Voltage-Dependent STM Images of Covalently Bound Molecules on Si(100) David F. Padowitz* Department of Chemistry, Amherst College, Amherst, Massachusetts 01002 Robert J. Hamers Department of Chemistry, UniVersity of WisconsinsMadison, 1101 UniVersity AVenue, Madison, Wisconsin 53706 ReceiVed: May 12, 1998; In Final Form: August 17, 1998 Several alkenes chemisorbed on silicon(100)-(2 × 1) have been studied by scanning tunneling microscopy. Images of these molecules are strongly bias dependent, typically changing from mounds to depressions as the sample bias voltage is changed from -2.5 to -1.0 V. Image contrast involves both a reduced barrier for tunneling through the molecule and alteration of the silicon surface states by bonding. I. Introduction Scanning tunneling microscope images of molecules often strikingly match the chemist’s intuition, but exactly how molecules are imaged is not obvious. Adsorbed molecules alter surface electronic structure, produce complex couplings between tip and substrate, and can be directly involved in resonant tunneling. Theoretically, the fundamentals of tunneling are well established and many computational approaches have been developed. 1-3 Detailed calculations of STM imaging have often been done for atoms on surfaces, but molecular adsorbates greatly increase the complexity and cost of computation. 4-6 There is a need for systematic experimental results on well- characterized systems to which theory can be compared. It is also worthwhile to develop intuitive viewpoints to complement computational methods and suggest new avenues for research. Of the many STM experiments on adsorbed molecules, a number have specifically addressed the mechanisms of image contrast. Recently an extensive experimental and computational study of alkanes physisorbed on graphite examined the influence of different functional groups to illuminate the interplay of “topographic” and “electronic” effects. 7 Other recent STM experiments bear on general problems of electron transport through molecules, such as conduction through “molecular wires” or conditions for resonant tunneling. 8-10 This paper presents experimental observations of dramatic voltage-dependent variations in images of organic molecules chemisorbed on silicon. The experimental system is well- characterized and versatile. For the clean silicon(100)-(2 × 1) surface, the connection between electronic structure and STM images is thoroughly understood. 11 Organic molecules can be covalently bound to silicon by the reaction of alkenes with the Si(100)-(2 × 1) surface in a vacuum. Though this chemistry is new and still being actively explored, the bonding and resulting geometries have been established for a number of molecules. 12 This reaction is very clean, applicable to a variety of molecules, and the resulting silicon-carbon bonds are very stable. Most molecules examined gave reproducible images for low coverages at room temperature during rapid bias voltage changes. This allowed precise location of the molecules with respect to the silicon lattice and parallel imaging of multiple biases during a single scan. A simple qualitative model explains our results and suggests a variety of further experiments. II. Experimental Section Substrates were cut from commercial silicon(100) wafers (Wacker-Chemtronic, <0.10 Ω-cm n-type, P- or Sb-doped), rinsed in methanol (Baker certified electronic grade), and mounted on a tungsten holder with tantalum clips and silicon spacers. Surface contaminants were oxidized by a 15 min UV- ozone treatment. The sample was introduced into the UHV system through a load-lock. After outgassing in UHV for 8 h at 575 °C, the silicon was flashed briefly to 1150 °C to remove surface oxide while maintaining the chamber pressure below 5 × 10 -10 Torr. To further reduce reactive gases, a titanium sublimation pump with a LN 2 cold trap was run prior to flashing and the trap kept filled throughout imaging. Dosing was typically done 1 h after flashing the sample. Alkenes (cyclopentene, 98%; cis-cyclooctene, 95% (remainder cyclooctane); 1,5-cyclooctadiene, 99 + %; and 1,3,5,7-cyclooc- tatetraene, 98%, all Aldrich) were degassed by several freeze- pump-thaw cycles. The alkenes were admitted through a leak valve and dosing tube a few centimeters from the surface. Chamber pressures of 1 × 10 -9 to 1 × 10 -8 Torr for 10-20 s resulted in coverages on the order of 5% of saturation. Imaging began shortly after dosing and continued for several hours. The STM used a Burleigh “inchworm” coarse approach mechanism, a tube scanner, and custom electronics. Tips were electrochemically etched tungsten, annealed in a vacuum. Tunneling currents ranged from 100 pA to 2 nA. All images were taken in constant current mode. Images at multiple biases could be obtained simultaneously by switching the bias on alternate scan lines. Sets of images were also taken sequentially at differing bias, adjusting the current for constant gap resistance. Barrier height images were obtained by applying a small modulation voltage to the z-piezo to vary the sample-tip distance and using a lock-in amplifier to extract the resulting variation in tunneling current. Because the piezoelectric had not been calibrated against modulation frequency, barrier heights were approximate. * Author to whom correspondence should be addressed. Telephone: (413) 542-2660. Fax: (413) 542-2735. E-mail: dfpadowitz@amherst.edu. 8541 J. Phys. Chem. B 1998, 102, 8541-8545 10.1021/jp982229v CCC: $15.00 © 1998 American Chemical Society Published on Web 10/02/1998