Hybrid Lithography/Diffusion Technique
P-GILD is showing potential for 0.18 µm generation devices.
Ruth DeJule, Associate Editor -- Semiconductor International, 1/1/1998
Hybrid Lithography/Diffusion Technique
A projection gas immersion laser technology that couples lithography with gas immersion laser processing for ultrashallow junction formation and annealing of 0.18 µm devices has been developed by researchers at Ultratech Stepper (San Jose, Calif.) in conjunction with the Defense Advanced Research Projects Agency (DARPA) and SEMATECH. This resistless, step-and-repeat doping process uses laser light to selectively dope regions of an integrated circuit.
Unlike traditional manufacturing methodologies, this approach delivers field by field, selective high-temperature processing, while eliminating multiple masking layers, thus potentially reducing chip manufacturing costs as device geometries continue to shrink.
Projection gas immersion laser technology has been demonstrated in two applications: the formation of ultrashallow junctions using projection gas immersion laser doping (P-GILD) and the formation of self-aligned silicides using gas immersion laser annealing (GILA).
By merging lithography and dopant incorporation into a single tool, P-GILD enables a resistless, step-and- repeat doping process that uses patterned 308 nm XeCl excimer laser light to selectively dope regions of the IC. The compact, modular equipment closely resembles a stepper in both structure and operation. An excimer laser illuminates the reticle with nanosecond pulses and is then imaged by a 4X projection lens onto the wafer, which is immersed in an ambient of the desired dopant gas. The exposed regions of the wafer are rapidly heated to the melting point by the laser, and dopant molecules on the wafer surface diffuse into the molten silicon.
Thermal anneals to nanoseconds and the elimination of 10 or more processing steps can mean ultrashallow junction formation of <30 nm and electrical resistances of <100 ohm/sq, up to one-tenth that is attainable by conventional single-step, nonthermal techniques for source/drain extensions, said Kurt Weiner, senior program manager at Verdant Technologies (San Jose, Calif.). Incorporated into existing process architectures, these ultrashallow and/or heavily doped junctions may reduce power dissipation and increase operational speeds compared to traditional ion implantation (Fig. 1), Weiner noted.
![]()
Fig. 1: P-GILD has demonstrated impurity profiles with junction depths as shallow as 30 nm, with high doping concentrations.
GILA, a derivative of P-GILD, was developed to extend the manufacturability of titanium silicide (TiSi2) to geometries 0.18 µm and below. Using GILA, microcrystalline C49TiSi2 precursor layers that scale easily down to 0.07 µm linewidths have been successfully developed but without the increases in resistivity that plague traditional rapid thermal annealing (RTA) techniques (Fig. 2).
![]()
Fig. 2: For laser-formed TiSi2, low resistivity to linewidths as small as 0.07 µm are indicated for various energies.
In this silicidation process, the initial RTA step for TiSi2 formation is replaced by a patterned laser annealing step. All other steps remain un-changed. A pulsed 308 nm laser irradiation is absorbed by the titanium overlayer. The heat generated promotes titanium diffusion into the silicon where melting occurs. Upon recrystallization, a thin layer of TiSi2 is formed.
This laser-induced silicide formation has inherent advantages over conventional thermal processing, according to Weiner. The metal is always the diffusing species; the thermal budget is essentially zero for the laser step; the silicidation process is insensitive to dopants and other impurities in the silicon; and no dopant segregation occurs into the silicide. Furthermore, there is no diffusion of metal atoms into the silicon, and smoother interfaces and differential silicide thickness over the source/drain and gate regions may be realized. This enables chipmakers to form thick silicide layers on the gate where low resistance is essential, while forming thin silicide layers on the source/drain regions.
Production tools are now beng developed to ensure that these new manufacturing techniques can be transferred into production by the end of the decade.