Bio-transport and Design
of BioMEMS Devices
(Bios 10 lecture)
James Gilchrist
Department of Chemical Engineering
September 30, 2009
an engineer
What is chemical engineering?
Chemical engineers design processes that perform
molecular transformations of raw materials
into products.
Intersection of Math, Physics, Chemistry, Biology
Bio
Phys
Math
but
Chem
Math + Physics + Chemistry + Biology ≠ ChE
Industries relevant to chemical engineering
and biology:
•
•
•
•
•
•
•
•
Agriculture
Chemical
Cosmetic
Food/beverage
Energy/Petroleum/Biofuels
Pharmaceutical
Medical devices/drug delivery
other…
What is chemical engineering?
(and why might biologists care)?
Core areas of chemical engineering:
• Reactor engineering:
Most biological processes are highly
efficient and selective
• Transport processes:
Organelles, cells, and organs are structured to
control transport rate of chemicals
• All of chemical engineering is rooted in
Material and Energy Balances
Cells that do not transport
material and energy
cannot survive.
Images from wiki
Transport processes across scales (distance and time)
Two primary modes of material transport:
• Diffusive
• molecular → cellular scale
• random motion of macromolecules discovered by
botanist Robert Brown in 1827
– explained by Albert Einstein in 1905)
• collisional diffusion
• Convective (flow)
• cellular → system scale
Péclet number (Pe): diffusion time-scale / flow time-scale
Pe << 1 – processes depend on diffusion
Pe >> 1 – processes depend on convection
Examples of diffusion:
A single blood cell in a
droplet of water will take
over 2 years to interact with
a 40x40 micron sensor by
diffusion alone.
Most molecules
within a cell diffuse
in microseconds.
Example: blood transport
Laminar to
turbulent flow
Increasing scale
Echocardiogram
of flow in heart
Developing Labs on a chip: why BioMEMS?
(MEMS – micro-electro-mechanical systems)
Why miniaturize?
“Better, faster, cheaper”
• Smaller scales optimize transport properties related to
macromolecular and cellular processes
• Faster and more accurate/reproducible analyses
Also
• Smaller sample size
• Less waste of reagents and lower fabrication costs
Developing Labs on a chip: why BioMEMS?
(MEMS – micro-electro-mechanical systems)
Two considerations:
• How do suspensions of macromolecules or cells
flow, mix, and segregate in these systems?
• How can we design microscale capture-and-release
platforms for disease detection?
• How can we use microscale processes to
„filter‟ out disease?
Blood flow in microchannels – Fahraeus-Lindqvist effect
In flow in a microchannel, the flow is fastest near the center and slowest
near the wall
water and other „normal‟ or Newtonian fluids develop a parabolic velocity
profile.
Blood has shear-dependent viscosity (Fahraeus-Lindqvist effect):
Less shear – higher viscosity
More shear – lower viscosity
„Separates‟ into cell-rich and cell-lean streams
?
Mixing at small scales
Smaller scales produce both quantitative and
qualitative differences.
No turbulence.
Invariant surfaces
Ottino, 1989
Pe >> 1, diffusion is slow.
Surface forces are significant.
Incorporation of conventional approaches to
manipulation difficult.
Only recently has design of the
channels begun to overcome limitations
in mixing by introducing chaotic
advection via breaking symmetries.
Liu et al, Mic. Mech.
Sys., 2000
Therriault et al, Nature
Mat., 2003
Geometric modulation
Time-periodic forcing
Even with chaotic advection, invariant
manifolds often exist, inhibiting mixing.
Strook and Whitesides, Science, 2001
(nm)
10-9
10-8
10-7
(mm)
10-6
10-5
10-4
(mm)
10-3 10-2 10-1 100 meters
Straight Channel (1D Flow), ~30 mm from entrance
Re ~ 0.04
Pe ~ 600
Re ~ 0.22
Pe ~ 3000
Re ~ 0.44
Pe ~ 6000
Re ~ 1.1
Pe ~ 15,000
f = 0.1
40 mm
f = 0.2
f = 0.3
No
Image
Herringbone Channel (2D Flow)
2D Concentration contours across the channels
Re = 8.2x10-7
Pe = 5,100
C. Gao, B, Xu and J. F. Gilchrist, PRE, 2009
3D Confocal Imaging – 2D channels, instability near entrance
0.02
tracking
0.04
0.06
0.08
0.1
0.12
0.14
25
20
15
10
5
0
0
10
20
30
40
50
60
0.03 0.04
0.04 0.05
0.05 0.06
0.060.070.070.080.08
0.02 0.03
0.09 0.09
0.1
70
80
0.1
0.11 0.11
0.12
25
25
20
20
15
15
10
10
5
5
0
00
0
10
10
20
20
30
40
30
50
40
60
50
70
60
80
70
0.12
Confocal Imaging – Blood flow in 2D channels (elastic particles)
tracking
Developing Labs on a chip: why BioMEMS?
(MEMS – micro-electro-mechanical systems)
Two considerations:
• How do suspensions of macromolecules or cells
flow, mix, and segregate in these systems?
• How can we design microscale capture-and-release
platforms for disease detection?
• How can we use microscale processes to
„filter‟ out disease?
Background - Fabrication of Colloidal Crystals
Method
Optical force /- Optical Tweezers
Gravitational force /- Epitaxy
Lee et al., Langmuir 20., 2004
Vossen et al., Mat. Res. Soc. Symp. Proc 705., 2002
Furst, Curr. Opin. Coll. Surf. Sci., 10., 2005
Electrical force /- Elec. field induction
Convective force /- Spin Coat
200- 8,000 rpm
Lumsdon et al., Appl. Phys. Lett. 82., 2003
Jiang et. al., Appl. Phys. Lett 89., 2006
Background - Fabrication of 2D Colloidal Crystals
Method-ConvecContinuous
Capillary Force / Convective Evaporation
Convective Assembly
Fcap
Dekov et al., Nature, 361,1993
Lazarov et al., J. Chem. Soc. Faraday Trans, 90, 1994
Fluid mechanics
neglected in
most previous studies
vw  v C
(k )
Prevo and Velev, Langmuir, 20, 2004
Dimitrov and Nagayama
Langmuir, 12, 1996
J e

