Micromachined Particles for Detecting Metal-Ion ... - IEEE Xplore
JOURNAL OF MICROELECTROMECHANICAL SYSTEMS, VOL. 15, NO. 5, OCTOBER 2006
1299
Micromachined Particles for Detecting Metal-Ion Concentration in Fluids Ryan Supino, Student Member, IEEE, and Joseph J. Talghader, Member, IEEE
Abstract—A remote microfluidic metal-ion sensor is developed using an electrochemical system integrated with a compact photovoltaic cell power supply. The sensor is designed to detect the sum of metal ions in a remote environment. The sensor uses electrodeposition to remove ions from the fluid around the sensor and deposit them on an electrode at the tip of a cantilever. The electrodeposited mass changes the resonant frequency of the cantilever, which can be determined upon read-out. The sensor is designed to be dropped in liquids or flow through microfluidic systems and can be used in parallel with many other similar sensors. The photovoltaic cells are directly integrated on the device and are capable of producing tens of microwatts of power at about 15% efficiency with laser excitation. However, the sensor operates at power levels of 50 nW with small voltages and currents using only scavenged daylight or room light. The complete device is integrated into a total volume below 3 , which is more than two orders of magnitude smaller 0.046 than other remote electrically-powered sensors reported to date. Although it is expected that multiple devices will be used in parallel to gain statistical data, individual particles detect metal-ion concentration within 24% of the actual concentration, making them suitable for safety testing and endpoint monitoring among other applications. [1714]
mm
Index Terms—Electrochemical, MEMS power supply, power scavenging, remote sensor.
I. INTRODUCTION XTENSIVE work has been performed in an effort to develop remote, self-powered microdevices [1]–[5]. Many of these devices have achieved impressive complexity, integrating a power source with sensing, communications, and even processing capabilities. Due to this level of complexity, especially the addition of communications and processing capabilities, the power requirements of such devices are often relatively substantial and involve dedication of large portions of device area to power supplies and communication, which places strict lower limits on device volume. As such, the smallest reported electriin cally powered remote device is only approximately 16 volume [3]. The primary difficulty associated with microscopic remote sensors is the inherent difficulty communicating with such a device. A number of different wireless communication schemes
E
Manuscript received October 25, 2005; revised March 16, 2006 Subject Editor M. Mehregany. R. Supino was with the Department of Electrical and Computer Engineering, University of Minnesota, Minneapolis, MN 55455 USA. He is now with Advanced Sensors and Microsystems, Honeywell International, Plymouth, MN 55411 USA (e-mail: [email protected]) The authors are with the Department of Electrical and Computer Engineering, University of Minnesota, Minneapolis, MN 55455 USA (e-mail: [email protected]). Digital Object Identifier 10.1109/JMEMS.2006.880246
have been developed for remote sensors typically based on either optical [6], [7] or RF [8]–[10] designs. Both types of communication schemes have drawbacks. Wireless RF communications tend to require a fairly large area antenna, on the order of , and typically have short data transmission distances on a the order of a few meters. Optical communication technologies are often designed to be passive (i.e., without an onboard light source). While they typically have longer data transmission distances, they require a large amount of sensor area to overcome diffraction limits on coupling and can often be fairly complex in design. Both types of communication schemes can achieve relatively low power consumption, on the order of less than 100 , however, this level of power usage is still difficult to pro. vide onboard a device with a total volume less than 1 Remote microdevices not only need a means of communicating data from the sensor to a receiving station, but also need a means of powering these operations. A variety of work has been performed on micropower generation for remote microsensors. Such work has included microsolar power supplies [11], [12], microbatteries [13], microfuel cells [14], low-frequency vibration energy scavenging [15], [16], in addition to others [17]. Many of these micro power supply devices are capable of pro. ducing modest amounts of power, on the order of 1–100 Unfortunately, most of these power sources are quite large, often or more. consuming volumes of 10 For a variety of applications including environmental sensing, drug or chemical delivery, industrial process flow monitoring, and microfluidic device evaluation, the development of extremely low-power fluidic sensors would be desirable. This paper introduces an electrochemical device architecture to detect metal-ion concentrations in electrolytes. The advantage of this type of architecture is that electrolytic reactions operate at low-voltages and require extremely low currents. The use of electrolysis allows the creation of metal-ion sensors using electrodeposition mechanisms or the creation of drug and chemical delivery systems using the electrochemical dissolution of thin film membranes [18]. In this paper, a fluidic metal-ion sensor is developed using electrodeposition as a sensing mechanism [19]. Since communication requires large amounts of sensor area, a means of storing data onboard the sensor could lead to large reductions in sensor volume. The use of electrodeposition allows the sensor device to store metal-ion concentration information on board. The electrodeposition sensing mechanism requires very low voltage and current operation allowing power to be delivered by three onboard photovoltaic (PV) cells. The sensor itself is comprised of a cantilever acting as a resonant mass sensor with an electrode at the tip where metal from the surrounding environment is electrodeposited as driven by the PV cell array. The total mass de-
1057-7157/$20.00 © 2006 IEEE
1300
JOURNAL OF MICROELECTROMECHANICAL SYSTEMS, VOL. 15, NO. 5, OCTOBER 2006
tabulated half-cell reactions. Summing these half-cell reactions can be deteran equilibrium cell potential of mined for the deposition of copper ions [20]. Under non-equilibrium conditions, or more specifically, during current flow, other modifications to the potential are necessary. These additional modifications, or overpotentials, can be added to the equilibrium cell potential giving [21]
(2)
Fig. 1. Illustration of copper electrodeposition. Cu is reduced at the cathode plating copper on the electrode surface and H O is oxidized at the anode giving off O gas.
