Semiconductor doping - Inside Mines

Semiconductor doping PHGN/CHEN/MLGN 435/535: Interdisciplinary Silicon Processing Laboratory

Si solar Cell

• Two Levels of Masks - photoresist, alignment • Etch and oxidation to isolate – thermal oxide, deposited oxide, wet etching, dry etching, isolation schemes • Doping - diffusion/ion implantation • Metallization - Materials deposition, PVD, CVD

What’s a metal, a semiconductor?

PHGN/CHEN/MLGN 435/535: Interdisciplinary Silicon Processing Laboratory

IV

How do we “dope” a semiconductor

Electrons and holes PHGN/CHEN/MLGN 435/535: Interdisciplinary Silicon Processing Laboratory

Conduction Band Ec

ED

EA

Ev Valence Band

Sheet Resistance, what is it? PHGN/CHEN/MLGN 435/535: Interdisciplinary Silicon Processing Laboratory

What is the Resistance of this bar of material with resistivity ρ? t W L R = ρ L/Wt We can rearrange to get a film dependent quantity called the Sheet Resistance Rs = ρ/t =R / (L/W) Notice L/W is unit less, but gives us the number of “squares” in the length of the bar. The units of Rs are ohms, but they are often given as Ω / .

Sheet Resistance - Four Point Probe PHGN/CHEN/MLGN 435/535: Interdisciplinary Silicon Processing Laboratory

If Probe spacing is: • Larger than film thickness • Smaller than distance to edge of film • Probe points are “small” Using a four point approach is a standard technique for eliminating the effects of contact resistance

Rs=4.53 V/I and ρ=Rst where t is thickness

How do we get the doping? PHGN/CHEN/MLGN 435/535: Interdisciplinary Silicon Processing Laboratory

Rs and t give us ρ, which gives us doping (but we must know t)

Another way to get doping - from C-V of a diode PHGN/CHEN/MLGN 435/535: Interdisciplinary Silicon Processing Laboratory

Formation of a p-n junction

Formation of a Schottky junction

1/C2 vs V PHGN/CHEN/MLGN 435/535: Interdisciplinary Silicon Processing Laboratory

C-2

Slope gives carrier Concentration

x-intercept give Vbiv Assumes an abrupt junction Schottky, p+n or n+p What if the line isn’t straight?

How about the thickness of our Oxide? PHGN/CHEN/MLGN 435/535: Interdisciplinary Silicon Processing Laboratory

Again, C = εA/W, so we should have another way to measure W. In practice, we must be careful about what C we use.

Corresponds to oxide thickness

What about trapped charge?

Inversion in an MOS structure PHGN/CHEN/MLGN 435/535: Interdisciplinary Silicon Processing Laboratory

accumulation (negative bias)

no bias

inversion (positive bias)

What about I-V Characteristics? PHGN/CHEN/MLGN 435/535: Interdisciplinary Silicon Processing Laboratory

Forward biased pn junction: Probability that carriers are over the barrier is like a Boltzmann factor

But, there is also an electric field pushing carriers back so at V = 0 there should be no current. We can write this in a simpler form as:

What about when light is shining on the device?

Note, there is a sign difference with respect to the capacitance analysis

How can we tell the carrier type


Thermovoltage

ee e

Hall Effect •  carrier type •  mobility •  sheet concentration

Hot Probe

V e

e

e

ee

Other methods of getting at the carriers PHGN/CHEN/MLGN 435/535: Interdisciplinary Silicon Processing Laboratory

•  •  •  •  • 

SIMS RBS – Rutherford Backscattering Polaron profiler Spreading Resistance ...

Doping - reminder PHGN/CHEN/MLGN 435/535: Interdisciplinary Silicon Processing Laboratory

Goal of Doping: Substitution of atoms with excess or deficiency of valence electrons e.g. B or P substituting for Si Diffusion doping (in fact most doping) is typically done in two steps: (Almost all doping is now ion implantation) Predeposition - Use a source to create the desired dose Drive in - Source at surface removed, additional diffusion to get desired distribution (in ion implantation the anneal also removes damage and activates the dopant).

Generic Predeposition Process PHGN/CHEN/MLGN 435/535: Interdisciplinary Silicon Processing Laboratory

Deliver Dopants to Partially Masked Substrates •  Diffusion (Hot) •  Ion Implantation (Cold)

Structure: Dopants Mask: Oxide, Nitride, Photoresist Silicon

Dopant delivery Options for Diffusion PHGN/CHEN/MLGN 435/535: Interdisciplinary Silicon Processing Laboratory

CB

Gas Source: •  Nasty Gases: AsH3, PH3, B2H6 •  Very similar to Deal –Grove Oxidation

Liquid Source:

δ

Cs

xj Co Ci

•  SOG: Spin-On Glass •  Doped SiO2 dissolved in solvents •  Apply exactly like Photoresist

Solid Source: •  Glass Discs (B2O3, P2O5) •  Close-space Sublimation •  Vapors sublime/diffuse/react

Which is Best?

