Article begins on next page

Transcription

1 Whispering-gallery mode composite sensors for on-chip dynamic temperature monitoring Rutgers University has made this article freely available. Please share how this access benefits you. Your story matters. [ This work is an ACCEPTED MANUSCRIPT (AM) This is the author's manuscript for a work that has been accepted for publication. Changes resulting from the publishing process, such as copyediting, final layout, and pagination, may not be reflected in this document. The publisher takes permanent responsibility for the work. Content and layout follow publisher's submission requirements. Citation for this version and the definitive version are shown below. Citation to Publisher Version: Frenkel, Matthew, Avellan, Marlon & Guo, Zhixiong. (2013). Whispering-gallery mode composite sensors for on-chip dynamic temperature monitoring. Measurement Science and Technology 24(7), Citation to this Version: Frenkel, Matthew, Avellan, Marlon & Guo, Zhixiong. (2013). Whispering-gallery mode composite sensors for on-chip dynamic temperature monitoring. Measurement Science and Technology 24(7), Retrieved from doi: /t3w66p17. Terms of Use: Copyright for scholarly resources published in RUcore is retained by the copyright holder. By virtue of its appearance in this open access medium, you are free to use this resource, with proper attribution, in educational and other non-commercial settings. Other uses, such as reproduction or republication, may require the permission of the copyright holder. Article begins on next page SOAR is a service of RUcore, the Rutgers University Community Repository RUcore is developed and maintained by Rutgers University Libraries

2 Whispering-Gallery Mode Composite Sensors for On-Chip Dynamic Temperature Monitoring Matthew Frenkel, Marlon Avellan, and Zhixiong Guo* Department of Mechanical and Aerospace Engineering, Rutgers, the State University of New Jersey, Piscataway, New Jersey, 08854, USA Whispering-gallery mode temperature sensors have been demonstrated to have extremely high accuracy. Previous experiments have been limited to indirect sensor heating by externally heating the local environment. In this paper, we coated PDMS films directly onto an electrical resistive wire as sensors, allowing on-chip dynamic temperature measurement. The effects of sensor size are discussed and verified through an expansion of the current theory of WGM resonance shifts to include composite materials. Finally, the WGM sensor s measurements are compared to the same measurements recorded by a thermocouple, demonstrating the great advantages of WGM sensors for on-chip real temperature monitoring. I. INTRODUCTION Research into optical whispering-gallery mode (WGM) sensors has continued to grow over the past decade because of their ability to act as highly accurate and extremely sensitive sensors in a number of different fields. Currently, molecular 1,2, temperature 3, pressure 4, and gas sensors 5, are just some of the active research topics into WGM devices. The WGM sensor consists of near-field light coupled into a dielectric micro-resonator and is sensitive to both *Corresponding Author. guo@jove.rutgers.edu 1

3 changes in the evanescent field surrounding 1-4, and within the resonator 5,6. Being that these measurements are frequency based, instead of intensity-based, the sensitivity of a WGM-based sensor is much higher than its intensity based counterpart. Molecular sensors have reached capabilities of detecting pico-molar chemical residues 6, individual RNA viruses 7, and temperature sensors have been developed for cryogenic temperatures with a resolution of the order of 10-3 K 8. The focus of this paper is to explore the possibility of using WGM sensors for on-chip dynamic temperature measurements. To date, experiments studying WGM-based temperature sensors involve shifting the temperature of a chamber surrounding the WGM working unit. This is very useful for verifying theoretical WGM models as well as determining the sensitivity of WGM devices but provides little information about their ability to be used for direct on-chip temperature measurements of components and devices. Here, we have coated WGM film sensors directly onto a heating component. In this way, we can show their innovative sensing capabilities for dynamic temperature measurements of heating/cooling components. II. EXPERIMENTAL SETUP The experimental setup, shown in Fig. 1, is primarily composed of four parts: the WGM sensor, the working electronic component, the chamber, and the data acquisition system. The WGM sensor consists of a dielectric micro-resonator and the optical coupling device. In our experiment, we used a tapered optical fiber (approximately 0.5µm in diameter) as the near-field coupling device. The fiber taper was fabricated through the heat and pull method 3. Any dielectric material may be used to fabricate micro-resonators. Fused silica beads are the most common type of resonator but the process involves high temperatures. On the other hand, 2

