eobt: from past to futurehomepages.laas.fr/nolhier/esref2015/session_c/ip_c.pdf1!! eobt: from past...

13
1 EOBT: from past to future Ludwig Josef Balk Faculty of Electrical, Information and Media Engineering University of Wuppertal, 42119 Wuppertal, Germany It is now fifty years ago that both electron beams and laser sources became commercially available to enable inspection techniques for all kinds of applications, but in special for the characterization and the testing of electronics devices. This happened more or less simultaneously with the beginning of integration of electronic components. While at the early times a simple imaging was done only, in the mid 60ies first work was carried out using all kinds of interaction products due to the impact of optical and electron beams to determine device features and their malfunctioning. Those interaction products could well be particles or photons as well as properties like electrical current and voltage. Although these kinds of testing techniques became more and more important, it took more than twenty year that a special conference on this topic was born: the 1 st European Conference on Electron and Optical Beam Testing of Electronic devices (EOBT), which was organized in the year 1987 in Grenoble by Bernard Courtois and Eckhard Wolfgang. From then on this conference continued till 1995 as an independent meeting that had attracted several hundreds of scientists and engineers. However, as quite often in research, if a new field becomes mature, the size of a conference reduces, which gave rise to the decision of merging the EOBT with ESREF due to their strong overlap in the field of failure analysis. There EOBT remained an important topic, sometimes as a special session or in other years merged with failure analysis in general. Over the years it had turned out that simple systems cannot always fulfill the tasks needed, due to the reduced sizes of structures and the demand on extreme spatial resolution as well as due to the more and more complicated vertical structure of devices, making it necessary to either prepare devices destructively or to go for sources with high vertical penetration such as for instance ion beams. And last not least socalled hybrid systems came into being enabling the simultaneous measurement of various device properties. The presentation will give a review of some of the important advantages presented in all of the EOBT conferences, without being complete, and it will try to give a view into what may be the needs for future developments. General remarks concerning failure analysis and device diagnostics If one wants to carry out a local analysis of properties of any kind of a sample, one has to carry out an experiment dedicated to the property in question and one has to couple this necessarily with a clear local definition. Whereas direct imaging techniques allow in insitu and fast studies of materials, such as optical or electron microscopes, they do not easily allow a detailed analysis of the data achieved. This is the main advantage of all kinds of scanning systems, in which a focused source is utilized to excite the specimen under test locally and where all sorts of interaction

Upload: others

Post on 07-Sep-2020

4 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: EOBT: from past to futurehomepages.laas.fr/nolhier/ESREF2015/SESSION_C/IP_C.pdf1!! EOBT: from past to future Ludwig Josef Balk Faculty of Electrical, Information and Media Engineering

 

1    

EOBT: from past to future

Ludwig Josef Balk

Faculty of Electrical, Information and Media Engineering University of Wuppertal,

42119 Wuppertal, Germany

It  is  now  fifty  years  ago  that  both  electron  beams  and  laser  sources  became  commercially  available  to  enable  inspection  techniques  for  all  kinds  of  applications,  but  in  special  for  the  characterization  and  the  testing  of  electronics  devices.  This  happened  more  or  less  simultaneously  with  the  beginning  of  integration  of  electronic  components.  While  at  the  early  times  a  simple  imaging  was  done  only,  in  the  mid   60ies   first   work   was   carried   out   using   all   kinds   of   interaction   products   due   to   the   impact   of  optical  and  electron  beams  to  determine  device  features  and  their  malfunctioning.  Those  interaction  products  could  well  be  particles  or  photons  as  well  as  properties  like  electrical  current  and  voltage.  Although   these   kinds   of   testing   techniques   became  more   and  more   important,   it   took  more   than  twenty   year   that   a   special   conference   on   this   topic   was   born:   the   1st   European   Conference   on  Electron   and   Optical   Beam   Testing   of   Electronic   devices   (EOBT),   which   was   organized   in   the   year  1987   in   Grenoble   by   Bernard   Courtois   and   Eckhard   Wolfgang.   From   then   on   this   conference  continued  till  1995  as  an  independent  meeting  that  had  attracted  several  hundreds  of  scientists  and  engineers.   However,   as   quite   often   in   research,   if   a   new   field   becomes   mature,   the   size   of   a  conference   reduces,  which   gave   rise   to   the  decision  of  merging   the   EOBT  with   ESREF  due   to   their  strong  overlap  in  the  field  of  failure  analysis.  There  EOBT  remained  an  important  topic,  sometimes  as  a  special  session  or  in  other  years  merged  with  failure  analysis  in  general.  

