1

1. Introduction and application.
2. Dopant solid solubility and sheet resistance.
3. Microscopic view point: diffusion equations.
4. Physical basis for diffusion.
5. Non-ideal and extrinsic diffusion.
6. Dopant segregation and effect of oxidation.
7. Manufacturing and measurement methods.

Chapter 7 Dopant Diffusion

NE 343: Microfabrication and thin film technology

Instructor: Bo Cui, ECE, University of Waterloo; http://ece.uwaterloo.ca/~bcui/

Textbook: Silicon VLSI Technology by Plummer, Deal and Griffin

2

Doping in MOS and bipolar junction transistors

Doping is realized by:

• Diffusion from a gas, liquid or solid source, on or above surface. (no longer popular)
• Ion implantation. (choice for today’s IC)
• Nowadays diffusion often takes place unintentionally during damage annealing…
• “Thermal budget” thus needs to be controlled to minimize this unwanted diffusion.

In this chapter, diffusion means two very different concepts: one is to dope the substrate from source on or above surface – the purpose is doping; one is diffusion inside the substrate – the purpose is re-distribute the dopant.

3

Application of diffusion

In the beginning of semiconductor processing, diffusion (from gas/solid phase above surface) was the only doping process except growing doped epitaxial layers.

Now, diffusion is performed to:

• Obtain steep profiles after ion implantation due to concentration dependent diffusion.
• Drive-in dopant for wells (alternative: high-energy implantation), for deep p-n junctions in power semiconductors, or to redistribute dopants homogeneously in polysilicon layers.
• Denude near-surface layer from oxygen, to nucleate and to grow oxygen precipitates.
• Getter undesired impurities.

Diffusion is the redistribution of atoms from regions of high concentration of mobile species to regions of low concentration.

It occurs at all temperatures, but the diffusivity has an exponential dependence on T.

4

Doping profile for a p-n junction

5

Diffusion from gas, liquid or solid source

Pre-deposition (dose control) Drive-in (profile control)

• Silicon dioxide is used as a mask against impurity diffusion in Silicon.
• The mixture of dopant species, oxygen and inert gas like nitrogen, is passed over the wafers at order of 1000^oC (900^oC to 1100^oC) in the diffusion furnace.
• The dopant concentration in the gas stream is sufficient to reach the solid solubility limit for the dopant species in silicon at that temperature.
• The impurities can be introduced into the carrier gas from solid (evaporate), liquid (vapor) or gas source.

6

Comparison of ion implantation with solid/gas phase diffusion

7

1. Introduction and application.
2. Dopant solid solubility and sheet resistance.
3. Microscopic view point: diffusion equations.
4. Physical basis for diffusion.
5. Non-ideal and extrinsic diffusion.
6. Dopant segregation and effect of oxidation.
7. Manufacturing and measurement methods.

Chapter 7 Dopant Diffusion

NE 343 Microfabrication and thin film technology

Instructor: Bo Cui, ECE, University of Waterloo

Textbook: Silicon VLSI Technology by Plummer, Deal and Griffin

8

Dopant solid solubility

Solid solubility: at equilibrium, the maximum concentration for an impurity before precipitation to form a separate phase.

Figure 7-4

9

Solid solubility of common impurities in Silicon

10

Solubility vs. electrically active dopant concentration

Not all impurities are electrically active.

As has solid solubility of 2×10²¹ cm^-3.

But its maximum electrically active dopant concentration is only 2×10²⁰ cm^-3 .

V: vacancy

Figure 7-5

As in substitutional site, active

Inactive

11

Resistance in a MOS

For thin doping layers, it is convenient to find the resistance from sheet resistance.

Figure 7-1

Sheet resistance R_S

• : (bulk) resistivity

x_j: junction depth, or film thickness…

R=R_s when l=w (square)

Figure 7-2

13

Ohm’s law:

Mobility μ:

By definition:

Therefore:

Finally:

Where:

Important formulas

σ: conductivity; ρ: resistivity; J: current density; E: electrical field

v: velocity; q: charge; n, p: carrier concentration.

14

Sheet resistance

N is carrier density, Q is total carrier per unit area, x_j is junction depth

For non-uniform doping:

This relation is calculated to generate the so-called Irvin’s curves. See near the end of this slide set.

15

1. Introduction and application.
2. Dopant solid solubility and sheet resistance.
3. Microscopic view point: diffusion equations.
4. Physical basis for diffusion.
5. Non-ideal and extrinsic diffusion.
6. Dopant segregation and effect of oxidation.
7. Manufacturing and measurement methods.

Chapter 7 Dopant Diffusion

NE 343 Microfabrication and thin film technology

Instructor: Bo Cui, ECE, University of Waterloo

Textbook: Silicon VLSI Technology by Plummer, Deal and Griffin

16

Diffusion from a macroscopic viewpoint

Fick’s first law of diffusion

F is net flux.

C is impurity concentration (number/cm³), D is diffusivity (cm²/sec).

D is related to atomic hops over an energy barrier (formation and migration of mobile

species) and is exponentially activated.

Negative sign indicates that the flow is down the concentration gradient.

This is similar to other laws where cause is proportional to effect (Fourier’s law of heat flow,

Ohm’s law for current flow).

Figure 7-6

17

Intrinsic diffusivity D_i

Intrinsic: impurity concentration N_A, N_D < n_i (intrinsic carrier density).

