ESI Oliveros 2012 - Carnegie Mellon...

21
CRYSTALLIZATION MODELING FOR SOLAR CELLS GERMAN OLIVEROS ENERGY SYSTEMS INITIATIVE MEETING CENTER FOR ADVANCED PROCESS DECISION MAKING 1

Transcript of ESI Oliveros 2012 - Carnegie Mellon...

  • CRYSTALLIZATION  MODELING  FOR  SOLAR  CELLS      

    GERMAN  OLIVEROS    ENERGY  SYSTEMS   INITIATIVE  MEETING    CENTER  FOR  ADVANCED  PROCESS  DECISION  MAKING  

    1

  • MOTIVATION  

    2

  • THE  DEMAND  FOR  SOLAR  ENERGY  

    q  Urgent  need  to  rely  less  on  tradiConal  energy  sources  •  Unsustainable  •  Contaminant  

    3

  • SILICON  SOLAR  CELL  SUPPLY  CHAIN  

    Silica  Quartz  (  

  • HORIZONTAL  RIBBON  GROWTH:  THE  MISSING  LINK?  

    •  First  conCnuous  design  patented  by  William  Shockley  in  1959:  

     

     

    Germanium  ribbons   Ice  ribbons  

    •  First  experimental  work  published  by  Carl  Bleil  in  1968:  

    •  Improvements  of  the  process  reported  by  B.  Kudo  in  1979:  

    5

  • RESEARCH  OBJECTIVE    

    To   design   and   analyze   a   con;nuous  wafering   system   using   the   concepts   of  Systems   Engineering   and   Materials   Science   to   find   desirable   design  parameters,  relevant  process  variables  and  op;mal  opera;ng  condi;ons  

     

    …under  the  hypothesis  that  a  smooth  and  uniform  film  is  formed  on  top  of  the  melt.  

    6

  • CONCEPTUAL  DESIGN  

    Challenges:  

    •  CrystallizaCon  and  stability  (micro-‐scale)  to  achieve  single  crystalline  Silicon  film  •  Complex  fluid  flow  and  heat  transfer  interacCon  •  Impurity  modeling  •  Lack  of  experimental  data  and  proof  of  concept  

    Vpull

    7

    •  ConCnuous  process  to  produce  150  –  250  micron  wafer    •  Silicon  floats  on  top  of  its  melt  •  No  sawing,  hence  no  material  losses   VcrystallizaCon  

  • FLUID  FLOW  AND  HEAT  TRANSFER  

    ( )

    ( )( )

    ( ) TTV

    AVqBBF

    FPVVV

    V

    2

    3

    2

    2

    PrRe1

    1Re1

    0

    ∇=∇⋅

    +

    −=

    +∇−∇=∇⋅

    =⋅∇Continuity:

    2D Navier-Stokes:

    Source Term:

    Energy Transport:

  • CRYSTALLIZATION  DYNAMICS  

    "↓$ &↓'$ ()↓$   /(, = -↓$ (↓↑2 )↓$ /(/↑2  

    "↓0 &↓'0 ()↓0   /(, = -↓0 (↓↑2 )↓0 /(/↑2   (&↓0   /(, =1(↓↑2 &↓0 /(/↑2  

    &↓$ =2&↓0 

    -↓$ ()↓$ /(/ − -↓0 ()↓0 /(/ =ρ∆345/4,  −1(&↓0 /(/ = 45/4, (&↓0∗ − &↓$∗ ) )↓0 = )↓8 + 4)↓9 /4& &↓0 

    Solid  

    Interface  CondiCons  

    -()↓$   /(, =σε()↑4 − )↓↑4 ↓∞ )

    (&↓$   /(: =0

    Top  

    Vy(y,t)  

    Qs(y,t)  

    Ql(y,t)  

    ΔH  

    Ny(y,t)

    Liquid  

    Closed  Domain  composed  of  two  moving  subdomains  

    Qrad

    “Hot”  