posited can be determined by making resonant frequency measurements of the cantilever beam after the sensor is removed from its liquid and correlating this information to the concentration of ions in the environment. Collecting the sensors involves taking a small volume of the electrolyte. Since a large number of sensors can be dispersed at once, statistically a small sample will have at least a few sensors. The sensors are dried onto a convenient surface (silicon and lab paper were used in this work) where they could be scanned by a laser beam to determine added mass. II. BACKGROUND The microparticle metal-ion sensor will be composed of three important elements, the electrochemical based sensor, a mass sensitive resonant cantilever structure, and a PV cell array power supply. This section will discuss the theoretical background associated with the electrochemical sensing system and the mass sensitive resonant cantilever. A. Electrochemistry Electrodeposition is the sensing mechanism employed by the metal-ion sensor. When driven by a power source, the device electrodeposits metal ions from the solution onto the tip of a cantilever beam. Such a system represents an electrochemical cell. This cell is composed of two electronic conductors, the anode and cathode, and an ionic conductor, the fluidic medium. When a voltage is applied across the anode and cathode, a system of oxidation-reduction reactions occurs. As an example, let two electrodes be immersed in an electrolyte containing copper ions as shown in Fig. 1. The reduction of copper occurs at the cathode and the oxidation of water takes place at the anode. The total chemical reaction is given as follows: (1) To determine the voltage necessary to drive the electroplating of copper, the chemical reaction can be broken down into well
is the equilibrium cell potential, is the ideal gas where constant, is absolute temperature, is the number of moles of electrons transferred in the reaction as given by (1), and is , Faraday’s constant. The overpotential terms are given by , and , while is the concentration of copper ions in solution. The concentration is in reference to a standard is a dimensionless term. concentration of 1 M, therefore , originates from the reThe first overpotential term, sistivity of the fluidic medium and the external electrical circuit. The ohmic overpotential is a geometrically dependent term and must be calculated over the complex geometry of the elec, detrochemical device. The second over-potential term, scribes the electron transfer kinetics associated with a reaction. In an electrochemical reaction, electrons are transferred to the reactant in the case of a reduction reaction and away from the reactant in the case of an oxidation reaction. The rate at which electrons are transferred, and thus the rate at which species are reduced or oxidized in an electrochemical reaction is dependent on the potential dropped across the electrode/electrolyte interface. , is a concentration dependent overpotential that Finally, is due to the consumption of metal ions at the cathode. As metal ions are plated at the surface, the solution adjacent to the electrode becomes depleted of metal ions and a concentration gradient is formed. The establishment of a concentration gradient sets up a diffusion limitation to the deposition reaction possibly limiting the rate of metal deposition. This deposition limitation will be discussed as the mechanism governing the operation of the metal-ion sensor device. and are considFor the purposes of this work, ered negligible. Consequently, the current limitations imposed by ohmic loss or reaction rate kinetics are expected to be large compared to the limitation on deposition current imposed by metal-ion diffusion. is exFirst, let us consider the ohmic overpotential. pected to be small due to the low-current operation of the sensor. For extremely low concentration solutions and low-current operation, rough calculations for the sensor geometry show that the expected overpotential due to ohmic loss to be less than 0.1V. Variations in the conductivity of the solution should not have a major impact on cell potential for the current levels anticipated. , is related to reaction Since the activation overpotential, rate constants, we expect it to be very dependent on temperature. However, it can be shown that the electron transfer kinetics for copper electrodeposition are very fast and therefore the limitation to current imposed by the reaction rate is much larger than
SUPINO AND TALGHADER: METAL-ION CONCENTRATION IN FLUIDS
1301
the current limit imposed by the concentration gradient. Specifically, the current limitation due to electron kinetics is predicted using the Butler-Volmer relationship [21]
The final important electrochemical theory for the sensors is a means of relating the total amount of deposited metal to the current flowing in the system. The relationship between reduced or oxidized metal and total charge transferred is given as
(3)
(5)
