University of Minnesota

EE 5141

Introduction to Microsystem Technology

2015 Spring Project Report

Fabrication Process of Microheater

Yi Ren

Project Goal

Try to make the world’s highest temperature microheater.

Component materials

1. Conducting material: We choose platinum(Pt) as the conducting metal. Because it has high

melting point (1,768.3 °C) and electrical resistivity (105 nΩ·m (at 20 °C)). Besides that,

platinum also has good chemical stability.

2. Dielectric material: We choose Al2O3 as the dielectric material. Al2O3 is a hard, durable and

chemically robust material. It has higher melting point than platinum (2,072 °C) and low thermal

conductivity (30 W·m-1·K-1) which means it can reduce the thermal dissipation of the

microheater and redue the power dissipation. Al2O3 can also be deposited by using ALD process,

which can well control the deposition thickness and have a smooth surface.

3. Electrode material: We choose gold (Au) as the electrode because of its good electrical

conductivity.

4. Wafer: We choose P-type (100) silicon wafer as our substrate. The reason is that its the most

commonly used wafer in the micro-fabrication. Since our microheater do not have specific

requirement of wafer and (100) is easy to be etched and cut off, we choose this wafer as our

substrate.

Fabrication Process

1. Al2O3 ALD process: First, we deposit 100nm Al2O3 thin film surround our wafer. This layer

will become the substrate of our microheater. The ALD process alternating pump

trimethylaluminum (TMA) and water into the chamber and pump out methane and unreactive

water vapor. After TMA pulse the TMA bonds to the surface hydroxyl groups, the remaining

TMA will not bind to the methyl groups on the surface therefore, the process is self-limiting

shown in figure 1 [1].

Fig.1 Self-limiting in Al2O3 ALD process [1]

Fig.2 Schematic after depositing Al2O3 surround the wafer

2. Lift-off process: In this step, we need to deposit platinum pattern on our substrate. We will

use photolithography to form the pattern on the substrate then use sputtering to deposit platinum

and finally release extra platinum that outside the pattern.

Before we spin LOR-3A on substrate, we need to clean the wafer by using Acetone, methanol,

isopropyl alcohol and DI water rinse. Then pre-bake the wafer for 60 sec in 150 °C. After we

spin LOR-3A on substrate, we need soft-bake LOR-3A for 2mins in 170 °C. Then we can spin

the photoresist 1805 on the LOR (Fig.3). The reason that we need 1805 is that LOR is not

photosensitive material.

Fig.3 Schematic after spinning LOR-3A and 1805 on the substrate

Then we put the wafer into UV light with mask, shown in figure 4.

Fig.4 Schematic of photolithography

After exposure to UV, the resolution of the 1805 will change. Then we can use developer S351

to resolve 1805 and CD-26 to resolve LOR to form pattern, shown in figure 5.

Fig.5 Schematic after developing and CD-26 etch

Then, sputtering 5nm Ti and 50nm Pt on the substrate. We should first sputtering Ti because Pt

can not bond stable on the Al2O3, we need Ti to bond them together. The reason that we use

sputtering is that it has better coverage and lower process temperature. If the process temperature

is too high, it might damage the LOR and 1805 and break the pattern. The schematic after

sputtering shown in figure 6 and the optical image shown in figure 7.

Fig.6 Schematic after sputtering

Fig.7 Optical microscope image before Pt lift-off

Finally, we use 1165 for 30sec to remove all the photoresist and release the extra Pt outside the

pattern, shown in figure 8 and 9.

Fig. 8 Schematic after Pt lift-off

Fig. 9 Optical microscope image after Pt lift-off

By using the same process, we can deposit the gold on the electrode, shown in figure10 and 11.

But this time, we use 1818 as photoresist. We deposit 10nm Ti and 200nm Au. Ti is also the

bond material between Au and Al2O3.

Fig. 10 Schematic after gold depositing on the electrode

Fig. 11 Optical microscope image after gold depositing on the electrode

3. Plasma Etch: Before we etch release silicon, we need to layout the release area and expose

the silicon from surrounding Al2O3. First, we need to encapsulate our microheater by 20nm

Al2O3 film. It can also protect the metal during the plasma etch. We also use ALD to deposit that

Al2O3 film. The schematic shown in figure 12.

Fig.12 Schematic after encapsulating

Then put the release area pattern by using photolithography. The process is the same in lift-off

process. After that, we use Bcl3 and Ar to do the plasma dry etch to etch Al2O3.We will over etch

a little bit to ensure we etch all the Al2O3, shown in figure 13 and 14.

