@prefix vivo: . @prefix edm: . @prefix ns0: . @prefix dcterms: . @prefix skos: . vivo:departmentOrSchool "Science, Faculty of"@en, "Physics and Astronomy, Department of"@en ; edm:dataProvider "DSpace"@en ; ns0:degreeCampus "UBCV"@en ; dcterms:creator "Kong, Luke"@en, "Wu, Samuel"@en ; dcterms:issued "2012-09-20T14:04:18Z"@en, "2012-04-02"@en ; dcterms:description """This project builds on work done by a previous APSC 459 group, who developed fluorescence-­ based droplet sorting device similar to that described by Agresti et al. up to the point where it could detect the fluorescence of droplets up to a rate of 1kHz. However, it was unable to sort droplets; this project sought to add droplet sorting functionality to their microfluidic chip design. Our project aimed to demonstrate droplet actuation with use of the existing experimental setup. This consisted of three main objectives: 1. Model effect of electrode design and geometry on droplet actuation using finite-­‐ element analysis (implemented in COMSOL). 2. Demonstrate actuation of droplets, redesigning microfluidic chip if necessary. 3. Optimize setup towards droplet sorting at speeds on the same order (1-2 kHz) as exisiting work (Agresti, 2010). Using COMSOL we determined the factors that will most strongly affect the dielectrophoretic force on a droplet in a channel. From these results we redesigned the droplet sorter chip from the previous group and fabricated chips with a new electrode design (microfluidic channels to be filled with low-melting-­point alloy). We then went on to demonstrate droplet redirection at a rate of at least 100Hz. Although droplet redirection, the main focus of this project, was successfully demonstrated, much work remains to be done on this project. The redirection needs to be coupled to the previous 459 group’s droplet detection setup in order for controlled droplet actuation. Recommendations were made about electrode fabrication, droplet transfer from generator to sorter and high voltage switching."""@en ; edm:aggregatedCHO "https://circle.library.ubc.ca/rest/handle/2429/43239?expand=metadata"@en ; skos:note """Ultrafast Microfluidic Droplet Sorter Extension Work Luke  Kong Samuel  Wu Project  Sponsor: Dr.  Carl  Hansen Project  Mentors: Tim  Leaver Adam  Quiring ENPH  459   Engineering  Physics The  University  of  BriLsh  Columbia April  2,  2012 Project  Number  1209 ! ExecuLve  Summary This  project  builds  on  work  done  by  a  previous  APSC  459  group,  who  developed  fluorescence-­‐ based  droplet  sorLng  device  similar  to  that  described  by  AgresL  et  al.  up  to  the  point  where  it   could  detect  the  fluorescence  of  droplets  up  to  a  rate  of  1kHz.    However,  it  was  unable  to  sort   droplets;  this  project  sought  to  add  droplet  sorLng  funcLonality  to  their  microfluidic  chip  de-­‐ sign. Our  project  aimed  to  demonstrate  droplet  actuaLon  with  use  of  the  exisLng  experimental   setup.    This  consisted  of  three  main  objecLves: 1. Model  effect  of  electrode  design  and  geometry  on  droplet  actuaLon  using  finite-­‐ element  analysis  (implemented  in  COMSOL).     2. Demonstrate  actuaLon  of  droplets,  redesigning  microfluidic  chip  if  necessary. 3. OpLmize  setup  towards  droplet  sorLng  at  speeds  on  the  same  order  (1-­‐2  kHz)  as  exist-­‐ ing  work  (AgresL,  2010). Using  COMSOL  we  determined  the  factors  that  will  most  strongly  affect  the  dielectrophoreLc   force  on  a  droplet  in  a  channel.    From  these  results  we  redesigned  the  droplet  sorter  chip  from   the  previous  group  and  fabricated  chips  with  a  new  electrode  design  (microfluidic  channels  to   be  filled  with  low-­‐melLng-­‐point  alloy).    We  then  went  on  to  demonstrate  droplet  redirecLon  at   a  rate  of  at  least  100Hz.     Although  droplet  redirecLon,  the  main  focus  of  this  project,  was  successfully  demonstrated,   much  work  remains  to  be  done  on  this  project.    The  redirecLon  needs  to  be  coupled  to  the  pre-­‐ vious  459  group’s  droplet  detecLon  setup  in  order  for  controlled  droplet  actuaLon.      Recom-­‐ mendaLons  were  made  about  electrode  fabricaLon,  droplet  transfer  from  generator  to  sorter   and  high  voltage  switching.   ! Table of Contents 1.0 Introduction! 1 2.0 Discussion! 4 2.1 Dielectrophoresis! 4 2.2 High-Voltage Switching! 4 2.3 Modeling! 5 2.3.1 The Construct! 5 2.3.2 Finite Element Analysis! 6 2.3.4 Computation! 7 2.3.5 Optimization! 8 2.4 Sorter Chip Design! 10 2.4.1 List of Modifications! 10 2.4.2 Design Description! 10 2.5 Sorter Chip Fabrication! 13 2.5.1 Fabrication! 13 2.5.2 Results! 13 2.6 Electrode Fabrication! 14 2.6.1 Procedure! 14 2.6.2 Results! 15 2.7 Electrode Characterization! 16 2.7.1 Procedure! 16 ! 2.7.2 Results! 17 2.8 Switch Characterization! 18 2.8.1 Procedure! 18 2.8.2 Results! 19 2.9 Droplet Actuation! 21 2.9.1 Procedure! 22 2.9.2 Results! 22 Conclusions! 24 3.0 Project Deliverables! 25 3.1 List of Deliverables! 25 3.2 Financial Summary! 25 4.0 Recommendations! 26 4.1 Electrode Fabrication! 26 4.2 Droplet Re-injection! 27 4.3 HV switching! 28 5.0 Appendices! 29 Appendix A: Fabrication Protocols! 29 A.1 Wafer Fabrication Procedure! 29 A.2 Wafer Fabrication Protocols! 30 A.3 PDMS Chip Fabrication! 32 ! Appendix B: SOP for generating droplets for use with 30um channel width sorter chips.! 33 Appendix C: HV Switch design by the UBC PHAS E-LAB! 34 Appendix D: Electrode alloy quotes! 36 6.0 References! 38 ! Table of Figures/Tables 1   Fig.  1.  Diagram  of  microfluidic  droplet  sorter 6   Fig.  2.  