where is known as the exchange current and represents the anodic and cathodic currents flowing at equilibrium, is the are constants total overpotential applied to the cell, and that describe the symmetry of the forward and reverse reactions. Using (3) in conjunction with typical copper electrodeposition is capable of flowing constants predicts that a current of 1.6 with an applied over potential of only 200 mV [22]. This current is much higher than is observed in devices used in this study, is not the rate limiting mechanism. indicating that is the basis for the metal-ion The third overpotential sensor operation. This overpotential is formed due to the depletion of reactant species in the proximity of the working electrode, which leads to a diffusion-based limitation on the transport of metal ions from the bulk solution to the cathode surface. Instead of viewing this mechanism as an overpotential, it can also be expressed as a diffusion limited current. For microelectrodes, the diffusion limited current is given as [19]
is the atomic mass, is current flow, is total rewhere is Faraday’s constant, and is the number of action time, electrons transferred per unit reaction. This equation is known as Faraday’s Law of Electrolysis. This relationship assumes that there are no parallel reactions occurring, such as the electrolysis of water, during the electrochemical reaction.
(4)
where is the beam thickness, is the length, is Young’s modulus, and is the material density of the cantilever. This resonant frequency equation is derived by solving the wave partial differential equation using boundary conditions of a beam with rectangular cross section and uniform thickness and width. As mass is added to the cantilever, a shift in the resonant frequency of the structure occurs. The change in resonant frequency can be approximated using Rayleigh’s Method [24] and is given by
where is the number of electrons transferred during the electrochemcial reaction, two in the case of copper ion reduction, is Faraday’s constant, is the electrode area, is the metal-ion diffusivity, is the concentration of ions in the bulk fluid is the electrode radius, and is an empirical conmedium, stant that represents geometrical deviations from an ideal circular microelectrode. Using typical copper electrodeposition parameters with a copper ion concentration of 1 mM and an electrode size of 30 it can be shown that the diffusion limited current 30 is on the order of 7 nA [22], which is almost three orders of magnitude lower than the current limit due to electron kinetics. Therefore, the device can be assumed to be diffusion limited at concentrations at or below 1 mM. The diffusion current expressed in (4) is directly dependent on the concentration of metal ions in solution. If an electrochemical device is operated with a power source that can handle the voltage requirements in addition to current levels greater than the diffusion limit, then the current will be limited by the diffusion process and will be proportional to the metal-ion concentration. In the case of a metal-ion sensor, this allows the integration of concentration data over a long period of time in order to measure small concentrations of metal ions. It is also important to notice that there is an initial time dependent transient in the diffusion current. This transient is due to the expansion of the depletion region and reaches a constant level as a stable depletion hemisphere forms. As long as the integration time of the sensor is much greater than this transient time, the error associated with the initial increased deposition rate will be negligible. For shorter integration times, the transient can be included directly since it has a well-quantified form.
B. Resonant Cantilever Mass Sensing The previous discussion of electrochemical theory is applicable to any type of electrochemical microparticle. This section considers the specifics of a sensor using cantilever mass detection. To begin, the fundamental resonant frequency of an undamped cantilever beam with one end clamped and the other end free is given by [23] (6)
(7) where is the shift in resonant frequency of the cantilever is the added mass, is the mass of the cantilever bebeam, fore loading, and is the cantilever resonant frequency before loading. The change in resonant frequency given by (7) assumes that the mass is added in a point at the tip of the cantilever. If the added mass is spread out along the cantilever length, a reduction in the sensitivity of the frequency shift due to mass loading occurs. Therefore, it is important to ensure that the mass loading is confined to a region at the tip of the cantilever beam. The resonant frequency shift can be measured in order to determine the total amount of mass added to the beam. In the case of a metal-ion sensor, this would be the total mass of metal electrodeposited at the cantilever’s tip. With an understanding of the frequency sensitivity of a cantilever beam to mass loading, the expected relationship between the resonant frequency shift of a metal-ion sensor and the concentration of ions in the fluidic medium can be determined. Combining (4), (5), and (7) the following relationship is obtained. (8)
1302
JOURNAL OF MICROELECTROMECHANICAL SYSTEMS, VOL. 15, NO. 5, OCTOBER 2006
Fig. 2. Drawing of complete sensor device. (a) PV cell mesa. (b) Cantilever with electrode at tip connected to PV cell array. (c) Counter electrode connected to PV cell array.