Over  the  years   it  had  turned  out  that  simple  systems  cannot  always  fulfill  the  tasks  needed,  due  to  the  reduced  sizes  of  structures  and  the  demand  on  extreme  spatial  resolution  as  well  as  due  to  the  more   and   more   complicated   vertical   structure   of   devices,   making   it   necessary   to   either   prepare  devices   destructively   or   to   go   for   sources   with   high   vertical   penetration   such   as   for   instance   ion  beams.   And   last   not   least   so-­‐called   hybrid   systems   came   into   being   enabling   the   simultaneous  measurement  of  various  device  properties.  

The  presentation  will  give  a  review  of  some  of  the  important  advantages  presented  in  all  of  the  EOBT  conferences,  without  being  complete,  and   it  will   try  to  give  a  view  into  what  may  be  the  needs  for  future  developments.  

 

General  remarks  concerning  failure  analysis  and  device  diagnostics  

If   one   wants   to   carry   out   a   local   analysis   of  properties  of  any  kind  of  a  sample,  one  has  to  carry   out   an   experiment   dedicated   to   the  property   in   question   and   one   has   to   couple  this   necessarily   with   a   clear   local   definition.  Whereas  direct  imaging  techniques  allow  in-­‐  

 

 

 

in-­‐situ   and   fast   studies   of   materials,   such   as  optical   or   electron   microscopes,   they   do   not  easily   allow   a   detailed   analysis   of   the   data  achieved.   This   is   the   main   advantage   of   all  kinds  of  scanning  systems,   in  which  a  focused  source  is  utilized  to  excite  the  specimen  under  test   locally   and  where   all   sorts   of   interaction  

Page 2: EOBT: from past to futurehomepages.laas.fr/nolhier/ESREF2015/SESSION_C/IP_C.pdf1!! EOBT: from past to future Ludwig Josef Balk Faculty of Electrical, Information and Media Engineering

 

2    

products   may   be   used   to   create   the  information   desired

 

Fig.1:  Overview  of  interaction  mechanisms  

The   local   correlation   can   be   achieved   by  scanning   the   location   of   the   impact   point   of  the  excitation  source,  the  spatial  resolution  by  modifying   source   parameters,   in   special  source  diameter  and  penetration  depth.  

Fig.2:   [1]   shows   all   possible   interactions   in  case   of   a   light   source   for   the   case   of   a  semiconductor  sample.  

 

If   one   uses   a   time   dependent,   in   special  pulsed,   beam   the   temporal   behavior   of   the  arising   products   and   by   this   temporal   device  properties  can  be  analysed,  too.  

Out   of   all   kinds   of   possible   scanning   sources  electron  and  optical  beams  gained  the  highest  importance,   mainly   due   to   the   easiness   for  scanning   and   due   to   a   very   mature  instrumentation.  

 

Scanning  electron  microscope  (SEM)  and  laser  scanning  microscope  (LSM)  

The   scanning   electron  microscope   (SEM)  was  invented   by     KNOLL   1935   [2]   following   the  invention   of   the   electron   microscope   by  RUSKA   1933   and   the   scanning   transmission  electron   microscope   by   VON   ARDENNE   1938  [3].  

Fig.3:  Knoll’s  electron  beam  scanner  

Although  further  development  was  carried  out  till  the  beginning  of  second  world  war,  it  took  a   long   time   till   an  SEM  became  a  commercial  product.   Essentially   it   was   OATLEY   [4],   who  finally   was   able   to   persuade   Cambridge  Instrument   Company   to   go   into   production  with   their   “Stereoscan”,   the   prototype   of   it  was   delivered   to   Dupont   Chemical  Corporation   in   the   U.S.A.   in   1964,   two   years  later   followed   up   by   the   Japanese   company  JEOL.  

 

Fig  4:  Cambridge  Stereoscan  (jpl.nasa.gov)  

Page 3: EOBT: from past to futurehomepages.laas.fr/nolhier/ESREF2015/SESSION_C/IP_C.pdf1!! EOBT: from past to future Ludwig Josef Balk Faculty of Electrical, Information and Media Engineering

 

3    

Most   important   for   the   further   use   in  electronics   was   the   improvement   in   spatial  resolution   due   to   the   invention   of   the   cold  field  emission  gun  by  CREWE  in  the  mid  60ies,  which  was  followed  up  by  the  first  commercial  FE-­‐SEM,  later  denominated  “critical  dimension  (CD)”-­‐SEM,  by  Hitachi  in  1972.  

 

Fig.5:  Hitachi,  1972,  FE-­‐SEM  HFS-­‐2  

In  case  of   laser   scanning  microscopy   (LSM)  at  first   the   laser   had   to   be   invented.   This  happened   to   be   in   1960   by  MAIMAN   for   red  light   following   up   the   original   invention   (for  microwaves)   by   TOWNES,  who   died   this   year  in  the  age  of  99.  

