next up previous contents
Next: 3. Advanced Oxidation Model Up: 2. Physics of Thermal Previous: 2.6 The Deal-Grove Model

Subsections


2.7 The Massoud Model

As described in the Section 2.6.6 the Deal-Grove model can not satisfy the so-called thin film oxidation. It should be taken into account that in the middle of the 60's, when Deal and Grove developed their model, oxide thicknesses under 30 nm were not fabricated in the semiconductor technology. Hence, there was no need to predict or simulate the growth for such thin oxide films. But with shrinking device geometry also the oxide thickness is decreasing. Hence, sometime in the 80's, MOS gates with thin thicknesses were grown and so the problem became important. In order to handle also thin film oxditation, in this time Massoud and other people, reengineered the Deal-Grove concept [56]. The yielded model is most suitable for thin oxide films, but it should be mentioned that it also works well for other oxide thicknesses. The price for this common validity is the higher complexity of this model.

2.7.1 Experimental Fitting

It was found that the SiO$ _2$ growth rate in the thin regime for a wide variety of experimental conditions can be expressed as [56]

$\displaystyle \frac{dx_0}{dt}=\frac{B}{2x_0+A}+C_1 \mathrm{exp}\Big(-\frac{x_0}{L_1}\Big)+C_2 \mathrm{exp}\Big(-\frac{x_0}{L_2}\Big).$ (2.33)

The first term on the right side of (2.33) is the linear-parabolic term where $ B$ and $ B/A$ are the parabolic and linear rate constants, respectively, as defined by Deal and Grove, but their values in the Massoud model are completely different [57]. In Arrhenius-expression the rate constants can be written in the form

$\displaystyle B=C_B \mathrm{exp}\Big(-\frac{{E_B}}{{kT}}\Big)$ (2.34)
$\displaystyle \frac{B}{A}=C_{B/A} \mathrm{exp}\Big(-\frac{{E_{B/A}}}{{kT}}\Big).$ (2.35)

The values for the pre-exponential constants $ C_{B}$, $ C_{B/A}$ and the activation energies $ E_{B}$, $ E_{B/A}$ for different crystal orientations are listed in Table 2.3. In this model it is adverse that $ C_{x}$ and $ E_{x}$ are not valid for the whole temperature range and so $ C_{x}$ and $ E_{x}$ differ from temperatures less and more than 1000$ ^{\circ }$C.

Table 2.3: Pre-exponential constants and activation energies for $ B$ and $ B/A$ [57].
Temperature Range T < 1000$ ^{\circ }$C T > 1000$ ^{\circ }$C
Crystal Orientation   (100) (111) (110) (100) (111)
$ C_B$ [nm$ ^2$/min] $ 1.70\times10^{11}$ $ 1.34\times10^9$ $ 3.73\times10^8 $ $ 1.31\times10^{5} $ $ 2.56\times10^5 $
$ E_{B}$ [eV] $ 2.22 $ $ 1.71 $ $ 1.63$ $ 0.68$ $ 0.76$
$ C_{B/A}$ [nm/min] $ 7.35\times10^{6}$ $ 1.32\times10^7$ $ 4.73\times10^8 $ $ 3.53\times10^{12} $ $ 6.50\times10^{11} $
$ E_{B/A}$ [eV] $ 1.76 $ $ 1.74 $ $ 2.10$ $ 3.20$ $ 2.95$


In (2.33) the two exponential terms represent the rate enhancement in the thin regime. They are defined in terms of pre-exponential constants $ C_{1}$ and $ C_{2}$ and characteristic lengths $ L_{1}$ and $ L_{2}$. The first decaying exponential has a characteristic lengths $ L_{1}$ in the order of 1nm, it is nonzero for the first 5nm of oxide growth, and vanishes for oxides thicker than 5nm. The second decaying exponential has a characteristic lengths $ L_{2}$ in the order of 7nm and it is present from the onset of oxidation to an oxide thickness of about 25nm, where it decays to zero and the growth becomes pure linear-parabolic.

