columns on this sample (Vt = 0.17 V, _t = 0.35 nA). Vertical periodic lines have a 9 _ with a molecular corrugation of around 2 _ . This former distance is consistent with the separations of silver or TCNQ rows in the bc plane of the AgTCNQ crystal structure [9,10].
3.4 Thin AgTCNQ films on KCl
We have studied this system in detail as high resolution electron microscopy (HREM) has been performed on a 30 _ epitaxially grown AgTCNQ film on KCl . We made a 100 _ thick pale-blue AgTCNQ film on KCl according to the reported procedure  with two silver strips for electrical contact. However, the sample was not sufficiently conductive to obtain stable tunneling. An AFM was required to image the surface of this film. An interesting screw dislocation was identified on this film, with 1000 _ tracks over 8 µm diameter.
The AFM could resolve the surface structure as small as a few hundred _ in an ambient environment. We also made a thicker (2000 _ multi-layered), more continuous AgTCNQ film on KCl which was conductive enough for further STM study. Three hundred angstrom TCNQ and 20 _ silver were deposited alternately up to about 2000 _. The resulting film was metallic-blue. Two silver strips were deposited on the AgTCNQ to provide conductive contacts for the STM measurement.
We have observed on this sample the same phenomenon we have ascribed to switching above.
We propose that first the low impedance state of AgTCNQ is imaged in order to obtain stable tunneling as discussed previously. Subsequent images (t=0 to t=237 sec.) reveal the continued switching of AgTCNQ with nanometer-scale resolution (Vt = -2.0 V, _t = 0.9 nA). We can clearly discern the neutral reaction products segregate as a hole forms and white streaks appear across the image. The more conductive regions (white streaks) are due to the formation of neutral silver and the less conductive area (widening hole) is due to the continued production of TCNQ¡ with time as earlier noted .
The field strength on these STM measurements is estimated to be 105 V/cm which is 10-100 times higher than typical switching threshold values (103 - 104 V/cm) of metal-TCNQ complexes . These experiments suggest that the STM when used in combination with the metal-TCNQ system may have potential as a molecular-based information storage system . Unlike HOPG  or metals , for which nanometer-size fabrications using the STM have been reported, the metal-TCNQ salts exhibit a reversible phase transition.
3.5 Reverse-transferred film of epitaxially grown AgTCNQ film on KCl
This film was prepared to examine the interface between the epitaxial AgTCNQ film and the KCl substrate. Because the epitaxy began on the KCl surface, we expected we might see a more distinct order at the interface than at the surface. The STM image revealed a very ordered region with molecular resolution. The transition from diagonally to vertically spaced periodicity is accompanied by a transition in the molecular corrugation from around 5 _, the width of a TCNQ molecule on edge, down to 1.7 _.
3.6 Thin AgTCNQ film on HOPG
This purple film showed AgTCNQ images at higher tunneling voltages (-0.5 V or more) and distorted HOPG images at a lower tunneling voltage (-0.15 V) as seen elsewhere with liquid crystal samples . Images with molecular resolution were obtained on this sample over a period of two days. Very detailed molecular structures were recognizable. Corrugations as small as 0.3 _ can be distinguished with typical molecular corrugations of 1.8 _. The contrast is reasonable for the ac or ab plane being imaged and agrees with the literature [15,22], i.e., the benzene and cyano structures appear as bright areas with the benzene ring having the highest contrast. Detailed analysis is proceeding using molecular orbital calculations and will be reported separately.
3.7 Copper-TCNQ (CuTCNQ)
We could not achieve molecular resolution on the AgTCNQ single crystal because of the poor conductivity of the unswitched high-impedance state of AgTCNQ. CuTCNQ is 1000 times more conductive than AgTCNQ in the unswitched high-impedance state . Our preliminary STM study on polycrystalline CuTCNQ showed clear surface topography without field-induced switching at a lower tunneling voltage (Vt = 0.55 V, It = 0.9 nA) than AgTCNQ and from the first scan. This result is important in the investigation of structural changes before and after switching. Crystals of CuTCNQ tend to grow as ribbon-like clusters, thus single crystals could not be obtained. The CuTCNQ samples imaged formed long twisted striations, quite unlike AgTCNQ. These striations were imaged via AFM on different samples. We are continuing this study to allow detailed comparison with AgTCNQ images. Field-induced switching experiments using the STM is also underway and will be reported elsewhere.