Basic  construct  of  COMSOL  electrode  model 7   Fig.  3.  Mesh  for  COMSOL  electrode  model 8   Fig.  4.  Y-­‐Z  slice  study  of   ∇  E 2halfway  through  oil  channel  in  x  direcLon. 11   Fig.  5.  AutoCAD  drawing  of  droplet  sorter  chips. 12   Fig.  6.  Flat  (top)  and  sharp  (bofom)  electrode  designs. 14   Fig.  7  Electrode  channels  following  injecLon  of  GaIn 16   Fig.  8.  Diagram  of  setup  for  electrode  characterizaLon. 17   Fig  9.  Electrode  voltages  when  a  0-­‐5V  square  wave  is  applied 18   Fig.  10.  SchemaLc  for  connecLng  HV  supply  to  electrodes 19   Fig  11.  Switch  output  when  10V  signal  is  applied  to  input   19   Fig.  12.  Switch  output  when  75V  signal  is  applied  to  input 20   Fig.  13.  Switch  output  measured  by  high-­‐voltage  probe  when  300V  signal  is  applied 21   Fig.  14.  Droplet  generaLng  device  generated  by  previous  APSC  459  group 22   Fig.  14.  Demonstrated  droplet  redirecLon 11   Table  1:  Droplet  Sorter  Design  SpecificaLons 13   Table  2:  Photoresist  specificaLons ! 1.0 Introduction   Our  ever-­‐growing  understanding  of  biology  allows  us  to  manipulate  and  engineer  bio-­‐ logical  systems  in  novel  ways,  with  such  fields  as  syntheLc  biology  and  protein  engineering.  As   biology  operates  on  small  scales  (proteins  to  cells)  and  across  large  orders  of  magnitudes  (tril-­‐ lions  of  molecules/cells),  effecLve  assaying  techniques  are  essenLal  for  successful  experimen-­‐ taLon  and  validaLon.    Screening,  however,  is  olen  a  rate-­‐limiLng  step  in  both  Lme  and  re-­‐ sources.   ExisLng  macro-­‐assaying  methods  involving  microwell  plates,  when  automated  roboL-­‐ cally,  can  achieve  processing  speeds  of  ~1  Hz,  but  this  does  not  allow  for  treatment  of  individual   cells  (AgresL  et  al.,  2010).     OpLmally,  it  is  desired  to  be  able  to  manipulaLng  individual  cells  in  a  high  throughput   manner,  requiring  technology  capable  of  operaLng  on  small  size  scales.    EncapsulaLon  of  bio-­‐ material  within  droplets  provides  such  a  planorm,  allowing  for  treatment  within  a  unique  picoli-­‐ tre  scale  microenvironment.    Each  droplet  could  potenLally  be  experimented  upon  individually,   able  to  be  split  and  merged  with  other  droplets  (Griffiths  &  Tawfik,  2006).     Fig.  1Diagram  of  microfluidic  droplet  sorter.    Fluorescent  droplets  are  seen  here  as  light-­‐colored  spheres,  non-­‐ fluorescent  drops  as  dark.  (AgresL  et  al.,  2010)     Sponsored  by  Dr.  Carl  Hansen  (UBC  Centre  for  High-­‐Throughput  Biology),  this  project   aims  to  replicate  a  droplet  sorter  developed  by  AgresL  et  al.    They  were  able  to  sort  droplets   ! 1 into  one  of  two  channels  by  applying  a  voltage  across  two  electrodes,  creaLng  a  non-­‐uniform   electric  field  and  a  subsequent  dielectrophoreLc  force  on  the  droplet.         If  successful,  this  device  could  be  applied  towards  directed  evoluLon  or  digital  PCR  ex-­‐ periments,  potenLally  reducing  Lme  and  reagent  costs  by  several  orders  of  magnitude.     The  project  objecLve,  as  stated  by  the  project  sponsor  Carl  Hansen: “Add  droplet  sorLng  funcLonality  to  the  microfluidic  chip.  This  should  be  accomplished  by   adding  an  electrode  upstream  of  a  juncLon  leading  to  two  collecLon  channels.  By  using  the   electrode  to  apply  an  electric  field  across  the  channel,  droplets  could  be  sorted  into  one  of   the  two  channels.  The  group  would  be  required  to  research  and  design  the  electrodes  (as   well  as  selecLng  a  suitable  fabricaLon  technique)  as  well  as  the  switching  electronics.  This   objecLve  could  involve  modeling  in  COMSOL  and  microfabricaLon  of  test  devices.” In  addiLon,  we  had  three  self-­‐formed  project  objecLves: 1.  Model  electrode  design  with  COMSOL   We  will  model  the  effect  of  electrode  design  on  actuaLon  of  droplets  using  COMSOL   (finite-­‐element  analysis).    Varying  different  factors  (electrode  posiLon,  geometry,  AC  vs.   DC  field,  field  strength),  we  will  produce  an  evaluaLon  for  the  electrode  design  which  will   exert  the  maximum  dielectrophoreLc  force  on  a  25  μm-­‐diameter  droplet.   2.  Demonstrate  droplet  actuaLon. We  will  demonstrate  actuaLon  of  droplets  using  the  electrode  design  chosen  from  the   modelling,  redesigning  the  microfluidic  chip  if  necessary.    SorLng  will  be  performed  at  1Hz   at  this  stage,  as  demonstraLng  that  actuaLon  is  controllable  and  reproducible  is  the  main   focus  of  this  objecLve. ! 2 3.  OpLmize  droplet  sorLng. Improve  experimental  setup  towards  sorLng  at  speeds  on  the  same  order  (1-­‐2  kHz)  as  ex-­‐ isLng  work  (AgresL,  2010).   This  report  will  outline  the  theory  used,  the  methods  and  results  of  various  experiments,   issues  encountered,  as  well  as  conclusions  and  recommendaLons  for  future  developments  re-­‐ garding  this  project.    This  report  aims  to  convey  project  findings  to  the  project  sponsor,  Carl   Hansen,  his  lab  personnel  and  the  Engineering  Physics  Project  Lab. ! 3 2.0 Discussion 2.1 Dielectrophoresis   ParLcles  present  in  an  applied  electric  field  exhibit  polarizaLon;  when  placed  in  a  non-­‐ uniform  electric  field,  the  parLcle  will  be  subject  to  a  dielectrophoreLc  force.    