Open Access
Add to BibSonomyAbstract
We present characterization results of microscopic platinum wires as bolometers. The wire lengths range from 16 μm down to 300 nm. Thus they are in many cases significantly smaller in size than the wavelength of the radiation from the 1200 K blackbody source they were exposed to. We observe a steep rise in both responsivity ℜ and detectivity D * with decreasing wire size, reaching ℜ = 3.1×104 V/W and D * = 2.7×109 cmHz1/2/W at room temperature for a 300×300 nm2 device. Two significant advantages of such small wires as bolometers are their low power requirement and fast response time. Our numerical estimations suggest response times in the order of nanoseconds for the smallest samples. They could help improve resolution and response of thermal imaging devices, for example. We believe the performance may be further improved by optimizing the design and operating parameters.
©2011 Optical Society of America
1. Introduction
Microbolometers are widely integrated in various technologies for both military and civilian applications. These include surveillance, security, and thermal imaging. They are made from different materials, e.g. metals or semiconductors [1]. The basic operating principle relies on the change of electrical properties of the device upon absorption of infrared (IR) radiation. Two highly desirable technical attributes for many applications are good sensitivity and fast response. Ease of use and cost can also play an important role for the end user. Semiconductor detectors that rely on excitation of electrons across a small band gap by IR photons usually have a high signal-to-noise ratio and a fast response time, but are selective in wavelength. Also such detectors often depend on complicated fabrication processes. Further, due to the small band gap they are operated at low temperature in vacuum, adding to inconvenience in use and price. Thermal detectors however, make use of the heating effect of the IR radiation. These can be as simple as a thin metallic strip or wire, thus quite straightforward to fabricate, low cost and reliable. They also operate at room-temperature but usually have a slower response (in the millisecond range) and a lower overall performance [1]. A great deal of work has been done on antenna-coupled microbolometers (see e.g. [2–5]) with time constants approaching 100 ns [6]. In this paper we present results of our study on thermal bolometers from lithographically patterned platinum micro- and nanowires. The smallest elements have dimensions significantly smaller than the wavelength of IR radiation being detected. They exhibit detectivity comparable with or higher than other bolometers and we argue that their response time is in the order of ten nanoseconds. Previously, we have studied thermal emission properties of such wires during resistive heating by a DC current. We observed a significant increase in the radiated signal for our narrowest wires [7,8]. With Kirchoff’s law of thermal radiation in mind the question naturally arose whether they would serve well as bolometers.
2. Characterization of responsivity, noise, detectivity and time response
2.1. Samples and experiment
We have fabricated two sets of platinum wires with different dimensions. Set A has widths ranging from 8 μm to 2 μm and lengths from 16 μm to 2 μm. Set B has lengths ranging from 300 nm to 20 μm with a fixed width of 300 nm. Both sets were fabricated on Si/SiO2 (105 nm) substrates using photolithography and e-beam lithography respectively, then DC sputtering and lift-off techniques. The SiO2 layer acts as thermal insulation between the wire and the Si substrate. Below the platinum we deposited 5 nm of chromium and titanium for set A and set B respectively, in order to improve adhesion. The thickness of the platinum layer was 50 nm for both sets. A typical sample is displayed in Fig. 1. It is designed with fine voltage sense leads, and source leads that broaden at the heater ends. This allows us to bias the wire by passing an electrical current through the outer source leads, and at the same time we monitor the resistance through the two inner sensing leads. We carefully characterize the current-voltage properties of our samples and record both their electrical and thermal resistance. The electrical resistance is approximated by:
where R 0 is a constant and α is the temperature coefficient of resistance (TCR) of the material. The resistance of our wires ranges from 6 to 300 Ω and their TCR is 0.002 K−1. Extensive studies have revealed that the thermal resistance remains constant, despite changes in resistance caused by e.g. changes in grain size if the wire is exposed to high temperatures. This is in agreement with our numerical modeling that shows a very weak dependence of the thermal resistance on the thermal conductance of the wire. Essentially this is because the heat dissipation away from the wire is governed by thermal properties of the surroundings, but not by the properties of the wire itself. Thus, after careful characterization, we can tell the temperature of the wire quite accurately by measuring its resistance. To characterize our bolometers we bias them at room temperature with a DC current source and we monitor their resistance. Samples were biased within the range of 300 μA to 1.9 mA. As a reference we chose to keep the Joule dissipation power constant at 24 μW, in order to limit the self-heating contribution to the temperature change, in comparison with the IR-heating. The self-heating contribution was in most cases limited to within 2 K (for samples longer than 2 μm), but rose gradually up to 9K with decreasing sample length. With a fixed bias current and with the resistance in a steady state the bolometer is suddenly exposed to a NiCr-Ni blackbody source at a temperature of 1200 K. The corresponding blackbody emission spectrum is in the near-IR and has a maximum at a wavelength of 2.4 μm. The irradiance of the source was 1 W/cm2. Exposure to the light source causes an abrupt resistance change in the wire that is easily detected with a four-point measurement without resorting to building the wire into a resistance bridge.
Fig. 1 Scanning electron microscope image of a 4 μm long by 300 nm wide wire showing the four-lead sample structure.