microscopy and atmomic force microscopy have been used to explore the surfaces
charge-transfer complexes. The STM has
successfully resolved the surfaces of copper and silver TCNQ films on various
substrates often with nanometer-scale resolution. The AFM has proved to be
advantageous when used to visualize nonconductive films such as very thin discontinuous
films of epitaxially grown metal-TCNQ on an insulating substrate, KCl. A field-induced
phase transition driven by the electric field produced at the STM tip was observed
on several samples to produce 100 nanometer-sized domains. This dimensionality
surpasses the diffraction limitation (_>400nm) of optical storage. The reversible
nature of the redox reaction makes these materials promising for the construction
of a molecular-based erasable STM information storage system with a very hig
h spatial resolution. Molecular level switching experiments are underway to
further explore this concept. Analysis of the molecular images of AgTCNQ using
molecular orbital calculations are also being conducted.
One of authors (S. Y.) wishes to thank H. Sasabe and M. Matsuzawa for encouragement, and also thanks to Mitsubishi Petrochemical Co., Ltd. for financial support.
* Current address : Advanced Materials Laboratory, Mitsubishi Petrochemical Co., Ltd., 1 Toho-cho, Yokkaichi, Mie 510, Japan
1. R. S. Potember, T. O. Poehler and D. O. Cowan, Appl. Phys. Lett., 34, 405 (1979)
2. R. S. Potember, T. O. Poehler and R. C. Benson, Appl. Phys. Lett., 41, 548 (1982)
3. E.I. Kamitsos, C.H. Tzinis, and W.M. Risen Jr., Solid State Commun. 42, (8) 561 (1982)
4. T. Sleator and R. Tycko, Phys. Rev. Lett.,60, 1418 (1988).
5. C. Schneiker, S. Hameroff, M. Voelker, J. He, E. Dereniak and R. McCuskey, J. Microsc.(Oxford), 152, 585 (1988)
6. D. M. Eigler and E. K. Schweizer, Nature, 344, 524 (1990)
7. Molecular Electronic Devices, ed. by F. L. Carter, Marcel Dekker, NY, NY, 1983 and 1987
8. Molecular Electronic Devices, ed. by F. L. Carter, R. E. Siatkowski and H. Wohltjen, Elsevier Science Publishers B. V., North-Holland, 1988
9. L. Shields, J. Chem. Soc., Faraday Trans. 2, 81, 1 (1985)
10. N. Uyeda, T. Kobayashi, K. Ishizuka and Y. Fujiyoshi, Nature, 285, 95 (1980)
11. R. C. Hoffman and R. S. Potember, Appl. Opt., 28, 1417 (1989)
12. N. Uyeda, T. Kobayashi and E. Suito, Microscopie Electronique 1970 Vol. 1 (ed. P. Favard) p 443 (Societ_ Fran_aise de Microscopie Electronique, Paris, 1970)
13. W. J. Siemons, P. E. Bierstedt and R. G. Kepler, J. Chem. Phys., 39, 3523 (1963)
14. R. S. Potember, T. O. Poehler, A. Rappa, D. O. Cowan and A. N. Bloch, Syn. Metals, 4, 371 (1982)
15. M. Hara, Y. Iwakabe, K. Tochigi, H. Sasabe, A. F. Garito and A. Yamada, Nature, 344, 228 (1990)
16. A. Koma and K. Yoshimura, Surf. Sci., 174, 556 (1986)
17. M. Hara, H. Sasabe, A. Yamada and A. F. Garito, Jpn. J. Appl. Phys., 28, L 306 (1989)
18. D. J. Thomson, J. Microsc.(Oxford), 152, 627 (1988)
19. R. H. Bernhardt, G. C. McGonigal, R. Schneider and D. J. Thomson, J. Vac. Sci. Technol., A8, 667 (1990)
20. R. Emch, J. Nogami, M. M. Dovek, C. A. Lang and C. F. Quate, J. Microsc.(Oxford), 152, 129 (1988)
21. W. Mizutani, M. Shigeno, Y. Sakakibara, K. Kajimura, M. Ono, S. Tanishima, K. Ohno and N. Toshima, J. Vac. Sci. Technol. A8, 675 (1990)
22. D. P. E. Smith, J. K. H. H_rber, G. Binnig and H. Nejoh, Nature, 344, 641 (1990) Research Toward Information Storage And Processing At The Molecular Level Using the Scanning Tunneling Microscope
21st, The VXM Network, http://www.vxm.com