Depending  on  a   difference  in  electric  properLes  (i.e.  dielectric  constant),  a  water  droplet  present  in  an  oil  will   move  towards  an  area  of  higher  electric  field,  experiencing  a  force  given  by  the  following  rela-­‐ Lon  (Ahn  et  al.,  2006): FDEP = 4πr3∇  E 2   Dielectrophoresis  is  the  actuaLon  mechanism  for  sorLng  droplets.    However,  if  a  DC   voltage  is  applied  screening  effects,  where  the  environment  around  a  polarized  parLcle  polar-­‐ izes  in  the  opposite  direcLon  to  offset  the  charge  accumulaLon,  reduce  the  effecLve  electric   field.    This  can  be  negated  by  applying  an  AC  signal  on  the  order  of  20kHz  (AgresL  et  al.,  2010).     Thus,  a  high  voltage  20kHz  AC  signal  of  at  least  1kV  across  the  electrodes  is  desired. 2.2 High-Voltage Switching To  produce  a  HV  AC  signal,  a  suitable  switching  mechanism  was  required  to  handle  voltages  of   at  least  1kV.    The  voltage  itself  is  provided  by  a  PS325  2500V-­‐25W  high  voltage  power  supply   made  by  Stanford  Research  Systems,  Inc.   A  solid  state  switch  from  Behlke  meeLng  technical  requirements  (HTS  61-­‐03-­‐GSM  -­‐  rise/ fall  Lme  of  10ns,  maximum  operaLng  frequency  of  2.5MHz)  was  idenLfied  and  subsequently   ordered,  but  due  to  a  number  of  complicaLons  on  part  of  the  company  we  were  not  able  to  ob-­‐ tain  the  switch  (Behlke,  2012).    A  switch  designed  and  constructed  by  the  PHAS  E-­‐LAB  was  for-­‐ tunately  able  to  be  obtained  from  Pavel  Trochtchanovitch  (E-­‐LAB  manager).    This  switch  is   ! 4 driven  by  a  0-­‐5V  signal  and  can  handle  switching  voltages  of  up  to  3kV.    A  schemaLc  of  the   switch  can  be  found  in  Appendix  C.   The  electrodes  were  characterized  in  order  to  determine  if  they  possessed  any  parasiLc   capacitance  which  would  limit  the  maximum  frequency  of  the  applied  AC  signal.    The  switch   was  characterized  to  determine  performance  with  our  system. 2.3 Modeling   The  cell  sorter  system  is  modeled  on  COMSOL  MulLphysics  solware  prior  to  its  fabrica-­‐ Lon.   The  governing  formula  for   the  dielectrophoreLc   force:   FDEP = 4πr3∇  E 2   is  a  funcLon  of   the  gradient  of  the  electric  field  squared.  The  goal  is  to  opLmize  the  system  on  the  solware  to   maximize   ∇  E 2 . 2.3.1 The Construct   This  system  contains  three  important  components:  Component  1  being  the  oil  channel,   where  the  droplets  travel;  Component  2  represents  the  ground  electrode;  Component  3   is  the   high  voltage  electrode.   These  three  components  are  in  the  same  x-­‐y   plane.   Above  this  plane,   lies  the  PDMS  top,  and  the  glass  base  is  sefled  below  it.   ! 5 Fig. 2. Basic construct of COMSOL electrode model   There  are  a  few  parameters  to  be  tested  in  this  stage.  By  speculaLon,  the  most  influen-­‐ Lal  parameter   would   be  the  distance  between   the  electrodes  and   the  oil   channel  in   the   x-­‐ direcLon  (d1).   Secondary  concerns  are  the  distance  between  the  electrodes  in  the  y-­‐direcLon   (d2),  and  the  thickness  of  the  electrodes  (d3).   2.3.2 Finite Element Analysis   The  modeling  process  provides  a  wide  range  of  accuracies.  COMSOL  basically  breaks  this   system  into  tens  or  hundreds  of  thousands  of  tetrahedrons  and  analyze  each  element  individu-­‐ ally.  To  achieve  an  accurate  esLmate,   it   is  necessary   to  have  a  higher   density   near   the  elec-­‐ trodes  and  the  oil  channel,  whereas  the  states  of  the  regions  far  away  do  not  contribute  much   to  the  accuracy  of  the  calculaLon.  The  system  is  therefore  broken  down  in  this  manner: ! 6 Fig.  3.  Mesh  for  COMSOL  electrode  model 2.3.4 Computation   Although  the  enLre  system  is  analyzed,  it  is  only  the  region  near  the  oil  channel  is  of  in-­‐ terest.  Therefore,  a  y-­‐z  study  plane  cusng  through  the  oil  channel  is  created  to  display  the  in-­‐ formaLon  in  a  more  clarified  style: ! 7 Fig.  4.  Y-­‐Z  slice  study  of   ∇  E 2halfway  through  oil  channel  in  x  direcLon.   In   the  figure  above,   the   colours  represent   the   strength  of   the  divergence  of  E-­‐norm   squared   in  the  x-­‐direcLon.   The  colour  darkens  as  the  strength   increases,   as  indicated  by   the   scale  bar.  The  maximum  value  occurs  in  the  area  of  the  oil  channel  perpendicular  to  the  signal   electrode. 2.3.5 Optimization   The  task  now  is  to  vary   electrode  spacing  and  opLmize  this  parameter.  The  procedure   involves  manually   decrement   the  value  of  a  parameter   that  determines  d1,   starLng  from  the   original  distance  used  in  the  previous  group’s  design.  Aler   plosng   the  results  on  Excel,   it   is   concluded  that  the  gradient  increases  as  d1  decreases. ! 8 Fig.  5.  Plot  of   ∇  E 2  as  a  funcLon  of  electrode  spacing  d1 According  to  the  data,  the  gradient  strength  increased    Lmes,  as  d1  is  decreased  to  a  fourth   of  its  original  value.    The  eventual  best  result  obtained  is   d1 = 7µm giving   ∇  E 2 = 3.77 ×1011 .     The  absolute  potenLal  difference  is  at  500  V,  in  DC  sesng.    The  secondary  parameters   are  then  tested,  and  they  demonstrated  lifle  contribuLon  to  the  gradient  strength.  Due  to  fab-­‐ ricaLon  limitaLons,  these  parameters  are  given  rather  modest  values,  which  will  be  elaborated   upon  in  the  next  secLon  of  this  report. ! 