Changing the Dopant Concentration. Diffusion Doping Ion Implantation

Size: px
Start display at page:

Download "Changing the Dopant Concentration. Diffusion Doping Ion Implantation"

Transcription

1 Changing the Dopant Concentration Diffusion Doping Ion Implantation

2 Step 11 The photoresist is removed with solvent leaving a ridge of polysilicon (the transistor's gate), which rises above the silicon wells.

3 Step 12 Chemical doping implants phosphorous (green) deep within the silicon wells surrounded by the silicon dioxide and polysilicon layers to produce negatively doped silicon.

4 Diffusion vs. Implantation

5 Changing the Dopant Concentration Diffusion Doping Ion Implantation

6 Diffusion: general considerations. Fick s first law: Diffusion of A molecules in pure B. J A D AB C A Where J A = net flux of A molecules at a point (cm -2 sec -1 ). D AB = diffusion coefficient of A in B (cm 2 sec -1 ). C A = concentration of A (cm -3 ). Note: C A = C A (x,y,z,t) Negative sign shows that movement is from higher concentration to lower concentration. One dimensional form: J x D C ( x, t) / AB A x Note: D is a phenomenological constant which can be calculated from appropriate physical models or determined from experiment.

7 Diffusion model: infinite source. Initial condition: C A (x,0)=0, x>0 Boundary condition: C A (0,t) = C 0 C A (,t)=0 Solution: C x, t) C A( 0 erfc 2 x D AB t where erfc( u) 1 2 u e 0 2 a da

8 Diffusion model: infinite source.continued. For large u, erfc( u) 1 e u u 2 Note also: 0 erfc( u) du 1 Concept of diffusion length. Can define diffusion length : L D 2 Dt CA( x, t) C0 erfc x L D

9 Concentration as a function of time according to infinite source model. Bxt ( ) erfc x 2 t 1 Bx j 0.1 Bx j 0.5 Bx j 1 Bx j 5 Bx j 10 Bx j x-axis units: sqrt(1/d) x j

10 Net dopant deposition and junction depth. Total dopant per square area: or Q 2 ( t) CA( x, t) dx C A 0 0 ( ) 1 QA t C0L D Increases with time as dopant supplied from infinite source. Junction depth: Junction forms at depth where dopant concentration equals background carrier concentration. C sub concentration of substrate background. Then junction depth, d j is given by d D 1 ( 2 Dt ) erfc ( Csub / C0) AB t

11 0 0 A ), ( 0 ), ( C dx t x C Q t A Diffusion model: Finite source. Initial condition: C A (x,0)=0, x> (in limit at goes to zero) Boundary condition: Solution: t D x t D Q t x C AB AB A 4 exp ), ( 2 0 (Fixed dopant per area)

12 Additional characteristics: finite source. Surface concentration (t>0) Q0 C A (0, t) Dt Junction depth: d j Q0 2 Dt[ln( C Dt sub )]

13 Dopant distributions from finite source. 6 Concentrations as function of depth and doping time. Cx j Cx j 0.01 Cx j 0.1 Cx j Linear Cxt ( ) exp x 2 4 t t x j 100 Logarithmic Cx j Cx j 0.01 Cx j 0.1 Cx j x j

14 Diffusion activation model. Thermal activation model: D AB = D 0 exp (-E A /kt) D 0 relates to frequency of vibrations. For simple model, D 0 ~ 0 d 2 where d is lattice constant and 0 is vibrational frequency -- approximately sec -1. Activation energy varies enormously depending on specific mechanism for diffusion. Substitutional vacancy diffusion. Interstitial displacement diffusion. Interstitial diffusion.

15 Diffusion coefficients in Silicon Substitutional diffusers D (cm 2 /sec) Activation energies Dopant E A (ev) P As Sb B Al 3.4 Ga 3.8 In

16 Diffusion coefficients in Silicon Interstitial diffusers D (cm 2 /sec) 10-8

17 Fast Diffusers in Silicon. Diffusion Coefficients of Fast Diffusants in Silicon at 900 o C Element D (cm 2 /sec) Ni 1.00E-01 O 2 Cu Pt H 2 Fe Au(s) Ag Na K Au Au(i) 7.00E E E E E E E E E E E-04 source: Wolf and Tauber, Silicon Processing. page 337. Activation energies in range of ev (relatively low).

18 Solubility limits for dopants in Silicon. Impurity concentration (cm -3 )

19 Practical methodology for silicon doping. A common strategy for defining dopant profile and concentration. Deposit known amount of dopant with narrow spatial profile (often at solubility limit) to determine net dopant quantity. This is often termed pre-deposition. Diffuse for known time at known temperature to determine profile. This is often termed drive-in. Can work backward: First: what is desired junction depth and dopant load? Second: what final profile will yield desired properties? Third: what initial conditions will give dose and profile based on a given dopant source?

20 Diffusion masking and three dimensional diffusion. Thermal oxide as diffusion barrier. erfc Gaussian

21 Some practical materials for thermal doping. Boron (p-type dopant). Normally the active dopant is B 2 O 3 deposited in situ via a number of techniques: Oxidation of boron nitride (wafer) followed by evaporation to silicon wafer. Oxidation of diborane or BCl 3 or other gaseous source. Spin-on glass materials containing boron (eg borosilicate glass chemistries). Arsenic and Antimony (n-type dopant). Gaseous sources such as arsine (AsH 3 ) or stybnine (SbH 3 ) are used (in spite of the high degree of toxicity of both). The active dopant can be As 2 O 3 or Sb 2 O 3 deposited in situ via direct evaporation in furnace (sometimes under sealed conditions). Spin-on glass materials containing arsenic or antimony can also be used. Phosphorus (n-type dopant). Phosphine (PH 3 ) can be used as gaseous source. Oxidation of gaseous POCl 3 to P 2 O 5 in situ. Transfer of P 2 O 5 from wafer sources. Spin-on glasses containing phosphorus can also be used.

22 General schematic: Diffusion furnace (bulk processing). Note: at high temperatures metals and alkalai metals diffuse rapidly through silicon. Thus fused quartz (polycrystalline SiO 2 ) is the preferred material for diffusion furnaces. Also used are silicon and silicon carbide.

23 Rapid Thermal Processing Usually single wafer Fast radiative heating of surface/ bulk is not brought to max temp. Less dislocations. Tube Furnace