    (&↓$   /(: =0

    )= )↓$  Booom  

    ASSUMPTIONS  

    a)  UnidirecConal  solidificaCon  b)  Neglect  convecCon  in  the  melt  

    c)  No  mass  diffusion  in  solid,  limited  mass  diffusion  in  liquid:  Ds  =  0,  0  <  DL  <  ∞   9

  • CRYSTALLIZATION  DYNAMICS  

    •  Slow  CrystallizaCon  VelociCes    avoids  formaCon  of  Impurity  Boundary  Layer  

    "↓$ &↓'$ ()↓$   /(, = -↓$ (↓↑2 )↓$ /(/↑2  

    "↓0 &↓'0 ()↓0   /(, = -↓0 (↓↑2 )↓0 /(/↑2   (&↓0   /(, =1(↓↑2 &↓0 /(/↑2  

    &↓$ =2&↓0 

    -↓$ ()↓$ /(/ − -↓0 ()↓0 /(/ =ρ∆345/4,  −1(&↓0 /(/ = 45/4, (&↓0∗ − &↓$∗ ) )↓0 = )↓8 + 4)↓9 /4& &↓0 

    Solid  

    -()↓$   /(, =σε()↑4 − )↓↑4 ↓∞ )

    (&↓$   /(: =0

    Top  

    Liquid  

    (&↓$   /(, =0

    )= )↓$  Booom  

    Interface  CondiCons  

    10

  • CRYSTALLIZATION  DYNAMICS  •  Contrast  this  with  faster  interface  velociCes  

    "↓$ &↓'$ ()↓$   /(, = -↓$ (↓↑2 )↓$ /(/↑2  

    "↓0 &↓'0 ()↓0   /(, = -↓0 (↓↑2 )↓0 /(/↑2   (&↓0   /(, =1(↓↑2 &↓0 /(/↑2  

    &↓$ =2&↓0 

    -↓$ ()↓$ /(/ − -↓0 ()↓0 /(/ =ρ∆345/4,  −1(&↓0 /(/ = 45/4, (&↓0∗ − &↓$∗ ) )↓0 = )↓8 + 4)↓9 /4& &↓0 

    Solid  

    )= )↓";

  • CRYSTALLIZATION  DYNAMICS  

    •  Impurity  effect  on  thermal  profile  is  seen  in  undercooling  effect  at  the  interface  

    •  High  levels  of  undercooling  lead  to  interfacial  instability!  

    "↓$ &↓'$ ()↓$   /(, = -↓$ (↓↑2 )↓$ /(/↑2  

    "↓0 &↓'0 ()↓0   /(, = -↓0 (↓↑2 )↓0 /(/↑2   (&↓0   /(, =1(↓↑2 &↓0 /(/↑2  

    &↓$ =2&↓0 

    -↓$ ()↓$ /(/ − -↓0 ()↓0 /(/ =ρ∆345/4,  −1(&↓0 /(/ = 45/4, (&↓0∗ − &↓$∗ ) )↓0 = )↓8 + 4)↓9 /4& &↓0 

    Solid  

    -()↓$   /(, =σε()↑4 − )↓↑4 ↓∞ )

    (&↓$   /(: =0

    Free  Boundary  

    Liquid  

    (&↓$   /(, =0

    )= )↓$  Booom  

    Interface  CondiCons  

    12

  • MULLINS-‐SEKERKA  STABILITY  

    •  Morphological   stability   theory  developed  by  Mullins  and  Sekerka  proved  condiCons  for  stability  under  diffusive  heat  and  mass  flow.  

    •  Stability  is  determined  by  temperature  gradients  in  both  phases,  concentraCon  

    gradients  on  the  liquid  side  and  surface  tension  effects.  

    Apply  PerturbaCon  Crystal  

    Melt   I(y,t)  =  I(y,t)  +  δ(t)  sin  (ωx)    

    Crystal  

    Melt  

    Grows?  

    Decays?  