9 2.4 Sorter Chip Design   The  chip  design  is  inherited  from  the  AutoCAD  file  drawn  by  the  previous  group.  The  de-­‐ sign  demonstrated  a  close  resemblance  to  the  original  design  by  AgresL  et  al.  However,  criLcal   changes  are  applied  according   to  the  COMSOL  model  to  maintain  consistency.    The  AutoCAD   file  can  be  found  on  the  USB  sLck  included  with  this  report. 2.4.1 List of Modifications 1. Channels  for  the  electrodes  are  added 2. Inlets  and  outlets  are  added  to  allow  the  liquid  form  of  the  metal  alloy  to  flow  into  the   channels  and  solidify 3. Replaced  the  second  design  with  the  current  design  to  have  in  total  8  similar  chips  on  a   single  wafer 2.4.2 Design Description   One  silicone  wafer   is  able  to  hold  designs  for  eight  PDMS  chips.   Four   different  designs   are  used,  resulLng  in  two  chips  per  design.    Therefore,  some  less  opLmal  designs  are  included   in  the  intenLon  of  result  comparison.  The  core  design,  inherited  from  the  COMSOL  model  plus   three  other  variaLons  make  up  the  four  designs.  Due  to  fabricaLon  limitaLons,  the  parameter   d1  as  described  in  SecLon  3,  has  been  increased  to  10µm . ! 10 Fig.  5.  AutoCAD  drawing  of  droplet  sorter  chips. Table  1:  Droplet  Sorter  Design  SpecificaLons Design d1 (μm) d2 (μm) 1 10 72 2 10 142 3 20 72 4 10 110 ! 11   As  stated  in  the  Table  3.01,  d1  in  the  third  design  is  twice  the  magnitude  of  the  others.  It   is  to  confirm  the  results  predicted  by  computer  modeling  discussed  in  SecLon  4.  The  fourth  de-­‐ sign  is  rather  unconvenLonal.  It  is  suspected  that  the  electrode  with  a  sharp  end  may  produce  a   more  powerful  electric   field.  This  idea  is  then  implemented  in  the  AutoCAD  design  for  tesLng   purposes. Fig.  6.  Flat  (top)  and  sharp  (bofom)  electrode  designs. ! 12 2.5 Sorter Chip Fabrication 2.5.1 Fabrication   FabricaLon  was  performed  with   the  microfabricaLon  faciliLes  in   the  NCE   cleanroom,   UBC.  The  process  took  in  total  4  days.  Detailed  protocols  can  be  found  in  Appendix  A. 2.5.2 Results   Two  wafers  were  fabricated  and  the  specificaLons  listed   in  Table  4.01.     VariaLons  in   channel  height  occurred  due  to  slight  differences  in  UV   exposure  Lme  and  fabricaLon  errors   (missing  UV  filter,  etc.). Table  2:  Photoresist  specificaLons Wafer 1 Height (μm) Width (μm) Output Channel 130 180 Oil Channel 40 30 Electrode 40 30 Wafer 2 Height (μm) Width (μm) Output Channel 142 185 Oil Channel 50 30 Electrode 50 30 ! 13 2.6 Electrode Fabrication   The  previous  group  fabricated  electrodes  with  etched  chrome  placed  below  the  PDMS   sorter  chip,  contrary  to  what  was  done  in  the  AgresL  paper,  where  channels  were  formed   alongside  that  of  the  sorter  and  filled  with  a  low-­‐melLng  point  solder.    We  opted  to  follow  the   method  in  the  AgresL  paper,  adding  channels  to  the  previous  sorter  chip  design.   Several  low-­‐melLng-­‐point  alloys  were  considered,  but  due  to  price,  eutecLc  Gallium  In-­‐ dium  was  chosen.    As  it  is  liquid  at  room  temperature,  it  can  be  injected  into  the  channel  inlets   with  a  syringe.    Quotes  from  Indium  and  AIM  Solder  can  be  found  in  Appendix  D. 2.6.1 Procedure   The  liquid  GaIn  is  withdrawn  into  a  1mL  syringe  and  injected  into  the  one  of  the  elec-­‐ trode  channel  inlets. ! 14 2.6.2 Results Fig.  7  Electrode  channels  following  injecLon  of  GaIn   GaIn  seemed  to  flow  well  through  the  PDMS  channels,  requiring  only  minimal  back   pressure  applied  to  the  syringe.    It  adheres  well  to  the  channel  walls  and  fills  the  channel  com-­‐ pletely  with  no  problems  at  corners  or  air  bubbles  without  any  need  of  priming.       However,  due  to  high  surface  tension  forces,  upon  removal  of  syringe  from  electrode   channel  inlet  a  large  amount  of  GaIn  will  spill  out.    As  it  does  not  spread  out  over  a  surface  at   room  temperature,  the  majority  can  be  withdrawn  back  into  the  syringe  and  the  residue  can  be   cleaned  off  by  water  or  HFE-­‐7500  oil,  but  it  cannot  be  removed  completely.       ! 15 2.7 Electrode Characterization   Electrodes  were  tested  to  determine  their  ability  to  conduct  an  AC  signal  by  determining   the  RC  Lme  constant.  This  Lme  constant  was  assumed  to  be  independent  of  the  voltage  applied   (resistance  and  capacitance  being  properLes  of  the  material).     2.7.1 Procedure prob e Fig.  8.  Diagram  of  setup  for  electrode  characterizaLon. 1. A  0-­‐6V  square  wave  was  applied  with  the  NI  DAQ  card  to  the  signal  electrode  inlet  and   the  output  signal  measured  at  the  signal  electrode  outlet.     2. Square  waves  at  200Hz,  20kHz  and  1MHz  were  applied  to  the  electrode  inlet  and  the   Lme  constant  was  measured  using  the  OSCILLOSCOPE  to  be  the  Lme  taken  for  the  volt-­‐ age  at  the  electrode  outlet  to  reach   1 e of  its  original  value  at  the  falling  edge. ! 16 2.7.2 Results Fig  9.  Electrode  voltage  (yellow)  when  a  0-­‐5V  square  wave  (200Hz,  20kHz,  1MHz  lel  to  right)  is   applied  (blue)   The  Lme  constant  at  the  signal  electrode  was  measured  to  be  30ns  at  each  applied  fre-­‐ quency,  while  that  of  the  applied  TTL  signal  was  25ns.    