Drive-in - estimating the profile PHGN/CHEN/MLGN 435/535: Interdisciplinary Silicon Processing Laboratory

Fick’s law - You need the PDE, but you also need the boundary conditions! C(z,0) = 0, Z ≠ 0 dC(0,t)/dz = 0 C(∞,t) = 0 ∞

∫ C(z,t)dz = Q

T

= constant

0

Solution: €

QT C(z, t) = e πDt

⎛ -z 2 ⎞ ⎜ ⎟ ⎜ 4 Dt ⎟ ⎝ ⎠

We can model the drive in step from our homework, here after a P predep with p8545 we had a sheet resistance of 12Ω/ and depth of 1.1µm. This gave a carrier concentration of 5x1019/ cm3 and a surface concentration of 5.5x1015/cm2

Characteristic Length Scale Diffusion Length

What about the diffusion Coefficient? PHGN/CHEN/MLGN 435/535: Interdisciplinary Silicon Processing Laboratory

Use first three terms in Fair’s vacancy model. 2 n − ⎡ n ⎤ 2− I told you to assume n~ni ~1019/cm3 o D = D + D + ⎢ ⎥ D Is this reasonable? ni ⎣ ni ⎦ From Campbell table 3.2 (1100C=1373K) Do = 3.9cm2/s e-(3.66/k1373) = 1.43 x 10-13cm2/s D- = 4.4cm2/s e-(4.0/k1373) = 9.13 x 10-15cm2/s D2- = 44cm2/s e-(4.37/k1373) = 4.00 x 10-15cm2/s D = 1.56 x 10-13cm2/s

Simulations


Suprem-IV is a process simulation tool developed at Stanford University

nanoHub TCAD tools PHGN/CHEN/MLGN 435/535: Interdisciplinary Silicon Processing Laboratory

https://nanohub.org/tools

Suprem simulation of boron predep and drive-in PHGN/CHEN/MLGN 435/535: Interdisciplinary Silicon Processing Laboratory

Boron Diffusion

Log10(Boron)

20

Boron Predep 1100C 30 min.

15 Boron drivein 1100C 30 min.

10

Boron predep in gas at 5 x 1020/cm3 concentration followed by drive-ins.

Boron drivein 1100C 60 min.

5 0 0

2

4

Depth in microns

Boron drivein 1100C 60 min 200 angstrom oxide cap

Effect of oxide cap on profile near the surface Boron Diffusion

21.0 20.5

Log10(Boron)

Why 5x1020/cm3? 1)  Damage threshold 2)  Solubility limit 3) B partial pressure 1)  Dimensional argument

20.0 19.5 19.0 18.5 18.0 0

0.5

1

1.5

Depth in microns

2

2.5

Solid Solubility, what is it?


1100C

5x1020/cm3

Oxide is an effective anti-diffusion barrier for Si VLSI? PHGN/CHEN/MLGN 435/535: Interdisciplinary Silicon Processing Laboratory

1)  2)  3)  4) 

For boron but not for phosphorus For phosphorus but not for boron It works well for both It depends

Final Topic on Diffusion: Oxide


How fast do dopants diffuse through oxide? Diffusivity important, Solubility important Consider Do of Boron Si prefactor 0.37cm2/s SiO2 prefactor 0.0003cm2/s

Activation Energy Activation Energy

3.46eV 3.53eV

Now Do of Phosphorous Si prefactor 3.9 cm2/s SiO2 prefactor 0.19 cm2/s

Activation Energy Activation Energy

3.66eV 4.03eV

• Oxide is often used as a diffusion mask- how thick does it need to be? • Oxide is used for isolation - does it isolate? What is the thermal load? • Oxide is also a gate dielectric with heavily B doped polysilicon gates diffusion through gate is an issue M Metal Doped polysilicon O

Oxide

S

Silicon

Suprem-IV Wet Oxide then Diffusion PHGN/CHEN/MLGN 435/535: Interdisciplinary Silicon Processing Laboratory

Oxide antidiffusion barrier

Log10(Boron)

20 15

60 min wet O2 at 1000C, 30 min boron predep at 1100C 30 minute boron predep at 1100C

10 5 0 0

Effect of oxide cap on 1 2 3 profile near the surface

4

Depth in microns

Substrate is P doped at 1 x 1014/cm3, Wet oxide growth at atmospheric pressure for 60 minutes at 1000C, Boron predep from 30 minutes at 1100C in gas with a concentration of 5 x 1020/cc.

Simulation of predep and drive-in to find junction depth PHGN/CHEN/MLGN 435/535: Interdisciplinary Silicon Processing Laboratory

1000°C P predep in p-type wafer doped at 1x1017/cm3. 1100°C drive in. How long to get a 4.0µm deep junction?