- Email: [email protected]
A low cost micro-heater for aerosol generation applications
Microelectronic Engineering 129 (2014) 46–52
Contents lists available at ScienceDirect
Microelectronic Engineering journal homepage: www.elsevier.com/locate/mee
A low cost micro-heater for aerosol generation applications Guannan Liu a, Daniel A. Lowy b, Amir Kahrim b, Chao Wang a, Zeynep Dilli b, Nick Kratzmeier a, Wei Zhao a, Martin Peckerar a,b,⇑ a b
Department of Electrical and Computer Engineering, University of Maryland, College Park, MD 20742, USA FlexEl, LLC, 387 Technology Drive, Suite 2104, College Park, MD 20742, USA
a r t i c l e
i n f o
Article history: Received 17 February 2014 Received in revised form 13 June 2014 Accepted 12 July 2014 Available online 19 July 2014 Keywords: Microsystem Micro-heater Nickel plating Selective wet etch Gas sensor Aerosol inhale drug
a b s t r a c t A low cost micro-heater for aerosol generation has been developed using thin film deposition, electroless metal plating and selective wet etching technology. The thermal and electrical behavior of micro-heater (MH) and micro-hot-plate (MHP) systems is reported. The microstructure of the MH consists of a free standing amorphous silicon membrane with a thin film micro-heater (Cr/Ni layer) on top. The heating element is fabricated by e-beam evaporation and solution based nickel plating. The metal pattern temperature reaches the boiling point of glycerol (290 °C) and higher. A wet etched formed reservoir storage space for the aerosol source liquid can be placed alongside or underneath the hot plate. The entire MH unit is fabricated on a thin glass coverslip. This structure and the use of a plating bath and wet etch render the device remarkably low cost (less than 18!) as compared to other silicon technology based heaters. This MH can be integrated with gas sensors to assist pre-heating to the working temperature. The chemical release cycles are well controlled and are ideally suitable for use in medical procedures, such as drug inhalation therapy. Ó 2014 Published by Elsevier B.V.
1. Introduction A low cost, fast-response, low-power micro-heater (MH) integrated with a high capacity thin film flexible power source is of great interest in medicine [1]. Immediate applications of such devices include aerosol generators for inhalation therapy [2,3]. Micro-heaters can be integrated with chemical sensors [4,5], gas sensors [6–11], humidity sensors [12,13], and micro-electrical mechanical system (MEMS) micro fluid pumps [14,15]. In these applications the MH assists in the system components in reaching the working temperature required to activate the sensing or pumping function. Moreover, the MHs can be powered by high-capacity, portable, flexible galvanic cells [16,17] or from multiple power sources, which take advantage of energy harvesting systems [18,19]. Several early CMOS compatible micro-heater designs used thin-films, such as poly-silicon [20,21]. Recently, platinum [12,22] or Pt-based metal combinations (Pt/Ti, Pt/Cr, Pt/Al2O3 or Pt/MgO) have been explored [4–9]. Nevertheless, low cost metals like nickel appear to be more suitable. Nickel has 34% less electrical resistivity and 27% more thermal conductivity than platinum. The ⇑ Corresponding author at: Department of Electrical and Computer Engineering, University of Maryland, College Park, MD 20742, USA. Tel.: +1 301 405 7187. E-mail address: [email protected] (M. Peckerar). http://dx.doi.org/10.1016/j.mee.2014.07.011 0167-9317/Ó 2014 Published by Elsevier B.V.
surface of Ni is remarkably unreactive (i.e., chemically inert) [23]. Electroless Ni plating is a fast deposition method (270 nm/min); and, it is low cost as compared to evaporation, plasma vapor deposition (PVD), and sputtering techniques. There is no waiting time for vacuum purging. The thickness and growth rate can be controlled simultaneously. Other techniques like PVD cannot deposit metal at thicknesses greater than a shadow mask aperture or the photo-resist used in a lift-off process. By contrast, a solution based nickel plating bath can grow metal as thick as 10 lm without metal spalling. One challenge in the MH hot-plate design process is to achieve uniform heating over the entire hot-plate area. Obviously, the choice of substrate materials affects the thermal uniformity. Silicon membranes are widely applied in MEMS designs owing to their excellent electrical and mechanical properties [1,7]. Literature discloses heaters based on Si3N4/SiO2 structures are amenable to relatively high operating temperatures, and act as heat spreaders to achieve temperature uniformity [4–6,8]. Other reports revealed that polyimide exhibits better mechanical and thermal properties in achieving free-standing membrane and high yield in fabrication, compensating for the fragile nature of the Si3N4 and SiO2 membranes [11,22]. Amorphous silicon is resistive to HF/HCl (10:1 volume ratio) etch; it can be used as a good shadow protective layer during wet-etching of the membrane supporting material and the
G. Liu et al. / Microelectronic Engineering 129 (2014) 46–52
47
reservoir trenches. While silicon nitride has a slower etch rate in HF/HCl acids, it can be used to form reservoirs for drugs stored on the micro-heater side of the assembly. The HF/HCl bath is a fast etchant for glass slides (25 lm/min etch rate.) It typically takes 5 min to etch away the 0.13–0.17 mm thick micro-cover-glass glass substrate and leave a free-standing membrane, which is pre-deposited on top of glass. AZ4620 photo-resist served as a shadow mask, to define the reservoir. AZ4620 was tested and proved to be a good protection mask in HF/HCl acid for at least 8 min. In this work, both silicon and silicon nitride membranes were used to achieve different functions. We report here a design that uses a free-standing silicon membrane or Si3N4-memberane support, low cost metals deposition and wet etch process techniques to fabricate large batches of micro-heaters for the purpose of medical drug aerosolizing. We describe how thin MHs can be made, and that they can be powered by means of flexible ultra-capacitor/galvanic cell hybrid power systems. The MHs fabricated for this paper were manufactured in a university cleanroom facility. We have summed total facility, labor and material costs and we have found our per-unit fabrication cost is 18!. We require no cost-intensive microelectronic processing equipment. Based on equipment costs, maintenance and material cost alone, we feel that this expense can be reduced by an order of magnitude (<2!/MH.)
2. Fabrication There are 3 types of micro-heaters designed and fabricated to meet the requirement of different applications: MHs with no reservoir and MHs equipped with a side and bottom reservoir. The no-reservoir MH is built for sensor assistance and integration with other CMOS applications. MHs equipped with reservoir storage are aiming for aerosol generation purposes. Side-reservoir MHs are for the storage of small amounts of medical drug and bottom-reservoir MHs are fit for storage of large doses. Shadow masks of both meander and double-spiral wires with different lengths were used for seed layer metal deposition. We used MacDermid nickel plating bath, at a plating temperature of 88 °C and magnetic stirring at 66 rpm, the output deposition rate being as fast as 270 nm/min.