 

Fig.6:  Nobel  Prize  winner  Charles  Townes  

Again  it  took  a  while  for  the  LSM  to  come  into  being   due   to   the   invention   by   BARTELL   and  RITZ,   of   the   Department   of   Chemistry   at   the  University  of  Michigan,   in   the  year  1975.  And  finally   in   1982   the   first   commercial   LSM   was  produced   by   Carl   Zeiss.   In   the   following  3dimensional  imaging  became  possible  by  the  

introduction   of   confocal   operation   allowing  lateral   and   vertical   resolution   down   to   20nm  (Fraunhofer-­‐Institut  für  Solare  Energiesysteme  ISE),  again  a  huge  gap  between  1957,  the  year  when  it  was  originally  patented  by  MINSKY.  

 

Fig.7:  Confocal  LSM  (taken  form  malone.bioquant.uni-heidelberg.de)

 

Integration  of  electronic  circuits  

Although   the   first   patent   for   an   integrated  device   by   JACOBI   (Wikipedia)   originates   from  1949,   it   lasted  till   several  authors  came  along  with   the   first   realization   at   the   end   of   the  50ies,   one   of   them,   KILBY,   was   awarded   the  2000  Nobel  Prize.  

 

Fig.8:  First  Integrated  Circuit  Invented  by  Jack  Kilby,  1958,  Texas  Instruments  

Pretty  soon  the  scale  of  integration  went  on  in  a   fantastic   manner,   as   one   can   see   in  comparison  of  Fig.10.  

Page 4: EOBT: from past to futurehomepages.laas.fr/nolhier/ESREF2015/SESSION_C/IP_C.pdf1!! EOBT: from past to future Ludwig Josef Balk Faculty of Electrical, Information and Media Engineering

 

4    

 

Fig  9:  Kilby,  Dallas  News  12  September  2014  

This  tremendous  increase  in  device  density  on  a  chip  was  later  on  described  by  MOORE,  well  known  as  Moore’s  Law.  

 

Fig.10:   Moore’s   Law   (2015   Pete   Carey,   San  Jose  Mercury  News  

Obviously   this   dependence   is   associated  with  a  decrease  of  node  size,  too.  

 

Fig:11:   Node   size   (Helen   Wills   Neuroscience  Institute1,  Berkeley,  2013  

Aside   from   this   an   even   higher   complexity  arose  with  the  invention  of  devices  with  more  than   just   electrical   and   electronical  functionality,   the   so-­‐called   micro-­‐electro-­‐mechanical   systems   (MEMS),   and   later   on  with   the   increase   of   device   density   by   three-­‐dimensional   packaging.   Both   gave   rise   to   the  need   to   characterize   local   thermal   and  mechanical  properties,  too.  

 

Fig.12:  Artist’s  view  of  3D-­‐integration  

Interaction  mechanisms  important  for  determination  of  electronical  properties  

As   already   mentioned,   electron   or   optical  beam   impact   causes   multiple   interaction  products   that   can   be   used   for   device  characterization.   Although   some   of   them   are  identical  for  the  two  different  beam  sources,  a  larger   variety   is   given   for   a   primary   electron  beam.   Out   of   these   the   probably   most  important   ones   are   secondary   electrons,  induced  current,  and  luminescence.    

Secondary   electrons   firstly   allow   precise  metrological   evaluation   and   secondly   via   the  so-­‐called   voltage   contrast   (VC)   a   local  potential   distribution   of   the   device.   Electron  beam   induced   current   (EBIC)   is   a   powerful  mode   for  determining   locations  of  electrically  active   areas,   such   as   pn-­‐junctions.  Cathodoluminescence   (CL)   can   be   used   to  analyze   optoelectronic   properties   in   devices  made  from  or  with  direct  band  gap  material.  

Page 5: EOBT: from past to futurehomepages.laas.fr/nolhier/ESREF2015/SESSION_C/IP_C.pdf1!! EOBT: from past to future Ludwig Josef Balk Faculty of Electrical, Information and Media Engineering

 

5    

The  years  before   the  EOBT   conference  came  into  being  

Confining   the   review   of   these   years   to  research   using   electron   beams   (though   the  situation   for   laser   based   experiments   was  quite   similar),   one   has   to   mention   that   from  1965   onwards   many   researchers   were  involved  in  understanding  the  new  SEM  based  modes   and   in   optimizing   the   information  gained   from   them.   And   it   is   important   to  mention   that   in   these   days   research,   both   in  university   and   in   industry,  was   very   thorough  and  detailed.  