4 polydimethylsiloxane (PDMS) is a liquid at room temperature until it is mixed with a curing agent. For this reason, PDMS was used in this experiment. PDMS was prepared in a 10:1 ratio, mixed thoroughly and then left in a vacuum for degassing. Once fully degassed, the PDMS was coated directly onto a tiny wire to form PDMS WGM annular micro-resonators. The wire is demonstrative of an electronic component to be heated through Joule heating. In our experiments, we used a 36-gauge (127µm diameter) nichrome wire (McMaster) as the heating element. The coated annular micro-resonators have two different geometry types formed around the cylindrical nichrome rod: the first type has a rapidly changing diameter along the wire and is named an ellipsoid shell; the second type has a nearly stable diameter and is referred to as a cylindrical annulus. These two different geometries were the product of different coating techniques. The ellipsoid shells were fabricated by directly placing a small droplet (200µm to 500µm) of PDMS onto the nichrome wire, by using the tip of a fiber optic wire, where as the cylindrical annuli were created by allowing a large PDMS droplet (> 500µm) to slide down the wire under gravity and leave a thin coating behind it. These two different techniques allowed for the examination of a larger range of resonator diameters. Thinner coatings were more easily fabricated as cylindrical annuli. Figure 2 shows two representative PDMS micro-annuli: cylindrical annuli were fabricated with thicknesses from ~ 10µm to ~100µm (25µm shown in the figure) and ellipsoid shells were fabricated with outer diameters ranging from ~200µm to ~500µm (311µm shown in the figure). In order to minimize thermal fluctuations due to free airflow, a chamber was constructed to contain the working unit. The chamber was created using a one-inch copper pipe coupling with four thin slits milled into it allowing for the fiber taper and nichrome wire to be put into place. The outside of the chamber was wrapped with a second nichrome wire for external 3

5 heating calibration, and covered with thermal paste (Omega omegatherm 201 ), before being surrounded with fiberglass insulation. The chamber was secured to another copper slab using thermal paste on the bottom, and a lid composed of both PDMS and fiberglass insulation was built for the top. The chamber is then raised into place around the tapered fiber, and the wire with PDMS coating, is clamped perpendicular to the tapered fiber. In addition, a thermocouple (Omega T-type 40 gauge, 79.9µm diameter bead) is also held inside the chamber by a clamp. This thermocouple is positioned 1-3mm away from the junction of PDMS film and fiber taper, and as close to contact with the nichrome wire as could be achieved. Data acquisition is done using a digital oscilloscope (Picoscope 3206B). A 1516nm distributed feedback laser (NEL NLK1556STG) is injected into one end of the fiber taper. Through a function generator (Agilent 33220A) and a laser controller (lightwave LCD-3724B) the light entering the taper follows a saw-tooth ramping with controllable amplitude and frequency. The other end of the fiber taper is directed towards a photo detector (Thorlabs PDA400). The Picoscope has two input channels, A and B, as well as an external trigger input. The trigger output from the function generator is plugged into the external trigger of the oscilloscope, the photo detector is plugged into channel A, and the signal from the thermocouple is amplified using a thermocouple amplifier (Omega Omni Amp IIB), before it is plugged into channel B. Using the Picoscope, both signals can be acquired simultaneously. The laser is set to ramp with a frequency of 100Hz and ten individual waveforms are collected with every acquisition. This frequency is fast enough to allow us to collect all waveforms in less than half a second, ensuring a stable temperature during the readings, while also being slow enough to allow the Picoscope to be set to sample at a rate of two million samples per waveform. 4

6 III. RESULTS AND DISCUSSION Before conducting the direct wire heating experiment, a calibration was performed on the PDMS sensors to determine their sensitivity. The calibration was conducted by slowly heating the chamber externally, without internal Joule heating. The chamber was heated to 10 C above room temperature at a rate of between 3-5 minutes per degree. This slow heating process was used so that the thermocouple and the WGM resonator are in thermal equilibrium. Data were collected every minute throughout the heating process. Multiple tests were performed to verify that the thermocouple and WGM resonator are in thermal equilibrium during the prescribed external heating. In these tests, multiple thermocouples were placed at the center of the chamber separated by distances of a few millimeters to a centimeter. The chamber temperature was slowly increased, as described above. The stability of the temperature inside the chamber was determined by the thermocouples. During external heating thermocouples within one centimeter of each other were in thermal equilibrium through the first 10 C of heating. Fig. 3 shows the relationship between the temperature change and the shift in resonance wavelength, of six different diameter resonators, 2 cylindrical annuli and 4 ellipsoid shells. In all cases, a strong linear correlation is found. It is also seen that, as the outer diameter of the resonator increases, the relationship between the temperature change and resonance wavelength changes, i.e. the resonator sensitivity depends on the diameter of the sensor. For the small resonator, a negative relationship exists, however, as the diameter increases this relationship becomes positive. Table 1 shows the values of the sensitivities found for these six resonators as well as the linear correlation coefficients (>0.998). It demonstrates that the change in sensitivity is 5

