Research NoteImplant dose monitoring by MOS C–V measurement
Introduction
Highly scaled CMOS technologies require the precise control of the channel implant dose during processing. The threshold voltage, the transconductance, and the effective channel length are all strongly influenced by the implanted channel dose. MOS capacitors with metal, polysilicon, or mercury gate electrodes are ideal test elements for the development, ramp up and routine monitoring of CMOS technologies. MOS structures are widely used to qualify and monitor important process modules, such as cleaning, epitaxy, oxidation, implantation, metallisation and anneals. Once the first polysilicon layer in a CMOS process is structured, the curriculum vitae of MOS structures, usually located in the scratch line area of process wafers, yields valuable information on irregularities in the process flow. For production monitoring, rapid and accurate determination of low and medium dose implants in the range <2×1012 cm−2 is still a challenging task. Four-point probe measurements lack the sensitivity for the middle and low dose range, and cannot be used on actual product wafers. Secondary ion mass spectroscopy and spreading resistance are destructive and time consuming, and therefore not suitable for in-line production monitoring. Thermal wave techniques quantify the lattice damage in the implanted wafer after ion implantation as a measure for the implanted dose, but cannot detect the electrically active dopant dose after anneals. MOS C–V is non-destructive, and yields a direct measurement of the majority charge carrier profile, but not the dopant profile. In the near surface region, i.e. in a depth range lower than about three Debye lengths from the Si–SiO2 interface, the depletion approximation, a prerequisite for MOS C–V profiling, fails. In this depth range, the approximately obtained carrier density profile differs significantly from the dopant profile. Therefore for the surface region a doping profile with reasonable accuracy cannot be determined by means of C–V.
C–V profiling is based on the measurement of the deep depletion C–V curve [1], [2]. A general prerequisite for the applicability of MOS C–V profiling is a low density of interface traps (<1010 cm−2 eV−1) to avoid an additional gate voltage stretch out in the C–V curve. Fig. 1 illustrates the strong deviation of C–V profiles from the corresponding dopant profiles in the near surface region. For several implant doses in the low and medium dose range, both the dopant profiles and the corresponding C–V profiles are shown. The surface Debye length limit implies that dose determination by integrating of MOS C–V profile is not applicable in the near surface region. Thus, this approach can be used only to determine a partial doping dose in the deeper bulk, neglecting the important surface contribution [3].
Here, we present a new approach, which overcomes this restriction. The new method enables the accurate determination of the entire dopant dose D from the surface into the bulk. The maximum dopant dose that can be measured amounts to approximately 2×1012 cm−2. This limitation results from the onset of avalanche breakdown at higher fields. Moreover, the method enables a self-consistent determination of further important MOS parameters, such as surface potential φs(Vg) characteristic and threshold voltage Vth.
Section snippets
Dopant dose
The majority charge carrier density Qsc,dep in deep depletion mode of the MOS structure is a direct measure for the electrically active dopant dose D in the semiconductor. The corresponding majority charge carrier dose Dsc can be written as
The basic idea is to determine Qsc,dep from a reference value for the majority charge carrier density in strong accumulation Qsc,acc and the change of the charge density ΔQsc, when the MOS structure is driven from accumulation towards depletion:
Experiment
To establish the validity of the method boron was implanted in p-type material with a substrate doping Nb=1015 cm−3 always at an energy of 40 keV at the following doses: 2×1011, 4×1011, 6×1011, 8×1011, 1012 cm−2. The physical oxide thickness was 4.1 nm. Fig. 2 shows the C–V characteristics for the implanted MOS samples and the non-implanted reference sample with the background doping Nb. The corresponding dopant profiles are shown in Fig. 1. Having determined the oxide capacitance Cox, Csc,acc
Summary and conclusions
We have demonstrated an accurate and practical method for threshold voltage implant dose monitoring. The reported MOS C–V technique is capable of detecting the entire electrically active implant dose in the low and medium dose range. The approach is based on the determination of the majority charge carrier density in deep depletion mode of the MOS structure. In contrast to other approaches, which only detect a partial implant dose in the deeper bulk, the new method yields the entire dose
References (4)
Non-quasistatic simultaneous HF/LF-CV measurements for rapid characterization of MOS structures
Solid-State Electron.
(1998)- et al.
Silicon impurity distribution revealed by pulsed MOS-CV measurements
J. Electrochem. Soc.
(1971)