Principal Investigator P V Smith Project g38

Physics Department, Machine VP

University of Newcastle

Co-Investigators M W Radny, P C L Stephenson and A J Dyson

Physics Department, University of Newcastle

Chemisorption and Etching on Silicon Surfaces

The main aim of this project is to study theoretically the chemisorption of different adsorbates on silicon surfaces and the associated desorption and etching processes. Understanding these processes is very important because of the crucial role played by silicon surfaces in most present-day electronic devices. Chemisorption studies on semiconducting surfaces are essential in helping us grow thin metallic or semiconducting films on semiconducting substrates and in determining the extent to which such surfaces can be modified for specific applications. It is well known that the presence of adsorbates on the surfaces of semiconductors can significantly alter the wear characteristics of such surfaces and their susceptibility to corrosion, reaction rates and usefulness as catalytic agents. Theoretical chemisorption studies are an essential precursor to any real understanding of these important industrial processes. Understanding the etching of silicon surfaces is also of immense fundamental and technological importance. This is due primarily to the direct application of these processes in the fabrication of semiconducting devices and the manufacture of patterned silicon substrates for the very-large-scale integrated (VLSI) circuits which underlie so much of modern day electronics.

What are the basic questions addressed?

Two main topics are currently being addressed within the broad context of the above project. These are:

(i) The etching of silicon surfaces by atomic fluorine and chlorine.

(ii) The chemisorption of boron onto silicon and its subsequent role in the adsorption of alkali-metals onto semiconductor surfaces.

The specific questions we are seeking to answer are:

(i) What are the chemisorption sites and equilibrium configurations for different exposures of atomic fluorine and chlorine on the (001) and (111) surfaces of silicon?

(ii) How do the species SiFx and SiClx (1<x<4) desorb from the Si(001) and Si(111) surface? That is, what are the actual mechanisms whereby etching of these surfaces occurs?

(iii) What are the main chemisorption sites and equilibrium structures appropriate to the adsorption of boron onto the (001) and (111) surfaces of silicon?

(iv) Why is the chemisorption behaviour of boron on these surfaces so different from that of the other Group III elements?

What are the results to date and the future of the work?

(i) We have determined the equilibrium topologies for the Si(001):F and Si(001):Cl systems up to exposures of 1.0 monolayer (ML). Some of these structures are believed to be important for the subsequent etching of the (001) surface of silicon by atomic chlorine and fluorine. We now plan to extend these calculations beyond monolayer coverage in order to study the formation and subsequent desorption of the various SiFx and SiClx (1<x<4) species on the Si(001) surface. This should enable us to delineate the actual mechanisms whereby this surface is etched by atomic fluorine and chlorine.

(ii) The minimum energy structures for the chemisorption of up to 5 fluorine or chlorine atoms on to the adatom and rest atom sites of the Si(111)7x7 surface have been determined. In addition, these studies have revealed the processes whereby the SiClx and SiFx species (1<x<4) are desorbed from this highly complex reconstructed surface. They thus represent the first theoretical determination at a truly fundamental level of the complete process of halogen etching of a silicon surface. We believe that this work is a major advance in this field. Our plan is to now perform accurate ab initio calculations to reliably determine the energetics of these etching processes as well as molecular dynamics calculations to study the behaviour of these processes at different temperatures and exposures (see (iv) below).

(iii) Preliminary calculations of the chemisorption of boron onto the Si(001)2x1 surface have confirmed structures suggested by experiment as well as predicting two novel topologies. In order to understand the unique behaviour of boron on silicon surfaces we plan to extend these calculations to include the (111) surface. Both cluster and periodic calculations will be employed to determine the minimum energy configurations and their associated electronic structure. Analogous calculations with aluminium and gallium may also be attempted in an endeavour to elucidate the difference with boron.

(iv) A molecular dynamics program suitable for studying chemisorption and desorption processes on semiconductor surfaces has been developed. A new model potential for silicon has also been formulated which accurately describes the Si(111)7x7 reconstructed surface, as well as the (001) and (111) 2x1 surfaces and bulk silicon, which are characterised by predominantly tetrahedral bonding. We now propose to employ this potential to simulate the interaction of atomic fluorine and chlorine with the (001)2x1 and (111)7x7 surfaces of silicon at different temperatures and exposures. The appropriate Si-F and Si-Cl potentials will be determined by fitting to energy curves derived from accurate first-principles calculations. These molecular dynamics calculations should prove highly significant in enabling us to simulate the actual chemisorption and desorption (etching) processes which occur on these technologically important silicon surfaces.