7 asymptotic. In the positive sensitivity region, the sensitivity increases as the resonator size increases. However, it is noticed that the absolute sensitivity at 172µm diameter is larger than that at 194µm. For resonators made of pure material in the past, diameter dependence was very slight 3,8. There was no evidence to demonstrate that the resonators geometry type had any effect on sensitivity except through diameter. As discussed earlier, cylindrical annuli could be more easily fabricated as thinner resonators. Additionally, cylindrical annuli have a more stable diameter along their length meaning that if the coupling point between the resonator and taper shifted, the effect on sensitivity would be small, whereas the same shift occurring between an ellipsoid shell and the tapered fiber would result in a dramatic change in sensitivity owing to the rapidly changing diameter of the shell. To understand the diameter-dependent sensitivity in composite sensor, we examine the fundamental theory used to explain the resonance shifts in a WGM sensor, described by the following equation 3 : dλ dt = λ # 1 0% $ n dn dt + 1 D dd& # 1 ( = λ dt ' 0 % $ n α + β & ( (1) ' where n is refractive index of resonator material, D is resonator diameter, λ 0 is the laser wavelength (1516nm), T is the temperature, α is the thermal optical coefficient, and β is the thermal expansion coefficient. The phenomenon of diameter dependence of sensitivity was also discovered in the past for composite sensors by Li et al. 9, when studying silica beads with PDMS coatings. They tried to explain changes in sensitivity by using an effective index of refraction for the system based on the energy factions of the resonating light found in each layer. In the current case, however, the concept of effective refractive index is not applicable because our inner core is a conductor, as such, no resonance will occur inside the core. Furthermore, the coated PDMS layer is relatively thick in our micro-resonators, one order of 6

8 magnitude larger than the laser wavelength, since almost all of the energy of the resonating light can be found near to resonators surface 10 we would not need to use an effective thermal optical coefficient regardless of the core material. It should also be mentioned that another apparent difference between the present study and Li et al. 9 is that the latter study did not actually realize and demonstrate on-chip temperature measurements; thus, the present study is novel. In order to explain the diameter dependence of sensitivity, we re-examine Eq. (1) from which it is clear that it is actually the thermal expansion coefficient that is affected by the diameter of the resonator and this coefficient should be an effective value for composite system. Tummala and Friedberg 11 investigated thermal expansion in a spherical composite system and we may adopt a similar relationship as follows: β eff = β 2 V 1 (1+υ 2 ) 2E 2 [(1+υ 2 ) 2E 2 ] +[(1 2υ 1 ) / E 1 ] (β 2 β 1 ). (2) where subscripts 1 and 2 represent nichrome wire and PDMS coating, respectively. E represents the Young s modulus, v represents the Poisson ratio, and V 1 is the volume fraction of the nichrome of interest region. This volume fraction is very important for understanding the asymptotic behavior of the sensitivity, as the PDMS thickness increases V 1 tends to zero and β eff = β 2. The thermal optical coefficient for PDMS 12 ranges from -1.5 to K -1. On average, the resonators used in this study had a thermal optical coefficient of ( ± 0.208) 10-4 K -1. This value was computed using Eq. (1) with the experimental sensitivity and the effective coefficient of thermal expansion determined from Eq. (2). In Fig. 4, we see the data collected during experiments plotted against the theoretical sensitivities based on the use of Eq. (2). The plot includes the sensitivity based on the average thermal optical coefficient of PDMS and the upper and lower bounds are found using the uncertainty in the thermal optical coefficient. It is 7