2.1. No-reservoir micro-heaters This type of micro-heater design is planar and suitable for integration with other devices. Fig. 1 shows the process steps of the device fabrication. An Oxford Plasma Lab plasma enhanced chemical vapor deposition (PECVD) system was used to deposit 1 lm of silicon nitride or 200 nm of amorphous silicon on top of 0.13– 0.17 mm or 0.17–0.25 mm micro-cover-glass wafers as a MH supporting film (Fig. 1a). 10 nm of Cr and 20 nm of Ni were deposited by E-Beam evaporation using a Denton with a stainless steel shadow mask to form the seed layer, Cr being used as an adhesion layer to avoid later thick Ni deposition peel-off issues. The shadow masks were made by laser cutting and have 100 lm width wires in meander and double-spiral shapes (Fig. 1b). AZ4620 was spin with 1.5 k rpm to yield a thickness of 15 lm, followed by a hard bake to protect heater patterns and the supporting film (Fig. 1c). The wafer was then flipped and spun with AZ4620 resist, followed by a soft-bake. The photolithography was accomplished using an exposure wavelength of 365 nm UV for 90 s and development in AZ400 K/DIW (1:3 volume ratio) for 90 s, followed by a hard bake (Fig. 1d). The work piece was etched in the HF/HCl (10:1 volume ratio) bath for 5 min (Fig. 1e). Photo-resist was stripped in heated Remover PG. This was followed by a
Fig. 1. Fabrication steps of the no-reservoir micro-heater with the ability to integrate other sensor devices. (a) Micro-cover-glass with MH supporting film layer. (b) Adhesion Cr layer and seed layer of Ni deposed and then thick Ni plated. (c) Heater pattern covered and protected by photo-resist AZ4620 layer. (d) Flip the sample over and spin photo-resist layer followed by photo-lithography to define the wet etch area. (e) HF/HCl wet etch the glass. (f) Remove photo-resist and flip over the sample.
standard acetone–methanol–isopropanol sequence rinse, and then dried by nitrogen gas (Fig. 1f).
2.2. Side trench reservoir micro-heater This type of micro-heater design is similar to the one described above, but it is intended for aerosol generation rather than CMOS compatibility. Inhaled drugs, initially in the liquid state, can be stored in a reservoir, and then transferred to the micro-hot-plate via a wicking fiber cloth (e.g., a filter paper) to evaporate, and generate an aerosol. Fig. 2 shows the process steps used. In order to form a trench, only a silicon nitride supporting film can be utilized, as it has a small etch rate in HF/HCl (Fig. 2a). Steps similar to those encountered in the no-reservoir design flow were used for step (b) to (f). But instead of hard baked AZ4620 to protect the metal, the side reservoir was defined by photolithography (Fig. 2c). At the end, the aerosol generator fluid was introduced in the top reser-
48
G. Liu et al. / Microelectronic Engineering 129 (2014) 46–52
Fig. 2. Fabrication steps of the side reservoir micro-heater with the ability to store liquid state drugs. (a) Micro-cover-glass with silicon nitride supporting film layer. (b) Adhesion Cr layer and seed layer of Ni deposited and then thick plated with Ni. (c) Spin photo-resist layer followed by photo-lithography to define the storage reservoir area. (d) Flip the sample over and spin photo-resist layer followed by photo-lithography to define the wet etch area. (e) HF/HCl wet etch both sides. (f) Remove photo-resist and flip over the sample. (g) Fill drug into the reservoir, stick a KimWipe tissue on top, and connect to the reservoir.
Fig. 3. Fabrication steps for the bottom reservoir micro-heater with the ability to store liquid state drugs in a sealed chamber. (a) Micro-cover-glass with amorphous silicon layer. (b) Adhesion Cr layer and seed layer of Ni deposed with thick Ni plated. (c) Heater pattern covered and protected by photo-resist AZ4620 layer. (d) Flip the sample over and spin photo-resist layer followed by photo-lithography to define the wet etch area. (e) HF/HCl wet etch the glass. (f) Remove photo-resist and flip over the sample; this becomes the top part of the micro-heater. (g) Take samples from step (a) and flip over, spin photo-resist layer, followed by photo-lithography to define the wet etch area for reservoir. (h) HF/HCl wet etch the glass and form the bottom reservoir. (i) Remove photo-resist, this becomes the bottom part of the micro-heater. (j) Seal the top and bottom part with seal glue, and use fiber cloth to connect reservoir, and then cover over the heater pattern.
49
G. Liu et al. / Microelectronic Engineering 129 (2014) 46–52
voir, and connected to the reservoir and micro-heater with a fiber cloth tissue (Fig. 2g).
2.3. Bottom seal reservoir micro-heater This type of design aims to optimize medical drug aerosol generation. As compared to side reservoir micro-heaters, these have a larger volume sealed reservoir storage space underneath the micro-heater. When the heater generates heat, the liquid in the reservoir vaporizes and the higher pressure will pump more fluid upward, transferring it through the fiber tissue to the MH. Fig. 3 presents the process steps of the device fabrication. In order to reduce the number of fabrication steps, only the amorphous silicon membrane is used, which has a high chemical resistivity to HF/HCl acid etch. Otherwise, an extra photo-resist layer had to be spun onto the surface, to protect the silicon nitride from being etched. The top part of the MH is made via the same process as the
no-reservoir MH component (Fig. 3a–f). The bottom part of the MH is formed by taking wafers from step (a) and flipping them over, spinning AZ4620, followed by a soft-bake, photolithography exposing with a dose of 365 nm UV for 90 s, and development in AZ400 K/DIW (1:3 volume ratio) for 90 s. This defines the bottom reservoir (Fig. 3g). The components are wet etched in HF/HCl (10:1 volume ratio) for 5 min (Fig. 3h). Then the photo-resist is removed in Remover PG with a gentle Q-tip swab, followed by cleaning via a standard acetone-methanol-isopropanol sequence rinse, and then drying by nitrogen gas blow (Fig. 3i). We combine the top and bottom part with a sealing glue, use fiber tissue to connect the reservoir, and cover up the heater’s metal pattern (Fig. 3j).
3. Results The Fig. 4 displays the fabricated devices: both silicon nitride and amorphous silicon supporting films are used. Four different
(a)
(b)
(c) Fig. 4. Photographs of fabricated micro-heaters: (a) no-reservoir MH with silicon nitride supporting film. (b) Side reservoir MH with silicon nitride supporting film. (c) Noreservoir MH with amorphous silicon supporting film (can be used as either top part of bottom reservoir MH or bottom part of top reservoir MH). Different lengths of metal pattern have been examined for optimization.
(a)
(b) Fig. 5. (a) Probe station setup for electrical, thermal and aerosol generation measurement. (b)Pattern resistance with respect to different plating times and thicknesses.
