LOW-VOLTAGE SCRATCH-DRIVE MICRO-SCALPELS

CONTROLLED BY A BINARY-ENCODED SIGNAL

Jung H. Cho and Mark G. Arnold

Lehigh University, Bethlehem, Pennsylvania, U.S.A.

Keywords: MEMS, SDA, Micro-robot, Micro-scalpel.

Abstract: A novel approach to applying multi-stylus MEMS SDA Scratch-Drive-Actuator (SDA) micro-robots to aid

in the diagnosis and treatment of dermatological conditions is presented. The operation of MEMS SDA has

been well demonstrated by the research of (Donald et al., 2008) (Donald et al., 2006)(Donald et al., 2003).

We assume that such SDAs may be applied to the skin and powered by a bandage-like substrate. A method

of controlling the turning operation of MEMS SDA Scratch-Drive-Actuator (SDA) micro-robots has been

developed previously by our research: adding an additional stylus arm to control left and right rotation as

well as using both arms to halt. In order to control multiple micro-robots without the complication of

different stress curling that requires high voltages incompatible with dermatological applications, an

alternative solution of controlling electrical connection between the parallel-plate body and the stylus arms

is presented that uses a binary-encoded signal. Also an additional beam added to the body of SDA to be

used as micro-scalpel can be controlled by this same signal

1 INTRODUCTION

Micro-Electro-Mechanical Systems (MEMS) are

now commonly used in mirrors, optical gratings,

variable capacitors, and accelerometers. Micro-

robots fabricated from MEMS need four basic

components: power supply, sensors, control and

motion transducers. One of the most widely

researched MEMS transducer is the Scratch Drive

Actuator (SDA), which is an L-shaped beam of

poly-silicon that moves across a powered substrate

as a voltage is applied and released. The electrostatic

attraction between the substrate and the poly-silicon

deforms the “L” shape, making the SDA act like a

spring, which is released when the external voltage

is removed. An un-tethered micro-robot has been

developed (Donald et al., 2003) utilizing such

electrostatic actuation. Although this design has only

been demonstrated in isolation from any biological

material on top of the artificial environment of a

powered substrate, this position paper argues that for

some dermatological procedures, SDA-based micro-

robots, which are placed between a flexible

(bandage-like) powered substrate and the skin, may

be a useful diagnostic aid and may also enhance

surgical precision.

The MEMS micro-robot built by (Donald et al.,

2006) has a dimension of 60μm by 250μm by 10μm.

Figure 1 shows the structure of this device proposed

in (Donald et al., 2006), which propels itself forward

(moving the brushing towards the viewer in Figure

1) along the powered surface by bending and

releasing its large rectangular scratch-drive plate

(Donald et al., 2006). A stylus steering arm provides

single-direction turning capability (counterclockwise

as shown in Figure 1) by holding the robot stationary

at the dimple while the release of the main plate

provides torque. The approach taken by (Donald et

al., 2008) encodes the four micro-robot states (stylus

up plate up, stylus up plate down, stylus down plate

up, stylus down plate down) with four different

voltage levels. Donald et al. have generalized the

multi-voltage-level encoding (Donald et al.,

2008)(Donald et al., 2006) to control multiple robots

from a single external signal by using hysteresis

built into the design of each SDA (different chip

dimensions corresponding to different activation

voltages). Having multiple robots operating

simultaneously would give more information and

greater control to the surgeon, but with the (Donald

et al., 2008) multi-voltage encoding, the full voltage

swing may be hundreds of volts, a level which may

not be safe or comfortable for the patient.

219

Cho J. and Arnold M. (2010).

LOW-VOLTAGE SCRATCH-DRIVE MICRO-SCALPELS CONTROLLED BY A BINARY-ENCODED SIGNAL.

In Proceedings of the Third International Conference on Biomedical Electronics and Devices, pages 219-223

DOI: 10.5220/0002760202190223

 SciTePress

The approach we propose is different. Unlike

the multiple-robot system of (Donald et al., 2008) in

which each robot obtains its motion command via a

unique external voltage, we assume that the robots

have on-board digital logic capable of receiving a

binary-encoded command over a series of clock

cycles, and that each possible motion for each robot

has a unique binary code. In our proposal, each robot

will have identical SDA dimensions, and use digital

logic to make the behavior of each robot unique.

