1. Transient Electrical and Thermal Characterization of InGaAlAs Thin Films with Embedded ErAs Nanoparticles
Tela Favaloro, Rajeev Singh, James Christofferson, Younes Ezzahri, Zhixi Bian, and Ali Shakouri
Electrical Engineering Department, University of California, Santa Cruz, California 95064, USA
Gehong Zeng, Je-Hyeong Bahk, and John E. Bowers
Department of Electrical and Computer Engineering, University of California, Santa Barbara, California , USA
Hong Lu, and Arthur C. Gossard
Materials Department, University of California, Santa Barbara, California 93106, US
Today I will talk about transient thermal and electrical measurement of cross plane material parameters of thin film thermoelectric devices
I represent the University of California in Santa Cruz, but this work was done with full colaboration with the university of California in Santa Barbara as well

2. Outline
High temperature apparatus for cross plane material characterization
Transient electrical measurements
Thermoreflectance imaging of thermoelectric devices in cooling and heating modes
Preliminary results
The material I am presenting today was taken with a recently developed high temperature characterization thermostat. With this particular configuration we are able to resolve transient electrical signal present in nanostructured thermoelectric devices. This thermostat also contains an integrated optical view port for thermoreflectance imaging. Images were taken on a thermoelectric device to temperatures of 700K. These results are given in the following presentation.
The viewgraph on the left shows a thermoelectric device in heating mode, while the figure on the right shows the same device excited with the opposite polarity current incooling mode. As you can see that both joule heating in the neck is readily apparant,; but the actual thermoelectric is still cooling. These images were taken with a 500 mA excitation pulse at 550K and are overlapped with the DC image of the device.
‘Merged’ Thermal Image -500ma, 550K
‘Merged’ Thermal Image +500mA, 550K

3. Thin Film Material Characterization
Material Figure of Merit
Established characterization techniques:
In-plane electrical conductivity and Seebeck (Van der Pauw, sample bars).
Need non-conducting substrate (difficult at high T).
Substrate transfer: stress issues
Cross-plane thermal conductivity
3 (min ~0.5-1m; need electrical isolation between heater and thin film)‏
Transient thermoreflectance (top microns of the sample, frequency dependent issues)‏ 1 3 2 4
I13
V24
As you all know, the efficiency of a thermoelectric material is given by what is called the material figure of merit
This quantity is comprised of sigma, the electrical conductivity; S, the seebeck coefficient, and kapp the thermal conductivity of the material.
It has been showed that through nanostructuring and restricting device dimensions to the micrometer range, one can tune these material properties to optimize device performance for different applications and temperature ranges
During this optimization process, it is important to accurately measure each of these parameters.
Established characterisation techniques each have there pros and cons and include: the vander pauw technique for in plane electrical cond. Direct measurement of in plane Seebeck coefficeint. And the 3 w method for measurement of the cross plane thermal conductivity.
However, due to material anisotropy is it quite useful to measure these parameters along the same direction; especially for the case of a superlattice configuraition. Cross plane measurements are necessry x
Also possible: Cross-plane determination of material parameters

4. High Temperature Characterization System
Sample mounted for thermal imaging
This is an image of the high temperature thermostat developed at UC santa cruz.
It contains an integrated optical viewport which is shown in the image on the left, from which on can see the sample being probed. In this image the smaple is mounted on a copper stage for thermal imaging; but the packaging of the sample allows for measurement of fast thermoelectric transients. The diagram below it outlines this. The sample is wire bonded to the transmission lines which are then connected coplanarly to high speed probes which are then impedance matched in a high speed circuit.
Currently, the thermostat has been able to maintain vacuum of 10 to the negative 6 mbar at temperatures around 800K.
Currently tested to vacuum of 10-6 mbar and temperatures above 800K
High speed measurement stage

5. Material System: Semimetal nanoparticles in semiconductor alloy
Thin film element: 50 m n-InGaAlAs, 0.6% ErAs
Substrate: 125 m AlN
Metal pad: 7 m Au
Mask: Dedicated voltage and current pads
The material system involved in these measurements is 50 micron thick n type InGaAlAs alloy with 0.6% randomly distributed ErAs quantum dots. Embedding quantum dots into the semiconductor material has been shown to decrease the thermal conductivity of the material while providing an enhancement to the material thermopower.
This material goes through specialized processing method for cross plane characterization where the material is flip chipped bonded to remove the growth substrate and thus any parasitics that may accompany it.
A device mask has been developed with separate pad for applied current and another for voltage sensing so that they dont interfere with each other
Mask design for cross-plane measurements