50
G. Liu et al. / Microelectronic Engineering 129 (2014) 46–52
lengths of pattern (16, 18, 20, and 22 mm, respectively) are processed to ascertain the best heating configuration.
In Fig. 5a we show the probe station setup for electrical, thermal, and aerosol generation measurements. Thick plated metal patterns are required to achieve low resistance and fast response. The metal thickness and plating time are linearly proportional, and the resistance is inversely proportional to the plate thickness, as shown in Fig. 5b. With the plating time and pattern thickness doubled, resistances decrease by a factor of 2. By adjusting plating times, we were able to achieve the targeted 8, 9, 10, and 11 X resistance values required for the 16, 18, 20, and 22 mm total length patterns, respectively.
amorphous silicon film. The lower blocks were set to account for the nickel emissivity. In this way, quantitatively accurate temperature profiling of the MH surface was possible in light of emissivity differences. The inset in the upper left of the Fig. 6a shows the ‘‘raw’’ thermal image without emissivity correction at room temperature. The default emissivity for entire imaging area is 0.34. Nickel and amorphous silicon thermal emissivity were set to 0.13 and 0.75, correspondingly. These values were acquired through emissivity calibration at 100, 150, and 200 °C. Fig. 6a shows the temperature distribution under 1.7 V bias. The Ni pattern can reach temperatures as high as 150 °C. The silicon film is heated by the metal pattern and tends to be about 20 °C cooler than the metal itself. Different voltage bias levels, their corresponding final operating temperatures and input powers are listed in Fig. 6b. Operating temperatures of 150, 200, 250, and 350 °C are shown.
3.2. Thermal measurements
3.3. Aerosol generation measurements
MHs temperature profile images were taken with an infrared camera (Optotherm Infrasight MI320). The results are shown in Fig. 6. The rectangular ‘‘block’’ structures in Fig. 6a are not directly obtained from the IR imager. But rather, the blocks are used to signify regions of different emissivities. One set of blocks (the upper set in Fig. 6a was set to account for the emissivity of the
In Fig. 7, we demonstrate the correlation between the aerosol volume quantities versus the specific power needed by the MH to generate the respective density of aerosol. Given that in several tests we used MHs connected in parallel, the power usage had to be normalized; this is why the units are W per MH. As shown in the graph, a single MH can generate abundant amounts of aerosol
3.1. Electrical measurements
(a)
(b) Fig. 6. (a) IR image of 2 identical MHs in parallel for both A-Si and Ni areas temperature measurements with emissivity correction (Ni emissivity set to 0.13 and A-Si emissivity set to 0.75). Inset figure at upper left is room temperature IR image without emissivity correction before heat measurement, the default emissivity for entire imaging area is 0.34. (b)Thermal measurement with respect to different input voltage bias.
Aerosol Volume [cm3]
G. Liu et al. / Microelectronic Engineering 129 (2014) 46–52
60
4. Discussion
50
In a series of 205 aerosol generating experiments, we obtained aerosol volumes in the range of 48.8 ± 7.9 cm3. A volume of 60 cm3 is the maximum volume that can be inhaled by a patient at one breath. In several cases we found that the total resistance of the MH would suddenly drop to 1–2X and reduce the yield and repeatability. This phenomenon is a consequence of the Ni-plating not only growing nickel on top of the pattern, but also on the sidewalls. When this happens, the pattern wires start merging with each other. This can be avoided by tightly bonding the shadow mask and wafer when depositing seed layers, or by using photolithography and lift-off processes for defining the seed layer. All the single, dual and triple MH systems generated consistently abundant amounts of dense aerosol. In a few instances, if one of MH of the dual systems burned out, the other one will continue operating with the remaining one. Similarly, the triple device was able to generate vapors with one remaining MH out of the initial three. In each scenario, the total quantity of aerosol was not affected, which enables the maintenance of smooth medical drug delivery, which remains constant over time. For our device, we set the voltage bias to 1.7 V to reach 150 °C for drug delivery purposes. Higher voltage bias (up to 3 V) and higher temperatures (up to 350 °C) can be applied for other forms of aerosol. For future development, we will further study the top reservoir MH. This type of design is aimed at the medical drug aerosol generation process. However, compared to the bottom reservoir MH, this design has the advantage of larger volume for liquid storage, it involves fewer fabrication steps, and incurs lower manufacturing cost owing to the mold made top part. Multiple MH units in parallel are used to vaporize a larger amount of liquid (or different liquids simultaneously), as shown in Fig. 9.
40 30 20
Multiple Heaters
10
Single Heater
0 0.8
0.9
1
1.1
1.2
1.3
1.4
1.5
Power [W/Microheater] Fig. 7. Aerosol volume quantity relative to specific power required for its aerosol generation.
70 60
Aerosol Volume[cm 3]
51
50 40 30 20
Dual Heater
10
Triple Heater
0 1.5
2.5
3.5
4.5
5.5
Power Supplied to the Entire Device [W] Fig. 8. Aerosol volume quantity relative to the total power needed by a device consisting of 2 MHs (blue data points) and 3 MHs (red data points), respectively. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
at low power (0.8–1.0 W) with 1.7 V voltage bias. For multiple MHs in parallel, the power requirement increased by ca. 50%. However, there was a gain in the kinetics of aerosol generation, as the aerosol was generated faster. In a series of tests, we have generated aerosol by means of the parallel designed dual and triple MHs. Results summarized in the graph displayed in Fig. 8 reveal a total power need of 2.05 ± 0.15 W for the dual MHs and of 2.35 ± 0.15 W for the triple MH systems, respectively.
5. Conclusions We were able to fabricate different types of low cost, fast responding micro-heaters for various applications. Nickel plating and wet etching baths were used to reduce costs and increase the speed of the process. The devices exhibited the desired heating characteristics. Both double spiral shaped and meander shaped heaters were investigated. While both geometries operated well, no significant differences in heat or aerosol generation were noted from one micro-heater design to another. On the basis of our experimental results, this device shows promise as an aerosol generator for drug delivery. Owing to their low cost (<18! per part), such micro-heaters can be used as disposable aerosol dispensers, although multiple other uses are also enabled.
Fig. 9. Design of the top reservoir micro-heater: (a) no-reservoir micro-heater with 3 double-spiral design in parallel. (b)PVC mold made top reservoir with wick and solvent inside.
52
G. Liu et al. / Microelectronic Engineering 129 (2014) 46–52
Appendix A. Detailed cost analysis
Appendix B. Metal Prices List
To perform a unit item cost analysis we must break the process flow into unit operations. The most significant unit operations for the micro-heater process are broken out in the table below and unit operation costs are listed.