VC   was   first   detected   by   SMITH   [5]   in   1955,  however,   more   detailed   information   was  reported   by   Everhart   [6]   in   1959,   WELLS   in  1968   [7],   and   LUKIANOFF   in   1972   [8].   The  further   implementation   of   so-­‐called   beam  blankers   or   choppers   allowed   to   link   the  investigation   of   local   device   properties   with  temporal   analysis   to   enable   local  determination   of   switching   behavior   of   a  device   under   test.   Here   important   work   was  carried  out  by  PLOWS  and  NIXON   in  1969   [9]  and  GOPINATH  in  1974  [10].  

EBIC   was   deeply   studied   by   DONOLATO   [11],  LEAMY   [12],   VAN   OPDORP   [13],   and   last   not  least  by  David  B  HOLT  [14],  who  died  last  year  at   the  age  of  86,  who  published  many  papers  on   this   topic.   His   last   publication,   a   book   co-­‐authored   with   BG   Yacobi   entitled   “Extended  Defects   in   Semiconductors   electronic  properties,  device  effects  and  structures”  was  published  in  2014.  

CL  was  originally  discussed  by  WITTRY  in  1965  [15],   followed   up   by   PANKOVE   in   1968   [16]  and  BALK  in  1973  [17].  Here  a  lot  of  work  was  involved  in  introducing  spectral  analysis  of  the  emitted   light  as  well  as   its   temporal  behavior  for   characterization   of   minority   carrier  properties.  

Later   in   the  early  80ies  MOUROU  utilized   the  electro-­‐optic  effects  (Pockels  and  Kerr  effects)  for   temporal   sampling   of   high   frequency  devices  with  picosecond  resolution  [18].  

 

The  beginning  of  EOBT  

As   there   were   great   activities   in   the   field   of  local   analysis   of   electronic   devices,   many  conferences   had   incorporated   this   topic   into  their  program.  Quite  often  these  were  physics  conferences   like   IITRI   Scanning   Electron  Microscopy,  IXCOM,  or  the  Oxford  Conference  of   Semiconducting   Materials.   The   IEEE  International   Electron  Device  Meeting   (IEDM)  included  sessions  accordingly.  Most  important  were   two   conferences   with   respect   to   beam  testing:  

the   IEEE   International   Reliability   Physics  Symposium  (IRPS),  already  existing  since  1952,  and   the   International   Symposium   for   Testing  and  Failure  Analysis  (ISTFA)  since  1974.  

However,   there   was   no   suitable   conference  existing   in   Europe   dedicated   to   this   topic   in  spite  of   its  great   importance   in  these  days.   In  this  respect   it  was  a  more  than  needed  event  that   in   the   year   1986   two   researchers   in   this  field   decided   to   organize   an   according  meeting.   These   were   Eckhard   WOLFGANG  from   Siemens   AG   in   Munich   and   Bernard  COURTOIS   from   TIMA-­‐CMP   in   Grenoble,   who  finally  created  in  a  very  successful  manner  the  “1st   European   Conference   on   Electron   and  Optical   Beam   Testing”   (EOBT)   which   took  place  9–11  December  1987  in  Grenoble.    

The  EOBT  was  planned  as  a  biannual  meeting  and  attracted  a   few  hundreds  of  participants.  It   took   place   as   an   independent   conference  following   the   Grenoble   event   in   1989   in  Duisburg,  Germany,   in  1991   in  Como,   Italy,   in  1993  in  Zurich,  Switzerland,  and  finally  in  1995  in  Wuppertal,  Germany.  

Page 6: EOBT: from past to futurehomepages.laas.fr/nolhier/ESREF2015/SESSION_C/IP_C.pdf1!! EOBT: from past to future Ludwig Josef Balk Faculty of Electrical, Information and Media Engineering

 

6    

 

Fig.13:  First  EOBT  cover  

It   has   to   be   mentioned   that   WOLFGANG  remained  member  of   the   steering   committee  from  the  very  beginning  till  today.  

 

Fig.14:  Eckhard  Wolfgang  2006  at  ESREF  in  Wuppertal  

Although   the   EOBT   steering   committee   had  decided   in   1995   to   continue   with   the  conference,  due   to   the  decreasing  number  of  participants   it  was  discussed  1996  to  go   for  a  joint   venture   with   a   related   conference.   The  only  reasonable  choice  in  this  respect  was  the  European  Symposium  on  Reliability  of  Electron  Devices,   Failure   Physics   and   Failure   Analysis  (ESREF)  which  was  founded  in  1990  and  which  had   a   strong   overlap  with   EOBT.   Therefore   it  was  decided  in  Enschede,  Netherlands,  during  the   1996   ESREF,   to   merge   EOBT   with   ESREF,  and   consequently   it   became   a   separate   topic  “session  C”  from  1997  onwards  till  nowadays,  although  there  remained  an  additional  overlap  with  the  topic    of  failure  analysis  of  session  D.  