In (8), it is assumed that for typical sampling times, the diffusion transient contributes a negligible amount to the overall charge transfer during metal deposition. Therefore, the transient portion of (4) is ignored. This is a reasonable assumption as long as sampling times are on the order of minutes. Typical diffusion , coefficients for metal ions in solution are around leading to a transient decay time on the order of 10 s. For increased accuracy during shorter sampling times, it is possible to include the transient portion of the diffusion curve. Additionally, it is assumed that sufficient light is present such that electrodeposition occurs the entire time the device is in solution such that any etching of the metal that may occur is negligible. This is a reasonable assumption as long as the device is not exposed to extremely acidic environments and is operating for the majority of the exposure time. III. DESIGN AND FABRICATION Some of the basic mechanisms of an electrochemical based microsensor were discussed in the previous section in addition to some specifics of the sensing mechanism used by the metal-ion sensor device. This section will present greater detail of the sensor architecture and discuss the fabrication process. The metal-ion sensor is an integrated device composed of a power supply combined with a cantilever resonant mass sensor. The cantilever has a Pt working electrode at the tip. Surrounding the cantilever is a Pt counter electrode. The cantilever tip is connected to the cathode side of a photovoltaic (PV) cell array and the counter electrode is connected to the PV cell anode. The PV cell array drives the electrodeposition process, gathering metal ions from solution and plating them on the electrode at the tip of the cantilever. Fig. 2 gives an illustration of the basic architecture of the sensor structure. The power supply is created using a unique SOI process. Individual PV cells are created using a simple - junction design. The PV cells are isolated from one another using anisotropic etching of the silicon while stopping on the SOI buried oxide layer. Corner compensation techniques are used to create square shaped mesas for the PV cells [25]. The 54.7 sloped side walls of the PV cell mesas allow easy metal coverage between the separate mesas in order to create a power supply capable of producing about 1.5 V. The metal interconnect lines are insulated from the PV cell using a thermal oxide layer and a silicon nitride layer. The thermal oxide layer provides two additional benefits. First, it helps to tie up surface states reducing the recombination
Fig. 3. Process flow of complete integrated metal-ion sensor device.
of photogenerated electron-hole pairs at the surface of the diffusion. Secondly, it provides a simple antireflection coating increasing the overall light gathering efficiency of the PV cells. thick with a (100) The p-type device layer of the SOI is 20 is deposited using orientation. A thin layer of low-stress low pressure chemical vapor deposition (LPCVD) and serves as a diffusion mask for the PV cell array. This film is patterned, regions of the PV cell where subsequently a defining the thick phosphosilicate glass (PSG) film is deposited to serve as the diffusion dopant source. The wafer is annealed at 1000 for a period of one hour. Subsequently, the films are stripped and layer is deposited and patterned another thin low-stress for PV cell mesa definition. Separate PV cells are defined using a KOH etch to remove the silicon between cells. The KOH etch is terminated on the buried oxide of the SOI wafer and uses 45% to etch through the 20- -thick device KOH solution at 80 etch mask is stripped. layer. Subsequently, the Next, a 1000 thick thermal oxide is grown at 1050 in a dry oxygen ambient. The oxide film is used to tie up surface region of the PV cell which increases efficiency. states in the The cantilever structural layer is next added by depositing and . Alupatterning another 6000 thick layer of low-stress minum interconnect metal is deposited by e-beam evaporation to form an ohmic contact between PV and is annealed at 450 cell devices. A 100- layer of Ti is followed by a 300- layer of Pt for both the counter and working electrodes. Fig. 3 gives a depiction of the process flow. The Pt metal serves as the deposition area on the cantilever beam, therefore it is necessary to insulate the metal connecting
SUPINO AND TALGHADER: METAL-ION CONCENTRATION IN FLUIDS
1303
Fig. 4. Photomicrograph of completed metal-ion sensor.