Chromium Price: 2.48 USD/KG = 0.07 USD/OZ (up to date of Apr 29th, 2014) Nickel Price: 8.26 USD/LB = 0.516 USD/OZ (up to date of Apr 29th, 2014) Platinum Price: 1453.50 USD/OZ (up to date of May 2nd, 2014) Some micro-heaters also use the thin-films such as poly-silicon [20,21] that is comparable to our PECVD membrane formation. PECVD is used to form a free-standing substrate for metal pattern, the price can be reached lower when batch fabrication. E-beam evaporation is only used for a very thin seed layer for later nickel growth in bath. To achieve low cost, thermal evaporator can be used.
Operation
Cost of MH without reservoir
Cost of MH with reservoir
Topside barrier film deposition (PECVD) Shadow mask plating seed (E-BEAM) Electroless up-plate Topside protect (Photoresist Spin) Shadow mask deposition for reservoir Wet-etch reservoir Strip topside protect ‘‘Scribe and dice’’ individual heaters Total
$200.00
$200.00
$400.00
$400.00
$50.00 $50.00
$50.00 $50.00
$0.00
$400.00
$50.00 $50.00 $200.00
$50.00 $50.00 $200.00
$1000.00
$1400.00
Material costs for the unit operation are included in the unit operation cost. We can easily process 10 ‘‘sheets’’ of material, 10 cm 10 cm = 100 cm2. Material costs are estimated to be $10/ ‘‘sheet’’ of starting material 10 sheets of starting material will cost $100. Thus the processing/material cost per run would be $1500.00. Next, we must assess the number of heaters that can be fabricated. We assume each heater image (taking ‘‘saw street’’ space into account) will be 0.5 cm 0.5 cm = 0.25 cm2. Ten 100 cm2 sheets contain 1000 cm2, yielding 8333 devices. Thus, the current unit cost is 18!. It must be emphasized, that these are premium prices, paid to an external service ‘‘vendor.’’ In this case, our vendor is the University of Maryland clean-room facility. Unit process costs role in university overheads that will not be incurred in an in-house operation. A fair fraction of the overhead is in clean-room maintenance. Our ‘‘in-house’’ operation will not require the stringent ‘‘class 10’’ clean-room specifications necessitating the use of HEPA filters, laminar air flow and other expensive measures. The most expensive operation (and the operation that limits the number of simultaneously processed substrates) is the shadow mask evaporation process. We will certainly be able to increase the number of simultaneously processed substrates with specially designed equipment. This alone could increase process throughput by an order of magnitude. We feel confident that in-house process costs will be less than 2! per part. Comparison with the platinum [12–22] or Pt-based metal combinations (Pt/Ti, Pt/Cr, Pt/Al2O3, Pt/MgO) [4–9] based micro-heaters shows our design with Cr/Ni is much less cost.
References [1] G.-S. Chung, Sens. Actuators, A 112 (2004) 55–60. [2] A.R. Bhashyam, M.T. Wolf, A.L. Marcinkowski, A. Saville, K. Thomas, J.A. Carcillo, et al., J. Aerosol Med. Pulm. Drug Delivery 21 (2008) 181–188. [3] P. Worth Longest, B.M. Spence, L.T. Holbrook, K.M. Mossi, Y.-J. Son, M. Hindle, Production of inhalable submicrometer aerosols from conventional mesh nebulizers for improved respiratory drug delivery, J. Aerosol Sci. (2012). [4] D. Mutschall, C. Scheibe, E. Obermeier, Basic micro module for chemical sensors with on chip heater and buried sensor structure, in: Solid-State Sensors and Actuators, 1995 and Eurosensors IX. Transducers’ 95. The 8th International Conference on, 1995, pp. 256–259. [5] J. Gardner, A. Pike, N. De Rooij, M. Koudelka-Hep, P. Clerc, A. Hierlemann, et al., Sens. Actuators, B: Chem. 26 (1995) 135–139. [6] Y. Mo, Y. Okawa, K. Inoue, K. Natukawa, Sens. Actuators, A 100 (2002) 94–101. [7] M. Casaletto, S. Kaciulis, G. Mattogno, L. Pandolfi, G. Scavia, L. Dori, et al., Appl. Surf. Sci. 189 (2002) 39–52. [8] U. Hoefer, G. Kühner, W. Schweizer, G. Sulz, K. Steiner, Sens. Actuators, B: Chem. 22 (1994) 115–119. [9] U. Dibbern, Sens. Actuators, B: Chem. 2 (1990) 63–70. [10] Y. Mo, Y. Okawa, M. Tajima, T. Nakai, N. Yoshiike, K. Natukawa, Sens. Actuators, B: Chem. 79 (2001) 175–181. [11] M. Aslam, C. Gregory, J. Hatfield, Sens. Actuators, B: Chem. 103 (2004) 153– 157. [12] C.-L. Dai, M.-C. Liu, F.-S. Chen, C.-C. Wu, M.-W. Chang, Sens. Actuators, B: Chem. 123 (2007) 896–901. [13] C.-L. Dai, Sens. Actuators, B: Chem. 122 (2007) 375–380. [14] W.Y. Sim, H.J. Yoon, O.C. Jeong, S.S. Yang, J. Micromech. Microeng. 13 (2003) 286. [15] H. Takagi, R. Maeda, K. Ozaki, M. Parameswaran, M. Mehta, Phase transformation type micro pump, in: Micro Machine and Human Science, 1994. Proceedings, 1994 5th International Symposium on, 1994, p. 199. [16] M. Peckerar, Z. Dilli, M. Dornajafi, N. Goldsman, Y. Ngu, R.B. Proctor, et al., Energy Environ. Sci. 4 (2011) 1807–1812. [17] M. Peckerar, M. Dornajafi, Z. Dilli, N. Goldsman, Y. Ngu, B. Boerger, et al., Fabrication of flexible ultracapacitor/galvanic cell hybrids using advanced nanoparticle coating technology, J. Vac. Sci. Technol. B Microelectron. Nanometer Struct. 29 (2011). 011008–011008-6. [18] W. Zhao, K. Choi, S. Bauman, Z. Dilli, T. Salter, M. Peckerar, A Radio-Frequency Energy Harvesting Scheme for Use in Low-Power Ad Hoc Distributed Networks, 2012. [19] W. Zhao, K. Choi, S. Bauman, T. Salter, D.A. Lowy, M. Peckerar, et al., Solid-State Electron. 79 (2013) 233–237. [20] M. Parameswaran, A.M. Robinson, D.L. Blackburn, M. Gaitan, J. Geist, Electron Dev. Lett., IEEE 12 (1991) 57–59. [21] E. Yoon, K.D. Wise, Electron Dev., IEEE Trans. 39 (1992) 1376–1386. [22] J.-S. Han, Z.-Y. Tan, K. Sato, M. Shikida, J. Micromech. Microeng. 15 (2005) 282. [23] J.P. Birk, M. Ronan, I. Bennett, C. Kinney, J. Chem. Educ. 68 (1991) 48.