6. Cross-Plane Transient Electrical Characterization
750K
700K
300K
Using this mask and high speed circuitry configuration we are able to resolve thermoelectric transient signal that is induced in a device at the electrical pulse edge. The primary viewgraph shows the amplitude of the Seebeck voltage taken at 300K to 700K with 50 degree increments. The inset shows that even at a couple hundred nanoseconds the signal is well defined so there is little error in the amplitude of the seebeck voltage.
300K
High speed circuitry enables below 100 nanosecond resolution

7. Extraction of Joule and Peltier Voltages
Through taking a bipolar measurement at multiple excitation currents, we are able to extract the Peltier and Joule components of the seebeck voltage, as Joule heating does not change sign with a bipolar measurement.
This data presented is for the 120x120um2 sample. The peltier component show a linear dependancy on the exitation current as expected, while the Joule voltage, albeit a smaller signal and thus subject to more noise interference, shows a quadratic dependance.
Through utilizing a bipolar measurement of the total pulse we are also able to determine the resistive voltage dropping across the sample and htus measure the cross plane electrical conductivity, the direct measurement of the figure of merit, and if total cooling is know the cross plane seebeck coefficient. This work will focus solely on determination of the Seebeck coefficient
2110
1370
798
70x70 μm2
1690
1120
560
120x120 μm2
700K
500K
300K
Vsp (V)‏

8. Thermoreflectance Imaging
T=300K
I = -95 mA
In order to determine the total cooling of these devices thermoreflectance images were taken which thermally map the sample surface.
The viewgraph on the left upper show a 50x50 um2 device in heating mode with an exitation current of 95 mA. The graph on the left shows the same device in cooling mode. Again, there is joule heating that occurs in the neck but the device is still cooling. The lower center image the resultant peltier cooling after the joule component has been subtracted out in a similar process as mentioned earlier for the electrical signal.
50x50μm2 device in heating mode
50x50μm2 device in cooling mode
50x50μm2 device Subtracted Image for Peltier Cooling

9. High Temperature Thermoreflectance Imaging
T=700K
I=- 95mA
This is the same device excited with the same current at 700k. The heating and cooling modes are both resolved, as well as the peltier cooling. There is some blurring in the images which may be due to changes in the coefficient of thernoreflectance that may occur at higher temperatures. Thus far the coefficient has been well calibrated for gold at room temperature but has not yet been done at these higher temperatures.
50x50μm2 device in heating mode
50x50μm2 device in cooling mode
50x50μm2 device Subtracted image for Peltier Cooling

10. Total Cooling Density Varies with Device Size
Current injection nonuniformity and current spreading within the sample result in decreased ΔT in larger samples
After the coefficient of thermoreflectance has been calibrated for at higher temperatures, there are still some discrepancies in the experiment design that must be accounted for.
These images were taken at room temperature fo r the different sample sizes-the data presented is after joule heating has been removed form the image.
As you can see the highest density of cooling and thus the largest change in temperature occurs in the smallest device size of 50x50um2. As we increase device dimensions: 100x100 and 150x150
curretn injection nonuniformity becomes more evident. Current spreading also occurs in the device, leading to decreased cooling in larger device dimensions.
50x50μm2 with 95mA excitation current
100x100μm2 with 140mA excitation current
150x150μm2 with 150mA excitation current
ΔT=-3.995
ΔT=-2.067
ΔT=-1.655

11. Results and Conclusions
Determination of Seebeck Coefficient for 70x70μm2 Sample:
Measured change in temperature after subtraction of Joule heating at 100mA excitation:
By utilizing these two transient techniques, we are able to determine the seebeck coefficient in the cross plane direction at high temperatures for the InGaAlAs alloy with embedded quantum dots.
The peltier component of the seebeck voltage is determined through transient electrical testing, while the overall cooling due to the peltier effect is quantified using transient thermoreflectance images.
Mention the viewgraph..
-6.656
-5.018
-4.215
ΔT (K)‏
2110
1370
798
Vsp (V)‏
700K
500K
300K