Matthieu Lagouge's PhD - Micro-extractor

This is the second part of my PhD research project on SU8 based actuators for micro-objects manipulation. The first part, design, fabrication and tests of microprobes has given a microtechnology process for fabricating polysilicon and SU8 made devices. This one use the aquired know-how for the development of a system combining a dielectrophoretic conveyer and an extractor-microactuator able to grab a micro-object out of a microchannel from liquid to air.


As said, the system consists in two parts that must work together. The first one is a dielectrophoretic conveyer, that is supposed to drive micro-object in a liquid to a micro-extractor. This one is able to get inside the microchannel and extract micro-objects out of the liquid.

Biocell conveyed by dielectrophoresis and lifted-up by the SU8 actuator
Overview of the micro-extractor system in run

Design and simulations

The micro-extractor system was thought so that it had to be as autonomous as possible, and still be efficient. Microactuators have greater chance to work if their design is kept as simple as possible. This means that the actuator had to do one thing: just extract the micro-object, and not reach it.
So a conveying system was needed to bring the targets to the actuator. Amongst all possible conveying system seen in literature, dielectrophoresis was the most interesting technology: it has no mobile parts, increasing the microtechnology process reliability, and it can move objects on large distances.

Dielectrophoretic conveyer

The dielectrophoresis system had to convey micro-object inside a «V-groove» channel. To perform dielectrophoresis, an electrodes network must be patterned on the way along target trip. The idea was to use these electrode both for the dielectrophoretic conveying system and as a impedance detector to warn a electronic logic command that the microactuator can be excited. This implies that the electrodes get inside the channel following the slope of the V-shape, but are separated one side of the channel from the other.

Rectangle electrodes following the slope of a V-channel orthogonally, but not linked at the channel's bottom
Schematics of the geometric configuration of the dielectrophoresis electrodes inside the microchannel

The dielectrophoretic force calculation is not given here, since it is already detailed in the electrostatic section.

Ansys simulations have been performed to evaluate the electric field gradient, that dielectrophoresis depends on, inside the channel. These simulations are done considering the electrodes geometric configuration given above, considering also that the actuator part with the micro-objects receptacle would be metallized and at bottom state would be between the 4 last electrodes. This means that a metallic zone would be located between the two last electrodes and disturb the electric fields. The surrounding medium chosen for the simulation is water, though almost any medium can offer the same kind of results, if you correctly manage signal frequency.

Ansys pictures showing voltage potential distribution following electrodes placement
Ansys simulation result showing voltage distribution

The voltage distribution shows the electrodes placement and the disturbed zone near the last electrodes.

Channel content cut along its length and height for electric field calculations
Electric field evolution with time, top view
Electric field evolution with time, lateral view
Ansys simulation result of the electric field evolution with time.

The disturbed zone can be easily found. But you can see that the disturbing volume does not reach the middle of the channel along Oz axis. So the dielectrophoretic effect will still be efficient above the actuator receptacle.


The microactuator was designed on the basis of the microprobes actuators technology. So it is mainly SU8 based, but this time there is no polysilicon layer under SU8. The residual stresses come from the SU8 itself and the gold layer used for actuation.
Since the e-beam evaporated gold has a strong compressive residual compressive stress, the structure has a tendancy to bend down at neutral state. Then, the SU8 has a larger thermal expansion coefficient, so heating the whole structure make it bend up. This is the principle of the extraction.
Another point to care about is the micro-object's receptacle, that must be at the channel's bottom, otherwise the target could get under the receptacle, and it would no longer be possible to extract it. So, electrostatic electrodes are planned to be combined with thermal actuation so that the structure can plan to the substrate.

schema of 4 SU8 arms with thermal and electrostatic gold electrodes
Schematics of the micro-extractor actuator


The fabrication process is done on a silicon wafer, and begins with the fabrication of the microchannel and the dielectrophoresis electrodes, then a sacrificial layer and the actuator, and finish with the releasing.

Microchannel and electrodes

As for the microprobes, the crystal quality of silicon combined with KOH etching allow the fabrication of triangle section microchannel in a bulk silicon wafer.

part of a silicon wafer

Silicon wafer

triangle section microchannel etched in the silicon wafer with a dioxide mask

Etching of microchannel with KOH using a dioxide mask

silicon wafer with etched triangle section microchannel

Removal of dioxide mask with fluorhydric acid

silicon wafer with microchannel and 2 more layers, top one is polysilicon

Deposit of one passivation layer and one polysilicon layer for the electrodes, both by LPCVD

polysilicon electrodes defined inside microchannel, with nitride under electrodes

Etching of polysilicon electrodes for dielectrophoresis by R.I.E.
Two recipes have been developed for this step, one based on Cl2, and one on SF6/CHF3

Polysilicon lines getting inside a triangular section microchannel
Scanning Electron Microscope picture of dielectrophoresis electrodes inside a channel

passivation layer deposited over electrodes and opened for access to polysilicon electrodes

Deposit of a passivation layer to avoid current flow between electrodes through liquid, and opening for electrical access by R.I.E. etching.

