1

1

Automated Design of Digital Microfluidics Lab-on-Chip

Krishnendu Chakrabarty

Department of Electrical and Computer EngineeringDuke UniversityDurham, NC

Connecting Biochemistry to Information TechnologyAnd Electronic Design Automation

2

AcknowledgmentsAcknowledgments• Students: Tianhao Zhang, Fei Su, William Hwang, Phil Paik, Tao Xu,

Vijay Srinivasan, Yang Zhao• Post-docs, colleagues, and collaborators: Dr. Vamsee Pamula, Dr.

Michael Pollock, Prof. Richard Fair, Dr. Jun Zeng (HP Labs), Dr. S. Krishnamoorthy (Baxter)

• Duke University’s Microfluidics Research Lab (http://www.ee.duke.edu/research/microfluidics/)

• Advanced Liquid Logic (http://www.liquid-logic.com/): Start-up company spun out off Duke University’s microfluidics research project

2

3

Motivation for LabMotivation for Lab--onon--ChipChip• Clinical diagnostics, e.g., healthcare for

premature infants, point-of-care diagnosis of diseases

• “Bio-smoke alarm”: environmental monitoring• Massive parallel DNA analysis, automated

drug discovery, protein crystallization

Conventional Biochemical Analyzer

ShrinkMicrofluidic Lab-

on-a-Chip

CLINICAL DIAGNOSTICAPPLICATION

20nl sample

Lab-on-a-chip forCLINICAL DIAGNOSTICS

Higher throughput, minimal human intervention, smaller sample/reagent consumption, highersensitivity, increased productivity

4

The Futility of Predicting ApplicationsThe Futility of Predicting ApplicationsKroemer’s Lemma of New Technology:

The principal applications of any sufficiently new and innovative technology have always been—and will continue to be—applications created by that technology.

Herbert Kroemer, Department of Electrical and Computer Engineering, University of California at Santa BarbaraNobel Prize winner for Physics, 2000

3

5

Tubes to Chips: Integrated CircuitsTubes to Chips: Integrated Circuits• Driven by Information Processing needs

IBM 701 calculator (1952)

IBM Power 5 IC(2004)

6

Tubes to Chips: Tubes to Chips: BioChipsBioChips• Driven by biomolecular analysis needs

Test tube analysis

BioMark™ Dynamic ArraysFluidigm

4

7

Why Do We Care?Why Do We Care?

2007

System Driver Beyond 2009: “Medical”

Intel Research Day 2007: Biochip prototypedemonstrated for point-of-care diagnostics andlab testing

8

Press Releases and News ItemsPress Releases and News Items

THE W

ALL STREET JO

URNAL

5

9

Why is BiochemistryWhy is Biochemistry--onon--aa--Chip Difficult?Chip Difficult?A

BC

A + B

A

BA + B

Synthesis

Analysis

Mixing

Reaction Separation

10

Why is BiochemistryWhy is Biochemistry--onon--aa--Chip Chip Difficult?Difficult?

6

11

By the way, what’s a biochip?It’s a miniature disposable for an

HTS - High-Throughput Screening -

(bio)analytical instrument

what does it do?Essentially the same operations you did in high school

chemistry class: dispensing,

mixing, detecting,

discarding,-just a lot cheaper and a lot faster than you did

12

Why do chips have to be small?

High-Throughput is why. If you do 106 assays in 10μl format,each time you do a reaction you’ll need 10 liters of reagents.

With the typical cost of biological reagents, even Big Pharma can’t afford this.

By the way, why High-Throughput?• Because you need a lot of raw data for many applications • Because, with the currently available technology, to produce

raw data that would keep a CPU busy for a few minutes ($0.1), you need a Ph.D. scientist and a couple of technicians for a month ($10,000)

7

13

Talk OutlineTalk Outline• Motivation• Technology Overview

– Microarrays– Continuous-flow microfluidics: channel-based lab-on-chip– “Digital” microfluidics: droplet-based lab-on-chip

• Overview of Fabrication Method• Design Automation Methods

– Synthesis and module placement– Droplet Routing– Pin-Constrained Design– Testing and Reconfiguration

• Conclusions

14

MicroarraysMicroarrays• DNA (or protein) microarray: piece of glass, plastic or silicon

substrate• Pieces of DNA (or antibodies) are affixed on a microscopic array• Affixed DNA (or antibodies) are known as probes• Only implement hybridization reaction

ATCGG

GATC

substrate

CATTGA

Hybridized array

Unhybridized array

DNA Sample

Laser

Optical ScanATCGG

GATC

substrate

CATTGA

TAGCC♦

GTAAC

♦T

CTAG♦

8

15

What are the main types of biochips?What are the main types of biochips?