Contents lists available at ScienceDirect
Microelectronic Engineering journal homepage: www.elsevier.com/locate/mee
A low cost micro-heater for aerosol generation applications Guannan Liu a, Daniel A. Lowy b, Amir Kahrim b, Chao Wang a, Zeynep Dilli b, Nick Kratzmeier a, Wei Zhao a, Martin Peckerar a,b,⇑ a b
Department of Electrical and Computer Engineering, University of Maryland, College Park, MD 20742, USA FlexEl, LLC, 387 Technology Drive, Suite 2104, College Park, MD 20742, USA
a r t i c l e
i n f o
Article history: Received 17 February 2014 Received in revised form 13 June 2014 Accepted 12 July 2014 Available online 19 July 2014 Keywords: Microsystem Micro-heater Nickel plating Selective wet etch Gas sensor Aerosol inhale drug
a b s t r a c t A low cost micro-heater for aerosol generation has been developed using thin film deposition, electroless metal plating and selective wet etching technology. The thermal and electrical behavior of micro-heater (MH) and micro-hot-plate (MHP) systems is reported. The microstructure of the MH consists of a free standing amorphous silicon membrane with a thin film micro-heater (Cr/Ni layer) on top. The heating element is fabricated by e-beam evaporation and solution based nickel plating. The metal pattern temperature reaches the boiling point of glycerol (290 °C) and higher. A wet etched formed reservoir storage space for the aerosol source liquid can be placed alongside or underneath the hot plate. The entire MH unit is fabricated on a thin glass coverslip. This structure and the use of a plating bath and wet etch render the device remarkably low cost (less than 18!) as compared to other silicon technology based heaters. This MH can be integrated with gas sensors to assist pre-heating to the working temperature. The chemical release cycles are well controlled and are ideally suitable for use in medical procedures, such as drug inhalation therapy. Ó 2014 Published by Elsevier B.V.
1. Introduction A low cost, fast-response, low-power micro-heater (MH) integrated with a high capacity thin film flexible power source is of great interest in medicine [1]. Immediate applications of such devices include aerosol generators for inhalation therapy [2,3]. Micro-heaters can be integrated with chemical sensors [4,5], gas sensors [6–11], humidity sensors [12,13], and micro-electrical mechanical system (MEMS) micro fluid pumps [14,15]. In these applications the MH assists in the system components in reaching the working temperature required to activate the sensing or pumping function. Moreover, the MHs can be powered by high-capacity, portable, flexible galvanic cells [16,17] or from multiple power sources, which take advantage of energy harvesting systems [18,19]. Several early CMOS compatible micro-heater designs used thin-films, such as poly-silicon [20,21]. Recently, platinum [12,22] or Pt-based metal combinations (Pt/Ti, Pt/Cr, Pt/Al2O3 or Pt/MgO) have been explored [4–9]. Nevertheless, low cost metals like nickel appear to be more suitable. Nickel has 34% less electrical resistivity and 27% more thermal conductivity than platinum. The ⇑ Corresponding author at: Department of Electrical and Computer Engineering, University of Maryland, College Park, MD 20742, USA. Tel.: +1 301 405 7187. E-mail address: [email protected] (M. Peckerar). http://dx.doi.org/10.1016/j.mee.2014.07.011 0167-9317/Ó 2014 Published by Elsevier B.V.
surface of Ni is remarkably unreactive (i.e., chemically inert) [23]. Electroless Ni plating is a fast deposition method (270 nm/min); and, it is low cost as compared to evaporation, plasma vapor deposition (PVD), and sputtering techniques. There is no waiting time for vacuum purging. The thickness and growth rate can be controlled simultaneously. Other techniques like PVD cannot deposit metal at thicknesses greater than a shadow mask aperture or the photo-resist used in a lift-off process. By contrast, a solution based nickel plating bath can grow metal as thick as 10 lm without metal spalling. One challenge in the MH hot-plate design process is to achieve uniform heating over the entire hot-plate area. Obviously, the choice of substrate materials affects the thermal uniformity. Silicon membranes are widely applied in MEMS designs owing to their excellent electrical and mechanical properties [1,7]. Literature discloses heaters based on Si3N4/SiO2 structures are amenable to relatively high operating temperatures, and act as heat spreaders to achieve temperature uniformity [4–6,8]. Other reports revealed that polyimide exhibits better mechanical and thermal properties in achieving free-standing membrane and high yield in fabrication, compensating for the fragile nature of the Si3N4 and SiO2 membranes [11,22]. Amorphous silicon is resistive to HF/HCl (10:1 volume ratio) etch; it can be used as a good shadow protective layer during wet-etching of the membrane supporting material and the
G. Liu et al. / Microelectronic Engineering 129 (2014) 46–52
47
reservoir trenches. While silicon nitride has a slower etch rate in HF/HCl acids, it can be used to form reservoirs for drugs stored on the micro-heater side of the assembly. The HF/HCl bath is a fast etchant for glass slides (25 lm/min etch rate.) It typically takes 5 min to etch away the 0.13–0.17 mm thick micro-cover-glass glass substrate and leave a free-standing membrane, which is pre-deposited on top of glass. AZ4620 photo-resist served as a shadow mask, to define the reservoir. AZ4620 was tested and proved to be a good protection mask in HF/HCl acid for at least 8 min. In this work, both silicon and silicon nitride membranes were used to achieve different functions. We report here a design that uses a free-standing silicon membrane or Si3N4-memberane support, low cost metals deposition and wet etch process techniques to fabricate large batches of micro-heaters for the purpose of medical drug aerosolizing. We describe how thin MHs can be made, and that they can be powered by means of flexible ultra-capacitor/galvanic cell hybrid power systems. The MHs fabricated for this paper were manufactured in a university cleanroom facility. We have summed total facility, labor and material costs and we have found our per-unit fabrication cost is 18!. We require no cost-intensive microelectronic processing equipment. Based on equipment costs, maintenance and material cost alone, we feel that this expense can be reduced by an order of magnitude (<2!/MH.)
2. Fabrication There are 3 types of micro-heaters designed and fabricated to meet the requirement of different applications: MHs with no reservoir and MHs equipped with a side and bottom reservoir. The no-reservoir MH is built for sensor assistance and integration with other CMOS applications. MHs equipped with reservoir storage are aiming for aerosol generation purposes. Side-reservoir MHs are for the storage of small amounts of medical drug and bottom-reservoir MHs are fit for storage of large doses. Shadow masks of both meander and double-spiral wires with different lengths were used for seed layer metal deposition. We used MacDermid nickel plating bath, at a plating temperature of 88 °C and magnetic stirring at 66 rpm, the output deposition rate being as fast as 270 nm/min.