The  electrode  signal  closely  follows  the   applied  voltage  waveform,  suggesLng  that  maximum  switching  frequency  across  the  electrodes   is  limited  by  the  applied  signal,  not  necessarily  by  the  electrode’s  capacitance.       This  suggests  a  maximum  frequency  of  at  least  33MHz,  well  above  that  of  the  desired   20kHz.    The  maximum  should  theoreLcally  be  even  higher,  since  the  electrode  Lme  constant   seemed  to  be  dependent  on  the  signal  Lme  constant. ! 17 2.8 Switch Characterization The  switch  was  characterized  to  determine  performance  with  our  system.     2.8.1 Procedure HV Switch =+ Fuse HV Supply 0V-5V TTL Signal prob e Fig.  10.  SchemaLc  for  connecLng  HV  supply  to  electrodes 1. A  10V  signal  (from  HV  source)  was  applied  to  the  electrodes  at  400Hz,  1kHz,  20kHz  and  the   switch  output  measured  on  the  scope. 2. A  75V  AC  signal  was  applied  to  the  electrodes  at  800Hz,  1kHz  and  20kHz  and  the  switch  out-­‐ put  measured. 3. 50V,  100V,  200V  and  300V  were  applied  to  the  switch  and  the  output  measured  with  a  high-­‐ voltage  probe  (which  reduces  a  voltage  by  1220x  to  be  readable  on  the  scope).    Due  to  some   yet  unknown  reason,  the  high-­‐voltage  probe  was  only  able  to  make  measurements  when  a   regular  scope  probe  was  also  afached  to  the  switch  output,  therefore  only  voltages  up  to   300V  (maximum  raLng  for  the  regular  probe)  were  able  to  be  tested. ! 18 2.8.2 Results Fig.  11.  Switch  output  (blue)  with  10V  signal  applied  at  input  (yellow)  at  400Hz,  1kHz  and  20kHz   lel  to  right   Output  signal  exhibits  a  long  fall  Lme,  which  results  in  the  signal  being  clipped  starLng   at  around  500Hz.    At  20kHz  the  voltage  is  similar  to  that  of  a  DC  voltage  signal. Fig.  12.  Switch  output  (blue)  with  75V  signal  applied  at  input  (yellow)  at  1kHz  and  20kHz  lel  to   right ! 19 Fig.  13.  Switch  output  measured  by  high-­‐voltage  probe  (purple)  and  regular  oscilloscope  probe   (blue)  when  300V  signal  is  applied  at  450Hz  and  20kHz   The  output  when  75V  and  300V  is  applied  exhibits  a  similar  behavior  to  that  of  10V,  with   the  falling  edge  clipping  upwards  unLl  it  is  similar  to  a  DC  signal  at  20kHz.    The  switch  can  there-­‐ fore  be  operated  in  AC  up  to  1kHz  before  clipping  becomes  a  large  factor  and  operaLng  the   switch  at  20kHz  can  let  us  mimic  a  DC  signal. ! 20 2.9 Droplet Actuation   Droplets  were  formed  using  the  exisLng  implementaLon  (Fig.  2);  a  PDMS  device  con-­‐ nected  to  input  syringe  pumps  coflows  HFE-­‐7500  oil  with  3%  w/w  PFPE-­‐PEG  block  copolymer   surfactant  (to  prevent  droplet  coalescence)  to  disperse  the  aqueous  droplets  at  the  flow-­‐ focusing  juncLon  (inset).    These  droplets  were  then  collected  in  a  microcentrifuge  tube  from   the  output  port  and  re-­‐injected  into  the  sorLng  device  (Mulholland  et  al.,  2011). Fig.  14.  Droplet  generaLng  device  generated  by  previous  APSC  459  group  (Mulholland  et  al.,   2011). The  sharp  and  flat  electrode  designs  were  tested  at  a  range  of  voltages  and  applied  frequencies,   with  actuaLon  detected  t  30  fps  by  the  CCD  camera  setup  developed  by  the  previous  459   group.    The  high-­‐voltage  source  is  connected  to  the  switch  via  a  high-­‐voltage  connector  and  the   switch  is  driven  by  a  0-­‐5V  TTL  signal  from  the  NI  DAQ  board.    The  switch  output  is  connected  to   the  signal  electrode  through  a  200mA  fuse  and  the  ground  electrode  is  connected  to  ground.     ! 21 2.9.1 Procedure 1. 30um  droplets  were  formed  using  20-­‐67-­‐31  (D1-­‐D2-­‐Height)  droplet  generator  chip  flowing   HFE-­‐7500  with  3%  w/w  RainDance  surfactant  at  250uL/hr  and  deionized  water  at  25uL/hr. 2. Droplets  were  withdrawn  into  a  1mL  syringe  and  re-­‐injected  into  the  sorter  chip  with  sharp   electrode  design  at  10uL/hr  with  an  accompanying  oil  flow  rate  of  100uL/hr. 3. A  HV  AC  signal  was  applied  to  electrodes  via  switch  at  500V,  1000V,  2000V,  2500V  with  fre-­‐ quencies  500Hz,  1kHz,  20kHz.     An  SOP  for  generaLng  droplets  used  with  this  test  can  be  found  in  Appendix  B. 2.9.2 Results Fig.  15.  Demonstrated  droplet  redirecLon  with  1000V/20kHz  applied  signal  (lel)  and  when  sig-­‐ nal  was  turned  off  (right)  -­‐  1000V-­‐20kHz.avi   Droplet  redirecLon  was  not  able  to  be  demonstrated  with  the  flat  electrode  designs  but   it  was  demonstrated  with  the  sharp  electrode  design  (see  data  files  in  USB  key  included  with   report).    Due  to  the  frame  rate  of  the  camera  (30  fps),  we  could  not  ascertain  the  rate  of  drop-­‐ let  actuaLon.    Slowing  down  the  video,  the  droplet  flow  rate  is  esLmated  to  be  on  the  order  of   at  least  100Hz.    Not  all  droplets  were  observed  to  flow  uniformly  into  one  channel,   ! 22   There  was  no  noLceable  difference  in  performance  of  droplet  redirecLon  when  the  ap-­‐ plied  signal  operated  at  500Hz,  1kHz  or  20kHz.    Here  20kHz  is  assumed  to  be  similar  to  a  DC   voltage,  since  the  output  signal  has  been  significantly  clipped  high.    To  determine  the  effect  of  a   higher  frequency  AC  signal  (20kHz),  a  more  suitable  switch  is  required.   