Fig. 13 Schematic after plasma etch

Fig. 14 Optical microscope image after plasma etch (The orange area is silicon)

4. Etch release process: There are three method to remove the silicon from the bottom. KOH

etch, XeF2 etch and bosch etch. We first tried the KOH etch. The principle is that KOH has much

higher etch rate for (100) silicon than (110) and (111) silicon. But the problem is KOH also etch

Al2O3. We found that in order to fully release the microheater, the Al2O3 encapsulation also be

etched, which is not what we want. In order to fix this problem,we should use SiNx as the

encapsulation or just use the other to method to etch the silicon.

Then, we tired XeF2 etch. The advantage of XeF2 is it can high select silicon to etch, almost no

affect on other materials. The schematic and SEM image after XeF2 etch shown in figure 15, 16.

Fig. 15 Schematic after XeF2 etch

Fig. 16 SEM image after XeF2 etch (not fully release)

The disadvantage is also easy to see in figure 16. It can not fully release the microheater, the

center of the microheater is still connecting with silicon, which will influence the performance of

our device.

Finally, we tried bosch etch. The main difference of bosch etch is that it etch from backside of

the wafer. The process shown in figure 17. First we make patterns at the backside of the wafer.

Then we use plasma SF6 and Ar to etch silicon for a while. After that, we use C4F8 to form a

layer to protect the edge. By repeating this process, we will finally etch all the silicon from

backside to top layer, shown in figure 18 and 19.

Fig. 17 Process of bosch etch [2]

Fig.18 Schematic after bosch etch

Fig. 19 SEM image after bosch etch (fully release)

Finally, we got 200nm Au electrode, 20nm Al2O3 encapsulation on the top, 100nm Al2O3

substrate and 50nm Pt heater wire. These numbers were measured by using Filmetrics ,

Ellipsometer and Atomic Force Microscope.

Mechanics and Characterization

1. Cantilever deflection in alumina structure

EIXLF )(1

(1)

zE

(2)

)3

1(

)(22

)(LxLx

EIxWxF

(3)

By combining equation (1), (2) and (3), we can get,

)3

1(

2)()()(2

LxLx

ExWxLzx

And the maximum stress occur at z=H/2

Hence,)

31(

)()()(2

max

LxLx

xWxLEHx

According to the data we got from the lab (shown below), L=0.24mm. H=120nm E=160GPa

Fig.20 Deflection of alumina cantilever beam

I choose 5 points,

X(mm) W(x) (μm) σ Stress(MPa)0 3 -

0.04 4 10.60.085 7 7.780.16 16 4.180.228 27 0.649

The figure shown below:

Fig.21 Stress Gradient in alumina cantilever

2. Cantilever deflection in alumina/platinum structure

Assume 50nm pt is on the 120nm alumina. According to stoney equation,

)1(6

2

AlPt

AlAlPt d

dE

AndAlAl

Al

Ed

21

We get )1(3 2

2

AlPt

AlAlPt d

d

σAl =300MPa, dAl= 120nm, dPt = 50nm and Poisson’s ratio of alumina is 0.21.

Hence, the stress in platinum is 729MPa

3. The curvature of the microheater should be the same as the alumina/platinum cantilever beam

which is around 0.91mm. The curvature of the microheater will cause the temperature is not

uniform distribute in the platinum beam, which leads to thermoelastic dissipation and reduces the

highest temperature that the microheater can reach.

4. The resistance versus temperature plots of two microheaters shown below,

Fig. 22 Resistance versus temperature figure of Microheater 1

Fig. 23 Resistance versus temperature figure of Microheater 2

According to the figure 22 and 23 and equation

)1(0 TRR

the TCR of microheater 1 is 0.00102 K-1, mircoheater 2 is 0.00108 K-1.

5. The 1/R versus I2 plots of two microheaters shown below,

FIg. 23 1/R versus I2plot of Microheater 1

FIg. 24 1/R versus I2plot of Microheater 2

The slope of these plot is α/G, so the effective thermal conductance of microheater 1 is

51025.6 W/K, microheater 2 is 51061.6 W/K. Hence, microheater 2 is unreleased and

microheater 1 is released.

6. The microheater we tested is bosch released and the maximun temperature we got is 1046 oC.

References

[1]. Dawson, Noel Mayur (2010, May 30th). “ATOMIC LAYER DEPOSITION OF

ALUMINUM OXIDE”. [Online]. Available:

http://dave.ucsc.edu/physics195/thesis_2010/noels_thesis.pdf

[2]. University of Pittsburgh. [Online]. Available:

http://www.pitt.edu/~qiw4/Academic/ME2080/lecture14.pdf