2.2. Responsivity
The performance of bolometers is characterized in terms of responsivity ℜ and noise. Responsivity is defined as the ratio of the output signal generated to the incident power [1], hence can be expressed as:
where V is the voltage along the wire and P inc is the incident IR power. In our case this can be expressed as [9]: where I is the bias current, ΔR is the resistance change, Irr is the irradiance of the blackbody IR source and A is the area of the wire. Figure 2 displays the responsivity as a function of the surface area of the wire. From these results it is clear that as the dimensions of the wires are reduced, responsivity increases. This is to be expected from Eq. (3). A maximum responsivity of 3.1×104 V/W is reached for the 300 nm long sample. The larger, micron-scale, photolithographically defined samples from set A yield a more modest responsivity of around 103 V/W or below.Fig. 2 Responsivity dependence on the area for the two sets of platinum nano- and microbolometers. Empty circles are samples from set A and filled ones are from set B.
2.3. Noise and detectivity
A standard figure of merit to evaluate the performance of bolometers is the specific detectivity D *, expressed as [10]:
where Vn is the root mean square (RMS) voltage fluctuation per unit bandwidth corresponding to the total voltage noise power spectral density Sn in the sample. It is evident from Eq. (4) that noise can severely limit the performance of the device. As it is quite challenging to measure noise in low resistance devices such as ours we estimate the contribution of the main sources: Resistance noise (i.e. Johnson-Nyquist noise), thermal conductance noise (TCN) and 1/ f noise. We use the results to calculate a detectivity for our bolometers. The resistance noise can be expressed as: where SV ,JN is the voltage power spectral density due to the resistance fluctuations, k B is the Boltzmann constant and T is absolute temperature. Thermal conductance noise arises from temperature variations in the detector caused by heat conductance fluctuations between the wire bolometer and its surroundings (mainly the substrate). The power spectral density of voltage fluctuations caused by this is expressed as [11]: where we have used for the thermal conductance to the surroundings. Excess, or 1/ f noise in metals is usually attributed to random hopping motion of scattering centers, i.e. impurities and defects [12]. It is a low-frequency, non-equilibrium phenomenon related to DC bias current. It can be expressed by Hooge’s empirical law [12]: where β is the Hooge constant for platinum films (β = 0.002, see Ref. [13]) and N 0 the numbers of carriers in the thermo-sensitive layer of the bolometer. Usually, the thermal conductance noise is negligible, and the preponderant noise between the Johnson noise and the 1/ f noise depends on material properties and operating conditions. When evaluating noise according to equations (5), (6) and (7), we chose f = 1 Hz in Eq.(7) as our measurements are done at DC and 1 Hz is a commonly chosen reference point for frequency dependent noise. Figure 3 displays the total noise voltage dependence on length for set B samples. The results indicate that under our measurement conditions the predominant noise source in our platinum thin film bolometers is resistance noise, i.e. Johnson-Nyquist noise. Increase in wire temperature and electrical resistance contribute to the increase in TCN with length. Even at the low frequency of f =1 Hz the 1/ f noise is an order of magnitude weaker than the resistance noise. Combining these results with our data for responsivity gives a measure of detectivity according to Eq. (4). The results are displayed in figure 4. It can be seen that the detectivity increases significantly as area decreases. A maximum detectivity of around 2.7×109 cmHz1/2/W was reached for the square 300 nm long sample. Our platinum bolometers show high responsivity, low noise, and thus high detectivity compared to other reported values, such as a VO2 bolometer (1.94×108 cmHz1/2/W) [10], poly SiGe (2.3×109 cmHz1/2/W) [14] or even carbon nanotubes (4.5×105 cmHz1/2/W) [15]. They also have a smaller size and a smaller resistance than most typical bolometers.Fig. 3 RMS noise voltage components and total noise dependence on length for set B samples (constant width of 300 nm).
Fig. 4 Detectivity dependence on the area for the two sets of platinum nano- and microbolometers. Empty circles are samples from set A and full ones from set B.
2.4. Time response
Another important advantage of using small bolometers is their short response time. A simple estimation of the thermal RC time constant τ is obtained by taking the ratio of thermal capacitance C to the thermal conductance to the surroundings G [11],
As an example, an estimation based on cp = 140 J/(kg K) and a sample with lateral dimensions 300×600 nm2 gives C = 2.7×10−14 J/K. With the measured value of thermal resistance Z th = G −1 = 150 K/mW this results in a response time of τ = 4 ns. This estimation agrees very well with our time dependent finite element simulation results [16]. Our previous results have shown that emission spectra from heated wires in the size range of the smaller samples reported here differs appreciably from the blackbody spectrum [8]. (Wires wider than ∼ 1μm do however appear to emit like blackbodies.) Nevertheless our smallest devices do emit very strongly in the energy range of interest, although this is limited to light polarized along the wire axis [7,8,17,18]. Therefore it is not surprising that they are sensitive as radiation detectors in the same regime although this may indicate that they are wavelength selective. Consequently we expect our devices to act as polarization sensitive bolometers. We are presently investigating the detectivity as a function of polarization, the results of which shall be published shortly. We believe that the performance can be improved by optimizing the bias current and the thermal insulation of the wires. Increasing the thermal impedance of the bolometers can be achieved by raising the thickness of the SiO2, using suspended structures or using substrates of low thermal conductivity, e.g. silica aerogel [6] or Si3N4 [5].