With  polydispersity  in  the  re-­‐injected  droplets,  when  500V  was  applied  the  chip  was   able  to  sort  smaller  droplets  (30um  diameter)  while  not  drawing  in  larger  droplets  (45um  di-­‐ ameter).        This  suggests  an  applicaLon  where  a  conLnuously  applied  voltage  could  act  as  a  filter   to  sort  droplets  based  on  size.   Videos  of  all  cases  tested  can  be  found  on  the  USB  sLck  included  with  this  report. ! 23 2.0 Conclusions   This  project  sought  to  add  droplet  redirecLon  to  a  microfluidic  sorter  chip  developed  by   a  previous  APSC  459  group.    An  technique  for  fabricaLng  electrodes  was  determined,  adding   addiLonal  microfluidic  channels  upstream  of  a  juncLon  leading  to  the  collecLon  channels  and   filling  them  with  low-­‐melLng-­‐point  alloy.    Electrode  designs  were  modeled  in  COMSOL  and  the   previous  group’s  sorter  chip  redesigned  accordingly  to  include  the  most  effecLve  designs.       A  high-­‐voltage  switch  was  obtained  from  the  PHAS  E-­‐LAB,  though  it  can  only  switch  high   voltages  at  a  maximum  of  1kHz,  much  less  than  the  desired  frequency  of  20kHz.    A  few  meth-­‐ ods  of  increasing  this  switching  frequency  were  suggested  by  the  PHAS  E-­‐LAB  and  included  in   the  recommendaLons.     Droplet  actuaLon  was  demonstrated  on  the  order  of  100Hz  with  an  applied  voltage  sig-­‐ nal  between  500V  -­‐  2500V  AC  with  a  frequency  range  of  0-­‐1kHz.     In  order  to  afain  fluorescence-­‐based  droplet  sorLng  at  appreciable  speeds  (1-­‐2kHz),   more  work  is  required  in  the  following  areas: 1. ActuaLon  needs  to  be  coupled  to  the  first  APSC  459  group’s  droplet  detecLon  system.     The  LabVIEW  scripts  would  need  to  be  modified  and  extended  to  drive  the  HV  switch. 2. A  new  HV  switch  which  could  operate  in  the  20kHz  range  would  need  to  be  sourced  or   the  current  switching  setup  improved.    The  effect  of  a  high-­‐frequency  high-­‐voltage  sig-­‐ nal  would  then  need  to  be  tested. 3. Droplet  generator-­‐sorter  coupling  would  need  to  be  improved  to  increase  control  over   droplet  speed  in  the  sorter  chip. ! 24 3.0 Project Deliverables 3.1 List of Deliverables Deliverable Medium Details COMSOL  Models Electronic Finite-­‐element  analysis  of  different  electrode   designs Droplet  ActuaLon Electronic Videos  of  droplet  actuaLon  (video)   Microfluidic  design Electronic AutoCAD  file  with  modified  chip  design Microfluidic  chips Physical Microfluidics  chips  with  electrode  channels Photolithographic  masks Physical Mask  fabricated  during  project Photoresist  molds Physical Molds  fabricated  during  project HV  switch Physical Was  borrowed  from  Pavel,  but  he  can  help   construct  copies Log  Books Paper Records  of  meeLngs,  rough  work RecommendaLon  Report Paper/Electronic Submifed  aler  project  is  finished. PresentaLon In  Person Given  aler  project  is  finished. 3.2 Financial Summary # DescripLon Quan-­‐ Lty Vendor Cost  ($   CAD) Purchased   by To  be   funded  by 1 EutecLc  Gallium  In-­‐ dium SKU-­‐pack  size: 495425-­‐5G 1 Sigma-­‐Aldrich 80.30 Hansen   Lab Hansen   Lab Total 80.30 ! 25 4.0 Recommendations 4.1 Electrode Fabrication   Currently  electrodes  are  fabricated  by  injecLng  eutecLc  Gallium-­‐Indium  (melLng  point   15.7  oC)  into  the  electrode  channels.    This  is  done  easily  with  a  1mL  syringe,  but  there  are  sev-­‐ eral  drawbacks: 1. GaIn  is  toxic  and  can  cause  severe  skin  burns,  making  fabricated  electrodes  more  diffi-­‐ cult  to  handle.    It  is  also  difficult  to  clean  off  surfaces. 2. GaIn  is  difficult  to  inject  –  flowing  the  GaIn  through  30um-­‐wide  channels  requires  a  large   enough  back  pressure  such  that  removal  of  the  syringe  from  the  channel  inlet  will  re-­‐ lease  a  large  amount  of  GaIn  from  the  syringe.    Not  only  a  waste  of  material,  even  when   cleaned  off  the  PDMS  remains  slightly  opaque,  reducing  image  quality  with  the  opLcal   system. 3. The  HV  and  ground  voltages  are  connected  to  the  electrode  channel  by  directly  inserLng   a  header  pin  into  an  electrode  channel,  which  causes  the  GaIn  near  the  oil  channel  to   shil  and  pull  away  if  the  pins  are  inserted/removed.    This  affects  the  uniformity  of  the   electric  field  generated. 4. GaIn  will  expand  and  contract  readily  with  changes  in  temperature.    Bringing  the  chip   into  an  environment  warmer  than  that  in  which  it  was  made  will  cause  the  GaIn  to  spill   out  of  the  electrode  channel  outlets  as  well  as  pull  away  from  the  walls.    If  the  header   pins  have  been  afached  to  the  chip  via  an  adhesive,  this  expansion  will  cause  the  adhe-­‐ sive  to  lil  from  the  chip.   Use  of  a  low-­‐melLng  point  alloy  whose  melLng  point  is  above  room  temperature  (60-­‐ 70oC)  could  easily  address  all  these  concerns.    Of  the  alloys  invesLgated,  Wood’s  alloy  seems  to   be  the  most  feasible  in  terms  of  cost  ($92.22  from  City  Chemical  LLC  as  opposed  to  $355  for  In-­‐ dalloy  19  from  Indium  for  the  minimum  order).    This  project  would  only  require  a  minimal   amount  and  the  PHAS  E-­‐LAB  has  agreed  to  purchase  any  excess  Wood’s  Alloy.     ! 26 4.2 Droplet Re-injection   Droplet  re-­‐injecLon  into  the  sorter  chip  is  difficult  and  erraLc  using  syringe  pumps.    Sev-­‐ eral  concerns  came  up: 1. Droplets  must  be  withdrawn  from  an  external  container  using  a  syringe.    