0.605 kd (1   )
l
Example Applications Varying Surface Morphology
Enhancement of Light Extraction Efficiency
Cell Capture and Release Substrates
with X. Cheng (Lehigh)
with N. Tansu (Lehigh)
Ee et al., Applied Physics Letters 2007.
Kumnorkaew et al., Langmuir 2008.
Ee et al., IEEE J. Selected Topics in Quantum Electronics, 2009.
Ee et al., Optics Express, 2009.
Dye-sensitized Solar Cells
Dye-coated
Nano/Mesoporous TiO2
–
Electrolyte (-)
Transparent
conductors
+
with M. Snyder (Lehigh) & G. Willing (U. Louisville)
Membranes for HIV
Separations/Sensing
Staggered Herringbone Channel (Chaotic Flow)
cycle = 25
25 ½
26
f = 0.3
Re ~ 0.44
Pe ~ 6000
20 mm
Multiple local maxima in
concentration profile.
Center dense region
“roping” through
channel.
Final concentration
profile is steady
W
2009 HHMI - BDSI summer research
Undergraduate Students:
Colleen Curley, Abbe Lefkowitz, Kristen
Mason, and D’Andre Watson
Graduate Students:
Pisist Kumnorkaew, Bu Wang, and Alex
Weldon
Professors:
Xuanhong Cheng and James F. Gilchrist
Our Microchips

We want to build devices that reversibly
capture cells
 There are currently no effective methods for
the isolation and release of specific cells

Releasing cells from the device allows
for research on targeted infected cells
Objectives
Deposit a well-ordered
monolayer of SiO2 magnetic
beads on a glass substrate
 Functionalize magnetic beads
with antibodies and build
microfluidic channels over the
monolayer
 Capture Jurkat and THP1
cells in the microfluidic
channels
 Controllably release captured
cells by magnetic force