Passive (array):all liquid handling functions are performed by

the instrument. The disposable is simply a

patterned substrate.

Active (lab-on-chip, μ-TAS):some active functions are performed by the

chip itself. These may include flow control,

pumping, separations where necessary, and

even detection.

16

Motivation for MicrofluidicsMotivation for Microfluidics

Test tubes

Robotics

MicrofluidicsAutomationIntegrationMiniaturization

AutomationIntegrationMiniaturization

AutomationIntegrationMiniaturization

9

17

MicrofluidicsMicrofluidics• Continuous-flow lab-on-chip: Permanently etched microchannels,

micropumps and microvalves• Digital microfluidic lab-on-chip: Manipulation of liquids as discrete

droplets(Duke University)

Control electronics (shown) are suitable for handheld or

benchtopapplications

Printed circuit board lab-on-a-chip –

inexpensive and readily manufacturable

Biosensors: Optical: SPR, Fluorescence etc. Electrochemical: Amperometric,

Potentiometric etc.

Mixing: Static, Diffusion Limited

Multiplexing

18

ElectrowettingElectrowetting• Novel microfluidic platform invented at Duke University• Droplet actuation is achieved through an effect called

electrowetting⎯ Electrical modulation of the solid-liquid interfacial tension

No PotentialA droplet on a hydrophobic surface originally has a large contact angle.

Applied PotentialThe droplet’s surface energy increases, which results in a reduced contact angle. The droplet now wets the surface.

10

19

What is Digital Microfluidics?What is Digital Microfluidics?

• Discretizing the bottom electrode into multiple electrodes, we can achieve lateral droplet movement

Droplet Transport (Side View)Note: oil is typically used to fill between the top and bottom plates to prevent evaporation.

20

What is Digital Microfluidics?What is Digital Microfluidics?

Transport25 cm/s flow rates, order of magnitude

higher than continuous-flow

methods

For videos, go to www.ee.duke.edu/research/microfluidics

11

21

What is Digital Microfluidics?What is Digital Microfluidics?

Splitting/Merging

22

Demonstrations of Digital MicrofluidicsDemonstrations of Digital Microfluidics

Droplet FormationDroplet Formation

Synchronization of many dropletsSynchronization of many dropletshttp://www.ee.duke.edu/research/microfluidics

12

23

What is Digital Microfluidics?What is Digital Microfluidics?

Droplet Formation8 droplets in 3.6s

24

What is Digital Microfluidics?What is Digital Microfluidics?

Mixing

13

25

AdvantagesAdvantages• No bulky liquid pumps are required

– Electrowetting uses microwatts of power– Can be easily battery powered

• Standard low-cost fabrication methods can be used

– Continuous-flow systems use expensive lithographic techniques to create channels

– Digital microfluidic chips are possible using solely PCB processes

Droplet Transport on PCB (Isometric View)

26

An ExampleAn Example• Detection of lactate, glutamate and pyruvate has also been

demonstrated.• Biochip used for multiplexed in-vitro diagnostics on human

physiological fluids

Fabricated microfluidic array used for multiplexed biomedical assays

14

27

CapabilitiesCapabilities• Digital microfluidic lab-on-chip

MIXERSMIXERSTRANSPORTTRANSPORT DISPENSINGDISPENSING REACTORSREACTORS

INTEGRATE

Digital Microfluidic

Biochip

DETECTIONDETECTION

Basic microfluidic functions (transport, splitting, merging, and mixing) have already been demonstrated on a 2-D arrayHighly reconfigurable system

Protein crystallization chip(under development)

28

Advantages of Digital MicrofluidicsAdvantages of Digital Microfluidics

• Very accurate droplet volumes– Droplet sizes in the 1 nanoliter to several

microliter range; droplet dispensing volume variation ~1%

• Programmable, software-driven electronic control

– No moving parts, tubes, pumps or valves • More efficient use of samples and reagents