With the proper choice of geometry, scratch-drive

activation can happen around ten volts. The

additional voltage swing needed to transmit our

binary code is on the order of a volt, so that the total

voltage of the signal applied to the patient skin will

be imperceptible. Since there can be a measurable

difference in conductance between normal and

tumorous tissue (Smith et al., 1986), having a large

swarm of robots performing such measurements in

real time will help the surgeon minimize the amount

of tissue removed.

Figure 1: Illustration of MEMS Micro-robot (Donald et al.

2006).

Figure 2: Illustration of SDA on power grid (Donald et

al., 2006).

2 DUAL-STYLUS SDA WITH

MICRO-SCALPEL

Figure 2 shows the schematic of the capacitively-

coupled power grid used by (Donald et al., 2006).

By using these electrodes, the SDA operates by

attracting its body to the electrodes when external

high voltage is applied, and jumping like a spring

when the voltage is removed. Therefore, in order to

be propelled, a clock-like voltage waveform has to

be applied (Donald et al., 2006). In our novel

approach, the voltage on the scratch-drive plate can

also be used to supply power to on-board digital

logic as shown in Figure 3. We have demonstrated

this in our previous research by developing a

Verilog-A model of the SDA and applying voltage

regulation to provide adequate voltage swing for 1V

40nm CMOS standard-cells from the plate voltage

as a power (Cho and Arnold, 2009).

Figure 3: Equivalent circuit showing V

platehi

and V

platelo

In order to change direction, Donald’s SDA (Donald

et al., 2006) uses a stylus steering arm and requires

higher voltage to achieve pull-in or snap-down

voltage (Donald et al., 2006) necessary to make

contact with the substrate. This requirement

introduces multi-voltage level encoded power

waveform, which is used by Donald’s SDA. In

order for many SDAs to interact as shown in

Donald’s (Donald et al., 2008), all of them need to

have different stresses applied during fabrication so

that the stylus arms can curl differently in order to

vary the pull-in voltage (Saha et al., 2006). Our

novel approach is to apply a switch or a large

transistor to control the conductivity between the

stylus arm and the parallel-plate body. This

eliminates the step needed for different stress

curling, and one voltage waveform can be used to

supply power to different SDA micro-robots. We

(a)

(b)

(c)

Figure 4: (a) Top view of proposed SDA modification

from (Donald et al., 2006) in order to perform both left

and right turn. (b) Frontal view of proposed SDA

modification from (Donald et al., 2006) in order to control

pull-in/snap-down voltage of the stylus steering arms. (c)

Top view of proposed SDA modification from (Donald et

al., 2006) in order to include micro-scalpel.

BIODEVICES 2010 - International Conference on Biomedical Electronics and Devices

220

also added a second arm, as seen in Figure 4a and

4b, to be able to turn left and right, and use both

arms to enforce a stationary position. Figure 4c

illustrates the addition of a micro-scalpel beam

which can be used to probe or cut samples as

necessary.

The modified SDA consists of 3 components

which are left and right stylus beams and parallel-

plate capacitor body. Figure 5 shows the high-level

illustration of this structure. The capacitance across

the parallel-plate is the most dominant energy

storage part of this circuit. Since the V1 and V2

electrodes are uniformly covering the whole area of

the parallel-plate, the voltage across the plate can be

summarized as (Donald et al., 2003),

2211

CVCV

plate

= (1)

When both arms are connected to the parallel plate

body, the voltage across the beams (V

left-arm

and

right-arm

in Figure 6), would be the same compared

to V

plate

Figure 5: Deformation in cantilever and parallel plate due

to applied voltage.

When the beams are electrically disconnected, then

each stylus beam becomes a well-known MEMS

cantilever beam (Saha et al., 2006)(Wei et al., 2002).

We applied the modeling technique and equations

from Wei’s research (Wei et al., 2002) to build a

Verilog-A model to work with our previously

developed model of SDA. Figure 6 shows a different

voltage across the beams and the plate when arms

are disconnected. Under ideal fabrication process,

left-arm

and V

right-arm

would be identical.

Figure 6: Illustration of different voltage across the beams

and the plate capacitor due to disconnection from the

parallel plate body.

left-arm

, V

right-arm

and V

plate

represent the pull-in

voltage needed to cause attraction. This equation is

defined as (Saha et al., 2006),

= (2)

Here ‘k’ is the spring constant, ‘d

’ is the initial gap

height and ‘A’ is the area coverage of the cantilever.

Previous research (Donald et al., 2008) influenced

based on careful selection of these parameters,

‘k’ and ‘d

’. We will show that these parameters can

remain constant, and by simply connecting and

disconnecting the cantilever arms from the parallel

plate body electrically we can control the V

. And

we will use this control in section V to present a

parallax algorithm which can be used to guide the

micro-robots with an on-board algorithm that does

not need external control.

