REVIEW OF SCIENTIFIC INSTRUMENTS 86, 053902 (2015)

Thin film cryogenic thermometers defined with optical lithography for thermomagnetic measurements on films J. Nelsona) and A. M. Goldmanb) School of Physics and Astronomy, University of Minnesota, 116 Church Street SE, Minneapolis, Minnesota 55455, USA

(Received 17 December 2014; accepted 22 April 2015; published online 6 May 2015) Resistance thermometers are common secondary thermometers in cryogenic applications. Bulk RuO2 thermometers are used in dilution refrigerators because of their low magnetoresistances in addition to their temperature sensitivity. Thermoelectric and thermomagnetic measurements require multiple thermometers to measure temperature differences. Here, we present a method to fabricate thin film RuO2 thermometers directly onto an experimental substrate. This enhances thermal contact between thermometers and films whose thermoelectric or thermomagnetic properties may be measured. Commercial thermometers have higher temperature sensitivities than the thermometers presented in this study, but commercial thermometers must be carefully heat sunk to the cryostat or sample to be useful. Thin film thermometers can be patterned with ultraviolet (UV) lithography. This allows both the size of the thermometer and its distance from the sample, when also patterned with UV lithography, to be on the order of micrometers. A universal calibration curve for these thin film thermometers has not been produced. The efficacy of these thermometers has been demonstrated through measurements of the Nernst effect in Nb. In this study, the thin film thermometers were calibrated using the cryostat thermometers. C 2015 AIP Publishing LLC. [http://dx.doi.org/10.1063/1.4919734]

I. INTRODUCTION

Thermoelectric and thermodynamic measurements are of interest because they allow access to physical properties other than those determined from simple electrical transport measurements. These measurements can be challenging because of the need for spatially separated temperature measurements and the necessity that the temperature difference across the sample be small in comparison with the absolute temperature. At cryogenic temperatures, there is the additional problem of the Kapitza boundary resistance which leads to large temperature gradients across interfaces between dissimilar materials. It is therefore difficult to heat sink a sample, and temperature gradients between the thermometer and area of interest can decrease measurement accuracy. To overcome some of these issues, we have developed a method to integrate resistance thermometers into the sample structure fabrication process allowing for very local temperature measurements near a sample. The thermometers are defined by ultraviolet (UV) lithography and fabricated by reactive DC magnetron sputtering. Figure 1 is a schematic of a device used to measure the Nernst effect in a superconducting film. The thermometers are 10 µm wide and are 10 µm from the film. The Nernst effect has been shown to be a useful probe of fluctuations and flux flow in superconductors.1,2 The response follows a characteristic tilted hill profile that has been observed in many materials where the resistance was not zero.3 This has led to interest in using thermomagnetic measurements to probe materials for superconducting fluctuations above the superconducting transition temperature. This is a)Electronic mail: [email protected] b)Electronic mail: [email protected]

0034-6748/2015/86(5)/053902/5/$30.00

especially interesting in the case of non-BCS superconductors. With the materials of interest, a dilution refrigerator is required to cool to the transition temperature. Resistance thermometers are typically employed as secondary thermometers in cryogenic measurements.4 The common commercial example for dilution refrigerator temperatures is RuO2 because of its high temperature sensitivity and low magnetoresistance (MR). It is desirable to have a thermometer that can be integrated with the sample substrate to minimize the problem associated with the Kapitza boundary resistance. Because of its lower magnetoresistance when compared to other cryogenic thermometers, RuO2 was chosen to test its value as an integrated thermometer. In addition, with the use of lithography, these thermometers could be placed within micrometers of a thin film sample. Recently, thick film Ru based thermometers have been created by curing a ruthenium oxide paste.5 Temperature sensitivity was comparable to commercial alternatives and the magnetoresistance was smaller. The thermometers could be attached to the substrate but were still very large. Several thermal cycles were required to stabilize the calibration curves. Ruthenium oxide nanopowder thermometers have also been shown to be a possible cryogenic thermometer.6 The thermometers fabricated this way were metallic and would not work below 70 K. One thermometer did show an upturn in resistance similar to that of sputtered thermometers, but was metallic and not insulating at higher temperatures. Commercial RuO2 thermometers may have a nonmonotonic temperature coefficient of resistance (TCR) at high temperatures but are most useful at temperatures below 4 K. Thin films of RuO2 can also be produced using a variety of methods ranging from deposition from solution to chemical

86, 053902-1

© 2015 AIP Publishing LLC

053902-2

J. Nelson and A. M. Goldman

Rev. Sci. Instrum. 86, 053902 (2015)

FIG. 1. Schematic of a device designed to measure the Nernst effect in thin film superconductors. All patterning for the device can be done with UV lithography. The thermometers are placed 10 µm from the sample space. The pictured device was used to measure the Nernst effect in Nb but could also be used for thermopower measurements.

and physical vapor deposition. Vapor deposition techniques are compatible with lithography. Reactive sputtering with oxygen provides an additional control over the resistivity and TCR of RuO2 films.7 Initial interest in the sputtering of RuO2 came from interest in producing resistors with zero TCR.8 From the published work, it is clear that there are a range of parameters with which resistors could be fabricated with a negative TCR. In these materials, the resistance is primarily due to grain boundary scattering.9 Additional oxygen inclusions can be incorporated into the grain boundaries.7 It is believed that the amorphous nature and additional oxygen are the source of the desired insulating behavior.

