Matthieu Lagouge's PhD - Microprobes

This is the first part of my PhD project on SU8 based MEMS microactuators for micro-objects manipulation.
This first part was done with Pr. Beomjoon Kim, from University of Tokyo - Institute of Industrial Sciences. This was the opportunity for me to learn from him his strong knowledge in microfabrication, and to take advantage of his expertize in SU8 polymer.

About Pr. Beomjoon Kim

Professor Beomjoon Kim is associate professor at University of Tokyo's Institute of Industrial Sciences and leads the Kim's lab. He focuses on micro/nanoelectromechanical systems and new ways to develop them at high speed and low cost. He has worked for French National Scientist Center (CNRS), University of Twente and has been co-director of CIRMM/CNRS (associate laboratory between CNRS and the University of Tokyo).
I'd like to express him again my acknowledgement for his friendly collaboration and for all the microfabrication know-how I've learnt from him.

Overview

The microprobes system consists in three bilayer thermal actuators orthogonally placed so that the tip at the end can get down inside a microchannel. The goal of this MEMS device is to be able to trap biocells inside a liquid flow between two microprobes, and use the middle one to peform electrical measurements.

Animation showing the microprobes system principle
Microprobes system principle

Considering a flow of biocells inside the channel, the two actuators on each side get down to trap them, then the middle one get down on the biocells for the electrical measurements.

Design and simulations

The microactuators are bilayer thermal actuators made of polysilicon and SU8 polymer. Gold lines are patterned over the SU8 as heaters. A current applied in the gold lines will heat them and the bilayer nature of the cantilever will make them get down.

The used polysilicon has a compressive residual stress, while SU8 combines both a tensile and a gradient-tensile residual stresses. This means that at neutral state, the cantilever bends up. Then, electrical current through gold heats the structure, and since SU8 has a larger thermal expansion coefficient than polysilicon, and since polysilicon is linked to the substrate so has a better heat conduction cooling, the structure bends down and the tip goes down in the microchannel.

Lateral view of the heat distribution in the structure
Schematic of the heat distribution model in the actuator for analytical calculations

For the analytical modeling, I considered that polysilicon was almost at the substrate temperature, and gold had uniform heating, so the heat gradient in inside SU8 layer.

Analytical calculations were input in Scilab (that is a vector oriented calculation software, developed by the INRIA. Results can be seen below:

Curves showing the profile of the actuator depending on the temperature
Profile curves of the cantilever along its length (analytical calculation)

These results were confirmed by a numerical simulations, using finite element modeling (fem simulation). The software used for these calculation was Ansys, that is a multiphysics finite element simulator.

Screenshots from ansys simulations showing distribution of strain and temperature in acutators
Ansys simulation results.
From top to bottom: strain at neutral state ; strain under electrical excitation ; temperature distribution during excitation.

These analytical calculations combined with numerical simulations led to the fabrication of the structure, that is working in theory.

Microfabrication

The microtechnology process starts with the making of the microchannel in bulk crystal silicon. A KOH etching allows the etch of V-shaped microchannel. The substrate is then covered with silicon nitride for passivation. Since the system is to be capable of electrical measurement, a polysilicon electrode is patterned inside the channel, under the position of the middle actuator.

microchannel etched in bulk silicon
Optical microscope pictures of the microchannel etched inside silicon substrate

Once the fixed part is over, the fabrication of the mobile actuators begin with the deposition of the oxide sacrificial layer. Then the first layer of the bilayer thermal microactuators is done in polysilicon. The second layer is made of SU8 polymer. A lift-off allows the fabrication of the heating electrodes in gold.
Finally, a second SU8 layer is patterned to cover electrodes to avoid heat loss in air, that is called convection.

complete structure with microchannel, buried electrodes, sacrificial layer, polysilicon-SU8 bilayer actuators, heating electrodes and 2nd SU8 layer
Scanning electrons microscope picture of the complete system before releasing

The last microfabrication step consists in etching the oxide so that the mobile parts will be free: this is the releasing. In this microtechnology process, there is no anchor to attach directly the actuators to the substrate. This means that an overetching release completely the actuators that will go away from the substrate. To avoid this, the «fixed» parts of the actuators are very larger in area than the mobile parts. Thus, the mobile parts are released before the fixed part, and by stopping the etching in time, only the cantilevers are free to move while fixed parts are still attached to the substrate by the resting oxide.

released structures, the cantilevers are free and bend up
Scanning electrons microscope picture of the released system

The system is complete and ready for tests!

Results

As can be seen in the releasing photo, the structures bend up at neutral state, due to the residual stress of polysilicon and SU8.
Some current input have succeeded in bending down the structure so that the tip can reach the bottom of the channel. And during the excitation, 3 schemes have been observed.

Curve relating tip deflection and input current. It shows 3 different zones in which rise of deflection has different rates.
Caracterization of the deflection of the tip at the end of the actuators depending on the input current

First zone correspond to a low input current, too much heat is lost by convection through the second SU8 layer, and the deflection is poor.
Second zone is the optimized scheme: heat get through SU8, and bottom polysilicon is still poorly heated.
Last zone corresponds to a kind of saturation, in which it becomes difficult to strain the structure while not so much more heat can be transfered.

Conclusion

This MEMS device has shown its efficiency thanks to its thermal actuators and offers potential for biocells measurements.
Pr. Kim has kept on working on this architecture at University of Tokyo, and made improved versions.
The following of my research uses the learnt know-how in SU8 and hot process to develop a new kind of structures, to get an extraction function.