Convective Deposition
Process by which a meniscus of
suspension is drawn across a
substrate to deposit a thin film.
 Used to obtain a well-ordered
monolayer of polystyrene
nanoparticles and SiO2 magnetic
bead microspheres

 Helps quantify the amount of magnetic
5 µm
beads in a given area
 Gives high surface coverage which
prevents cells from binding to substrate

Each concentration of suspension
(Φnano , Φmicro) has its optimum
.
condition (angle, speed,
temperature, humidity)
1 µm
Magnetic Bead Layers

Depending on deposition and surface conditions,
we obtain monolayers of magnetic beads
 Challenges arise due to non-uniform particle size
Sub-monolayer
Monolayer
(Optimum)
Multilayer
Optimizing
deposition speed
Optimizing
nanoparticle
concentration
Results
Ideal monolayer deposited with 20% SiO2, 6%
Polystyrene, and 30 µm/s deposition speed
Magnetic Bead Liftoff:
Surface Chemistry

Binding force between beads and
substrates must be optimized
 Must be strong enough for beads to endure
fluid flow through the microchannels
 Must be weak enough so that the beads can lift
off from the surface when magnetic force is
applied
Surface Chemistry
When deposited on the untreated substrate, magnetic
beads bind too strongly
 Silane treatment alters substrate surface chemistry to
weaken binding between the magnetic beads and the
substrate

 Silanes change liquid contact angle and particle affinity for
the substrate
○ Hydroxymethyl triethoxysilane (H-treated)
○ 2-[Methoxy(polyethyleneoxy)propyl] trimethoxysilane (P-treated)
○ 2-Cyanoethyl triethoxysilane(C-treated)
Suspension Droplet Contact Angles
Untreated Substrate
<10º
H Silane
47º
P Silane
41.5º
C Silane
37º
Surface Chemistry

We measured deposition speed
necessary to obtain a monolayer
of magnetic beads on slides
treated with each silane
Silane Treatment
Optimum Speed for a Monolayer
Clean Glass
12.5 µm/s
H
12.5 µm/s
P
13.3 µm/s
C
11.7 µm/s
All other parameters held constant
Lift-off Experiments
We aim to quantify when beads irreversibly bind to the
substrate and when they can be controllably released
under flow and magnetic forces
Experimental conditions:


•
Silane treatment, polystyrene removal in a toluene bath,
and drying under nitrogen flow and in air
• Solvent replacement and DI water wash experiments to
remove unknown stock solution

Treated slides rinsed under various flow rates to
simulate microchannel conditions
Results

Toluene baths preceding DI water baths promoted
uncontrollable liftoff (in DI water baths)
•
Samples dried between toluene and DI water baths
showed no liftoff
Increased time between deposition and rinse
strengthens binding between beads and substrate
 Liftoff on P-treated substrates was difficult to tune
 Solvent-replaced suspensions deposited on
H-treated substrates left at room temperature
to equilibrate showed most controllable
release

•
Optimum silane equilibration time is 1 hour
Results
Liftoff
No Liftoff
Cell Capture and Culture Objective
Our eventual goal is to capture infected cells
using a microfluidic channel and keep them
viable for analysis
 Preliminary Model

 Used cells that simulate white blood cells
 Fabricated and flowed cells through microchannels
 Analyzed cell capture using fluorescence microscopy
Deposition vs. Plain Glass





Microchannels made on slides with monolayer depositions
Cells pumped through channels at 2-5 µl/minute for 5
minutes
Cells (~7 µm diameter) fixed and stained to fluoresce green
Devices analyzed using fluorescence microscopy
Plain glass and defect areas captured more cells than areas
of monolayer deposition
Deposition
Clean Glass
100 µm
Geometric Factor in Cell Capture

