Physical Design Issues in Biofluidic Microchips Tamal Mukherjee - - PowerPoint PPT Presentation

physical design issues in biofluidic microchips
SMART_READER_LITE
LIVE PREVIEW

Physical Design Issues in Biofluidic Microchips Tamal Mukherjee - - PowerPoint PPT Presentation

Physical Design Issues in Biofluidic Microchips Tamal Mukherjee MEMS Laboratory ECE Department ECE Department Carnegie Mellon University Pittsburgh, PA, USA tamal@ece.cmu.edu http://www ece cmu edu/~mems http://www.ece.cmu.edu/ mems


slide-1
SLIDE 1

Physical Design Issues in Biofluidic Microchips

Tamal Mukherjee MEMS Laboratory ECE Department ECE Department Carnegie Mellon University Pittsburgh, PA, USA tamal@ece.cmu.edu http://www ece cmu edu/~mems http://www.ece.cmu.edu/ mems

Carnegie Mellon

ISPD, April 16, 2008 ISPD, April 16, 2008

g

slide-2
SLIDE 2

Tubes to Chips: ICs

Driven by Information Processing needs

2

IBM 701 calculator (1952) Intel 4004 Calculator IC (1971)

slide-3
SLIDE 3

Tubes to Chips: BioChips

Driven by Biomolecular Analysis needs

Image from Barnard College Archives

Test tubes & Beakers (1950) Agilent DNA analysis Lab on a Chip (1997)

3

(1950) Lab on a Chip (1997)

slide-4
SLIDE 4

Portable Analysis y

New knowledge of molecular basis of biology

e.g. Human Genome Project Massively parallel analysis infrastructure

Integration and miniaturization will drive

biomolecular analysis instrumentation

Biomolecular S k ith T i d

4

Biomolecular “mainframes” Spock with Tricorder Sensor + computer Burns Science 2002

slide-5
SLIDE 5

Typical Biological Lab Functions y g

Synthesis Analysis

A C A A + B B C A + B B A + B A + B Mixing Reaction Separation

5

Mixing Reaction Separation

slide-6
SLIDE 6

Microdevice Technology Summary gy y

BioMEMS Bi Chi BioChips Droplet Lab-on-a Chip Droplet Ch l P Channel Pressure Electrokinetics

6

Electrokinetics

slide-7
SLIDE 7

Channel-based LoC: EK drive

What is Electrokinetics ?

V lt d i fl

Flow direction

Voltage driven flow

Wh Electrokinetic flo ?

Why Electrokinetic flow?

Plug velocity profile Portable kV sources

EK flow Pressure flow

Portable kV sources EK flow can be used for

electrophoresis p

EK flow already

used in complex designs

Serial Mixer

designs

Mixer ORNL

7

slide-8
SLIDE 8

Microdevice Technologies: LoC g

Miniaturized Bio-chemical Lab-on-a-Chip Individual functional units demonstrated

Analyzer, Reactor, …

Research driven by integration

Design aids needed to handle complexity!

Design aids needed to handle complexity!

DNA Analysis Chemical Synthesis Amino-Acid Analysis Immunoassay y y

8

  • U. C. Berkeley
  • U. Hull

ORNL

  • U. Alberta
slide-9
SLIDE 9

Outline

Introduction Motivation for Design Automation Design Hierarchy

g y

Multi-function System Simulation Multi plex Physical Synthesis Multi-plex Physical Synthesis Summary

9

slide-10
SLIDE 10

Multiplex Lab-on-a-Chip

Same subsystem, integrated for redundancy,

bi t i l i t combinatorial experiments

integration

10

year

slide-11
SLIDE 11

Multifunction LoCs

Example: Immunoassay

1 Load sample {Ag* Ag} loading reagent

  • 1. Load sample {Ag*, Ag}
  • 2. Mix with reagents {Ab}

3 R

A * + Ab A * Ab

mixing

  • 3. Rxn: Ag* + Ab Ag*-Ab
  • 4. Inject sample plug

S t l t

mixing reaction

  • 5. Separate analytes
  • 6. Detection

{ [Ag* Ag] Ab Ag*-Ab}

injection buffer

{ [Ag ,Ag], Ab, Ag -Ab}

separation sample waste p detection waste

11

buffer waste

slide-12
SLIDE 12

Complexity Hierarchy y y

Subsystem System Element Functional Subsystem System Element Functional Component Increasing Integration

