Effects of "fast" rapid thermal anneals on sub-keV boron and BF2 ion implants

The events of "fast" ramp-rates (up to 425(deg)C/s) and spike anneals are investigated for 025 keV 05 keV and 10 keV ^sup 11^B^sup +^ and for 11 and 22 keV BF^sub 2^ at a dose of 1e15/cm^sup 2^ Below an implant activity threshold where no extended defaults form, fast ramp-rates become important in minimizing the thermal diffusion constituting and reducing the junction profundity Above this implant energy doorsill TED minimizes the advantages of these fast ramp-rates. Annealing in a depressed and controlled O^sub 2^ ppm in N^sub 2^ ambient further brings diffusion by minimizing/ eliminating oxygen related enhanced diffusion events while simultaneously optimizing anneal reproducibility and across-the-wafer uniformity.

Key words: Fast ramp rates, spike anneals, sub-keV implants, enhanced diffusion, boron

INTRODUCTION

Ion implant and annealing requirements to achieve shallow source-drain extensions that satisfy down to 010 micron technology has been reported extensively by dint of us.1-21 In an earlier publication 11 upon the effect of ramp-rates, it was shown that for implants at energies above a defect-free activity threshold, ramp-rates alone do not matter; on the contrary that cool-down rates, and time at temperature were important. It was subsequently shown by dint of us18 that below an might threshold for a dose of 1e15/cm^sup 2^ and for samples annealed in a depressed O^sub 2^ ppm in N^sub 2^ ambient, all enhanced diffusion was eliminated/minimized. It is the intent of this research to investigate the consequences of fast ramp-rates (up to 425(deg)C/s) in this defect-free implant regime where no reach outed defects were ever observed to form (and no T observed) to determine whether ramp-rates do matter.

EXPERIMENTAL

Implants were performed upon a Varian VIISion-80 PLUS ion implanter into 200 mm n-type CZ wafers of resistivity 10-20 (Omega)-cm using either ^sup 11^B^sup +^ or ^sup 49^BF^sub 2^+ ions at energies of 025 keV 05 keV or 20 keV and 11 keV and 22 keV respectively. All implants were at a dose of 1e 15/cm^sup 2^ and at a tilt and twist angle of 0(deg) To eliminate any variability in the native oxides upon the wafers, all wafers were dipped in a 40:1 H^sub 2^O: HF (49%) acid for 30 sec within 1 h before implant. After implantation, the wafers were annealed employing a STEAD AST SHS-3000 rapid thermal processing (RTA) a whole All implant sets were annealed using the identical anneal matrix of conditions, consisting of: a "spike" anneal at 1050 deg C in either a 33 ppm O^sub 2^ in N^sub ^2 or 1000 ppm O^sub 2^ in N^sub 2^ ambient at ramprates of 50 deg C 250 deg C or 425 deg C/ and a cool-down rate of 87 deg C/ Ramp circle of times were controlled from 530 deg C for each of the ramp rates reported. To render certain no temperature "over-shoot" at the high temperature anneal pace the ramp rates were gradually adjusted downward above the last 30 deg C The thickness of the resulting oxide formed during the anneal stage was measured using a PLASMOS SD3000 ellipsometer. All wafers were probed using a Prometrix OmniMap RS35c fourpoint probe a whole following a BOE strip. Secondary ion mass spectrometry (SIMS) deepness profiling was performed at Evans East Inc. using a Physical Electronics phi 6600 quadrupole SIMS spectrometer to determine the junction profundity (using the O^sub 2^ leak technique) and to determine the post-anneal retained dose.7

Result and Discussion:

Figure 1 is a SIMS overlay of a 10 keV ^sup 11^B^sup +^ 1e15/ CM^sup 2^ ion implant at 0 deg tilt. Depicted in this figure are the profiles for the "as-implanted" wafer and those from sum of two units 1050 deg C "spike" RTA wafers annealed at sum of two units different ramp-rates, 50 deg C/ and 425 deg C/ the pair wafers were annealed in an ambient of 33 ppm O^sub 2^ in N^sub 2^ and moderately colded at a rate of 87 deg C/ The sheet resistance (R^sub s^) retained dose (D^sub r^) and junction motion at a concentration of 7e17/cm^sup 3^ for these implants are listed in Table I. Referring to Fig. 1 it is observ that the ramp-rates do not significantly affect the overall dopant profile and final junction deepness (i.e. 6 A at 7e17/cm^sup 3^) This is in agreement with our earlier findings, 111618 that in the might regime where TED is occurring, ramp-rates do not matter. Referring to Table I, it is seen that the faster ramp-rate yields a slightly higher retained dose and sheet resistance value. The increase in sheet resistance for the faster anneal could issue from either a lower perforation mobility or a lower electrical activation. This is commonly being researched in more detail. As the efficiency of the implant is decreased to 500 eV and then to 250 eV (Figs. 2 and 3) the results of faster ramp-rates in creating shallower junctions is now observ In addition, as the might is lowered, the overall dopant diffusion is reduc and the difference between the observ junction profundity for faster ramp-rates, particularly above 1e18/cm^sup 3^ increases. through the time an energy of 250 eV is reached, all diffusion appears to be thermal18 and a 55 A difference in junction motion (at 3e18/cm^sup 3^) is observ between the sum of two units ramp-rates (compared to 38 A for the 500 eV implant). Table I also displays that R^sub S^ and D^sub r^ are higher for all the faster ramp-rates.