Fig.15:  5th  EOBT:  Bernard  Courtois,  Erich  Hödl  (rector  of  Wuppertal  University),  Ursula  Kraus  (Lady  Mayoress  of  Wuppertal),  Ludwig  Josef  Balk,  Eckhard  Wolfgang  (from  left  to  right)  

The  years  1987  –  1995  

One  main  topic   in  1987  was  the  development  of   testing  by  means  of  VC  as  underlined  by  a  still   important   overview   by   LUKIANOFF   [19].  Typical   topics   were   development   of   electron  energy   spectrometers   for   quantitative  measurements,   see   the  work   by  DINNIS   [20],  and   the  early   stages  of   linking  electron  beam  test  systems  with  CAD  data  bases.  

 

Fig.16:  Spectrometer  simulation,  DINNIS  

 

Fig.17:  Electron  beam  test  system  linked  with  a  CAD  database,  KOMATSU  et  al.  [21]  

Page 7: EOBT: from past to futurehomepages.laas.fr/nolhier/ESREF2015/SESSION_C/IP_C.pdf1!! EOBT: from past to future Ludwig Josef Balk Faculty of Electrical, Information and Media Engineering

 

7    

A   second  major   topic  was  OBIC   of   integrated  circuits,  as  reported  by  WILSON  [22].  

 

Fig.18:  OBIC  image  of  dislocations  in  a  silicon  bipolar  transistor  due  to  red  light  impact  [22]  

The  influence  of  the  color  of  the  exciting  laser  was  discussed  by  ZIEGLER  and  FEUERBAUM  [1]  

 

Table  1:  taken  from  [1]  

Fig.19:  Influence  of  laser  color  on  OBIC  results  [1]  

The   years   after   the   first   EOBT   were   strongly  involved   in   improvement   of   existing  techniques   as   well   as   in   the   introduction   of  new   methods   rather   than   in   applications   or  case  studies.   In  this  manner  RAO  reported  on  design   and   implementation   of   a   high  performance   e-­‐beam   tester,   fig.17   being   a  photograph  of  the  system  [23].  

 

Fig.20:  E-­‐beam-­‐tester  from  Intel  

At   these   days   still   a   lot   of   own   design   of   the  electron  beam  column  was  necessary,  too.  

 

Fig.21:  Electron  beam  column  [23]  

Fig.22   demonstrates   the   temporal   resolution  as   well   as   the   application   for   logic   state  mapping  [23]  

 

Fig.22:  Temporal  sensitivity  of  e-­‐beam-­‐tester  

A  comprehensive  discussion  of  the  different  operation  modes  of  VC  was  given  by  REINERS  [24]  in  1989.  

Page 8: EOBT: from past to futurehomepages.laas.fr/nolhier/ESREF2015/SESSION_C/IP_C.pdf1!! EOBT: from past to future Ludwig Josef Balk Faculty of Electrical, Information and Media Engineering

 

8    

Although  e-­‐beam  testing  became  mature,  still  laser  beam  applications  were  developed  such  as  the  laser  beam  assisted  logic  state  analysis.  

 

Fig.23:   Automatic   test   equipment   for   laser  beam  testing  (FOUILLAT  1991)  [25]  

As   an   important   technique   for   ultra-­‐high  frequency  applications  the  method  of  electro-­‐optic   probing   was   introduced   by   WHITAKER.  Fig.  20  gives  a  good  description  of  the  system  and   demonstrates   the   extreme   temporal  sensitivity  [26].  

 In  parallel   to   this,  attempts  were  undertaken  to  achieve  picosecond-­‐resolution  with  e-­‐beam  systems   as   well   either   by   improved   beam  blankers  or  by  use  of  laser  stimulated  electron  emission  (see  Fig.  25)..  

Approaching  the  end  of  a  “stand  alone”-­‐EOBT  conference,  papers  on  applications  of  e-­‐beam  testing   gained   importance,   and,   whereas   at  the   beginning   only   IC   applications   were  discussed,   the   new   applications   were   more  versatile,   such   as   the   investigation   of   solar  cells  by  MIL’SHTEIN  [28].  

a)  Schematic  of  e-­‐o-­‐sampling  system  

b)  Exciting  and  probing  beams    

c) Temporal response of GaAs switch

Fig.24:   Electro-­‐optic   sampling   of   integrated  circuits  [26]  

 

Fig.25:   E-­‐beam   testing   with   laser   induced  photoemission  (FIXL,  1993)  [27]  

 

Fig.26:  Potential  drop  over  solar  cells  for  different  illumination  conditions  [28]  

Finally,  the  1995  conference  was  the  onset  of  techniques   using   scanning   probe  microscopy,  

Page 9: EOBT: from past to futurehomepages.laas.fr/nolhier/ESREF2015/SESSION_C/IP_C.pdf1!! EOBT: from past to future Ludwig Josef Balk Faculty of Electrical, Information and Media Engineering

 

9    

in  special  scanning  force  microscopy  (SFM),  in  order  to  achieve  highest  spatial   resolution,  as  demonstrated  by  two  examples.  