3 INTERMITTENT POWER

AND MAGNETIC TUNNEL

JUNCTION NON-VOLATILE

FLIP-FLOP

As described in the previous section, a MEMS SDA

micro-robot is driven by external voltage using the

electrodes underneath to create an electrostatic field

and to cause actuation which is transferred into

forward or turning motions. This external voltage is

applied in clocked fashion around 1 KHz, which we

will refer to as a major cycle. Since the voltage is

applied intermittently, there needs to be a solution to

hold important states needed for continuous

operation of the on-board logic. This leads to

applying non-volatile flip-flops developed using

Magnetic- Tunnel-Junction (MTJ) technology (Zhao

and Belhaire, 2007). This flip-flop works like a

standard flip-flop but information is stored in MTJs;

therefore, when the SDA's major power cycle

occurs, the MTJ flip-flop restores to its previously

saved state. In order to demonstrate this we

simulated the MTJ flip-flop (Zhao and Belhaire,

2007) and developed Verilog-A model of a dynamic

storage behavior of MTJ. We then simulate this MTJ

flip-flop in Cadence AMS environment to co-

simulate both transistors and Verilog-A/Verilog-

AMS models.

4 DUAL-STYLUS SDA

SIMULATION

In order to demonstrate the operation of the dual-

LOW-VOLTAGE SCRATCH-DRIVE MICRO-SCALPELS CONTROLLED BY A BINARY-ENCODED SIGNAL

221

stylus SDA, we have developed a Verilog-A model

to capture the voltage across the beams and the

parallel plate. First, we verified that V

plate

can still be

used to supply power to the on-board CMOS digital

logic when two arms are connected and

disconnected. We have chosen 4 bit counter to

demonstrate the operation along with storage

behavior of the MTJ flip-flop from section III. A 4-

bit counter was synthesized from Verilog RTL into

40nm standard cells using positive edge flip-flops

and they were replaced with a MTJ non-volatile flip-

flop Verilog-A/AMS model. We then integrated all

the models to simulate in the Cadence AMS

environment. The setup is shown in Figure 7. This

type of system-level simulation utilizing Verilog-

A/Verilog-AMS models has been accepted in

research (Mateu and Moll, 2007).

Figure 7: Top level simulation setup.

Figure 8 shows the simulation result of the counter.

The first waveform shows the intermittently-applied

and voltage-regulated output to drive the ring

oscillator and the counter. The next 4 waveforms are

output of the counter x[3:0]. As the power was

removed, the counter state was saved in the MTJ

flip-flop as 0011. When the power returned, it

restored the state 0011 and continued counting. This

result provided assurance that more elaborate state

machines can be pursued with this architecture.

Figure 8: Simulation result of 4-bit counter.

Then, we used the counter output bit x[3] to control

the left stylus and inversion of x[3] to control the

right stylus. Figure 9 shows the voltage across V

left-

arm

and V

right-arm

changing as x[3] toggles. As

expected, V

plate

remained constant during x[3]

change since it has the most capacitance to hold the

charge.

Figure 9: V

left-arm

, V

right-arm

and V

plate

result.

5 BINARY-ENCODED CONTROL

Section II through IV described support circuits

developed in order to control one dual-stylus micro-

robot. However, in order to control multiple micro-

robots there needs to be a global communication

channel necessary to operate them. We propose that

using the V1/V2 power grid described in section II

as a communication channel would be a solution to

this problem. The original V1 or V2 functional goal

has not changed and will continuously provide

propulsion and supply power to CMOS digital logic

on board the micro-robot. The architecture we

propose is while V1 or V2 is in a high state and

supplying power to on-board logic we can apply

higher frequency serial data onto V1 or V2 to

convey control information to each of the micro-

robots on the power grid. This idea is in line with

technology used to transfer data through power-lines

using frequency division multiplexing (Hensen,

1998). Figure 10 illustrates an example of using an

eight-bit binary signal to control the motion of four

robots (from a swarm that could contain up to thirty-

two robots). In this example, none of the micro-

scalpels are engaged (indicated by the most

significant bit of the 8-bit code); each robot number

(indicated by the middle 5-bit value) performs a

different move (indicated by the least two significant

bits).

BIODEVICES 2010 - International Conference on Biomedical Electronics and Devices

222

Figure 10: V1 / V2 are used as channels to apply binary

encoded controls to each micro-robot.

