BF/sub 3/ and AsH/sub 3/ plasma immersion ton implantation (PIII) are used for formation of ultra-shallow p+/n and n+/p junctions. Both PIII processes are found to result in energetic ion implantation and dopant deposition on the wafer surface. Retained doses after implant and anneal were measured using nuclear reaction analysis (NRA), secondary ion mass spectroscopy (SIMS) and Rutherford back-scattering (RBS). Boron deposition from 2 kV BP/sub 3/ PIII allows for higher retained doses and lower sheet resistance than possible with mass-analyzed 2 keV BF/sub 2//sup +/ implantation, for which the retained dose is sputter-limited. However, for the conditions used here, this lower resistance was also accompanied by a deeper junction depth, so that the benefit from lower resistance was lost. For arsenic implants, and for the conditions used in this work, PIII 2 kV AsH/sub 3/ and mass-analyzed 2 keV As/sup +/ implantation produced a junction with similar sheet resistance and junction depth, whereas 10 kV AsH/sub 3/ PIII produced a shallower, more abrupt junction than the mass-analyzed counterpart of similar sheet resistance.

A Faraday cup dosimetry system was developed and characterized to address the issues of plasma immersion ion implantation (PIII) dose measurements. Pure ion current was measured by using an electrostatic suppression mechanism combined with high-bandwidth fiber-optic electronics to isolate high-voltage pulses and eliminate the primary and secondary electron and displacement currents. The ion-current waveform measured by the Faraday cup was verified by an XPDP1 particle-in-cell simulation. All of the positively charged ions striking the target surface were counted for implant dose by the Faraday cup so that both high-energy implant dose during the pulse and low-energy implant dose between pulses can be separately determined. The dose of the high-energy implant during pulses, which is more influential on the junction depth, can be measured with a fairly good accuracy, although the low-energy implant dose cannot be accurately measured due to more complicated surface effects. Compared with other dosimetry methods for PIII doping processes, the Faraday dosimetry technique offers better repeatability and controllability for PIII processes due to its direct, in-situ manner.

The deposition-etch characteristics of BF3 plasmas were quantitatively measured and analyzed using a deposition monitor, and were correlated with plasma parameters. It was found that by controlling pressure and rf power, the source could be operated in regimes which were either deposition or etch dominant. This data was then applied to a −2 kV plasma immersion ion implantation BF3 process to explain retained dose characteristics. A good qualitative agreement between the deposition-etch data and implanted retained dose data was obtained.

Charge control during the source-drain implants in the CMOS process is critical to avoid damage to the thin gate dielectric. PIII has always been touted as having good charge control, since the charge deposited on the gate during the implant pulse is neutralized by the plasma electrons between pulses, thus having a ‘built in plasma flood’. However, unless the time between pulses Tbet is long, the gate can float to a positive potential in excess of 15 V. We show that Tbet can be arbitrarily reduced without inducing any gate voltage stress by applying a positive bias during this time. This requires a special ‘Charge Balance Modulator’, as well as an additional electron source. The balance of charge is measured by an in situ charge monitor, which simulates the charge on a CMOS gate. Data from this charge monitor, as well as SPIDER wafers are compared to theory. The PIII process can be considered to be a blend of ion implantation as well as plasma processing. Depending on process conditions, combinations of ion implantation, as well as plasma deposition and etch combine to yield the final dopant profile. Some of these components may be advantageous or disadvantageous, depending on the process. An important component in controlling the relative amounts of these components is the pulse width and frequency. The Charge Balance Modulator allows control of these parameters independent of charging issues.

In plasma immersion ion implantation, the wafer is negatively pulsed while immersed in a dc ambient plasma. During this high voltage pulse, the sheath expands, and plasma ions are accelerated to the wafer. The essential character of this plasma sheath expansion can be described by a simple mathematical model, first proposed by Lieberman. In this article, we build on Lieberman’s model, extending it to describe the ion current before and after the pulse. We find that a dip in ion current is predicted immediately after the pulse, due to the depletion of ions within the sheath. This simple model is tested using Faraday cup data, and is also compared to a particle-in-cell simulation.

Much has been published recently stating the effectiveness of plasma immersion ion implantation (PIII) in producing ultra-shallow junctions (USJ). Although PIII is capable of producing USJ implants as effectively as conventional mass-analyzed ion implantation, there are several differences. One such difference, due to sheath dynamics and pulse characteristics, is that PIII implants always contain low energy components. Although such (shallower) components are not deleterious for USJ formation, there is much concern that they can be controlled repeatably. Others have developed sheath models which relate pulse current to ion energies, and have compared predicted currents to measurements of the total platen current. However, the total platen current includes contributions due to displacement currents (needed to create the high voltage negative pulse) as well as both plasma and secondary electron currents. We developed a Faraday cup embedded within the wafer platen which is capable of measuring the ion current alone. This ion current data, along with the platen voltage data has been be used to test and further refine the sheath model. The model was then used to predict the energy content of the ions, which was further tested by comparison with as-implanted SIMS profiles

Many applications of ion sources (e.g., ion implantation in the semiconductor industry) are very sensitive to particles and have tight specifications on allowable particle number and size. Among the sources of particles are the ion source itself (due to either nucleation in the plasma, or ion bombardment of the surfaces), and the extraction electrodes (due to ion bombardment). This article investigates the processes to which such particles are subjected during their flight through the extraction electrodes. They travel at much lower velocity than the accelerated ions due to their much larger mass, and so are bombarded by these increasingly energetic ions. The processes considered during the trajectory of the particle are: charging, acceleration in the electrode fields, entrapment within the suppression gap, heating from ion bombardment, radiation cooling, melting, vaporization, and Coulomb explosion. These processes are all modeled simultaneously as the trajectory of the particle is followed. A general condition for particle trapping is developed, and it is found that in the case of silicon particles, these particles can be annihilated by a combination of Coulomb disintegration and vaporization.

A method is proposed to prepare ultrathin silicon oxynitride films for gate dielectrics used in deep submicron metal–oxide–semiconductor field effect transistor device structures, namely plasma immersion N implantation into SiO2 films. Plasma immersion implantation pulse voltages in the range 200–1000 V, and fluences from 1016 to 1017 N cm−2 were implanted into thermally grown SiO2 films, with thicknesses between 3 and 6 nm. The areal densities of N and O in the resulting oxynitride films were determined by nuclear reaction analysis, before and after annealing in high-vacuum. N, O, and Si profiles in the films were determined with subnanometric depth resolution by medium energy ion scattering. The results indicate that plasma immersion ion implantation allows for shallow and controlled deposition of significant amounts of nitrogen (up to 3.8 nm of equivalent Si3N4 thickness). Implantation is accompanied by moderate damage at the oxynitride/Si interface which can be recovered by thermal annealing.