    Crystal  

    Melt  

    Crystal  

    Melt  

    13

  • MULLINS-‐SEKERKA  CONDITION  

    ( )

    ( ) c

    cM

    mGpDVGG

    DVmGp

    DVGGp

    DVTV

    ωω

    ωωωωω

    δδ

    2*'

    *2*'*2 2

    +⎥⎦

    ⎤⎢⎣

    ⎡⎟⎠

    ⎞⎜⎝

    ⎛−−

    ⎭⎬⎫

    ⎩⎨⎧

    ⎥⎦

    ⎤⎢⎣

    ⎡⎟⎠

    ⎞⎜⎝

    ⎛−+⎥⎦

    ⎤⎢⎣

    ⎡⎟⎠

    ⎞⎜⎝

    ⎛−+−⎥⎦

    ⎤⎢⎣

    ⎡⎟⎠

    ⎞⎜⎝

    ⎛−Γ−

    =

    INSTABILITY   =          CAPILLARITY            +   THERMAL  GRADIENTS  

    CONCENTRATION  GRADIENTS  

    +  

    14

  • PROOF  OF  CONCEPT  Challenges:    How  to  make  it  work?  

    1.  CrystallizaCon  should  begin  at  the  center  of  the  melt  surface  

    2.  Once  film  is  formed,  how  to  develop  a  pulling  mechanism  that  resembles  the  operaCon  with  

    silicon?  

    •  Start  seeding  process  inside  of  the  bath  Strategy:  

    •  Flat  non-‐conducCng  solid  surface  to  facilitate  nucleaCon  

    •  HeaCng  on  the  sidewalls  

    •  Water  inlet  

    Start  pulling  …and  hope  for  the  best  

    15

  • PROOF  OF  CONCEPT  

    16

  • THANK  YOU  

    17

  • EXISTING  TECHNOLOGIES  

    Edge-‐Defined  Film  Fed  Growth  (EFG)   Molded  wafer  (MW)  

    Shaping die

    Silicon melt Substrate

    Silicon sheet

    Crystal growth Ribbon

    pulling

    Ribbon-‐Growth  on  Substrate  (RGS)  String  Ribbon  

    18

  • EXISTING  TECHNOLOGIES  

    Method   Thickness†  (µm)  

    Pulling  speed  

    (cm/min)  

    Throughput  (cm2/min)  

    Furnaces/  100  MWP  

    EFG   300   1-‐2   136   100  SR   300   1-‐2   13   600  RGS   300   650   10140   2-‐3  MW   450   300   4680   6-‐7  

    19

  • RELEVANT PROPERTIES AND SYSTEM PARAMETERS

    Property Value

    Density of Liquid Silicon 2293 [kg/m3]

    Density of Solid Silicon 2570 [kg/m3]

    Thermal Conductivity of Liquid Silicon 18 [W/m ⁰C]

    Thermal Conductivity of Solid Silicon 58 [W/m ⁰C]

    Heat Capacity of Liquid Silicon 968 [J/kg ⁰C]

    Heat Capacity of Solid Silicon 1040 [J/kg ⁰C]

    Mass Diffusivity of Aluminum in Silicon 7e-8 [m2/s]

    Segregation Coefficient of Aluminum in Silicon 2.8e-3

    Slope of the Liquidus Line in Al-Si Phase Diagram -0.000095 [⁰C/ppm] (or -0.95 [⁰C/%wt]

    Melting Point of Pure Silicon 1414 ⁰C

    Temperature of “Cold” Wall 1300 [⁰C]

    Temperature of “Hot” Wall 1500 [⁰C]

    Initial Temperature 1500 [⁰C]

    Initial Concentration of Aluminum in Silicon Melt 5000 [ppm]

    Height of System 3 [cm]

    Desired Thickness 1.2 millimeters

  • CURRENT  TECHNOLOGY:  THE  CZOCHRALSKI  AND  WIRE  SAW  PROCESS  

    •  Process  discovered  by  Jan  Czochralski  in  1916  •  SCll  one  of  the  most  common  techniques  to  produce  the  silicon  wafers    •  High  purity  product  with  minimal  amount  of  imperfecCons  •  Extensive  modeling  by  Robert  Brown  at  MIT    

    Disadvantages  

    û   Final  product  has  circular  cross-‐secConal  area  û   Batch  Process  

    û   Tremendous  amount  of  material  losses    

    Half  of  “technological  effort”  is  lost   21