2.1. No-reservoir micro-heaters This type of micro-heater design is planar and suitable for integration with other devices. Fig. 1 shows the process steps of the device fabrication. An Oxford Plasma Lab plasma enhanced chemical vapor deposition (PECVD) system was used to deposit 1 lm of silicon nitride or 200 nm of amorphous silicon on top of 0.13– 0.17 mm or 0.17–0.25 mm micro-cover-glass wafers as a MH supporting film (Fig. 1a). 10 nm of Cr and 20 nm of Ni were deposited by E-Beam evaporation using a Denton with a stainless steel shadow mask to form the seed layer, Cr being used as an adhesion layer to avoid later thick Ni deposition peel-off issues. The shadow masks were made by laser cutting and have 100 lm width wires in meander and double-spiral shapes (Fig. 1b). AZ4620 was spin with 1.5 k rpm to yield a thickness of 15 lm, followed by a hard bake to protect heater patterns and the supporting film (Fig. 1c). The wafer was then flipped and spun with AZ4620 resist, followed by a soft-bake. The photolithography was accomplished using an exposure wavelength of 365 nm UV for 90 s and development in AZ400 K/DIW (1:3 volume ratio) for 90 s, followed by a hard bake (Fig. 1d). The work piece was etched in the HF/HCl (10:1 volume ratio) bath for 5 min (Fig. 1e). Photo-resist was stripped in heated Remover PG. This was followed by a
Fig. 1. Fabrication steps of the no-reservoir micro-heater with the ability to integrate other sensor devices. (a) Micro-cover-glass with MH supporting film layer. (b) Adhesion Cr layer and seed layer of Ni deposed and then thick Ni plated. (c) Heater pattern covered and protected by photo-resist AZ4620 layer. (d) Flip the sample over and spin photo-resist layer followed by photo-lithography to define the wet etch area. (e) HF/HCl wet etch the glass. (f) Remove photo-resist and flip over the sample.
standard acetone–methanol–isopropanol sequence rinse, and then dried by nitrogen gas (Fig. 1f).
2.2. Side trench reservoir micro-heater This type of micro-heater design is similar to the one described above, but it is intended for aerosol generation rather than CMOS compatibility. Inhaled drugs, initially in the liquid state, can be stored in a reservoir, and then transferred to the micro-hot-plate via a wicking fiber cloth (e.g., a filter paper) to evaporate, and generate an aerosol. Fig. 2 shows the process steps used. In order to form a trench, only a silicon nitride supporting film can be utilized, as it has a small etch rate in HF/HCl (Fig. 2a). Steps similar to those encountered in the no-reservoir design flow were used for step (b) to (f). But instead of hard baked AZ4620 to protect the metal, the side reservoir was defined by photolithography (Fig. 2c). At the end, the aerosol generator fluid was introduced in the top reser-
48
G. Liu et al. / Microelectronic Engineering 129 (2014) 46–52
Fig. 2. Fabrication steps of the side reservoir micro-heater with the ability to store liquid state drugs. (a) Micro-cover-glass with silicon nitride supporting film layer. (b) Adhesion Cr layer and seed layer of Ni deposited and then thick plated with Ni. (c) Spin photo-resist layer followed by photo-lithography to define the storage reservoir area. (d) Flip the sample over and spin photo-resist layer followed by photo-lithography to define the wet etch area. (e) HF/HCl wet etch both sides. (f) Remove photo-resist and flip over the sample. (g) Fill drug into the reservoir, stick a KimWipe tissue on top, and connect to the reservoir.
Fig. 3. Fabrication steps for the bottom reservoir micro-heater with the ability to store liquid state drugs in a sealed chamber. (a) Micro-cover-glass with amorphous silicon layer. (b) Adhesion Cr layer and seed layer of Ni deposed with thick Ni plated. (c) Heater pattern covered and protected by photo-resist AZ4620 layer. (d) Flip the sample over and spin photo-resist layer followed by photo-lithography to define the wet etch area. (e) HF/HCl wet etch the glass. (f) Remove photo-resist and flip over the sample; this becomes the top part of the micro-heater. (g) Take samples from step (a) and flip over, spin photo-resist layer, followed by photo-lithography to define the wet etch area for reservoir. (h) HF/HCl wet etch the glass and form the bottom reservoir. (i) Remove photo-resist, this becomes the bottom part of the micro-heater. (j) Seal the top and bottom part with seal glue, and use fiber cloth to connect reservoir, and then cover over the heater pattern.
49
G. Liu et al. / Microelectronic Engineering 129 (2014) 46–52
voir, and connected to the reservoir and micro-heater with a fiber cloth tissue (Fig. 2g).
2.3. Bottom seal reservoir micro-heater This type of design aims to optimize medical drug aerosol generation. As compared to side reservoir micro-heaters, these have a larger volume sealed reservoir storage space underneath the micro-heater. When the heater generates heat, the liquid in the reservoir vaporizes and the higher pressure will pump more fluid upward, transferring it through the fiber tissue to the MH. Fig. 3 presents the process steps of the device fabrication. In order to reduce the number of fabrication steps, only the amorphous silicon membrane is used, which has a high chemical resistivity to HF/HCl acid etch. Otherwise, an extra photo-resist layer had to be spun onto the surface, to protect the silicon nitride from being etched. The top part of the MH is made via the same process as the
no-reservoir MH component (Fig. 3a–f). The bottom part of the MH is formed by taking wafers from step (a) and flipping them over, spinning AZ4620, followed by a soft-bake, photolithography exposing with a dose of 365 nm UV for 90 s, and development in AZ400 K/DIW (1:3 volume ratio) for 90 s. This defines the bottom reservoir (Fig. 3g). The components are wet etched in HF/HCl (10:1 volume ratio) for 5 min (Fig. 3h). Then the photo-resist is removed in Remover PG with a gentle Q-tip swab, followed by cleaning via a standard acetone-methanol-isopropanol sequence rinse, and then drying by nitrogen gas blow (Fig. 3i). We combine the top and bottom part with a sealing glue, use fiber tissue to connect the reservoir, and cover up the heater’s metal pattern (Fig. 3j).
3. Results The Fig. 4 displays the fabricated devices: both silicon nitride and amorphous silicon supporting films are used. Four different
(a)
(b)
(c) Fig. 4. Photographs of fabricated micro-heaters: (a) no-reservoir MH with silicon nitride supporting film. (b) Side reservoir MH with silicon nitride supporting film. (c) Noreservoir MH with amorphous silicon supporting film (can be used as either top part of bottom reservoir MH or bottom part of top reservoir MH). Different lengths of metal pattern have been examined for optimization.
(a)
(b) Fig. 5. (a) Probe station setup for electrical, thermal and aerosol generation measurement. (b)Pattern resistance with respect to different plating times and thicknesses.
