VOLUME 70, NUMBER 13 PH YSICAL REVI EW LETTERS 29 MARCH 1993 Site-Specific Measurement of Adatom Binding Energy Differences by Atom Extraction with the STM Hironaga Uchida, Dehuan Huang, Franqois Grey, and Masakazu Aono Aono Atomcraft Project, ERATO, JRDC, 5-9-9 Tohkohdai, Tsukuba sh-iJb, araki 300 26,-Japan (Received 21 October 1992) Using a scanning tunneling microscope, single adatoms can be extracted from a Si(111)7X7 surface by field evaporation, when the sample voltage is pulsed at 4 V or more in either polarity. Statistically, adatoms at the center of the 7X 7 unit cell are more frequently removed than those near the corner holes, by a ratio of 1. 6:1. This difference can be explained by assuming that the binding energy of center ada- toms is approximately 0. 1 eV less than for corner adatoms. The relationship of this result to previous ob- servations of greater chemical reactivity at center adatom sites is discussed. PACS numbers: 79.70.+q, 68.35.DV, 73. 40.Gk Atomic-scale modification of surfaces by scanning tun- neling microscopy (STM) provides a unique means to probe the local physics and chemistry of such surfaces [1], as well as a promising technology for the fabrication of novel electronic devices [2]. Modification of semicon- ductor surfaces is of particular interest for practical ap- plications. In their pioneering work, Becker, Golovchen- ko, and Swartzentruber [3] made atomic-scale protuber- ances on Ge(111) by briefiy increasing the bias between tip and sample above 3 V. The proposed mechanism was field ion emission. However, Si(111) could not be modified in the same way, even for biases up to 20 V. Lyo and Avouris [41 have succeeded in manipulating sin- gle atoms on Si(111) by moving the tip to within a few A units of the surface and applying a 3 V pulse. Because of the close proximity of tip and sample, this process is be- lieved to depend on direct chemical interaction as well as electric field. Recently, grooves several nanometers wide were made on Si(111) at bias voltages in the range 3-6 V, without first moving the tip towards the sample [5]. The process was directly dependent on electric field, rather than on current or voltage, indicating that the mechanism was field evaporation. In this Letter, we show that a similar process can be used to modify Si(111) on the atomic scale and remove single Si adatoms from the surface. The statistics of adatom vacancy creation yield informa- tion about the diff'erences of binding energy for diAerent adatom sites in the Si(111)7&&7 unit cell. The approach presented here for studying surface energetics, by directly comparing how easily diAerent atoms are extracted from a surface, should prove widely applicable. The experiments were made with a commercial ul- trahigh vacuum (UHV) STM (JEOL JSTM-4000 XV). Samples cut from wafers of p-doped Si(111) were cleaned in UHV by repeated Hash heating to 1200 C. The base pressure in the chamber was 1 x 10 Pa. The STM tip was a 0. 1 mm W wire, sharpened by electrolytic etching using a 0.5N solution of KOH. Electron bom- bardment heating of the tip to above 1200 C was per- formed in the UHV chamber. This step has been shown to be critical for obtaining reproducible modification [5]. Silicon adatoms were extracted as follows. A clean, flat area of the surface was imaged at +2 V and — 2 V at a tunneling current of 0.6 nA, revealing the regular pat- tern of Si adatoms of the Si(111)7&7 reconstructed sur- face. The tip was moved to a point in the imaged region and a voltage pulse of either +6 V or — 6 V was applied to the sample for 10 ms in constant current mode [6]. After the pulse was applied, the same area was again im- aged at 2 V in both polarities. Figure 1 illustrates the typical result of a single — 6 V pulse. Certain adatom positions are modified, the dominant form of modification being a dark spot in both polarities. Commonly adsorbed gases from the rest gas in the UHV chamber, such as H and 0, are known to appear light in one or both polarities [7,8], as are deposited metal atoms [9]. Thus the dark appearance in both polarities is evi- dence that these modifications are not due to adsorption, but due to vacancy formation. This is also consistent with the observation of groove formation when high volt- age is applied for longer periods [5]. Other sorts of defects, which do not appear dark in both polarities, are occasionally generated by the voltage pulse (see Fig. 1). The exact origin of these defects is not known yet, though it is likely that they are due in part to W field evaporated from the tip. In field ion emission studies, the critical field for the onset of W ion emission is 5. 7 V/4 compared to 3. 0 V/A for Si [10]. This difference might explain why Si extraction is more likely than W deposition, even though the field at the tip is expected to be larger than at the sample. We note, though, that direct comparison with results for field ion emission from isolated tips is not very reliable, since recent calculations show that the field ion emission process itself is strongly affected by the close proximity of the sample [11]. The extent of modification on the Si surface varies with the magnitude of the pulse. Of the order of 10 vacancies are produced by a — 6 V pulse to the sample, as shown in Fig. 1, and the result is similar for a +6 V pulse. For a — 4 V pulse, fewer vacancies are formed and frequently a single adatom can be extracted from a predetermined po- 2040 1993 The American Physical Society