At this stage, the dielectrophoresis system is ready for use.
The following is the fabrication of the actuator.

SU8 micro-actuator

The fabrication of the micro-actuator begins with a sacrificial layer, that will be used to release mobile parts, and the structural layers composing the actuator.

glass layer deposited over the dielectrophoretic system

Deposit of a low temperature oxide by LPCVD as a sacrificial layer

glass layer deposited over the dielectrophoretic system with openings where actuator is to be anchored to the substrate

Opening of the sacrificial layer to allow anchors for the actuator

SU8 layer patterned forming the micro-actuator

Patterning of the SU8 structural layer for the microactuator. SU8 is deposited and patterned directly by photolithography

gold layer forming excitation electrodes on SU8 layer

Gold layer patterned by lift-off to form both the thermal and electrostatic electrical lines

complete structure not released
Scanning electron microscope of the structure with gold electrodes before releasing.
You can see the thermal lines at the middle of the structure and the electrostatic electrodes on the sides very clearly.

actuator slightly above the surface of the device, glass has disappeared

Releasing of the micro-actuator by etching all of the sacrifical layer with B.H.F. (Buffered Fluorhydric Acid)

released structure, curving so that the receptacle gets beside the channel
Scanning electron microscope picture of a released structure.
Due to residual stress, the receptacle goes beside the microchannel end. The actuation cycle of the extractor has to correct it.


As for the rest of the demonstration, the tests will concern the dielectrophoresis conveyer first, and the micro-actuator next.


The dielectrophoresis effect was quite tough to obtain, as electrical relative permittivity of liquid and particles used for the tests were unknown, and at this time this was the first dielectrophoresis conveying experiment of the laboratory. (This is not any more the case!)
Nowadays, I succeeded in getting a displacement of the particles at a speed different from the liquid, proving a dielectrophoretic force added to the electro-osthmosis effect.

yellow phosphorecent particles are moving inside a circle from one picture to another
Microscope pictures of phosphorecent latex particles above dielectrophoresis electrodes. The red circle shows a particular region where you can follow evolution in the position of particles.

The dielectrophoresis system was proven to be working. Optimizations are possible in the electrodes design, so that, for example, the dielectrophoretic effect would be more visible inside the channel and less next to the channel.


The microactuator was tested both in a purely thermal actuation, to evaluate its capacity to bend up (and so extract the micro-object out of the microchannel), and in a combined thermal and electrostatic actuation to ensure its ability to put the receptacle down the microchannel so that it can grab the micro-object.

For the thermal actuation, a particular actuator has been chosen. This one has no attachment at its end, so that the residual stress completely define its neutral shape, and the effect of thermal excitation is directly measurable.

thermal microactuator bending up step by step
Animation of a thermal actuator heated step by step thanks to Joule's effect

The electrostatic actuation was done under a light thermal excitation. As you could see at the end of the microfabrication section, the released structures have a bending, and electrostatics is efficient if surfaces are as close as possible. The idea is to make them more plan, and then to used electrostatics to zip the structure to the substrate.

photos of almost plan actuators with difference of planeity due to electrostatic actuation
Microscope pictures of microactuator, both thermally excited, and in the second picture, electrostatically excited.
The red circle shows an area where the contact area with the substrate is larger with electrostatics.

As you can see, I have not succeeded in putting the thermal electrostatic actuator completely plan with the receptacle in its designed position.
Anyway, the microactuator could be corrected quite easily, as could the rest of the design slightly changed so that the receptacle would be able to reach the depth of the microchannel.


Through the complete system is not functional, the separated elements have been proven to work.
The dielectrophoresis system works and is able to displace latex microbeads, and the thermal electrostatic micro-actuator can be bent up by thermal actuation or planned by thermal-electrostatic actuation.
The SU8-2000 residual stress is undoubtly a trouble when it comes to be used as a movable part, since its residual stresses are quite complicated to manage.
At the time of this writing, the SU8-3000 serie must be available and may be more controllable.