50
G. Liu et al. / Microelectronic Engineering 129 (2014) 46–52
lengths of pattern (16, 18, 20, and 22 mm, respectively) are processed to ascertain the best heating configuration.
In Fig. 5a we show the probe station setup for electrical, thermal, and aerosol generation measurements. Thick plated metal patterns are required to achieve low resistance and fast response. The metal thickness and plating time are linearly proportional, and the resistance is inversely proportional to the plate thickness, as shown in Fig. 5b. With the plating time and pattern thickness doubled, resistances decrease by a factor of 2. By adjusting plating times, we were able to achieve the targeted 8, 9, 10, and 11 X resistance values required for the 16, 18, 20, and 22 mm total length patterns, respectively.
amorphous silicon film. The lower blocks were set to account for the nickel emissivity. In this way, quantitatively accurate temperature profiling of the MH surface was possible in light of emissivity differences. The inset in the upper left of the Fig. 6a shows the ‘‘raw’’ thermal image without emissivity correction at room temperature. The default emissivity for entire imaging area is 0.34. Nickel and amorphous silicon thermal emissivity were set to 0.13 and 0.75, correspondingly. These values were acquired through emissivity calibration at 100, 150, and 200 °C. Fig. 6a shows the temperature distribution under 1.7 V bias. The Ni pattern can reach temperatures as high as 150 °C. The silicon film is heated by the metal pattern and tends to be about 20 °C cooler than the metal itself. Different voltage bias levels, their corresponding final operating temperatures and input powers are listed in Fig. 6b. Operating temperatures of 150, 200, 250, and 350 °C are shown.
3.2. Thermal measurements
3.3. Aerosol generation measurements
MHs temperature profile images were taken with an infrared camera (Optotherm Infrasight MI320). The results are shown in Fig. 6. The rectangular ‘‘block’’ structures in Fig. 6a are not directly obtained from the IR imager. But rather, the blocks are used to signify regions of different emissivities. One set of blocks (the upper set in Fig. 6a was set to account for the emissivity of the
In Fig. 7, we demonstrate the correlation between the aerosol volume quantities versus the specific power needed by the MH to generate the respective density of aerosol. Given that in several tests we used MHs connected in parallel, the power usage had to be normalized; this is why the units are W per MH. As shown in the graph, a single MH can generate abundant amounts of aerosol
3.1. Electrical measurements
(a)
(b) Fig. 6. (a) IR image of 2 identical MHs in parallel for both A-Si and Ni areas temperature measurements with emissivity correction (Ni emissivity set to 0.13 and A-Si emissivity set to 0.75). Inset figure at upper left is room temperature IR image without emissivity correction before heat measurement, the default emissivity for entire imaging area is 0.34. (b)Thermal measurement with respect to different input voltage bias.
Aerosol Volume [cm3]
G. Liu et al. / Microelectronic Engineering 129 (2014) 46–52
60
4. Discussion
50
In a series of 205 aerosol generating experiments, we obtained aerosol volumes in the range of 48.8 ± 7.9 cm3. A volume of 60 cm3 is the maximum volume that can be inhaled by a patient at one breath. In several cases we found that the total resistance of the MH would suddenly drop to 1–2X and reduce the yield and repeatability. This phenomenon is a consequence of the Ni-plating not only growing nickel on top of the pattern, but also on the sidewalls. When this happens, the pattern wires start merging with each other. This can be avoided by tightly bonding the shadow mask and wafer when depositing seed layers, or by using photolithography and lift-off processes for defining the seed layer. All the single, dual and triple MH systems generated consistently abundant amounts of dense aerosol. In a few instances, if one of MH of the dual systems burned out, the other one will continue operating with the remaining one. Similarly, the triple device was able to generate vapors with one remaining MH out of the initial three. In each scenario, the total quantity of aerosol was not affected, which enables the maintenance of smooth medical drug delivery, which remains constant over time. For our device, we set the voltage bias to 1.7 V to reach 150 °C for drug delivery purposes. Higher voltage bias (up to 3 V) and higher temperatures (up to 350 °C) can be applied for other forms of aerosol. For future development, we will further study the top reservoir MH. This type of design is aimed at the medical drug aerosol generation process. However, compared to the bottom reservoir MH, this design has the advantage of larger volume for liquid storage, it involves fewer fabrication steps, and incurs lower manufacturing cost owing to the mold made top part. Multiple MH units in parallel are used to vaporize a larger amount of liquid (or different liquids simultaneously), as shown in Fig. 9.
40 30 20
Multiple Heaters
10
Single Heater
0 0.8
0.9
1
1.1
1.2
1.3
1.4
1.5
Power [W/Microheater] Fig. 7. Aerosol volume quantity relative to specific power required for its aerosol generation.
70 60
Aerosol Volume[cm 3]
51
50 40 30 20
Dual Heater
10
Triple Heater
0 1.5
2.5
3.5
4.5
5.5
Power Supplied to the Entire Device [W] Fig. 8. Aerosol volume quantity relative to the total power needed by a device consisting of 2 MHs (blue data points) and 3 MHs (red data points), respectively. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
at low power (0.8–1.0 W) with 1.7 V voltage bias. For multiple MHs in parallel, the power requirement increased by ca. 50%. However, there was a gain in the kinetics of aerosol generation, as the aerosol was generated faster. In a series of tests, we have generated aerosol by means of the parallel designed dual and triple MHs. Results summarized in the graph displayed in Fig. 8 reveal a total power need of 2.05 ± 0.15 W for the dual MHs and of 2.35 ± 0.15 W for the triple MH systems, respectively.
5. Conclusions We were able to fabricate different types of low cost, fast responding micro-heaters for various applications. Nickel plating and wet etching baths were used to reduce costs and increase the speed of the process. The devices exhibited the desired heating characteristics. Both double spiral shaped and meander shaped heaters were investigated. While both geometries operated well, no significant differences in heat or aerosol generation were noted from one micro-heater design to another. On the basis of our experimental results, this device shows promise as an aerosol generator for drug delivery. Owing to their low cost (<18! per part), such micro-heaters can be used as disposable aerosol dispensers, although multiple other uses are also enabled.
Fig. 9. Design of the top reservoir micro-heater: (a) no-reservoir micro-heater with 3 double-spiral design in parallel. (b)PVC mold made top reservoir with wick and solvent inside.
52
G. Liu et al. / Microelectronic Engineering 129 (2014) 46–52
Appendix A. Detailed cost analysis
Appendix B. Metal Prices List
To perform a unit item cost analysis we must break the process flow into unit operations. The most significant unit operations for the micro-heater process are broken out in the table below and unit operation costs are listed.
