NE 203 Assignment-3 2020

1 Oxygen Solubility in Silicon [1, 2]

Figure 1: Oxygen Solubilityand Diffusivity in Silicon.

Cs(T) = 9 · 1022 · exp(−1.52eV/kT) cm−3

15ppm wafer has oxygen solubility of 15 · 10−6 × 5 · 1022cm−3 = 7.5 · 1017cm−3. Wafer is
super-saturated at 1234oC, as seen from the graph.

2 Denuding
Heating the wafer in Oxygen free atmosphere causes out diffusion of the dissolved Oxygen
from the wafer. Nitrogen ambient (oxygen free) keeps the surface concentration of
Oxygen in Silicon negligible.

3 Denuded Zone: Oxygen out-diffusion [1, 2]

DO = 0.13 · exp(−2.53eV/kT) cm2/s
Solution of diffusion differential equation with infinite sink boundary condition: C(x)

= Cberf(
Cb is the concentration in the bulk. In our case, Cb = 15ppma = 7.5 · 1017cm−3. C(x < 10µm)
< 1ppma = 5 · 1016cm−3 to ensure no precipitate formation in the active area.
C(10µm) < 5 · 1016cm−3 = 7.5 · 1017erf(10µm/2LD)cm−3

Minimum time required for denuding can be found by substituting the diffusion constant
in the diffusion length for a given temperature. (tmin(T) = L2D/D(T))

-8 Oxygen out-diffusion from Silicon 12 Comparison between denudation recipe

Figure 2: (a) Minimum denuding time for different temperatures (b) Comparison
between 1ppma and 10ppma DNZ Oxygen concentration (without considering out-
diffusion from the back surface)

Assuming availability of a high temperature ( ), 10hr anneal at

1355oC can give required denuding in the active area ([O]i < 1ppm)! Compared to that, the
recipe given in Fig.1 denudes the active area to [O]i < 10ppma, which would give a small
precipitate density in the active area, and 1100oC is generally the maximum temperature
achievable in many furnaces. Note: In our case, DNZ with [O]i < 10ppma is x < 100µm.

4 Distance between Nuclei

For a nucleus density of 1011cm−3, assuming negligible volume of the nucleus, the average
inter-nucleus distance d = (1011cm−3)−1/3 = 2.15µm

5 Super-Saturation
For a concentration that is saturated at a given temperature, becomes super saturated, as
the saturation solubility decreases with decreasing temperature.
6 Nucleation
Gibbs Free Energy of a spherical precipitate of radius r is given by:

Maxima occurs at r = r∗ = −2γ/∆GV , and the Gibbs Free Energy barrier G(r∗).

Figure 3: GFE vs nucleus size at different temperatures

Temperature(oC) 900 700 500

r∗(Ao) 4.6 3.2 2.5
G(r∗) = GCrit(eV) 2.41 1.15 0.67
Table 1: Nucleation Barrier

Nucleation Barrier reduces with increasing super-saturation (decreasing

temperature), as G(r) reduces for all r, because of reduction in volume free energy of
formation (Gv). Those (r < r∗) nuclei, which are on verge of stability at higher temperature,
are stable at lower temperature because of reduced volume free energy of formation (Gv),
and hence the critical radius also reduces.

7 Equilibrium Nucleus Density

Concentration of nucleation [O]i = no = 15ppm = 7.5 · 1017cm−3.
Substituting r∗ in G: The critical barrier
Equilibrium nucleus density : n(T) = noexp(−Gcrit/kT) Note: the
nucleation barrier also varies with the temperature.
 Figure 4: Equilibrium Nucleus Density at different temperatures

From the figure, we can see that nucleus density n > 1012cm−3 for T < 690oC.