What computational techniques are used and why is a supercomputer required?

Experience has shown that in order to gain a full understanding of these chemisorption/desorption systems it is necessary to employ a number of different and complementary techniques. Our general approach is thus to first employ a semiempirical technique, which is quite fast computationally, to delineate the various possible minimum energy chemisorption sites. More rigorous geometry optimisation calculations will then be conducted using the ab-initio GAUSSIAN92, GAMESS'92 or CRYSTAL'92 Hartree-Fock (HF) programs or the local density functional (LDF) program that we have recently acquired from the Fritz-Haber Institute in Berlin. These programs allow for both cluster and periodic calculations as well as the incorporation of correlation effects. Electronic structure determinations are performed using either CRYSTAL'92 or the Berlin LDF code. At present, the bulk of these calculations are being conducted on our own workstation network which includes 3 IBM RISC6000 machines and 2 DEC-ALPHA's. To obtain sufficiently accurate results to discriminate between different possible structures or reaction pathways, however, it is often necessary to employ very sophisticated basis sets (HF) or very large numbers of plane waves (LDF). Such calculations are beyond the capacity of our workstations and demand the facilities of a supercomputer. Fortunately, optimised versions of both the GAUSSIAN92 and GAMESS'92 programs are already available on the VP2200. Our approach to date has thus been to obtain the best HF geometries possible on our workstations and then use these structures as the starting point for more accurate calculations on the VP2200. This has worked very successfully and provided the results for two separate publications in the last six months. At some point, we may consider setting up a suitably optimised version of CRYSTAL'92 and/or the Berlin LDF program so that we can also compute definitive electronic structures. At present, however, all of our allocated time is being used in computing state-of-the-art HF minimum energy configurations for these semiconductor:adsorbate systems.

Publications

Chemisorption of fluorine on the silicon (001) 2x1 surface, M W Radny and P V Smith, Vacuum 45, 293 (1994).

Half monolayer and monolayer chemisorption of fluorine on the silicon (001) 2x1 surface, M W Radny and P V Smith,Surf. Sci. 301, 97 (1994).

A SLAB-MINDO study of half monolayer and monolayer chemisorption of chlorine on the silicon (001) surface, M W Radny and P V Smith, Surf. Sci. 319, 232 (1994).

An ab-initio Hartree-Fock study of the B/Si(001)2x1 adsorption system, M W Radny and P V Smith, accepted by Vacuum.

Theoretical study of the chemisorption and etching of fluorine on the Si(111)7x7 surface, P L Cao and P V Smith, submitted to Surface Science.

Chemisorption and etching mechanism of chlorine on the Si(111)7x7 surface: a theoretical study, P V Smith and P L Cao, submitted to Surface Science.

Hartree-Fock analysis of the interaction of fluorine with the Si(111)7x7 surface, M W Radny, P V Smith and P L Cao, submitted to Surface Science.

Ab initio Hartree-Fock study of the B/Si(001)2x1 adsorption system, M W Radny and P V Smith, Eighteenth ANZIP Solid State Physics Conference, Wagga Wagga, NSW, Australia, February, 1994.

An ab initio Hartree-Fock Study of the B/Si(001)2x1 Adsorption System, M W Radny and P V Smith, 14th General Conference of the Condensed Matter Division of the European Physical Society, Madrid, Spain, March, 1994.

An ab initio Hartree-Fock study of the B/Si(001)2x1 adsorption system, M W Radny and P V Smith, 17th International Seminar on Surface Physics, Kudowa, Poland, June, 1994.

Chemisorption and Etching Mechanism of F on Si(111)7x7: a theoretical study, P L Cao and P V Smith, 6th APPC and 11th AIP Congress, Brisbane, Australia, July, 1994.

A theoretical study of the etching of the Si(111)7x7 reconstructed surface due to the chemisorption of atomic chlorine, P L Cao, M W Radny and P V Smith, Nineteenth ANZIP Solid State Physics Conference, Wagga Wagga, Australia, February, 1995.

An ab initio Hartree-Fock study of the F/Si(111)7x7 adsorption system, M W Radny, P V Smith, and P L Cao, Nineteenth ANZIP Solid State Physics Conference, Wagga Wagga, Australia, February, 1995.

A new empirical potential for molecular dynamics simulations of silicon, P C L Stephenson, M W Radny and P V Smith, Nineteenth ANZIP Solid State Physics Conference, Wagga Wagga, Australia, February, 1995.