12

slide-13
SLIDE 13

Outline

Introduction Motivation for Design Automation Design Hierarchy

g y

Multi-function System Simulation Multi plex Physical Synthesis Multi-plex Physical Synthesis Summary

13

slide-14
SLIDE 14

Simulation Techniques

Computational fluid dynamics

B uf f er B uf f er

buffer

One single turn Complimentary turns

B uf f er S am pl e B uf f er S am pl e

buffer sample

Flow direction Flow direction

Reduced order models

~ 10 Hours 2~3 days ~ 10 hours

educed o de

  • de s

Hierarchical decomposition

and parameterization

Serial Mixer (ORNL)

p

Capture geometric effects Amenable for use in design

(ORNL)

14

slide-15
SLIDE 15

Hierarchy Example: Immunoassay y y

V+ V+ Ag* Ag

Buffer

V+

Sample

V+ Ag , Ag

Flow Direction

V+ V-

Mixing and Reaction

V+

g Ag* + Ab Ag*-Ab

Ab

Sample Waste Ag*, Ag, Ab, Ag*-Ab Pinching

V+

Ag Ab Buffer

15

V+

Buffer Waste

slide-16
SLIDE 16

Synthesis Phase: Steady State y y

V+ V+

Buffer

V+

Flow Direction Mixing and Reaction

V- V+

Pinching g Ag* + Ab Ag*-Ab Sample Waste Ag*, Ag, Ab, Ag*-Ab Ag Ab Buffer

16

V+

Buffer Waste

slide-17
SLIDE 17

Analysis Phase: Transient y

V+

Buffer

V-

Flow Direction

V- V-

Sample Plug Sample Waste Sample Plug Mixture of Ag*, Ag, Ab, Ag*-Ab

  • n

l

V-

Separatio Channe Buffer S

17

V-

Buffer Waste

slide-18
SLIDE 18

Analysis Phase: Transient y

V+

Buffer

V-

Flow Direction

V- V-

Sample Waste

  • n

l

V-

Separatio Channe Ag*-Ab Ab Buffer S Ag*,Ag

apart distance R l ti

18

V-

Buffer Waste

broadness band p Resolution =

slide-19
SLIDE 19

Component Library

Library of LoC Unit Compose Topology Operations Compose Topology Function Type

well mixer

Function Type

mixer reactor injector separator splitter

19

slide-20
SLIDE 20

Composition Examples

Sample waste Buffer Sample A B

Serpentine separation hi (ORNL) M lti t i (M K h t l ) chip (ORNL)

System waste aste System waste

Multi-stream mixer (M. Koch, et al.)

ste-1 Buffer Sample wa Buffer e Was S B Sample A1 A2 A3 A4 A5 Waste 2 Sample

20

Spiral chip (ORNL) Serial Mixing network (S.C. Jacobson, et al.)

Waste-2

slide-21
SLIDE 21

Outline

Introduction Motivation for Design Automation Design Hierarchy

g y

Component Models Multi function System Simulation Multi-function System Simulation Multi-plex Physical Synthesis Summary

21

slide-22
SLIDE 22

Simulating a Multifunction Design g g

Real

(Cheim, Clin. Chem., 44:3, 591-598, 1998)

Immunoassay Chip from U. Alb t Alberta

Operation

Mixing/Dilution

g

Reaction Injection

j

Separation Detection

22

Wang et. al. Transducers ‘05

slide-23
SLIDE 23

Simulation Results

10 mm after injection 10 mm after injection 10 mm after injection

Calibration curve

1.0 Schematic E i l

Th*

1.0 Schematic E i l

Th*

Ag*

ntration c

Before turn

Th*

ntration c

Before turn

ntration c ntration c

Before turn

Th*

0.6 0.8 Experimental

a ratio Th

0.6 0.8 Experimental

a ratio Th

Ag Ag*

elative concen

Th*-Ab complex Th*

elative concen elative concen elative concen

Th*-Ab complex Th*

0.2 0.4

Area Ab-Th*

0.2 0.4

Area Ab-Th*

Ab-Ag* Ab-Ag* Ag

Re

After turn

Re

After turn

Re Re

After turn

10 20 30 40 0.0

Theophylline Th (mg/L)

10 20 30 40 0.0

Theophylline Th (mg/L)

Antigen (Ag) (mg/L)

Electropherogram

5 10 15 20 25 30 35

Time (s)