9 seen that the measured dependence of diameter is consistent with the trend in theoretical analysis. This figure demonstrates that the influence of the resonator diameter dominates, rather than the thermal optical coefficient in a composite sensor system; and Eqs. (1) and (2) establish a proper relationship between diameter and sensitivity. It is worthy of mentioning that the thermal optical coefficient of PDMS may vary during different fabrication processes; and this is why not all data points in Fig. 4 are located within the theoretical prediction range. Another possible explanation is that Eq. (2) was originally proposed for spherical composite system, but we used it in cylindrical and ellipsoid systems since we could not find better expression in literature. After calibrating the sensor and investigating the influence of diameter, we proceed to conduct the experiment involving internal Joule heating of the wire with temperature measured by the coated PDMS resonators. A low electrical current, amps, is used to keep the wire temperature from exceeding the bounds of the calibration. Data is collected as quickly as possible; the limiting factor is recording the data to the computers hard drive, which normally takes around 2 seconds to save 10 waveforms at a sampling rate of 2 million samples per waveform. Figs. 5 (a) and (b) show the heating and corresponding cooling curves based on three different current values (0.03A, 0.04A, and 0.05A respectively). Fig. 5 illustrates that when compared, the thermocouple and WGM sensor give different temperatures. As the current in the wire increases, this temperature difference increases. For example, at a current of 0.03A a difference of 0.8 C is seen at steady state, while at a current of 0.05A the difference reaches 2.15 C. In all cases, it is the WGM sensor that shows the higher measured temperature. In order to examine which sensors is more accurate, we investigate the theoretical temperature variation in the wire. A transient heat transfer analysis in the heated wire will lead to the following equation: 8

10 [ ] dt π 4 D 2 1 ρ 1 C 1 + D 2 2 ( 2 D 1 )ρ 2 C 2 dt = 4I 2 σ e 2 πd 1 πd 2 h( T T ) (3) In which subscripts 1 and 2 are the nichrome wire and PDMS coating respectively. ρ and C are the density and specific heat respectively, I is the current in the nichrome, σ e is the electrical resistivity of the nichrome wire, calculated as ( ± 0.105) 10-6 (Ωm), h is the heat transfer coefficient, T is the temperature of the ambient air, and T is the temperature of the nichrome- PDMS composite. During heating, the wire is assumed to have an initial temperature equal to the ambient air and during cooling the initial temperature is the same as the steady-state temperature reached in Fig. 5 (a). In order to properly determine the theoretical temperature variation, we need to determine the heat transfer coefficient, h. Churchill and Chu s correlation for natural convection in horizontal cylinders is the following 13 : Nu D = h theoryd 2 k " $ Ra 6 = $ D " $ $ Pr # # ( ) 9 16 % ' & 8 27 % ' ' ' & 2 (4) where Pr and k are the Prandtl number and thermal conductivity of air, respectively, Nu D is the Nusselt number, and Ra D is the Rayleigh number, Ra D = gβd 3 2 (T ηγ max T ), in which, g represents the gravity, γ and η are the thermal diffusivity and kinematic viscosity of air, respectively. Through an iterative process of Eq. (4), the theoretical heat transfer coefficient can be determined and then used with Eq. (3) to determine the dynamic temperature of the nichrome. The solid lines in Figs. 5 (a-b) are such theoretical predictions. The measurements by the WGM sensor match the theoretical temperatures much closer than the thermocouple does. 9

11 In steady state, energy generated through Joule heating must be equal to the energy lost due to free convection. Using the steady-state temperature reached, T max, we can compute the measured heat transfer coefficient as follows: h meas = 4I 2 σ e π 2 D 1 2 D 2 (T max T ) (5) Table 2 compares the experimentally measured and theoretically predicted heat transfer coefficients. The measurements were conducted at the 295µm ellipsoid shell of the wire heated with different currents. It is seen that the measurements match well with the predictions. In general, our experimental results demonstrate the superiority of the WGM sensor to a thermocouple for dynamic and steady state temperature measurements. We have demonstrated that PDMS resonators can be easily fabricated directly onto an element of interest as a microannular resonator. Results in Figs. 5 (a-b) and Table 2 clearly verify that these resonators can precisely and accurately monitor real time temperature changes throughout the heating and cooling processes of a component of interest, unlike their thermocouple counterparts. IV. CONCLUSIONS The results of this study show a clear advantage in the WGM sensor versus a thermocouple for dynamic on-chip temperature measurements. As components get smaller it becomes increasingly difficult to ensure that there is good contact between the component and the thermocouple. In addition, as the thermocouple bead becomes large compared to the component, more of it will be exposed to ambient air or nearby component surfaces. In these cases, thermocouples readings become very unreliable, but the coated WGM sensor provides an attractive alternative. We have shown that the measurements taken by a calibrated WGM sensor are accurate, when compared with theoretical predictions, in both the transient and steady state. 10