The idea: smaller beads will result in more
contact area. Therefore will make cell capture
more efficient.
5 µm
Small beads: more contact area
Big beads: less contact area
 The
test:
Number of Cells
Captured on Deposition
0.5 um silica beads
1um silica beads
47
14
Ongoing Work
Refine and optimize deposition, surface
treatment, and cell capture
 Continue parametric studies on magnetic
bead liftoff and cell capture
 Optimize cell release in microfluidic channels

 Modeling forces involved for release

Proof of Concept
 Transition to capturing rare instead of model cells
Acknowledgements
We gratefully acknowledge funding from the
Howard Hughes Medical Institute as well as
those in connection with the Biosystems
Dynamics Summer Institute
Acknowledgements
FABRICATION OF NANOPOROUS MEMBRANES FOR
BIO-NANO-PARTICLE FILTRATION
Sherwood Benavides
Colleen Curley
Jonathan Cursi
Meaghan Phipps
OBJECTIVES

Deposit monolayers of binary suspensions
to be used in membrane fabrication

Fabricate membrane 6-10 microns in
thickness and functionalize its surface
with antibodies
Purification

Concentration
Develop platform to test efficacy of
nanoporous membrane in
microfluidic device
Detection

Model filtration and sensitivity analysis
OBJECTIVES

Deposit monolayers of binary suspensions to be used in membrane
fabrication

Fabricate membrane 6-10 microns in thickness and functionalize its
surface with antibodies

Develop platform to test efficacy of nanoporous membrane in
microfluidic device

Model filtration and sensitivity analysis
MEMBRANE TEMPLATING VIA SELF-ASSEMBLY

Convective deposition was used to obtain
a well-ordered monolayer

A meniscus of suspension drawn across a
substrate deposits a thin
film

Binary suspensions of polystyrene
nanoparticles or PEG pre-polymer
were deposited with silica
microspheres
MEMBRANE TEMPLATE VIA SELF-ASSEMBLY
20% SiO2 and 8% polystyrene
Unmelted
20% SiO2 and 8% polystyrene
Melted
MEMBRANE TEMPLATE VIA SELF-ASSEMBLY
6% PEG
30% PEG
10% PEG
15% PEG
20% PEG
50% PEG
MEMBRANE FABRICATION
1/6 Vmono
MEMBRANE FABRICATION
To form the pores of the membrane, silica spheres were removed
by etching with HF and KOH
 Once etched, the membrane will be functionalized with antibodies
to selectively capture viral particles

Etched with HF
Etched with KOH
ETCHING WITH KOH
½ Vmono – 3 layers
80 degrees Celsius
7 hours
ETCHING WITH KOH
½ Vmono – 3 layers
90 degrees Celsius
17 hours
ETCHING WITH KOH
½ Vmono – 3 layers
90 degrees Celsius
24+ hours Etching
ETCHING WITH KOH
½ Vmono – 3 layers
90 degrees Celsius
24+ hours Etching
PLATFORM DEVELOPMENT

Flow liquid through the device at a
constant pressure and flow rate
will be measured at the outlet

As more viral particles are captured
on the membrane, the flow rate
will decrease

The number of particles captured in
the devices, and ultimately in
the patient’s blood, can be
calculated
MODELING: BEGINNING STAGES
Velocity Field
Pressure Field
p0 = 1.1e5 Pa; T_in = 298K
Pf = 1.0e5 Pa
MODEL SENSITIVITY AND FILTRATION ANALYSIS

SiO2 microspheres randomly arrange into
square close packed and hexagonally
close packed (HCP) structures during
convective deposition

The HCP arrangement maximizes crystal
array density

Polystyrene/PEG particles fill the
interstices between SiO2 particles
and remain after etching in order to
form a membrane
Hexagonally
close packed
unit cell
Etched HCP
unit cell
MODEL SENSITIVITY AND FILTRATION ANALYSIS

Velocity field data for fluid flow through a single HCP-based pore for
a pressure drop of 1.5 psi:
Surface Plot
Cross Section
Higher Value
Lower Value