II. EXPERIMENT A. Thermometer fabrication

Thin films of RuO2 were sputtered in a DC magnetron system in a reactive environment. A recipe was adapted from a report discussing a process for making RuO2 films with a zero temperature dependent resistance.8 In the report, parameters were presented for making films with a negative TCR. Film growth was performed at a constant power of 50 W applied to a 3′′ diameter target of 99.95% ruthenium. The pressure during deposition was maintained at 10 mTorr with an O2 concentration of 20% of the total Ar/O2 environment. The substrates were not heated during deposition. Optical lithography was used to mask off the substrate which was a Si wafer with a SiO2 surface layer. Working thermometers have also been fabricated on strontium titanate substrates. The thermometers were 10 µm wide, 80 nm thick, and roughly 10 µm from the sample space. Contact to the sample space and thermometer was made with platinum leads 20 nm thick, which were predeposited onto the substrate. An integrated thin film heater was also fabricated. After the thermometer growth, the wafer was capped with 120–150 nm of SiN, leaving only the sample space and contact pads open. Tests of the design were conducted on a Nb metal film sputtered onto the wafer and patterned using a standard liftoff technique. The capping layer can protect the thermometers in case a shadow mask is required for the thin film sample deposition step. Resistance of the thermometer was measured

FIG. 2. Calibration curve of a thin film RuO2 thermometer down to dilution refrigerator temperatures. (Inset) Data from a second RuO2 thermometer with a different dimension showing the high temperature behavior. In each case, a commercial bulk RuO2 thermometer was used to produce the temperature axis.

in a four-wire linear configuration, while a van der Pauw configuration was used for the sample. A new calibration curve was obtained for each thermal cycle to room temperature. The resistance at fixed temperatures was reproducible over the course of the series of measurements below 20 K. Preliminary measurements were carried out in a Quantum Design Physical Properties Measurement System (PPMS). The secondary thermometer of the cryostat was used to produce the calibration curve for the on-wafer thermometers. An example of a calibration curve obtained in a dilution refrigerator is displayed in Fig. 2. Above 1 K, the data were collected as the system slowly cooled, but below 1 K, stable temperatures were used as the system was warmed incrementally. The resistance of the thin film thermometer becomes independent of temperature below 50 mK. The current-voltage characteristics of our thermometer sample were linear at excitation currents above those used for the measurement, so we believe this observed effect is due to poor contact to the mixing chamber or growing thermal isolation of the electron system from the lattice. The inset of Fig. 2 displays the high temperature behavior of another RuO2 film measured in a PPMS. All thermometers are insulating above 100 K but some show metallic behavior between 10 and 100 K. All samples show a strong upturn in resistance below 10 K. We use the figure of merit S=

d(logR) d(logT)

(1)

as a measure of the thermometer sensitivity. At 1 K, several samples have S = 2 − 3 × 10−2. Commercial RuO2 sensors have 0.1 < S < 1 at this temperature.4 Resistance thermometers for cryogenic applications are typically semiconductors which display some form of variable range hopping (VRH) that follows the form R(T) = R0exp[(TA/T) x ],

(2)

with TA being the activation energy and x representing the hopping exponent. Theory for VRH predicts hopping

053902-3

J. Nelson and A. M. Goldman

Rev. Sci. Instrum. 86, 053902 (2015)

FIG. 3. Magnetoresistance (MR) at 2 K. At magnetic fields below 30 mT, a slight negative magnetoresistance is observed. The higher field MR is positive and linear.

exponents of 1/4 or 1/3 corresponding to 3D or 2D Mott VRH, or 1/2 for Efros-Shklovskii VRH.10 Experimentally hopping exponents are seen in the range of 1/4–1/2 in addition to multiple parameter functions; however, the resistance change for the thin film thermometers was too small to verify the conduction mechanism. The MR for a thin film RuO2 thermometer can be seen in Fig. 3, where MR =

R(B) − 1. R(0)

(3)

Some samples show an initial negative MR below 30 mT. At higher fields, the MR appears to be linear and this behavior was observed in both cryostats. A wide range can be found for MR sensitivity in commercial RuO2 sensors. The thin film samples have a MR similar to the high end of the commercial alternatives.

signal. The measurement was made in a PPMS with a base temperature of 2 K. Samples were cooled by a combination of thermal contact to the sample holder and residual He exchange gas in the sample space. The PPMS was outfitted with a cryopump that was used to remove the exchange gas and reduce the cooling power of the system. This was done to better control the direction of the temperature gradient along with thermally isolating the wafer from the sample stage. The chip was thermally anchored at the far end of the heater. The thermal link to the sample was achieved with thermally conductive silver paint. Without the use of the silver paint, a reduction of the critical field was observed when the exchange gas was removed. The observed behavior was attributed to heating of the sample due to an insufficient thermal link to the sample stage. Figure 4 shows a measurement of the Nernst voltage vs. magnetic field of Nb. The voltmeter was attached such that the transverse voltage would be positive. The temperature differences were typically

Thin film cryogenic thermometers defined with optical lithography for thermomagnetic measurements on films.

Resistance thermometers are common secondary thermometers in cryogenic applications. Bulk RuO2 thermometers are used in dilution refrigerators because...
1MB Sizes 0 Downloads 6 Views