12 When these WGM sensors are coated onto a device, they can no longer be treated as pure materials, but instead need to be examined as composites; as a result, effective thermal optical and thermal expansion coefficients need to be considered. We found that the effective thermal expansion is more important than effective thermal optical coefficient. We have demonstrated that PDMS based WGM sensors can be easily fabricated directly onto an area of interest to create micro resonators. Both ellipsoid shells and cylindrical annulus are equally effective as resonators, but due to the rapidly changing diameter of the ellipsoid, the WGM sensitivity can be greatly influenced by the coupling location of resonator, whereas, the stable diameter of the cylindrical annulus reduces the influence of the coupling location, making it the more attractive sensor choice. V. ACKNOWLEDGMENTS We acknowledge support to the work by the National Science Foundation under Grant No. CBET We would also like to thank John Petrowski and Joseph Vanderveer for their assistance in machining the chamber. REFERENCES 1 F. Vollmer and S. Arnold, Whispering-gallery-mode biosensing: label-free detection down to single molecules. Nature Methods 5, (2008). 2 H. Quan and Z. Guo, Simulation of single transparent molecule interaction with optical microcavity. Nanotechnology 18, (2007). 3 Q. Ma, T. Rossmann, and Z. Guo, Temperature sensitivity of silica micro-resonators. J. Phys. D. App. Phys. 41, (2008). 11

13 4 T. Ioppolo, M. Kozhevnikov, V. Stepaniuk M. Otugen, and V. Sheverev, Micro-optical force sensor concept based on whispering gallery mode resonators. App. Opt. 47, (2008). 5 Q. Ma, L. Huang, L., and Z. Guo, Spectral shift response of optical whispering-gallery modes due to water vapor adsorption and desorption. Meas. Sci. Technol. 21, (2010) 6 L. Huang and Z. Guo, Nanofiltration and sensing of picomolar chemical residues in aqueous solution using optical porous resonator in microelectrofluidic channel, Nanotechnology 23, (2012). 7 V. Dantham, S. Holler, V. Kolchenko, Z. Wan, and S. Arnold, Taking whispering gallery-mode single virus detection and sizing to the limit. Appl. Phys. Lett. 101, (2012). 8 Q. Ma, T. Rossmann and Z. Guo, Whispering-gallery mode silica microsensors for cryogenic to room temperature measurement, Meas. Sci. Technol. 21, (2010). 9 B. Li, Q. Wang, Y. Xiao, X. Jiang, and Y. Li, On chip, high-sensitivity thermal sensor based on high-q polydimethylsiloxane-coated microresonator. App. Phys. Lett. 96, (2010). 10 H. Quan and Z. Guo, Simulation of Whispering-gallery mode resonance shifts for optical miniature biosensors. J. Quant. Spec. and Rad. Trans. 93, (2005). 11 R. R. Tummala, A. L. Friedberg, Thermal Expansion of Composite Materials. J. Appl. Phys. 41, (1970). 12 W. Yeung and A. Johnston, Effect of temperature on optical fiber transmission, Appl. Opt. 17, (1978). 12 A. Bejan, Convection Heat Transfer (Wiley, 3 rd ed., Hoboken, NJ, 2004). 12

14 FIG 1. Schematic of the experimental setup. 13

15 FIG 2. Representative coated resonators: (a) an ellipsoid shell with PDMS thickness 92µm; and (b) a cylindrical annulus with PDMS thickness 25µm. The nichrome wire has a diameter of 127µm in both images. (Due to the curved surface of the PDMS and the difference in refractive index to the surrounding air, the wire appears magnified inside of the PDMS resonator). 14

16 FIG. 3. Calibration of temperature rise vs. wavelength shift for different resonator diameters. The 172µm and 194µm resonators are cylindrical annuli, the rest are ellipsoid shells. 15

17 FIG. 4. Influence of resonator diameter on sensitivity at room temperature. The two left most data points are cylindrical annuli and the rest are ellipsoid shells. 16

18 FIG 5. Plots of the experimentally measured and theoretically predicted dynamic temperature changes in the nichrome wire with an coated with an ellipsoid shell of 295µm diameter: (a) heating processes with three different currents, and (b) the corresponding natural cooling processes. 17

19 TABLE 1. The sensitivity of six different diameter micro-resonators. The 172µm and 194µm resonators are cylindrical annuli, the rest are ellipsoid shells. Resonator Diameter (µm) Sensitivity (nm/k) Correlation Coefficient ± ± ± ± ± ±

20 TABLE 2. Comparison of experimentally measured and theoretically predicted heat transfer coefficients for the wire at 295µm ellipsoid shell, heated directly with different currents. Current 0.03 A 0.04 A 0.05 A WGM-measured 2.01 ± ± ± 0.27 temperature rise at steady state ( C) h theory (W/m 2 K) ± ± ± 0.18 h measured (W/m 2 K) ± ± ± 5.38 Difference (%)