8 Nucleation Rate
Oxygen impingement frequency on critical nucleus (ripening rate): ν = 4π(r∗)2ZnoD/a
Where, Z = 0.001, a = 2.35Ao, D: diffusivity of Oxygen in Silicon. Nucleation Rate:

R(T,Ci) = n(T,Ci)ν(T,Ci)

Nucleation rate is maximum at 728oC. For neq = 1012cm−3, nucleation rate R(690oC) =
3.66 · 106cm−3s−1. It will take 76 hours to form 1012cm−3 nuclei. During this time, DNZ
(Denuded Zone), which has 1ppm concentration is considerably less supersaturated (TE
= 950oC), hence has lower nucleation rate. From the graph, we can see that no nucleus
will form in 10hr in DNZ. This (690oC, 76 hours) does not compare well with the recipe
given in the figure (700oC, 4hr). Note: In that case, DNZ concentration is 10ppm, which
causes a nucleation rate of 2.4·103cm−3/s, or 6.9·107cm−3 nuclei, which are still negligible
compared to 1012cm−3.

9 Fick’s Laws in Spherically Symmetric Coordinates

First Law: J = −D∇rC(r,θ,φ,t)
Second Law: ∂C∂t = −∇rJ(r,θ,φ,t)
Hence,

Assuming the concentration to be symmetric in θ,φ all the time. (Unnecessary, but
simplifying)
Figure 5: (a) Nucleation Rate and Nucleus Density using TE = 1350oC (b) Nucleation Rate
at 728oC for different initial concentrations, with TE taken from solid solubility curve [2]

10 Steady state solution of the Oxygen concentration

Boundary Conditions: C(R)=

∂ 2∂C
(r )=0
∂r ∂r
2∂C
r =A
∂r
C(r) = B − A/r
C(R) = B − A/R = Cs
C(∞) = B = Ci
R
C (r )= C i − (C i − C s )
r

11 Flux at the interface

12 Precipitate Growth Rate
In dt time, J(R)4πR2dt atoms of Oxygen enter the precipitate, increasing the volume of the
precipitate by 4πR2dR. If the number density of the Oxygen in precipitate is CP, J(R)4πR2dt
= 4πR2dR(CP − Cs) added atoms increase the radius by dR. Note: Cs atoms are already
present in the expected incremental volume!!

13 Precpitate Growth at 700oC

substituting the values, t700oC = 4.6 · 107s = 13000hr Good

luck with that !!

14 Precpitate Growth at 1000oC

substituting the values, t1000oC = 4.3 · 104s = 12hr

That is a reasonable time. Compared to that, the recipe has 15hr of growth step at the
same temperature.

15 Gettering the Impurities

Impurities get attracted near/in the precipitate and hence, the precipitates act as sink for
diffusion! Diffusion time :, where Ro(= 1.33µm for N = 1011cm−3) is radius of
spherical volume per precipitate such that = 1. From the derivation in the
appendix, we can see that τo is enough to getter the impurities from the bulk.

To remove the impurities from the DNZ, the impurities need to diffuse from the active
area towards the precipitates. Assuming the concentration at the edge of the DNZ: C(L) =
0, and no flux at the surface = 0, the general solution of 1D diffusion equation by
variable separable method:
where, τa = L2/D, on substituting boundary and initial conditions becomes
πx t
C (x,t )= C i cos ( )exp (− 2. 5 )
2L τa
Relative solubilities of different metals in Silicon (extended for 600-1000oC) Diffusivities of different metals in Silicon (extended for 600-1000oC)

Figure 6: Impurity Solubility and Diffusivity for different metals [4]

Note: n > 1 harmonics are ignored as they decay quickly, leaving only n = 1 after an initial
transient. In t = τa, C/Ci < 10−9, where, L = 120µm. Time required to clean different
impurities from the bulk and the DNZ:
Metal Fe Cu Aus
τo 10ms 0.1ms 100ms
8τa 10min 12s 3.2hr
Table 2: Gettering time for different impurities at 1000oC

References
[1] J. C. Mikkelsen, Jr., ”Oxygen, Carbon, Hydrogen, and Nitrogen in Silicon”, Materials
Research Society, Princeton, NJ, p. 419, 1986.

[2] Patel, Dasharath, ”Precipitation and denuded zone formation in silicon single crystal
during high temperature anneal”. Master’s Theses. Paper 789, 1994.

[3] A. Borghesi et. al. ”Oxygen precipitation in silicon”,Journal of Applied Physics, Vol 77,
1995.

[4] Klaus Graff, ”Metal impurities in Silicon-Device Fabrication”, Springer-Verlag Berlin

Heidelberg, 1995.