– No liquid is wasted priming channels• Extremely energy efficient

– Nanowatts of power per single step of actuation

• Development cycles are short, and assays can be implemented with software changes

• Compatible with live biologic and most other materials

• Pump fluids through channels• Must adapt assays to channel-

based format• Complex or multiplexed assays

become a plumber’s nightmare• Off-chip pumps and valves mean

large, expensive equipment and low reliability

• Expensive, time consuming, up-front investments required for most chip developments

• Designs are fixed in the development process

Other Microfluidic TechnologiesDigital Microfluidics

•Droplets moved in “virtual channels” defined by electrodes•Programmable electrodes directly control discrete droplet operations

Caliper Technologies’LabChip

15

29

Glass Chip Platform DevelopmentGlass Chip Platform DevelopmentTop Plate (Optional) (i.e. glass or plastic)

Gasket Layer (100 to 600 µm) (proprietary)

Hydrophobic Layer (50 nm) (i.e. Teflon dip coated)

Insulator Layer (1 to 25 µm) (i.e. parylene)

Patterned Metal on Substrate(i.e. chrome on glass via lift-off process)

Top plate is either glued or fixed in place by pressure

Contacts are made either through the top or bottom

Droplets are either dispensed by hand or formed from on-chip reservoirs

Chip Assembly

30

PCB Chip Platform DevelopmentPCB Chip Platform Development

Fabrication ProcessFlash Plating

(Copper)

PCB

• PCB Material – Mitsui BN300 – 64 mil• Top Metal Layer (Electrodes) – Cu – 15µm• Bottom Metal Layer (Contacts) – Cu – 15µm• Dielectric – LPI Soldermask – 25 µm• Via Hole Filling – Non-conductive Epoxy• Hydrophobic Layer – Teflon AF – 0.05 to 1.0 µm• Gasket (spacer) – Dry Film Soldermask (Vacrel 8140) – 4 mils (~95µm after processing)

Gasket Layer(Dry Soldermask)

Hydrophobic Layer(Teflon AF)

Dielectric(LPI Soldermask)

Top Metal Layer(Copper)

Bottom Metal Layer(Copper)

Via Hole Filling(Non Conductive Epoxy)

16

31

ComputerComputer--Aided Design: VisionAided Design: Vision• Automate labor-intensive tasks, reduce burden on chip users

– Map bioassays to a fabricated chip: schedule fluidic operations,determine droplet flow pathways, configure fluidic modules dynamically, etc.

– Monitor the chip for defects that require remapping of bioassays• Role of computer-aided design (CAD) tools

– Reduce setup time associated with the use of these chips– Allow automatic reconfiguration of a faulty chip and remap the

remaining steps of bioassay. – Develop capabilities that mirror compiler and operating system support

provided to software programmers– Obviate the need for tedious remapping of assays to the chip by hand

for each target application.• Similar to an FPGA? Logic Interconnects

32

CAD = ?CAD = ?

cad n. An unprincipled, ungentlemanly person

CAD abbr. Computer-Aided Design

Better to be in CAD than to be a cad?

17

33

The Road Not TakenThe Road Not Taken……I shall be telling this with a sighSomewhere ages and ages hence:Two roads diverged in a wood, and I-I took the one less traveled by,And that has made all the difference.

Robert Frost, The Road Not Taken

Agilent’s Protein LabChip Nanogen’s NanoChip™

Microelectronic Array Cartridge

i-STAT Biodiagnostic μ-system

34

Similar to Concurrency in a PC!Similar to Concurrency in a PC!

January 28, 2007 Prof. Radu Marculescu Attn: Outstanding Ph.D. Dissertation Award Carnegie Mellon University Department of Electrical and Computer Engineering 5000 Forbes Avenue Pittsburgh PA 15213-3890

Dear Prof. Marculescu:

I am very pleased to write this letter of nomination for Dr. Fei Su for the ACOutstanding Ph.D. Dissertation Award in Electronic Design Automation. received his Ph.D. degree from Duke University in May 2006 and his thesis wowas carried out under my supervision.

Fei’s PhD dissertation is titled “Synthesis, Testing, and ReconfiguratTechniques for Digital Microfluidic Biochips”. It is focused on design automatiand test methods for emerging lab-on-a-chip devices that rely on the principleelectrowetting-on-dielectric. By exploiting the reconfigurability inherent droplet-based “digital” microfluidics, these devices are revolutionizing a wrange of applications, such as high-throughput sequencing, paraimmunoassays, blood chemistry for clinical diagnostics, DNA sequencing, aenvironmental toxicity monitoring.