Another formulation of (2.33), where the two terms which represent the rate enhancement in the thin regime are decaying exponentially with time, can be expressed as [58]

$\displaystyle \frac{dx_0}{dt}=\frac{B+K_1 \mathrm{exp}\big(\frac{- t}{\tau_1}\big)+K_2 \mathrm{exp}\big(\frac{- t}{\tau_2}\big)}{2x_0+A},$ (2.36)

where all four parameters $ K_1$, $ K_2$, $ \tau _1$ and $ \tau _2$ were fitted to an Arrhenius-type expression

$\displaystyle K_1=K_1^0 \mathrm{exp}\Big(- \frac{E_{K1}}{kT}\Big),$ (2.37)
$\displaystyle K_2=K_2^0 \mathrm{exp}\Big(- \frac{E_{K2}}{kT}\Big),$ (2.38)
$\displaystyle \tau_1=\tau_1^0 \mathrm{exp}\Big(- \frac{E_{\tau 1}}{kT}\Big),$ (2.39)
$\displaystyle \tau_2=\tau_2^0 \mathrm{exp}\Big(- \frac{E_{\tau 2}}{kT}\Big).$ (2.40)

The pre-exponential constants and activation energies in the above expressions (2.37)-(2.40) for different crystal orientations and dry oxidation in the temperature range form 800-1000$ ^{\circ }$C are listed in Table 2.4.

Table 2.4: Arrhenius-expression parameters for the pre-exponential constants $ K_1$ and $ K_2$, and the time constants $ \tau _1$ and $ \tau _2$ in the 800 - 1000 $ ^{\circ }$C range [58].
Crystal Orientation   (100) (111) (110)
$ K_1^0$ [nm$ ^2$/min] $   2.49\times10^{11}  $ $    2.70\times10^9  $ $   4.07\times10^8 $
$ E_{K1}$ [eV] $ 2.18 $ $ 1.74 $ $ 1.54$
$ K_2^0$ [nm$ ^2$/min] $   3.72\times10^{11} $ $   1.33\times10^9 $ $   1.20\times10^8 $
$ E_{K2}$ [eV] $ 2.28 $ $ 1.76 $ $ 1.56$
$ \tau_1^0$ [min] $   4.14\times10^{-6} $ $   1.72\times10^{-6} $ $   5.38\times10^{-9} $
$ E_{\tau 1}$ [eV] $ 1.38 $ $ 1.45 $ $ 2.02$
$ \tau_1^0$ [min] $   2.71\times10^{-7} $ $   1.56\times10^{-7} $ $   1.63\times10^{-8} $
$ E_{\tau 2}$ [eV] $ 1.88 $ $ 1.90 $ $ 2.12$

2.7.2 Analytical Oxidation Relationship

As already mentioned in Section 2.6.2, it would be convenient to have an analytical expression for the oxide thickness $ x_0$. For this purpose (2.36) is rewritten as

$\displaystyle (2x_0+A)dx_0=\Big[ B+K_1 \mathrm{exp}\Big(-\frac{t}{\tau_1}\Big)+K_2 \mathrm{exp}\Big(-\frac{t}{\tau_2}\Big)\Big]dt.$ (2.41)

Integration of (2.41), from time 0 where the native oxide thickness is x$ _i$ to an oxidation time t where the oxide thickness x$ _0$ results in [58]

$\displaystyle x_0^2 +Ax_0=Bt+M_1\Big[ 1-\mathrm{exp}\Big(-\frac{t}{\tau_1}\Big)\Big]+M_2\Big[ 1-\mathrm{exp}\Big(-\frac{t}{\tau_2}\Big)\Big]+M_0$ (2.42)

with the substitutions

$\displaystyle M_0=(x_i^2+Ax_i), \qquad M_1=K_1\tau_1, \qquad M_2=K_2\tau_2.$ (2.43)

The equation (2.42) is quadratic and can be solved obtaining an analytic expression for the oxide thickness as a function of the oxidation time of the form

$\displaystyle x_0=\sqrt{\Big(\frac{A}{2}\Big)^2 +Bt+M_1\Big[ 1-\mathrm{exp}\Big...
...)\Big]+M_2\Big[ 1-\mathrm{exp}\Big(-\frac{t}{\tau_2}\Big)\Big]+M_0}-\frac{A}{2}$ (2.44)

This relationship describes the oxide growth in dry oxygen from the onset oxidation with an smaller than 1-2% compared to the measured data. An example in Fig. 2.19 shows the growth in thin regime for (100) oriented silicon in temperature range 900-1000$ ^{\circ }$C. The oxide thicknesses were calculated by (2.42) with the parameters from Table 2.3 and 2.4.

Figure 2.19: Oxide thickness versus oxidation time for (110) oriented silicon in dry oxygen at 900 $ ^{\circ }$C, 950 $ ^{\circ }$C, and 1000 $ ^{\circ }$C.
\includegraphics[width=0.75\linewidth,bb=29 52 716 528, clip]{curv/mass/drymas}


next up previous contents
Next: 3. Advanced Oxidation Model Up: 2. Physics of Thermal Previous: 2.6 The Deal-Grove Model

Ch. Hollauer: Modeling of Thermal Oxidation and Stress Effects