SPRENGEPIEL   reported   on   measurements   on  passivated   submicron   IC   by   electro-­‐force  probing  [29].  

 

Fig.27:  Principle  of  SFM-­‐testing  

 

MERTIN   optimized   the   method   of   electro-­‐optic   sampling   by   the   introduction   of   near-­‐field  detection  by  means  of  a   fiber  as  used   in  scanning   near-­‐field   optical   microscopy  (SNOM)  [30)  

Fig.28:  Comparison  of  SFM  result  with  sampling  oscilloscope  [29]  

a)  Pockels-­‐cell  experiment  

b)  Indirect  and  direct  (back  side)  probing  

c)  SNOM  fiber  detection  for  high  spatial  resolution  

Fig.29:  Electro-­‐optical  system  with  SNOM  fiber  detection  [30]  

Summarizing  this  period  is  best  by  referring  to  the   preface   of   the   1995  meeting:   “if  we   look  back  at   the  years   that  have  elapsed  since   the  first   EOBT,   the   changes   are   clearly   apparent.  In   the   first   place,   electron   beam   testing   has  developed   to   become   a   standard   procedure  for   design   verification   and   failure  analysis…..Optical   techniques   have   also  developed   further,   but   no   commercial  equipment   for   them   is   available…..scanning  probe   techniques,   such   as   the   electric   force  microscope,   have   developed   very   well.”   This  indicates   the   relative   high   importance   of   e-­‐beam  testing  at  that  time.  

ESREF  Session  C  –  since  1997  

With   the   merger   of   EOBT   into   the   ESREF  conference  automatically   the  total  number  of  contributions   dedicated   to   this   topic   reduced  significantly.   Moreover,   due   to   the   fact   that  many   techniques   had   become   quite   mature  the   number   of   papers   with   new   scientific  content   shrank   as  well,   and   case   studies   and  applications   made   a   much   stronger  

Page 10: EOBT: from past to futurehomepages.laas.fr/nolhier/ESREF2015/SESSION_C/IP_C.pdf1!! EOBT: from past to future Ludwig Josef Balk Faculty of Electrical, Information and Media Engineering

 

10    

contribution  than  in  earlier  days.  Nevertheless  there   still   is   a   good   amount   of   interesting  developments  given.    

With   respect   to   new   scanning   probe  techniques  the  analysis  of  quantum  devices  by  MONTELIUS  (1998)  has  to  be  mentioned  as  an  example   for   one-­‐dimensional   electron  transport  [31].  

 

Fig.30:   Local   measurement   of   electrical  behavior  of  single  electron  device  for  different  temperatures  [31]  

In   2009   TIEDEMANN   reported   on   Finite  Element   Analyses   assisted   Scanning   Joule  Expansion   Microscopy   on   Interconnects   in  order   to   understand   the   thermo-­‐mechanical  stress  in  devices[32].  

 

Fig31:  a)temperature  distribution,  b)  vertical  and  c)  lateral  displacement,  d)modulus  of  total  displacement  vector  [32].  

This   research   followed   up   local   thermal  measurements   in   devices   by   means   of  scanning     thermal   microscopy,   as   already  reported  by  FIEGE  in  1998  [33].  

Measurement   of   mechanical   properties   was  also  reported  by  DILHAIRE  (1999)  by  means  of  goniometric   probing   of   thermally   induced  waves,  a  technique  being  applied   later  by  the  same  author  on  electronic  devices  [34].  

 

Fig.32:   Non-­‐invasive   probing   of   thermo-­‐  mechanical   features   of  MOS   transistor:   a)   40  seconds,   b   )10   minutes,   and   c)   45   minutes  after  excitation  [34]  

Finally   GRAUBY   showed   (2005)   that   this  technique   is   applicable   to   non-­‐invasive  analysis  of  devices  [35].  

 

Fig.33:   a)   optical   image,   b)  magnitude   and   c)  phase   shift   of   normal   surface   displacement  [35]  of  ESD  protection  circuit  

Aside  of  this  optical  method,  other  techniques  of  optical  beam  testing  still  remained,  and  for  other   applications   rather   than   for   devices   of  extremely   high   integration,   they   gained   new  importance.  DE  WOLF   introduced   in  1997  the  use   of   micro-­‐Raman   spectroscopy   for   the  

Page 11: EOBT: from past to futurehomepages.laas.fr/nolhier/ESREF2015/SESSION_C/IP_C.pdf1!! EOBT: from past to future Ludwig Josef Balk Faculty of Electrical, Information and Media Engineering

 

11    

analysis   of   conduction   lines   [36].   Laser   beam  backside   probing   of   CMOS   circuits   was  reported   by   KASAPI   in   1999.   Due   to   short  laser,  pulse  waveforms  of  about  10GHz  could  be  analyzed  stroboscopic  [37].  