5 10 15 20 25 30 35

Time (s)

5 10 15 20 25 30 35

Time (s)

5 10 15 20 25 30 35

Time (s)

Theophylline Th (mg/L) Theophylline Th (mg/L)

Antigen (Ag) (mg/L)

Ag↑ ⇒ unreacted Ag* ↑ ⇒ Ab-Ag* ↓ Simulation matches experiment

23

Simulation matches experiment Simulation time is a few CPU seconds

Wang et. al. Transducers ‘05

slide-24
SLIDE 24

Optimizing the design: NLP g g

) ( i : f bj

x x

) ( ) ( . . ) ( min : = <

i i

y h y g t s y f

  • bj

) , ( PARAMS x SIM y

i i =

i

x

i

y

) ( =

i

y h

*

x

7.6 cm

10x less space S f

mixer 0.75 cm

Same perf

2 cm

reactor injector separation channel detector 1.22 cm 1.73 cm 1.14 cm 2.23 cm

7.6 cm 24 2.5 cm

2.33 cm

7

Wang et. al. Transducers ‘05

slide-25
SLIDE 25

Outline

Introduction Motivation for Design Automation Design Hierarchy

g y

Multi-function System Simulation Multi plex Physical Synthesis Multi-plex Physical Synthesis Summary

25

slide-26
SLIDE 26

Multiplex Physical Synthesis y y

Family of subsystems

Simultaneously determine: placement dimensions # of sections voltage Input: Design Specs

Intermediate Placement

voltage Overall Chip Dimensions Species/Buffer properties Operational constraints Operational constraints Chip fabrication Subsystem performance

Final Routed Layout

Route subsystems to wells:

26

single layer, planar

  • min. length, bends

Pfeiffer et. al., TCAD ‘06

slide-27
SLIDE 27

Placement Features

Subsystem optimization:

NLP

“System-on-Chip” extensions*:

a d

c

s t e a c d e

Orientation:

b f s b c f

* Murata, H. et al., IEEE Trans. on CAD. 1996

Orientation: Well placement:

and

ET

a b c d e f

a b c d e f EL ER

a b c d e f

Overlap constraints:

b f

EB

a Never

a d e

Penalty

b f

27

a b Never

a b c d f

Penalty

Pfeiffer et. al., TCAD ‘06

slide-28
SLIDE 28

Routing Features g

a d e d a e

Routing grid graph*:

Expand

a b c d e f d a c f b

Expand

* Lengauer, T., Combinatorial Algs. for IC Layout, 1990

f f 8 7 9

fl i fl t

Node constraints:

5 4 6

flows in = flows out 1 flow in/out of node (single layer, planar)

1 2 3

(single layer, planar)

8 7 9

+1

Bend reduction:

5 4 6

+1 +0

penalize bends favor straight paths

+1

28 1 2 3

+0

favor straight paths

Pfeiffer et. al., TCAD ‘06

slide-29
SLIDE 29

Multiplex Synthesis Example y

Placed and Routed Design

P&R By Hand Automated Improvement

Place: 20 min

Time

5+ hrs.

Place: 20 min Route: 3 min Total: 23 min

> 10X faster

Di i

1 67 8 8 1 61 3 79 2 5X ll

29

Dimensions

1.67cm x 8.8cm 1.61cm x 3.79cm ~ 2.5X smaller

Pfeiffer et. al., TCAD ‘06

slide-30
SLIDE 30

Summary

Lab on a Chips integrate bioanalysis functions Hierarchically decomposition used for

development of fast, accurate, reusable, t i d d l parameterized models

Relatively few types of band profile on a chip

Profile representation to simplify PDE into ODEs

Models are reuseable Separation models integrated with P&R

algorithms for simultaneous model based a go t s o s u ta eous

  • de based

placement followed by routing

Focus still at ‘circuit’-level, need to consider

30

Focus still at circuit level, need to consider

architecture, protocol optimization

slide-31
SLIDE 31

Acknowledgements g

Collaborators

  • Prof. James F. Hoburg (ECE)
  • Prof. Steinar Hauan (ChE)
  • Prof. Qiao Lin (ME, now at Columbia Univ)

Students

Anton Pfeiffer, Yi Wang, Ryan Magargle, Xiang He,

Bikram Baidya

Funding

DARPA DSO SIMBIOSYS Program NSF ITR Program

31

g