Note that n_i is quite high at typical diffusion temperatures, so "intrinsic" actually applies under many conditions. E.g. at 1000^oC, n_i =7.14×10¹⁸/cm³.

E_a: activation energy

D⁰(cm²/s) E_a(eV)

B 1.0 3.46

In 1.2 3.50

P 4.70 3.68

As 9.17 3.99

Sb 4.58 3.88

Figure 7-15, page 387

18

Fick’s second law

The change in concentration in a volume element is determined by the change in fluxes in and out of the volume.

Within time Δt, impurity number change by:

​

​

During the same period, impurity diffuses in and out of the volume by:

​

​

Therefore:

​

​

Or,

​

​

Since:

​

​

We have:

If D is constant:

Figure 7-7

19

Solution to diffusion equation

At equilibrium state, C doesn’t change with time.

Diffusion of oxidant (O₂ or H₂O) through SiO₂ during thermal oxidation.

20

Gaussian solution in an infinite medium

C→0 as t →0 for x>0

C →∞ as t →0 for x=0

∫C(x,t)dx=Q (limited source)

This corresponds to, e.g. implant a very narrow peak of dopant at a particular depth, which approximates a delta function.

Important consequences:

• Dose Q remains constant
• Peak concentration (at x=0) decreases as 1/ √t
• Diffusion distance from origin increases as 2 √Dt

Figure 7-9

At t=0, delta function dopant distribution.

At t>0

21

Gaussian solution near a surface

A surface Gaussian diffusion can be treated as a Gaussian diffusion with dose 2Q in an infinite bulk medium.

Note: Pre-deposition by diffusion can also be replaced by a shallow implantation step.

1. Pre-deposition for dose control

2. Drive in for profile control

Figure 7-10

22

Gaussian solution near a surface

Surface concentration decreases with time

Concentration gradient

Junction depth

At p-n junction

23

Error function solution in an infinite medium

An infinite source of material in the half-plane can be considered to be made up of a sum of Gaussians. The diffused solution is also given by a sum of Gaussians, known as the error-function solution.

This corresponds to, e.g. putting a thick heavily doped epitaxial layer on a lightly doped wafer.

At t=0

C=0 for x>0

C=C for x<0.

erfc: complementary error function

Figure 7-11

24

Error function solution in an infinite medium

Evolution of erfc diffused profile

Important consequences of error function solution:

• Symmetry about mid-point allows solution for constant surface concentration to be derived.
• Error function solution is made up of a sum of Gaussian delta function solutions.
• Dose beyond x=0 continues to increase with annealing time.

Figure 7-12

25

Error function solution in an infinite medium

Properties of Error Function erf(z) and Complementary Error Function erfc(z)

For x << 1

For x >> 1

26

Boundary condition: C(x,0)=0, x≠0; C(0,t)=C_s; C(∞,t)=0

Error function solution near a surface

Constant surface concentration at all times, corresponding to, e.g., the situation of diffusion from a gas ambient, where dopants “saturate” at the surface (solid solubility).

Pre-deposition dose

Cs is surface concentration, limited by solid solubility, which doesn’t change too fast with temperature.

27

Successive diffusions

• Successive diffusions using different times and temperatures
• Final result depends upon the total Dt product

When D is the same (same temperature)

When diffused at different temperatures

As D increases exponentially with temperature, total diffusion (thermal budget) is mainly determined by the higher temperature processes.

For example, the profile is a Gaussian function at time t=t₀, then after further diffusion for another 3t₀, the final profile is still a Gaussian with t=4t₀=t₀+3t₀.

(The Gaussian solution holds only if the Dt used to introduce the dopant is small compared with the final Dt for the drive-in i.e. if an initial δ/delta function approximation is reasonable)

1-

Irvin’s curves

Motivation to generate Irvin’s curves: both N_B (background carrier concentration), R_s (sheet resistance) and x_j can be conveniently measured experimentally but not N₀ (surface concentration). However, these four parameters are related by:

​

​

​

​

Irvin’s curves are plots of N₀ versus (R_s, x_j) for various N_B, assuming erfc or half-Gaussian profile. There are four sets of curves for (n-type and p-type) and (Gaussian and erfc).

1-

Irvin’s curves

Four sets of curves: p-type erfc, n-type erfc, p-type half-Gaussian, n-type half-Gaussian

Explicit relationship between: N₀, x_j, N_B and R_S.

Once any three parameters are know, the fourth one can be determined.

Figure 7-17

30

Example

Design a boron diffusion process (say for the well or tub of a CMOS process) such that ρ_s=900Ω/square, x_j=3μm, with C_B=1×10¹⁵/cm³.

From (half-Gaussian) Irvin’s curve, we find

C_s << solubility of B in Si, so it is correct to assume pre-deposition (here by ion implantation) plus drive-in, which indeed gives a Gaussian profile.

The average conductivity of the layer is

31

Example (cont.)

Assume drive-in at 1100^oC, then D=1.5×10^-13cm²/s.

Pre-deposition dose

32

Now if we assume pre-deposition by diffusion from a gas or solid phase at 950^oC, solid solubility of B in Si is C_s=2.5×10²⁰/cm³, and D=4.2×10^-15cm²/s.

The profile of this pre-deposition is erfc function.

However, the pre-deposition time is too short for real processing, so ion-implantation is more realistic for pre-deposition.

Example (cont.)