Microsystems for biomedical and sensing applications are often referred to lab-on-a-chip or biochips. These are typically centimeter-sized chips, with on-chcomponents having micrometer feature lengths. These components are createdIC fabrication technology (surface or bulk micro-machining), and they are diverse functionality. Just as bioscience is sometimes called “wet” science, tapplication of biochips relies primarily on its ability to work with fluids throuits on-chip components. Biochemical samples are placed on the biochip inliquid form. A pre-programmed analysis is then carried out automatically andparallel. Miniaturization enables minute sample volume, thus it speeds chemical reactions and analytical detection; automation and parallelization mait possible to carry out a massive number of different tests simultaneously. Thecharacteristics, especially the delivery of results for a large number of tests witha short amount of time, are especially relevant for clinical diagnostienvironmental monitoring, and bio-defense applications.

Digital microfluidics has heralded the second (and remarkably advancgeneration of biochips. It utilizes tiny droplets as on-chip chemical compoucarriers. An on-chip array of electrodes that are individually addressable throuCMOS electronics can manipulate each droplet electrically. A set of programmaCMOS instructions can induce the merging of two droplets. Such mergoperations constitute the key operations in on-chip chemical reactions. Multi-s

The operating system manages complexity, allows multi-tasking!

18

35

Constraints:Area <…Delay <…

Top-down design method

System Level

…..Module Level

….. Gate Level

…..

Circuit Level

Gate Level

Module Level

VLSI chip wanted

Bottom-up design method

Design MethodologyDesign Methodology• VLSI design

…..

Circuit Level

System LevelIf constraints are not met

Re-design

36

Biochip wanted

Bottom-up design method

If constraints are not met

Re-design all sub-blocks !

Biochip Design Methodology Biochip Design Methodology • Bottom-up vs. top-down biochip design

Component 1 designed and verified

Component ndesigned and verified

Anticipated to be needed

…..

…Module 1

designed and verifiedModule m

designed and verified

Anticipated to be needed

Biochip designedBiochip wanted

Constraints

Biochip designed

Module 1 designed

Module m designed

Constraints Constraints

…..

Component 1 designed

Component ndesigned

…..

ConstraintsConstraints

Top-down design method

19

37

Biochip Design AutomationBiochip Design Automation

38

Design Automation: Biochip SynthesisDesign Automation: Biochip Synthesis• Full-custom bottom-up design Top-down system-level design

Scheduling of operationsBinding to functionalresourcesPhysical design

S1: Plasma, S2: Serum,S3: Urine, S4: Saliva

Assay1: Glucose assay, Assay2: Lactate assay, Assay3: Pyruvate assay, Assay4: Glutamate assay

S1, S2, S3 and S4 are assayed for Assay1, Assay2, Assay3 and Assay4.

20

39

Sequencing Graph ModelSequencing Graph Model

Sequencing graph model for multiplexed bioassays

40

Mathematical Programming Model Mathematical Programming Model

• First define a binary variable

⎩⎨⎧

=ijX 1 if operation vi starts at time slot j.

0 otherwiseStarting time of operation vi :

∑=

×=T

jiji XjSt

1

Completion time of operation:C = max {Sti + d(vi) : vi ∈D1, …, Dn}

Objective function: minimize C

Dependency constraintsStj ≥ Sti + d(vi) if there is a dependency

between vi and vj

Resource constraintsReservoirs/dispensing ports

Nr reservoirs/dispensing ports assigned to each type of fluid (Nr = 1)

… : 1≤ j≤ T,11:

∑∈

≤Ivi

iji

X ∑+∈

≤nmi Ivi

ijX:

1

Reconfigurable mixers and storage units

Nmixer(j) + 0.25 Nmemory(j) ≤ Na 1 ≤ j ≤ T

Optical detectorsNd detectors are assigned to each

bioassay (Nd = 1)

