Atomic force microscope tip-induced local oxidation of silicon: kinetics, mechanism, and nanofabrication

Phaedon Avouris, Tobias Hertel, Richard Martel
1997 Applied Physics Letters  
Atomic force microscope induced local oxidation of silicon is a process with a strong potential for use in proximal probe nanofabrication. Here we examine its kinetics and mechanism and how such factors as the strength of the electric field, ambient humidity, and thickness of the oxide affect its rate and resolution. Detection of electrochemical currents proves the anodization character of the process. Initial very fast oxidation rates are shown to slow down dramatically as a result of a
more » ... result of a self-limiting behavior resulting from the build up of stress and a reduction of the electric field strength. The lateral resolution is determined by the defocusing of the electric field in a condensed water film whose extent is a function of ambient humidity. There is currently great interest in the possible use of proximal probes as tools for the fabrication of nanoelectronic devices. Among the different approaches tested so far, the most promising has been the atomic force microscope ͑AFM͒ induced oxidation of silicon and metals. The first report of tip-induced oxidation of silicon was a scanning tunneling microscope ͑STM͒ study by Dagata et al. 1 In this study, an H-passivated Si surface was scanned in air by a positively biased tip to generate surface oxide features. A different approach, where an AFM with a conducting tip biased negative with respect to the sample is used to induce oxidation, was demonstrated by a number of workers. [2] [3] [4] [5] With this approach, thicker insulating oxides can be produced. Oxide lines as narrow as 10 nm have been generated, 4 and nanoelectronic devices have been fabricated using this process. [6] [7] [8] [9] The mechanism of the tip-induced oxidation has been addressed by several authors. The dependence of the tip-induced oxidation on tip bias provides evidence that the process is affected by the generated electric field. Furthermore, Gordon et al. 10 suggested that the initial density of surface OH groups is rate limiting, while Teuschler et al. 11 attempted, unsuccessfully, to explain the tip-induced oxidation in terms of the Cabrera and Mott model 12 of fieldinduced oxidation. The fact that water from the ambient is necessary for the oxidation has been interpreted by Sugimura et al. 13 as an indication that the process is analogous to electrochemical anodization. No current flow has, however, been detected during the reaction. To optimally use this oxidation process in nanofabrication requires that we understand the factors that control its characteristics. With this in mind, we address questions involving the mechanism of the process, the rate of the reaction, and its dependence on electric field strength and oxide thickness, and try to identify the factors that control the thickness and lateral extent of the oxide, ͑i.e., the lithographic resolution͒. The n-type ͑8-12 ⍀ cm͒ Si͑100͒ samples were cleaned by removal of the native oxide in aqueous 10% HF solution. Local oxidation was performed in the ambient using con-ducting p ϩϩ -Si tips ͑radius Ͻ100 Å͒ and a commercial AFM microscope ͑M5, Park Scientific Instruments͒. The relative humidity was kept constant during experiments at values ranging from 10% to 95%. In Figs. 1͑a͒ and 1͑b͒, we show two grids of oxide lines written at a rate of 0.3 m/s using Ϫ10 V tip bias. In Fig. 1͑a͒ , the relative humidity was 61% and the linewidth ϳ90 nm. Lowering the humidity to 14% reduces the width by a factor of ϳ4. The height of the oxide barely changes. Such results demonstrate the strong influence of external factors on the characteristics of the oxidation process. One of the most important pieces of information needed in considering the use of the process in fabrication is its intrinsic rate. To determine the reaction kinetics, we applied voltage pulses with the tip stationary over a surface site. The applied voltage was then varied between Ϫ2 and Ϫ20 V and the pulse duration was increased from 10 ms to 1000 s while the tip was moved to a new position before each pulse. The width and height of the resulting oxide dots was obtained from AFM images. To obtain the total amount of Si oxidized, we also imaged the indentations left after the oxide was selectively etched away by aqueous HF. In this way, we found an apparent volume expansion upon oxidation of 3.0Ϯ0.4, which is higher than the expected increase by a factor of 2.27 anticipated for formation of amorphous SiO 2 . Kinetic results are shown in Fig. 2͑a͒ where the height of the oxide dots is plotted versus voltage pulse duration t. The fits show a rapid decrease of the growth rate with time as 1/t. No clear bias threshold was observed. For sufficiently a͒ Electronic mail: avouris@watson.ibm. com FIG. 1. The aspect ratio ͑height/width͒ of oxide lines improves significantly when the relative humidity is lowered from 61% to 14%.
doi:10.1063/1.119521 fatcat:a4xreuxmordarfqbxo65evsady