Page 148 - Microsensors, MEMS and Smart Devices - Gardner Varadhan and Awadelkarim
P. 148

128    SILICON MICROMACHINING:  BULK




                                  PP(n)           PP(p)
                                            ::::•:•:•  •:• !&*(  •^ - -  -—  -=
                                    j  Oxide  gr<)Wth | :!:|R:j: ^  Oxide  growth
                                      on n  Si a -  Mi  S] on p-Si  and
                                      pass vat c)n  /!!!!!  IE  passivation
                                    [  p-S'i s etched  m  m  m^-4-.  -__-__-_


                                                    ^fllrSSvt


                                    I**,:  , , ,|  , i  f_  - .^.-C. _=CH— S  fi
                                     ^^\
                                    Suitable voltage for p-n junction etch  stop.

     Figure 5.11  Current-voltage characteristics of n-Si and p-Si in KOH. No current  /  flows  at the
     OCP  and  the  current  stops above  the passivating potential  (PP) (Linden  et al. 1989)

       The  growth  of  an  anodic oxide  is  believed  to result  from  the  progressing  competition
     between  the  oxidation  of  the  silicon  and  the  dissolution  of  the  oxide  products  at  the
     silicon-solution  interface.  Ellipsometric  measurements  have  given  evidence  of  such  an
     oxide  layer (Palik et al.  1985a).  The results have been fitted with  a multilayer  model  that
     suggests a graded  connective  layer  of SiO x.
        Measurements  on  etch  rates  in  KOH  as  a  function  of  applied  potential  (Palik  et  al.
     1985a)  show  that  at  the  OCP  there  is  very  little  difference  in  the  etch  rates  between
     the  n-  and  p-type  substrates,  whereas  there  is  a  marked  difference  at  other  potentials.
     Furthermore,  the  etch  rate  is  not  proportional  to  the  current,  and  in  fact,  the  etch  rate
     attains  a  maximum at  the  OCP  (current is  zero)  and  slows  down  as  the  PP  (current is a
     maximum) is approached; eventually, the etch  stops when the current drops. The etch rates
     at the  OCP,  therefore,  seem  to be independent of free-carriers  concentration, and  it  seems
     reasonable  to  suggest  that  the  chemical  mechanism  is  dominant.  However,  at  the  other
     potentials,  in  which the  etch  rates  for  the  n-  and  p-type  dopants  differ,  this  is probably
     not  the  case.  At  these  potentials,  it  is  possible  that  a  combined  chemical  and  electro-
     chemical  mechanism  is responsible  for  the etching process.  A  chemical-electrochemical
     mechanism  is  proposed  (Palik  et al.  1985b)  in  which  a  chemical  reaction  takes  place  at
     the  silicon  surface  with  sequential  attacks  of  Si-Si  bonds  by  H 2O  and  OH,  resulting
     in  the  discharge  of  the  OH  into  the  etching  layer  giving  soluble  silicate  of  the  form
     Si(OH) 2O 2.
       It is then followed  by the  more rapid  electrochemical  reaction  that oxidises underlying
                                          –
     bonds.  The  reduction-oxidation  couple OH /H 2 O  is  assumed  to  supply  the  species  for
     etching, electrolysis,  and  oxidation.
       The  hydrogen  gas  produced  per  Si  atom  has  been  measured  (Palik et al.  1985b)  and
     was  found  to  be  2H 2  per  Si;  this  indicates  a  chemical  reaction  for  a  Si—Si  bond  at  the
     OCP  as  follows:

                         Si -  Si + H 2O    Si -  H + Si - OH             (5.10)
   143   144   145   146   147   148   149   150   151   152   153