,11: )(

∑ ∑∈ −=

≤Dvi

j

vdjlij

i i

X ∑ ∑∈ −=

≤1: )(

1ni iDvi

j

vdjlijX… 1≤ j≤ T

Objective Constraints

21

41

Physical Design: Module PlacementPhysical Design: Module Placement• Placement determines the locations of each module on the

microfluidic array in order to optimize some design metrics • High dynamic reconfigurability: module placement 3-D

packing modified 2-D packing

Reduction from 3_D placement to a modified 2-D placement

42

Unified Synthesis MethodologyUnified Synthesis Methodology

22

43

Protein Assay: Dilution StepsProtein Assay: Dilution StepsSequencing graph model

• Maximum array area: 10x10

• Maximum number of optical detectors: 4

• Reservoir number: 1 for sample; 2 for buffer; 2 for reagent; 1 for waste

• Maximum bioassay time: 400 s

44

Synthesis ResultsSynthesis Results

Bioassay completion time T: 363 seconds

Biochip array: 9x9 array

23

45

Synthesis Results (Cont.)Synthesis Results (Cont.)• Defect tolerance

Bioassay completion time T: 385 seconds (6% increase)

46

Droplet RoutingDroplet Routing• A key physical design problem for digital microfluidic

biochips• Given the results from architectural-level synthesis and

module placement:– Determine droplet pathways using the available cells in the

microfluidic array; these routes are used to transport droplets between modules, or between modules and fluidic I/O ports (i.e., boundary on-chip reservoirs)

• To find droplet routes with minimum lengths– Analogous to the minimization of the total wirelength in VLSI

routing• Need to satisfy critical constraints

– A set of fluidic constraints– Timing constraints: (the delay for each droplet route does not

exceed some maximum value, e.g., 10% of a time-slot used in scheduling)

24

47

Static fluidic constraintStatic fluidic constraint Dynamic fluidic constraintsDynamic fluidic constraints

Fluidic ConstraintsFluidic Constraints

Rule #1: |Xi(t+1) − Xj(t+1)| ≥ 2 or |Yi(t+1) − Yj(t+1)| ≥ 2, i.e., their new locations are not adjacent to each other.

• Assume two given droplets as Diand Dj, and let Xi(t) and Yi(t) denote the location of Di at time t

How to select the admissible locations at time t +1?

Rule #2: |Xi(t+1) − Xj(t)| ≥ 2 or |Yi(t+1) − Yj(t)| ≥ 2, i.e., the activated cell for Di cannot be adjacent to Dj.

Rule #3: |Xi(t) − Xj(t+1)| ≥ 2 or |Yi(t) − Yj(t+1)| ≥ 2.

Directly adjacent

Diagonally adjacent

48

Experimental VerificationExperimental Verification

(a) Experimental verification of Rule #1: droplets begin on electrodes 1 and 4; (b) Electrodes 2 and 3 are activated, and 1 and 4 deactivated; (c) Merged droplet.

(a) Experimental verification of Rule #2: droplets begin on electrodes 2 and 4; (b) Electrodes 1 and 3 are activated, and 2 and 4 deactivated.

25

49

Experimental Verification (Cont.)Experimental Verification (Cont.)

(a) Experimental verification of Rule #3: droplets begin on electrodes 4 and 7; (b) Electrodes 3 and 6 are activated, and 4 and 7 deactivated; (c) Merged droplet.

• To demonstrate that adherence to Rule #1 is not sufficient to prevent merging. Both Rule #2 and Rule #3 must also be satisfied during droplet routing.

• These rules are not only used for rule checking, but they can also provide guidelines to modify droplet motion (e.g., force some droplets to remain stationary in a time-slot) to avoid constraint violation if necessary

50

Design of PinDesign of Pin--Constrained BiochipsConstrained BiochipsDirect Addressing• Each electrode connected to an independent pin

• For large arrays (e.g., > 100 x 100 electrodes)– Too many control pins ⇒ high fabrication cost– Wiring plan not available

PCB design: 250 um via hole, 500 um x 500 um electrode

Via HolesVia HolesWiresWires

Nevertheless, we need high-throughput and low cost:DNA sequencing (106 base pairs), Protein crystallization (103 candidate conditions)

Disposable, marketability, $1 per chip

26

51

PinPin--Constrained Biochip DesignConstrained Biochip Design• Cross-referencing

Orthogonally placed pins on top and bottom plates

Advantagek = n x m pins n + m pins for an n x m microfluidic array

DisadvantageSuffer from electrode interference

52

Electrode InterferenceElectrode Interference• Unintentional Electrode Actuation

Selected column and row pins may intersect at multiple electrodes

• Unintentional Droplet Manipulation

1 2 3 4 5 6 7 8 9 10

1 2 3 4 5 6 7 8 910

3

2

1

destination cellsdestination cells

Unintentional Unintentional destination cellsdestination cells

27

53

Efficient (Concurrent) Droplet Efficient (Concurrent) Droplet ManipulationManipulation• Goal: Improve droplet manipulation concurrency on

cross-referencing-based biochips.