3. Conclusion
In summary we have investigated bolometric properties of Pt wires with dimensions in the order of and smaller than the IR wavelength being detected. These are uncooled devices that exhibit responsivity of 3.1×104 V/Wand detectivity of 2.7 ×109 cmHz1/2/W. The design and fabrication of these devices is very simple and can be achieved with state-of-the-art photolithography. The resistance of our devices ranges from several Ω’s to a few hundred Ω. They display comparable detectivity to many more complex bolometers. We estimate numerically their time response in order of nanoseconds. This, together with their smallness should help improve e.g. imaging devices such as scanning thermal microscopes or thermal cameras. Further improvement in their detectivity could be made by optimizing the bias current and the thermal insulation of the wires.
Acknowledgments
This research was funded in part by the Icelandic Research Fund and the University of Iceland Research Fund.
References and links
1. A. Rogalski, “Infrared detectors: status and trends,” Prog. Quantum Electron. 27, 59–210 (2003). [CrossRef]
2. E. N. Grossman, J. A. Koch, C. D. Reintsema, and A. Green, “Lithographic dipole antenna properties at 10 μm wavelength: comparison of methods-of-moments predictions with experiment,” Int. J. Infrared Millim. Waves 19, 817–825 (1998). [CrossRef]
3. I. Codreanu, F. J. González, and G. D. Boreman, “Detection mechanisms in microstrip dipole antenna-coupled infrared detectors,” Infrared Phys. Technol. 44, 155–163 (2003). [CrossRef]
4. F. J. González and G. D. Boreman, “Comparison of dipole, bowtie, spiral and log-periodic IR antennas,” Infrared Phys. Technol. 46, 418–428 (2005). [CrossRef]
5. F. J. González, B. Illic, and G. D. Boreman, “Antenna-coupled microbolometers on a silicon-nitride membrane,”Microwave Opt. Technol. Lett. 47, 546–548 (2005). [CrossRef]
6. F. J. González, C. S. Ashley, P. G. Clem, and G. D. Boreman, “Antenna-coupled microbolometer arrays with aerogel thermal isolation,” Infrared Phys. Technol. 45, 47–51 (2004). [CrossRef]
7. S. Ingvarsson, L. J. Klein, Y.-Y. Au, J. A. Lacey, and H. F. Hamann, “Enhanced thermal emission from individual antenna-like nanoheaters,” Opt. Express 15, 11249–11254 (2007). [CrossRef] [PubMed]
8. Y.-Y. Au, H. S. Skulason, S. Ingvarsson, L. J. Klein, and H. F. Hamann, “Thermal radiation spectra of individual subwavelength microheaters,” Phys. Rev. B 78, 085402 (2008). [CrossRef]
9. A. Kosarev, M. Moreno, A. Torres, and C. Zuniga, “IR sensors based on silicon-germanium-boron alloys deposited by plasma: fabrication and characterization,” J. Non-Cryst. Solids 354, 2561–2564 (2008). [CrossRef]
10. C. Chen, X. Yi, X. Zhao, and B. Xiong, “Characterizations of VO2-based uncooled microbolometer linear array,” Sens. Actuators, A 90, 212–214 (2001). [CrossRef]
11. R. Smith, F. Jones, and R. Chasmar, The Detection and Measurement of Infra-red Radiation (Oxford Univ. Press, 1957).
12. S. Kogan, Electronic Noise and Fluctuations in Solids (Cambridge Univ. Press, 1996). [CrossRef]
13. D. Fleetwood, J. Masden, and N. Giordano, “1/f Noise in platinum films and ultrathin platinum wires: evidence for a common, bulk origin,” Phys. Rev. Lett. 50, 450–453 (1983). [CrossRef]
14. S. Sedky, P. Fiorini, K. Baert, L. Hermans, and R. Mertens, “Characterization and optimization of infrared poly SiGe bolometers,” IEEE Trans. Electron Devices 46, 675–681 (1999). [CrossRef]
15. R. Lu, Z. Li, G. Xu, and J. Wu, “Suspending single-wall carbon nanotube thin film infrared bolometers,” Appl. Phys. Lett. 94, 163110 (2009). [CrossRef]
16. S. Æ. Jónsson, “Nonlinear thermal electric analysis of platinum microheaters,” Master’s thesis, University of Iceland (2009).
17. H. F. Hamann, J. A. Lacey, and S. Ingvarsson, “Progress towards a thermally driven, infra-red near-field source using nanoheaters,” J. Microsc. 229, 512–516 (2008). [CrossRef] [PubMed]
18. L. J. Klein, S. Ingvarsson, and H. F. Hamann, “Changing the emission of polarized thermal radiation from metallic nanoheaters,” Opt. Express 17, 17963–17969 (2009). [CrossRef] [PubMed]
