Ionization and Mass Spectrometry of Decaborane for Shallow Implantation of Boron into Silicon

Abstract

Implantation of energetic ions into solids has played a critical role in the development of semiconductor technology over the last 25 years. It has made possible precise doping of the semiconductor layers in various parts of electronic devices, as well as fine adjustment of the threshold voltage of metal-oxide-semiconductor (MOS) transistors. Fabrication of today's multilayer silicon integrated circuits may utilize 20 implantation steps with ions of different species, ranging in energy from 5 keV to a few megaelectronvolts. As the quest for higher circuit density and smaller device dimensions continues , not only lateral but also vertical dimensions of the implanted layers become critical. In particular, a further decrease in the size of MOS transistors requires the creation of very shallow junctions in the source, drain and the source and drain extension regions. The International Technology Roadmap for Semiconductors projects the need for 25 to 43 nm deep junctions in the 0.13 m semiconductor devices, expected by the year 2002 and for 20 to 33 nm junctions in the 0.1 m devices projected for the year 2005. 1 The difficulty of forming such junctions emerges as a major roadblock in further progress of silicon MOS technology. Extension of the standard ion implantation technology to very shallow depths encounters fundamental problems. The very small projected ion range requires low ion energy, particularly for light ions. The energy of B ions required for p-type ultrashallow junctions in silicon is of the order of 100 eV. Transport of ion beams of such low energy is hindered by the coulombic forces (beam space-charge), to the extent that standard ion implanters cannot deliver sufficient ion currents for commercial semiconductor implantation. New designs of implanters based, for example, on the deceleration of energetic beams in front of the target, have been introduced and are being evaluated by the industry. 2 Another solution to the problem of low energy implantation may be plasma immersion. 3 This technique , however, does not discriminate among various ion species, and its potential for precise dose control remains uncertain. An alternative approach, discussed in this paper, is based on energetic beams of cluster ions, which produce implantation effects equivalent to those of monomer ions at lower energy. In a cluster of n identical atoms impacting on a surface with kinetic energy E, each of the constituent atoms carries the energy E/n, which defines its range in the solid. Also, the ion beam of n atom cluster transports n times the mass of a monomer beam at the same charge. Thus space-charge problems in the beam transport as well as problems of target charging are minimized. At the ion source, the limit imposed by the Child-Langmuir law on the extracted beam current density is proportional to the extraction voltage to the 3/2 power but inversely proportional to the square root of ion mass. 4,5 Since the extraction voltage for a given energy per atom is proportional to n, the extracted fluence increases as n 2. These effects make cluster ion beams an attractive tool for shallow implantation. Considerable interest has been aroused recently by the reports of ultrashallow junctions formed by implantation of clusters formed by ionization of decaborane (B 10 H 14). Experimental MOS devices with such junctions were made by Fujitsu in collaboration with Kyoto University. 6,7 There was no information, however, about the implanted species, as the ion beam had not been mass analyzed. The ioniza-tion and breakup properties of a decaborane molecule subjected to energetic electron bombardment in an ion source have not been well known beyond the fact that attempts to use the compound in conventional implanter ion sources have not produced cluster ions. In a beam of decaborane (B 10 H 14), each boron atom carries approximately one-eleventh of the beam energy, and the boron dose per unit charge is ten times larger than in the case of a monomer B ion beam. Thus implantation of decaborane may be an attractive alternative to ultralow energy atomic B implantation. These kinetic energy and implantation depth considerations do not imply that the effects of boron and decaborane ions are the same. An impact of a cluster of atoms on a surface delivers simultaneously a number of atoms into a limited volume of a solid. Dynamics of collisions, involving collective motion of the cluster atoms is expected to be quite different from the collision cascades following an impact of a single atom. This may lead to differences in crystal damage , backscattering coefficients, and sputtering yields. Differences in damage may affect amorphization dose and possibly diffusion in postimplantation annealing. Backscattering and sputtering may determine the maximum concentration of the implanted species. Decaborane ions will implant not only boron but also hydrogen. Since hydrogen is substitutionally insoluble in silicon, it is not expected to contribute to excess self-interstitials and thus to transient-enhanced diffusion. If there were any interstitials remaining after incomplete recombination of Frenkel pairs generated by hydrogen impacts, they would very likely diffuse to the surface. A hydrogen atom carries only a small fraction of the decaborane beam energy, about 40 eV for 5 keV implantation, and will be implanted very close to the surface. It has been shown that H diffuses out of Si from even larger depths at temperatures around 750C. 8 Annealing after implantation should expel the shallow hydrogen implanted by de-Future generations of Si electronic devices will need very shallow p-n junctions, in the tens of nanometer range. Implantation of B to form p-type junctions of such low depth requires very low energies, below 1 keV, where the ion beam formation and transport are hindered by space-charge effects. Shallow implantation also can be achieved using higher energy beams of ionized large molecules , such as decaborane (B 10 H 14), since the atoms are implanted with only a fraction of the beam energy. Measurements of electron impact ionization and breakup of decaborane in the electron energy range, 25-260 eV, and temperatures up to 350C are reported here. Ions containing 10 B atoms were found to be the dominant component in all measured mass spectra. In another set of experiments, the beams of the B 10 H x cluster ions were generated in an electron impact ionization source, mass analyzed, transported through a 2.5 m long ion beam line, and implanted into Si. No significant breakup of the ions and no neutral beam component were found. Beams of ions with ten B atoms were formed more easily and are more robust than initially thought. The results confirm the potential of decaborane cluster ions for low energy implantation of boron.