Each micro-robot is assigned a unique 5-bit number

and as part of the on-board logic there is an 8-bit

UART which recovers the serial communication and

programs each micro-robot for the operation it needs

to perform. When a particular micro-robot receives a

5-bit number that does not match its assigned robot

number, the robot in question ignores the three

command bits. If, on the other hand, the 5-bit

numbers match, the robot in question will latch the

three command bits into MTJ flip-flops so that at the

next major cycle the robot in question will perform

the command specified. The UART is clocked every

minor cycle, which operates at a much higher

frequency than the major cycle. Since transmitting

each 8-bit code requires a start and stop bit, the

bandwidth required on the global communications

channel for a swarm of 32 robots is at least

10*32=320 times the major cycle, and the UART

operates at some multiple of this. For example, if the

major cycle is 1 KHz and the UART requires 8

minor cycles per bit received, the minor cycle needs

to be about 2.5 MHz because the channel needs to

transmit at least 320,000 bits/second.

6 CONCLUSIONS

We have proposed a novel approach of applying

MEMS SDA micro-robots to assist in

dermatological procedures on the assumption power

may be applied via a bandage-like substrate. We

discussed features needed on a MEMS micro-robot

to achieve this. It needs improvement from Donald

et al. (Donald et al., 2006) to provide uniform

control for turning MEMS SDAs using a much

lower voltage signal that used by Donald et al. By

adding an additional stylus arm the robot can now

turn both left and right as well as use both arms to

stop. A third stylus arm provides a micro-scalpel.

Fabricating a transistor connection between each

stylus and the parallel-plate body allows the micro-

robot to control the pull-in voltage. Using this

control capability, we also presented a new approach

to using the power grid to communicate to each

micro-robot using a binary-encoded signal which

operates at much lower voltages than previous multi-

robot SDA systems.

REFERENCES

Bruce R. Donald, Christopher G. Levey, and Igor

Paprotny, 2008.

Planar Microassembly by Parallel Actuation of MEMS

Microrobots. In IEEE – Journal of Microelectro-

mechanical Systems, vol. 17, no. 4.

Bruce R. Donald, Christopher G. Levey, Craig D.

McGray, Igor

Paprotny, and Daniela Rus, 2006. An Untethered,

Electrostatic, Globally Controllable MEMS Micro-

Robot. In IEEE – Journal of Microelectro-mechanical

Systems, vol. 15, no. 1.

Bruce R. Donald, Christopher G. Levey, Craig D.

McGray, and

Daniela Rus, 2003. Power Delivery and Locomotion of

Untethered Micro-actuators. In IEEE – Journal of

Microelectromechanical Systems, vol. 12, no. 6.

W. Zhao, E. Belhaire, C. Chappert, 2007. Spin-MTJ based

Non-Volatile Flip-Flop. In International Conference on

Nanotechnology.

Loreto Mateu and Francesc Moll, 2007. System-level

Simulation

of a Self-powered Sensor with Piezoelectric Energy

Harvesting. In International Conference on Sensor

Technologies and Applications.

Shimul Chandra Saha, Ulrik Hanke, Geir Uri Jensen, and

Trond Saether, 2006. Modeling of Spring Constant and

Pull-down Voltage of Non uniform RF MEMS

Cantilever. In Behavioral Modeling and Simulation

Workshop, Proceedings of the 2006 IEEE

International, pp. 56-60.

Leow Cheah Wei, Abu Baker Mohammad and

Norazan Mohd. Kassim, 2002. Analytical Modeling For

Determination Of Pull-In Voltage For An Electrostatic

Actuated MEMS Cantilever Beam. In Proceedings of

ICSE2002.

Smith, Susan Rae, Foster, Kenneth R., Wolf, Gerald L.,

1986.

Dielectric Properties of VX-2 Carcinoma Versus Normal

Liver Tissue. In IEEE Transactions on Biomedical

Engineering, BME-33, 5, 522-524.

C. Hensen, 1998. Data transmission applications via low

voltage

power lines using OFDM technique. In 1998 IEEE 5th

International Symposium on Spread Spectrum

Techniques and Applications, 1, 210 – 214.

Jung H. Cho and Mark G. Arnold, 2009. Powering

Embedded

CMOS Logic on MEMS-based Micro-Robots. In IEEE

International Behavioral Modeling and Simulation

Conference, 73-77.

LOW-VOLTAGE SCRATCH-DRIVE MICRO-SCALPELS CONTROLLED BY A BINARY-ENCODED SIGNAL

223