with Sub-Microsecond Response Times
Principle Scientist: Megan Bourg
M.E. Bourg, W.E. Van der Veer, A.G. Guell, and R.M. Penner*, Electrodeposited Submicron Thermocouples With Microsecond Response Times, Nano Letters 7 (2007) 3208.[PDF]. See also Science, Editor's Choice, Oct. 5, 2007.
In order to understand processes such as turbulent combustion and detonation, and to achieve real-time monitoring of these processes within internal combustion engines, for example, temperature measurements with a time resolution in the sub-microsecond range are required. The smallest commercially available thermocouples have time-responses in the millisecond range. Faster response times (Table 1) have been achieved by forming junctions using evaporated metal films. Measurements on such thin-film thermocouples (TFTCs) has exposed a serious problem: The response times of thin-film thermocouples (TFTCs) are directly proportional to the total thickness of the metal films and a response time of 1.0 microsec or less requires films that are less than 100 nm in thickness. For such TFTCs, the Seebeck coefficient, which is the slope of the output voltage versus temperature response function, is depressed relative to its value for macroscopic junctions of the same two metals. The resulting loss of sensitivity for the TFTC can be as high as 85%. The Seebeck coefficient is depressed because the thermopower of metal films decreases in proportion to 1/t as the thickness, t, is reduced. Unavoidably for TFTCs, temperature sensitivity must be traded off for improved response time.
Thermocouples based upon submicron diameter wires (SMTCs) contend with the same physics as TFTCs, but the thermal mass of a cylindrical wire is proportional to (radius)2, with the result that response times decrease faster with decreasing size than is the case for TFTCs. Does this mean that SMTCs can achieve microsecond response times without sacrificing sensitivity? The answer to this question is not known: In spite of the technological maturity of TCs and RTDs (resistance temperature detectors) versions of these devices based upon submicron wires have not been evaluated to our knowledge. We take a first step in this direction in this paper, which has three objectives: 1) To describe an unconventional approach for fabricating silver-nickel thermocouples that are smaller than 1.0 micron in diameter, 2) to measure the temporal properties of SMTCs for measuring temperature, and, 3) to measure the Seebeck coefficients of the SMTCs.
Figure 1. Six step process flow for fabrication of Ag/Ni submicron thermocouple ensembles by electrochemical step edge decoration.
Our procedure for preparing submicron scale silver-nickel thermocouples (TCs) using Electrochemical Step Edge Decoration on graphite surfaces is shown above. These procedure produced end-butted junctions between a nickel and silver wire with precise alignment of the two wire segments because both nucleate on the same linear step edge defect feature. Each fabrication operation produced ensembles of 2 - 20 TCs with diameters in the 1.0 micron to 500 nm range.
Figure 2. (a) Low magnification scanning electron microscope (SEM) image of an ensemble of Ag/Ni submicron junctions on HOPG. The wires grow on the step edges of the HOPG and therefore form parallel arrays. At the interface of the Ag/Ni junctions the Ag particles are large, 500 nm - 1 micron, but the diameter of these wires tapers with increasing distance from this compositional interface. (b) High magnification SEM image of a single Ag/Ni junction. (c) A high magnification SEM image of a single etched Ag/Ni junction. (d) Energy dispersive X-ray microanalysis spectra for both the Ag and Ni wires. The absence of Ni peaks in the spectrum for the Ag wire (blue) confirms that the n-alkanethiol SAM effectively insulates the Ag segment from contact with the nickel plating solution. The carbon (C) peak seen in both the Ag (blue) and Ni (green) spectra is from the HOPG substrate.
The interface between silver (top) and nickel (bottom) regions is shown at low magnification in the SEM image of Fig 2. The vertical lines seen in this image are step edges that have been decorated with continuous silver or nickel wires. Upon closer inspection of this SEM image, it is apparent that wires remain continuous as they cross the silver/nickel compositional interface. Junctions between silver and nickel wire segments are shown at high magnification in the SEMs of Fig 2c. It is important to note that the n-alkane thiol SAM passivation of the silver wires in step 4 completely suppressed nickel electrodeposition on these wires in step 5. We base this conclusion on energy dispersive x-ray microanalysis spectra acquired on either side of these junctions before transfer off the HOPG. A typical pair of these spectra, Fig 2b, shows clean silver and nickel surfaces immediately adjacent to one another for the junction shown in Fig 2b.
These "submicron TCs" or SMTCs produced linear voltage versus temperature output over the range from 20 - 100 oC characterized by a Seebeck coefficient of 20 microV/oC, equal to the 21 microV/oC that is theoretically expected for a junction between these two metals. The time response of SMTCs was evaluated using two different laser heating methods and compared with the smallest mechanically robust commercially-available type J TCs. Electrochemical etching of the silver wire introduced constrictions at grain boundaries that reduced the thermal mass of the junction without altering its integrity or its overall diameter, producing a decrease of the measured rise time for SMTCs by 5-20%.
Figure 3.(a) Schematic diagram of the response time measurement system. (b) Vout versus time traces obtained at a chopping frequency of 20Hz for a Ag/Ni SMTC and a 75 micron type J TC. (c) Vout versus time traces for the same TCs shown in (b), but at a chopping frequency of 1800Hz. (d) Plot of Vout normalized to output voltage at 20Hz, Vout,20Hz, versus the chopping frequency for 75 micron and 125 micron type J TCs, Ag/Ni SMTCs, and etched Ag/Ni SMTCs. (e) Log-normal plot of the same data shown in (c). The response time was defined as the time (1/f) when the signal was attenuated to Vout/Vout,20Hz = 0.05..
Collectively, these data demonstrate that SMTCs can alleviate an important short-coming of TFTCs - specifically, the loss of temperature measurement sensitivity with decreasing device time response. These results provide motivation to produce thermocouples using wires in the nanometer range, and to probe the wire diameter-dependence of the TC sensitivity for these devices. Achieving this objective will require a nanofabrication methodology having a higher degree of precision...like LPNE!