This  causes  some   droplets  to  be  lel  as  residue  in  the  container  as  well  as  risks  having  some  droplets  coalesce   when  drawn  through  the  syringe  Lp. 2. Back  pressure  during  re-­‐injecLon  is  difficult  to  control,  complicated  by  relaLvely  long  tubing   between  the  syringe  pump  and  the  chip  with  changes  in  height.    Olen  Lmes  it  was  necessary   to  apply  a  large  back  pressure  then  switch  it  off  and  have  the  residual  pressure  drive  the   droplets  in  order   3. Droplets,  being  less  dense  than  the  HFE-­‐7500  oil  environment  are  posiLvely  buoyant  and  will   clump  at  the  highest  point  in  a  loop  of  tubing.    This  requires  a  larger  back  pressure  to  drive   droplets  and  results  in  the  droplet  injecLon  speed  being  someLmes  unpredictable.   The  system  in  the  AgresL  paper  has  both  droplet  generator  and  sorter  on  the  same  chip;   droplets  are  formed  and  flow  through  a  stretch  of  polyetheretherketone  tubing  directly  to  the   sorter.    Having  an  integrated  generator/sorter  chip  would  address  some  of  the  issues  surround-­‐ ing  re-­‐injecLon. ! 27 4.3 HV switching   Currently,  the  maximum  operaLng  frequency  of  the  switch  provided  by  the  PHAS  E-­‐LAB   is  approximately  1kHz  at  best,  due  to  the  large  Lme  constant  of  the  signal  output.    In  order  to   properly  test  dielectrophoresis  with  a  HV  AC  signal  (on  the  order  of  20kHz  or  greater),  an  alter-­‐ naLve  switching  method  is  required.    Pavel  Trochtchanovitch  from  the  PHAS  E-­‐LAB  has  several   alternaLves  which  could  be  explored: 1. Using  a  second  switch  to  sink  current  from  the  signal  electrode  as  soon  as  the  first   switch  is  turned  off.    This  would  involve  an  addiLonal  interface  circuit  which  prevents   the  two  switches  from  being  turned  on  at  the  same  Lme,  with  some  tunable  “dead-­‐ Lme”  between  when  the  operaLon  of  the  switches.    Of  course,  this  dead-­‐Lme  will  be  a   limiLng  factor  on  the  maximum  operaLng  frequency  of  the  dual-­‐switch  setup. 2. Building  a  switching  circuit  from  gate  drivers  instead  of  opto-­‐isolaters.    This  design   would  be  limited  to  ±  600V,  so  a  bipolar  HV  source  would  be  required. ! 28 5.0 Appendices Appendix A: Fabrication Protocols A.1 Wafer Fabrication Procedure 1. Clean the surface of the silicone wafer a. Pour IPA solution onto both sides of the silicone wafer b. Use nitrogen gas to dry the wafer 2. Pour photoresist onto the wafer 3. Set wafer on the spin-machine a. Turn on the vacuum and ensure the wafer is secured b. Set up the protocol 4. Start the spin-machine and wait 5. Bake the wafer in the oven 6. Ultraviolet exposure a. Attach the mask on a glass slide b. Place wafer in the frame c. Line up the wafer with the mask using the microscope 7. Expose the wafer in ultraviolet 8. Develop the wafer a. Pour the developer in a container b. Place the wafer in the liquid and rinse c. In a separate container, rinse the wafer again with the same developer 9. Bake the wafer in the oven 10. Examine the wafer under a microscope to make sure no unwanted dust particles are present 11. Measure the height of the oil channel ! 29 A.2 Wafer Fabrication Protocols Name: Adam Q / Luke Date: 07-Feb-12 Photoresist: SU-8 3050 Lot# 11040263 Expiriry 5/1/2012 Previous layers: Should have done IPA coating before resist Spin speed and time 500rpm spread cycle for 10s, 3250 rpm for 30 seconds Pre-bake (time and temperature) 95 C 15 min Exposure time: 9s w/ filter Post-bake time: 1-5-1 65-95-65 Developer: SU8 Developer Development time: 2 min Solvent to wash off developer: IPA Hardbrake: Ramp to 120C, hold 10 mins, ramp down Thickness (using alpha-step) 37-40 (μm) Other notes: Some of the electrodes are touched the oil channel. May be the result of under- developing Name: Adam Q / Luke / Sam Date: 07-Feb-12 Photoresist: SU-8 100 (lot 11080530, exp 9/1/2012) Previous layers: Su8 3050 Spin speed and time 500 rpm for 30s, 1600 rpm for 50s Pre-bake (time and temperature) 65 for 20 min, 95 for 50 min, 65 for 5 min Exposure time: 26 sec with filter Post-bake time: n/a Developer: SU8 developer Development time: 8 min Solvent to wash off developer: Ipa Hardbrake: None Thickness (using alpha-step) n/a ! 30 Other notes: Forgot to tape over alignment marks on pre- vious layer so alignment was very difficult Name: Adam Q / Luke / Sam Date: 09-Feb-12 Photoresist: SU-8 100 (lot 11080530, exp 9/1/2012) Previous layers: SU8 3050, SU8 100 (removed) Spin speed and time 500 rpm for 30s, 1600 rpm for 50s Pre-bake (time and temperature) 65 for 20 min, 95 for 50 min, 65 for 5 min Exposure time: 26 sec with filter Post-bake time: 65 for 2 min, 95 for 12 min, 65 for 2 min Developer: SU8 developer Development time: 8 min Solvent to wash off developer: IPA Hardbrake: None Thickness (using alpha-step) 130μm Other notes: ! 31 A.3 PDMS Chip Fabrication 1. Use TCMS to remove moisture from silicone wafers under a fume hood 2. Make a foil dish for each wafer 3. Place each wafer in foil dish 4. Apply PDMS cleaning layer to remove any particles a. Mix PDMS using 10:1 ratio of RTV A:B with 20g A and 2g B per wafer b. Mix in centrifuge (1 minute mix plus 2 minute degas) c. Pour onto wafer 5. Bake at 80C for 25 minutes 6. Remove cleaning layer a. Cut along edge of water using a scalpel b. Peel off PDMS 7. Apply thicker PDMS layer a. Mix PDMS using 10:1 ratio of RTV A:B with 50g A and 5g B per wafer b. Mix in centrifuge (1 minute mix + 2 minute degas) c. Pour onto wafer 8. Place PDMS with wafers in a vacuum chamber for 1 hour a. Bubbles will form, so alternate at beginning between vacuum and atmospheric pressure to ensure PDMS will not bubble over sides of foil dish b. After bubbling subsides, wafers can be left in vacuum chamber for the re- mainder of the hour 9. Bake at 80C for 1 hour 10. Punch holes for channel inlets and outlets ! 