METHODS FOR QUANTIFYING DOPANT ACCUMULATION

It is important to quantify profile smearing produced by accumulation, at any given temperature and growth rate, both to characterise the incorporation behaviour and to establish favourable regimes for the production of sharp doping

profiles. A number of methods are possible. Most in-situ studies such as Auger and

XPS, used in the assessment of dopant accumulation, determine the surface enrichment of dopants directly. For such work the segregation ratio rd is used.

However most in-situ methods tend to have poor detection limits and limited depth

resolution, and are therefore not ideal for the quantitative determination of boron accumulation from elemental sources. This is especially so for boron at low growth temperatures. Most require some form of growth interrupt and, as has already been shown in Chapter 3, surface accumulation under dynamic and static conditions are different, making this a dubious procedure. In the production of devices growth interrupts are undesirable, due to incorporation of impurities from the residual gas (such as carbon and oxygen) into the active part of the device, so that boron incorporation studies during continuous growth are more relevant.

For this work the most suitable parameter, from the point of view of characterisation of the accumulated phase during growth, is the degree of profile smearing in depth profiles as the accumulated phase depletes. This was obtained using high resolution SIMS (see Appendix I). The SIMS nomenclature is used throughout the present work, ie trailing edge slopes refer to the upslope in the

boron profile (formed by opening the cell shutter), and leading edge slopes refer to the downslope.

Slopes in SIMS boron depth profiles tend to be linear on a logarithmic (concentration) vs depth plot. The exponential dependence of concentration versus depth in decay slopes is a characteristic of ’first-order’ incorporation and has been observed for all dopants investigated in Si MBE on Si(100) substrates (eg Sb [Allen

et al 1982], In [Knall et al 1984], Ga [Nakagawa et a l 1988] and boron [Tatsumi et

al 1988 B]. It appears from the present work that this is also the case for boron,

with the SIMS dopant profiles showing linear behaviour over a wide range of temperatures, growth rates, and doping levels. The SIMS leading edge slope is characterised by the exponential decay constant A (nm) that is defined as the growth distance for the boron concentration, n, to fall from an initial value n„ to n0/e. The boron concentration in the SIMS leading edge slope can be expressed as

boron concentration n = n0 exp ( — x/ A ) (4.3)

It is usually more reliable to determine the degree of profile smearing in the leading edge slopes, since these are least affected by SIMS induced broadening (see Appendix I), allowing determination of A with good accuracy for A ^ 1.3nm for

the case o f boron [Dowsett et a l 1992]. However the boron trailing edge slopes are

also of interest [Mishima et al 1990] and qualitative discussions of these are

included in this Chapter.

A was measured by first fitting the inverse slopes (using a linear regression), to give a value in nm/decade, and then converting this into an exponential decay length, by multiplying by log10e. For boron coevaporation at the temperatures used in the present work the A parameter and the surface enrichment parameter rd (see section 4.3.3) are equivalent. It should be noted that, in general, the two will be equivalent whenever dopant is not desorbing in significant quantities from any surface-enriched layer. Evidence from the areas o f boron doped regions grown at

temperatures between 900 and 450°C (see Fig. 3. la) suggests that boron desorption is not significant during the temperatures commonly used for MBE.

4.5 BORON ACCUMULATION IN Si MBE (PREVIOUS WORK)

During the period of this study, there developed a strong world-wide interest in coevaporation p-type doping, especially using elemental boron. This was because early work suggested that elemental boron did not show significant profile smearing or the temperature dependent incorporation properties inherent to other dopants

[Kubiak et al 1985A, Kubiak et al 1985B and see Chapter 1]. However the high

crucible temperatures necessary to evaporate the element dissuaded many workers from its use, most preferring to use compound boron species (see Chapter 1). A summary of these boron incorporation studies is presented here.

4.5.1 PROFILE SMEARING USING COMPOUND BORON SOURCES

Previous work using compound sources demonstrated boron profile

smearing, as seen from Auger, XPS, and SIMS studies [de Frisart et al 1988,

Tatsumi et al 1988 B, Jackman et al 1988 and 1989]. De Frisart et al observed a

temperature dependence of boron profile smearing over a wide range of temperatures, but their results were closely correlated with oxygen incorporation

from the use of a compound (B203) source that was also temperature dependent.

Above 700°C boron profile smearing was determined to decrease with increasing temperature, in agreement with equilibrium segregation theory, with an activation energy o f *>— 0.33eV. Below this temperature boron profile smearing was sharply reduced but the doped material became polycrystalline and it was not clear whether the transition was associated with the corresponding reduction in material quality.

This phenomenon was also observed by Jackman et al (1988 and 1989) who

observed a sharp reduction in HBOz profile smearing at temperatures below 700°C,

for which oxygen incorporation was » lO ^cnr3. Jackman et al could not obtain any

quantitative data on the temperature dependence of profile smearing, since the boron doped layers were grown too close together to resolve the leading edge slopes.

Tatsumi et al used XPS to determine the degree of boron surface enrichment

at 7S0 and 800°C using a HB02 source. Growth at these elevated temperatures avoided oxygen incorporation, but meant that much o f the observed profile