9 steps needed if 9 steps needed if moving one droplet moving one droplet at a time (too slow)at a time (too slow)

54

Efficient Droplet ManipulationEfficient Droplet Manipulation• Observation

– Droplet manipulations whose destination cells belongs to the same column/row can be carried out without electrode interferences.

destination cellsdestination cells

94

28

55

Efficient Droplet ManipulationEfficient Droplet Manipulation• Methodology

– Group droplet manipulations according to their destination cells– All manipulations in a group can be executed simultaneously

The goal is to find an optimal grouping plan which results in the minimum number of groups.

56

Efficient Droplet ManipulationEfficient Droplet Manipulation• Problem formulation

Destination cells NodesDestination cells in one column/row a clique Grouping Clique partitioningOptimal grouping Minimal clique-partitioning (NP-Complete)

29

57

Broadcast ElectrodeBroadcast Electrode--AddressingAddressing• Observation

“Don’t-Cares” in Electrode-Actuation SequencesElectrode control inputs: 3 values“1” –- activated “0” –- deactivated“x” –- can be either “1” or “0”Therefore, activation sequences can be combined by interpreting “x” Floating electrodeFloating electrode

Example: A droplet routed counterclockwise on a loop of electrodes Corresponding electrode activation

sequences

58

Solution Based on Clique PartitioningSolution Based on Clique Partitioning• Idea

– Combining compatible sequences to reduce # of control pins

• Clique partitioning based methodElectrodes NodesElectrodes with compatible activation sequences a clique Optimal combination Minimal clique-partitioning

30

59

Solution Based on Clique PartitioningSolution Based on Clique Partitioning

Bioassay synthesis results

Scheduling & droplet routing plan

Activation sequence for each electrode Undirected graph

Extract

Calculate

Map

Clique partitioning Result

Generate

Reduced number of control pins

Combine

60

Application to a Multiplexed BioassayApplication to a Multiplexed Bioassay

A biochip target execution of a multiplexed assay

Sequencing graph model of the multiplexed assay

• A glucose assay and a lactate assay based on colorimetric enzymatic reactions • 4 pairs of droplets – {S1, R1}, {S1, R2}, {S2, R1}, {S2, R2}, are mixed in the mixer in

the middle of the chip, the mixed droplets are routed to the detector for analysis

31

61

Results Results

303525# of control pins

Cross-referencing-based method

Array-partitioning-based method

Broadcast addressing

Addressing methods

Comparison of bioassay completion time using different addressing methods

73 s 73 s

132 s

62

Application to MultiApplication to Multi--functional functional ChipChip• Multi-functional Chip

– biochips targeting the execution of a set of (multiple) predetermined bioassays

• Application of Broadcast Addressing to Multi-functional Chips Key idea: treat the union of the target bioassays as a single bioassay– Collect droplet routing and schedule information for each target

bioassay– Calculate activation sequences for each bioassay – Merge the activation sequences from the different assays and obtain a

collective activation sequence for each electrode – Note that merging of activation sequences can be carried out in any

arbitrarily-chosen order

32

63

Addressing Results Addressing Results

Sequencing graph model of the multiplexed assay

Sequencing graph model of protein dilution

Sequencing graph model of Polymerase Chain Reaction (PCR)

64

Addressing Results Addressing Results Chip layout and broadcast-addressing result for the multi-functional chip for

1. Multiplexed assay2. PCR assay 3. Protein dilution assay

Total number of control pins: 37

The addition of two assays to the biochip for the multiplexed assay leads to only 13 extra control pins

33

65

ReconfigurabilityReconfigurability• Common microfluidic operations

– Different modules with different performance levels (e.g., several mixers for mixing)

– Reconfiguration by changing the control voltages of the corresponding electrodes