32 Appendix B: SOP for generating droplets for use with 20-67-16 (D1- D2-Height) droplet generator chip. 1. Flush  droplet  generator  with  disLlled  water. 2. Check  for  presence  of  debris;  if  present,  flush  again  with  water. 3. Set  up  syringe  pumps a. Withdraw  disLlled  water  and  HFE-­‐7500  oil  with  3%  w/w  Raindance  surfactant   into  1mL  syringe. b. Afach  tubing  long  enough  to  reach  generator  chip  to  end  of  syringes;  inject  liq-­‐ uid  to  Lps  of  tubing. c. Connect  tubing  to  appropriate  inlets. d. Set  syringe  diameter  (4.83mm  if  using  1mL  syringe). e. Set  flow  rate  to  25uL/hr  for  water,  250uL/hr  for  oil. 4. Turn  on  pump  for  water,  ensuring  that   5. Aler  5  min,  turn  on  pump  for  oil. 6. With  generator  outlet  open,  use  opLcal  setup  and  LabVIEW  program  to  ensure  that  cor-­‐ rectly  sized  droplets  are  being  outpufed. 7. Afach  a  short  piece  of  tubing  to  generator  outlet  and  redirect  droplets  into  a  micro-­‐ centrifuge  tube  (or  container  of  choice)  for  storage. ! 33 Appendix C: HV Switch design by the UBC PHAS E-LAB ! 34   The  switch  transistors  Q1  and  Q2  are  driven  by  an  applied  TTL  signal  (0-­‐5V),  opLcally  iso-­‐ lated  with  U1.    The  signal  is  propagated  through  to  the  3N400S  high-­‐voltage  MOSFET  whose   drain-­‐to-­‐source  voltage  is  rated  as  being  up  to  4kV.    The  MOSFET  begins  to  conduct  when  the   gate-­‐to-­‐source  voltage  exceeds  2~4V,  resulLng  in  the  voltage  at  P5/P6  being  the  same  as  that  at   P1/P2.    C2  has  been  removed  to  allow  for  faster  switching. ! 35 Appendix D: Electrode alloy quotes AIM Solder 12-04-02 7:37 AMGmail - Low melting-point solder quote. Page 1 of 1https://mail.google.com/mail/u/0/?ui=2&ik=288eee50d0&view=pt&q=solder&qs=true&search=query&msg=135ca9579228772c Samuel Wu Low melting-point solder quote. Gaby Melki Wed, Feb 29, 2012 at 11:28 AM To: Samuel Wu Cc: Claude Carreau Hello, Following is the quote for the solder wire you requested: Alloy: 49Bi/21In/18Pb/12Sn Diam.: 0.030" Min qty: 10 feet Price: $45.90/ft Lead time : 10 working days or sooner FOB : Montreal Terms: COD Regards, Gaby Melki Customer Service Manager AIM Metals & Alloys L.P. Tel : 514- 494-5502 * 800-361-0783 Fax: 514- 494-6133 * 800-363-7754 gmelki@aimsolder.com www.aimsolder.com “Solder Plus Support” From: Samuel Wu [mailto:samuelwu90@gmail.com] Sent: Wednesday, February 29, 2012 12:56 PM To: Gaby Melki Subject: Low melting-point solder quote. [Quoted text hidden] ! 36 Indium Page: QUOTATION Customer: ICA Territory: ICA Estimate: ICA Contact: Customer Currency: Net Amount ________________________________________________________________________________ ISO 9001 R E G I S T E R E D Customer RFQ #: Terms: Delivery Terms: Quote Date: Expiration Date: Unit PriceQuantityItem / Description UM Y002-007.RVF 10 FEB 2011 19027944 Samuel W. University of British Columbia ble Dept. of Physics Free car AMPEL, Room 245 Utica, NY 2355 East Mall Vancouver BRC V6T 1Z4 Canada Phone: 604-961-1530 E016093 Prepay (c/card 02/23/12 03/02/12 BritColm US$ US Dollar Ind#19 Solder Wire 28.000 12.71000 355.88 .030" dia. wire GM Min. order 28gm @ $12.71/gm lead time: estimated 12-15 working days after ARO single release Sale Amount: 355.88 USA Tax: 0.00 GST/VAT: 0.00 Misc: 0.00 Total Amount: 355.88 Please contact us immediately to discuss any issues with this correspondence. Thank you for your business. ! 37 6.0 References AgresL,  J.  J.,  AnLpov,  E.,  Abate,  A.  R.,  Ahn,  K.,  Rowat,  A.  C.,  Baret,  J.-­‐C.,  Marquez,  M.,  et  al.   (2010).  Ultrahigh-­‐throughput  screening  in  drop-­‐based  microfluidics  for  directed  evoluLon.   Proceedings  of  the  Na1onal  Academy  of  Sciences  of  the  United  States  of  America,  107(9),   4004-­‐9.  NaLonal  Acad  Sciences.  doi:10.1073/pnas.0910781107 Ahn,  K.,  Kerbage,  C.,  Hunt,  T.  P.,  Westervelt,  R.  M.,  Link,  D.  R.,  &  Weitz,  D.  A.  (2006).  Dielectro-­‐ phoreLc  manipulaLon  of  drops  for  high-­‐speed  microfluidic  sorLng  devices.  Applied  Physics   Le=ers,  88(2),  024104.  doi:10.1063/1.2164911 Behlke.  Fast  High  Voltage  Transistor  Switch:  61-­‐31-­‐GSM  -­‐  Datasheet.  Retrieved  April  1,  2012       from  Behlke  website:  hfp://www.behlke.de/pdf/61-­‐03-­‐gsm.pdf Griffiths,  A.  D.,  &  Tawfik,  D.  S.  (2006).  Miniaturising  the  laboratory  in  emulsion  droplets.  Trends   in  biotechnology,  24(9),  395-­‐402.  doi:10.1016/j.Lbtech.2006.06.009 Mulholland,  B.,  da  Costa,  D.,  &  Eldridge,  D.  (2010).  Ultrafast  Microfluidic  Drop  Sorter.  Alterna-­‐ 1ves. ! 38"""@en ; edm:hasType "Report"@en ; edm:isShownAt "10.14288/1.0074470"@en ; dcterms:language "eng"@en ; ns0:peerReviewStatus "Unreviewed"@en ; edm:provider "Vancouver : University of British Columbia Library"@en ; dcterms:rights "Attribution-NonCommercial-NoDerivatives 4.0 International"@en ; ns0:rightsURI "http://creativecommons.org/licenses/by-nc-nd/4.0/"@en ; ns0:scholarLevel "Undergraduate"@en ; dcterms:isPartOf "University of British Columbia. ENPH 459"@en ; dcterms:title "Ultrafast microfluidic droplet sorter extension work"@en ; dcterms:type "Text"@en ; ns0:identifierURI "http://hdl.handle.net/2429/43239"@en .