Last  not   least  POGANY  demonstrated   in  2002  the   ability   of   electric   field   mapping   in   InGaP  HEMTs  and  GaAs  Terahertz  emitters  by  means  of  backside  infrared  OBIC  [38].  

Fig.34:   a)   band   diagram   of   gate   region,   b)  cross   section,   c)   OBIC   scan   in   mesa   area:      OBIC  of  InGaP-­‐HEMT  [38]  

On   the   other   hand   the   minute   structures   of  todays   integrated   devices   often   do   not   allow  either  a  non  destructive  analysis  or  techniques  on   bulk   material   at   all,   as   the   interaction  volume  of   an   electron  beam  would   be   by   far  too   large.   This   is   why   the   “old   fashioned”  transmission   electron   microscope   gained   its  importance   back   in   failure   analysis   (see   for  instance  ENGELMANN,  2000).  Quite  often  the  TEM  (or  STEM)  analysis  is  accompanied  with  a  FIB   (focused   ion   beam)   system   for   optimum  specimen   preparation,   as   in   the   example   of  980nm   SL   SQW   InGaAs/AlGaAS   pump   laser  diode  (VANZI;  2000)  [40].  

Fig.35:  Analysis  of  InGaAs/AlGaAs  pump  laser  a)  expected  hottest  area,  b)  enlarged  view  of  overstressed  facet  [40]  

But   coming   to   an   end   of   this   capture,   two  opposite   effects   can   be   noted.  New  methods  based   on   optical   probes   gain   further  importance,   and   on   the   other   hand   the   old  SEM   still   remains   an   important   and   reliable  tool,  as  shown  by  two  examples.  Using  a  laser  SQUID   microscope   NIKAWA   (2011)   could  visualize  defects  in  an  Vdd  line  [41].  

a)  basic  concept  

b)  schematic  of  system  

Page 12: EOBT: from past to futurehomepages.laas.fr/nolhier/ESREF2015/SESSION_C/IP_C.pdf1!! EOBT: from past to future Ludwig Josef Balk Faculty of Electrical, Information and Media Engineering

 

12    

 

c)  localization  of  open  defect                                            Fig.36:  SQUID  measurement  of  conduction  line  [41]  

And,   combining   3D   simulation   with   SEM  images  allows  high  quality  topography  without  the   disturbing   influence   of   the   electric   and  magnetic   fields   being   present   (CIAPPA,   2014)  [42]..  

 

Fig.37:  3D  simulation  of   field   influence              a)  isolines  of  electric  field  and  mesh  structure,  b)  trajectories  of  electrons  [42]  

 

Future  aspects  

As   already   mentioned   the   number   of  contributions   with   innovative   contents   has  decreased,   and   more   and   more   case   studies  are  being  reported  lately.  This  is,  however,  not  really   a  must,   as   one   can   see   by   comparison  with   other   conferences.   The   recent   ISTFA  symposia,   as   an   example,   contained   a   lot   of  new  approaches,  especially   in  connexion  with  the   needs   for   3D   integrated   devices.   And,  some   of   these   papers   are   authored   by  European   groups,   such   as   the   work   on   “3D  void   imaging   in   through   silicon   vias   by   X-­‐ray  nanotomography   in  a  SEM”  by  LALOUM  from  ST   Microelectronics   [43].,     Here   it   may   be  necessary   to   motivate   researchers,   in   special  

those   from   Europe,   to   participate   at   “ESREF-­‐session  C”.  A  further  issue  it  to  move  to  other  excitation   sources   rather   than   electrons   or  light,  in  order  to  achieve  very  high  penetration  depths  for  the  analysis  of  devices  without  the  need   for   decapsulation.   This   may   be  important   for   assessment   of   high   power  devices,   as  a   sample  preparation  may  change  the   device   behavior   significantly.   In   this  respect   X   rays   are   one   option,   another   one  can   be   the   use   of   ions   in   the   MeV   energy  regime.   And   last   not   least,   as   physical   failure  analysis  becomes  more  complicated,  it  may  be  necessary   to   join   it   more   intensively   with  computer  assisted  methods.  