66

Reconfiguration and Graceful DegradationReconfiguration and Graceful Degradation• Reconfigure the faulty module

– Avoid defects (faulty cells)• Reconfiguration: bypass faulty cells

– No spare cells; use fault-free unused cells• Defect tolerance in design procedure (increase in design complexity)

– Incorporate physical redundancy in the array• Spare cells replace defective cells (local reconfiguration,

application-independent)•

34

67

Testing of Digital Microfluidics BiochipsTesting of Digital Microfluidics BiochipsStimuli: Test droplets; Response: Presence/absence of droplets

Fragmentation of droplets and their motion is prevented

Dielectric islands(islands of Teflon coating)

1Non-uniform dielectric layer

Coatingfailure

Droplet transportation without activation voltage

Pressure gradient (net static pressure in some direction)

1Misalignment of parallel plates (electrodes and ground plane)

Excessive mechanical force applied to chip

Unintentional droplet operations or stuck droplets

Electrode-stuck-on (electrode remains constantly activated)

1Irreversible charge concentration on electrode

Electrodeactuation for excessiveduration

Droplet undergoes electrolysis; preventsfurthertransportation

Droplet-electrode short (short between the droplet and the electrode)

1Dielectric breakdown

Excessiveactuationvoltageapplied toelectrode

Observableerror

Faultmodel

No.cells

Defect type

Cause of defect

68

More Defects in Digital Microfluidic BiochipsMore Defects in Digital Microfluidic BiochipsObservableerror

Faultmodel

No.cells

Defect type

Cause of defect

Assay results are outside the range of possible outcomes

Contamination

Droplet transportation is impeded.

Resistive open at electrode

1Sample residue on electrode surface

Protein absorption during bioassay

Electrode short2Particle connects two adjacent electrodes

Particle contamination or liquid residue

A droplet resides in the middle of the two shorted electrodes, and its transport cannot be achieved

Electrode short (short between electrodes)

2Metal connectionbetween adjacent electrodes

Failure to activate theelectrode for droplettransportation

Electrode open (actuation not possible)

1Broken wire to Control source

Failure of droplet transportation

Floating droplets (droplet not anchored )

1Grounding failure

Abnormalmetal layerdepositionand etchvariationduringfabrication

35

69

Electrical Detection MechanismElectrical Detection Mechanism• Minimally invasive• Easy to implement (alleviate

the need for external devices)• Fault effect should be

unambiguous

Capacitive changes reflected in electrical signals (Fluidic domain to electrical domain)• If there is a droplet,

output=1; otherwise, output=0

• Fault-free : there is a droplet between sink electrodes Faulty: there is no droplet.

Electrically control and track test stimuli droplets

Droplet

150 pF

74C14

5 K

1N914

1N52315.1V

1N914

Gnd

+ 5 V

10 KOutputPeriodic

square waveform

70

DefectDefect--Oriented Experiment Oriented Experiment • Understand the impact of certain defects on droplet flow, e.g., for

short-circuit between two electrodes• Experimental Setup

– To evaluate the effect of an electrode short on microfluidic behavior

36

71

ConclusionsConclusions• Digital microfluidics offers a viable platform for lab-on-chip for clinical

diagnostics and biomolecular recognition• Design automation challenges

– Automated synthesis: scheduling, resource binding, module placement; droplet routing; testing and reconfiguration

• Bridge between different research communities: bioMEMS, microfluidics, electronics CAD and chip design, biochemistry

• Growing interest in the electronics CAD community– Special session on biochips at CODES+ISSS’2005 (appears in CFP now)– Special issue on biochips in IEEE Transactions on CAD (Feb 2006), IEEE Design &

Test of Computers (Jan/Feb’07)– Workshop on biochips at DATE’06– Tutorials on digital microfluidic lab-on-chip at DATE’07, ISCAS’08, VDAT 2007;

embedded tutorial at VLSI Design 2005– Other notable activities in digital microfluidics: University of California at Los

Angeles, University of Toronto, Drexel University, IMEC (Belgium), Freiburg (Germany), Philips (Netherlands), Fraunhofer Institute (Berlin, Germany), and many more….

72

37

73

2006 X, 406 p. HardcoverISBN-10: 1-4020-5122-0, ISBN-13: 978-1-4020-5122-7

ISBN: 0849390095Publication Date: 10/5/2006, 248 p.