Chromium Price: 2.48 USD/KG = 0.07 USD/OZ (up to date of Apr 29th, 2014) Nickel Price: 8.26 USD/LB = 0.516 USD/OZ (up to date of Apr 29th, 2014) Platinum Price: 1453.50 USD/OZ (up to date of May 2nd, 2014) Some micro-heaters also use the thin-films such as poly-silicon [20,21] that is comparable to our PECVD membrane formation. PECVD is used to form a free-standing substrate for metal pattern, the price can be reached lower when batch fabrication. E-beam evaporation is only used for a very thin seed layer for later nickel growth in bath. To achieve low cost, thermal evaporator can be used.
Operation
Cost of MH without reservoir
Cost of MH with reservoir
Topside barrier film deposition (PECVD) Shadow mask plating seed (E-BEAM) Electroless up-plate Topside protect (Photoresist Spin) Shadow mask deposition for reservoir Wet-etch reservoir Strip topside protect ‘‘Scribe and dice’’ individual heaters Total
$200.00
$200.00
$400.00
$400.00
$50.00 $50.00
$50.00 $50.00
$0.00
$400.00
$50.00 $50.00 $200.00
$50.00 $50.00 $200.00
$1000.00
$1400.00
Material costs for the unit operation are included in the unit operation cost. We can easily process 10 ‘‘sheets’’ of material, 10 cm 10 cm = 100 cm2. Material costs are estimated to be $10/ ‘‘sheet’’ of starting material 10 sheets of starting material will cost $100. Thus the processing/material cost per run would be $1500.00. Next, we must assess the number of heaters that can be fabricated. We assume each heater image (taking ‘‘saw street’’ space into account) will be 0.5 cm 0.5 cm = 0.25 cm2. Ten 100 cm2 sheets contain 1000 cm2, yielding 8333 devices. Thus, the current unit cost is 18!. It must be emphasized, that these are premium prices, paid to an external service ‘‘vendor.’’ In this case, our vendor is the University of Maryland clean-room facility. Unit process costs role in university overheads that will not be incurred in an in-house operation. A fair fraction of the overhead is in clean-room maintenance. Our ‘‘in-house’’ operation will not require the stringent ‘‘class 10’’ clean-room specifications necessitating the use of HEPA filters, laminar air flow and other expensive measures. The most expensive operation (and the operation that limits the number of simultaneously processed substrates) is the shadow mask evaporation process. We will certainly be able to increase the number of simultaneously processed substrates with specially designed equipment. This alone could increase process throughput by an order of magnitude. We feel confident that in-house process costs will be less than 2! per part. Comparison with the platinum [12–22] or Pt-based metal combinations (Pt/Ti, Pt/Cr, Pt/Al2O3, Pt/MgO) [4–9] based micro-heaters shows our design with Cr/Ni is much less cost.
References [1] G.-S. Chung, Sens. Actuators, A 112 (2004) 55–60. [2] A.R. Bhashyam, M.T. Wolf, A.L. Marcinkowski, A. Saville, K. Thomas, J.A. Carcillo, et al., J. Aerosol Med. Pulm. Drug Delivery 21 (2008) 181–188. [3] P. Worth Longest, B.M. Spence, L.T. Holbrook, K.M. Mossi, Y.-J. Son, M. Hindle, Production of inhalable submicrometer aerosols from conventional mesh nebulizers for improved respiratory drug delivery, J. Aerosol Sci. (2012). [4] D. Mutschall, C. Scheibe, E. Obermeier, Basic micro module for chemical sensors with on chip heater and buried sensor structure, in: Solid-State Sensors and Actuators, 1995 and Eurosensors IX. Transducers’ 95. The 8th International Conference on, 1995, pp. 256–259. [5] J. Gardner, A. Pike, N. De Rooij, M. Koudelka-Hep, P. Clerc, A. Hierlemann, et al., Sens. Actuators, B: Chem. 26 (1995) 135–139. [6] Y. Mo, Y. Okawa, K. Inoue, K. Natukawa, Sens. Actuators, A 100 (2002) 94–101. [7] M. Casaletto, S. Kaciulis, G. Mattogno, L. Pandolfi, G. Scavia, L. Dori, et al., Appl. Surf. Sci. 189 (2002) 39–52. [8] U. Hoefer, G. Kühner, W. Schweizer, G. Sulz, K. Steiner, Sens. Actuators, B: Chem. 22 (1994) 115–119. [9] U. Dibbern, Sens. Actuators, B: Chem. 2 (1990) 63–70. [10] Y. Mo, Y. Okawa, M. Tajima, T. Nakai, N. Yoshiike, K. Natukawa, Sens. Actuators, B: Chem. 79 (2001) 175–181. [11] M. Aslam, C. Gregory, J. Hatfield, Sens. Actuators, B: Chem. 103 (2004) 153– 157. [12] C.-L. Dai, M.-C. Liu, F.-S. Chen, C.-C. Wu, M.-W. Chang, Sens. Actuators, B: Chem. 123 (2007) 896–901. [13] C.-L. Dai, Sens. Actuators, B: Chem. 122 (2007) 375–380. [14] W.Y. Sim, H.J. Yoon, O.C. Jeong, S.S. Yang, J. Micromech. Microeng. 13 (2003) 286. [15] H. Takagi, R. Maeda, K. Ozaki, M. Parameswaran, M. Mehta, Phase transformation type micro pump, in: Micro Machine and Human Science, 1994. Proceedings, 1994 5th International Symposium on, 1994, p. 199. [16] M. Peckerar, Z. Dilli, M. Dornajafi, N. Goldsman, Y. Ngu, R.B. Proctor, et al., Energy Environ. Sci. 4 (2011) 1807–1812. [17] M. Peckerar, M. Dornajafi, Z. Dilli, N. Goldsman, Y. Ngu, B. Boerger, et al., Fabrication of flexible ultracapacitor/galvanic cell hybrids using advanced nanoparticle coating technology, J. Vac. Sci. Technol. B Microelectron. Nanometer Struct. 29 (2011). 011008–011008-6. [18] W. Zhao, K. Choi, S. Bauman, Z. Dilli, T. Salter, M. Peckerar, A Radio-Frequency Energy Harvesting Scheme for Use in Low-Power Ad Hoc Distributed Networks, 2012. [19] W. Zhao, K. Choi, S. Bauman, T. Salter, D.A. Lowy, M. Peckerar, et al., Solid-State Electron. 79 (2013) 233–237. [20] M. Parameswaran, A.M. Robinson, D.L. Blackburn, M. Gaitan, J. Geist, Electron Dev. Lett., IEEE 12 (1991) 57–59. [21] E. Yoon, K.D. Wise, Electron Dev., IEEE Trans. 39 (1992) 1376–1386. [22] J.-S. Han, Z.-Y. Tan, K. Sato, M. Shikida, J. Micromech. Microeng. 15 (2005) 282. [23] J.P. Birk, M. Ronan, I. Bennett, C. Kinney, J. Chem. Educ. 68 (1991) 48.