 

References  

[1]   E.   Ziegler   and   H.P.   Feuerbaum,   ME   7  (1987),  309-­‐316  

[2]  M.  Knoll,  Z.  Tech.  Phys.  11  (1935),  467-­‐475  

[3]  M.  von  Ardenne,  Z.  Phys.  109  (1938),  553-­‐572  

[4]  see  for  instance:  C.W.  Oatley,  JAP  53  No  2  (1982),  R1  

[5]  K.C.A.  Smith.and  C.W.  Oatley,  Brit.  J.  Appl.  Phys.  6  (1955),  391-­‐399  

[6]  T.  E.  Everhart  et  al.,  J.  Elec.  Cont.  7  (1959),  97-­‐111  

[7]  O.G.  Wells  and  C.G.  Bremer,   J.Phys.  E:  Sci.  Instr.  1  (1968),  902-­‐906  

[8]   G.V.   Lukianoff   and   T.R.   Touw,   IITRI/SEM  (1975),  465-­‐571  

[9]   G.S.   Plows   and  W.C.   Nixon,   J.Phys.   E:   Sci.  Instr.  1  (1968),  595-­‐600  

[10]   A.   Gopinath   and   M.S.   Hill,   IITRI/SEM  (1973),  197-­‐204  

[11]  C.  Donolato,  Optik  52  (1978),  19-­‐36  

[12]  H.J.  Leamy,  JAP  53  No  6  (1982),  R51-­‐R80  

[13]   C.   van   Opdorp,   Philips   Res.   Rept.   32  (1977)  

Page 13: EOBT: from past to futurehomepages.laas.fr/nolhier/ESREF2015/SESSION_C/IP_C.pdf1!! EOBT: from past to future Ludwig Josef Balk Faculty of Electrical, Information and Media Engineering

 

13    

[14]  D.B.  Holt,  see  for   instance:  DB  Holt  et  al.  “Quantitative   Scanning   Electron  Microscopy”,  Academic   Press,   London,   New   York,   San  Francisco  (1974)  

[15]   D.B.  Wittry   and   D.F.   Kyser,   JAP   36   No   4  (1965),  353-­‐358  

[16]   J.I.   Pankove,   JAP  39  No  12   (1968),   5368-­‐5371  

[17]   see   for   instance:   LJ.  Balk  and  E.  Kubalek,  IITRI/SEM  (1977),  739-­‐776  

]18]  J.A.  Valdmanis  et  al.,  APL  41  (1990),  211  

[19]  G.V.  Lukianoff,  ME  7  (1987),  115-­‐129  

[20]  A.R.  Dinnis,  ME  7  (1987),  139-­‐146  

[21]  F.  Komatsu  et  al.,  ME  7  (1987),  267-­‐274  

[22]  T.  Wilson,  ME  7  (1987),  297-­‐307  

[23]   V.R.M.   Rao   and   P.  Winer,  ME  12   (1990),  295-­‐302  

[24]  W.  Reiners,  ME  12  (1990),  325-­‐340  

[25]  P.  Fouillat  et  al.,  ME  16  (1992),  287-­‐294  

[26]   J.F.   Whitaker   et   al.,   ME   12   (1990),   369-­‐397  

[27]   A.J.   Fixl   and   K.A.   Jenkins,  ME   24   (1994),  81-­‐88  

[28]  S.  Mil’shtein,  ME  31  (1996),  3-­‐12  

[29]   J.   Sprengepiel   et   al.,  ME  31   (1996),   181-­‐186  

[30]  W.  Mertin,  ME  31  (1996),  365-­‐376  

[31]  L.  Montelius  et  al.,  MR  38  (1998)  943-­‐950  

[32]   A.-­‐K.   Tiedemann   et   al.,   MR   49   (2009),  1165-­‐1168  

[33]   G.B.M.   Fiege   et   al.,   MR   38   (1998),   957-­‐962  

[34]  S.  Dihaire  et  al.,  MR  39  (1999),  981-­‐985  

[35]  S.  Grauby  et  al.,  MR  45  (2005),  1482-­‐1489  

[36]  I.  de  Wolf  et  al.,  MR  37  (1997),  1591-­‐1594  

[37]  S.  Kasapi  et  al.,  MR  39  (1999),  957-­‐961  

[38]  D.  Pogany  et  al.,  MR  42  (2002),  1673-­‐1677  

[39]   H.J.   Engelmann   et   al.,   MR   40   (2000),  1747-­‐1751^  

[40]  M.  Vanzi  et  al.,  MR  40  (2000),  1753-­‐1757  

[41]  K.  Nikawa  et  al.,  MR  51  (2011),  1624-­‐1631  

[42]  M.  Ciappa  et  al.,  MR  (2014)  

[43]  D.  Laloum  et  al.,  ISTFA  Proceedings  (2013)  

 

Abbreviations:  

IITRI/SEM:  IITRI    Scanning  Electron  Microscope  Symposium,  Chicago,  U.S.A.  

ME:  Microelectronics  Engineering  

MR:  Microelectronics  Reliability