the tip of the cantilever to the PV cell so that no metal is deposited in this region. If metal is deposited along the entire length of the cantilever, the effect of mass loading is counteracted by an increase in the stiffness of the beam. This leads to a reduction in the resonant frequency shift, or sensitivity, due to the added mass of the deposited metal. To prevent this, a 3000 thick layer is deposited using a plasma vapor deposition system and is patterned to cover all metallization other than the counter and working electrodes. After completion of the front side of the wafer, a thick photoresist layer is deposited on the front in order to protect the sensors during backside processing. The backside is then patterned using a thick photoresist in order to define sensor separation and cantilever pit areas. This pattern is etched using a deep reactive ion etch (DRIE) through the full thickness of the SOI handle thick, and terminates on the buried layer, approximately 300 oxide layer. The buried oxide layer is then removed in the open areas using 10:1 buffered oxide etchant for a period of 20 min. Care is taken to ensure that the etch is stopped after the oxide is removed from the sensor separation areas and the cantilever pit areas while not completely undercutting the area between the PV cell mesas and the silicon handle layer. Finally, the sensors are released by immersing the wafer in acetone dissolving the photoresist on the front side holding the sensors together. Fig. 4 shows an optical micrograph of the completed device. IV. EXPERIMENTAL RESULTS Each of the various aspects of the metal-ion sensor were evaluated. This includes the PV cell characteristics, the current demands of an electrochemical sensing or delivery mechanism, and resonant mass sensing driven by electrodeposition. In analysis of the power supply, arrays consisting of three PV cells connected in series were tested. These arrays were separate test structures from the arrays integrated with the sensor and were about four times larger in area. Individual cells had an and the total PV cell mesa size was active area of 0.032 on a side. The PV cell array was characterized using 228 a laser diode with a wavelength of 660 nm. Fig. 5 shows the general I–V behavior of the PV cell array under varying laser illumination intensities. Also included in Fig. 5 is an I–V plot of
Fig. 5. I–V curve of PV cell array at three distinct laser intensity levels and one under white light illumination.
Fig. 6. PV cell array open circuit voltage versus light intensity.
a single photovoltaic cell under broadband microscope illumination. In addition to measuring the general I–V behavior of the PV cell arrays, measurements of open cell voltage and short circuit current were made over a range of light intensities. Fig. 6 shows that the open cell voltage varies from about 1 V with a light to 1.52 V at an intensity of 2780 . intensity of 79.8 This corresponds to a range of short circuit currents from 0.4 to 13.2 for the respective light intensities (Fig. 7). The diffusion limited current of the sensor was shown to be dependent on the metal-ion concentration in solution as given by (4). The current was measured as a function of time using a long cantilever beam with a 30 by 36 Pt 200 electrode at the tip. The experiment was performed by putting a solution on the canfew drops of various concentration tilever and then applying a deposition voltage of 1.5 V. Fig. 8 conshows the current–time (I–t) curves for a range of centration from 0.1 to 1 mM. The I–t curves exhibited the general inverse root time dependence predicted by (4) and show an increase in steady-state current with concentration. Fig. 9 is a
1304
JOURNAL OF MICROELECTROMECHANICAL SYSTEMS, VOL. 15, NO. 5, OCTOBER 2006
Fig. 7. PV cell short circuit current versus light intensity. Fig. 9. Plot of the steady-state diffusion current and total integrated charge versus concentration during electrodeposition in various CuSO solutions.
Fig. 8. I–t curves of Cu electrodeposition on 200 various CuSO solutions.
m cantilever structure in
plot of the steady-state electrodeposition current and the total integrated charge versus solution concentration. Both are linearly dependent on ion concentration as expected. To give an idea of the current requirement of the sensor, during electrodeposition solution, the average deposition current was in 1 mM 35 nA with a peak current of 0.16 . Mass detection was performed using a resonant cantilever structure. The resonant frequency was measured before and after electrodeposition in order to determine the amount of copper deposited at the tip of the beam. Measurement of the mechanical resonance was performed using an optical lever technique [26]. A laser was focused onto the cantilever electrode. The reflected light illuminated a position sensitive detector (PSD) which measured the cantilever displacement. The PSD signal was sent to a current amplifier then a spectrum analyzer, providing the frequency spectrum of the vibrating cantilever. Vibration of the cantilever was achieved due to thermal mechanical and ambient noise and therefore no active actuation mechanisms were used. Fig. 10 provides the results of the mass detection experiment solution for after electrodeposition in 0.25 and 1 mM 600s as previously shown in Fig. 8. In 1 mM solution,
Fig. 10. Frequency spectrum of 200 m long cantilevers before and after metal electrodeposition in 0.25 mM and 1 mM CuSO solution. A frequency shift of 580 Hz and 2.62 kHz were observed, respectively.
the resonant frequency of the cantilever was seen to shift from 31.848 kHz prior to deposition down to 28.968 kHz after metal deposition for a total freqency shift of 2.88 kHz. In 0.25 mM solution, the frequency shifted from 31.720 to 31.140 kHz for a total shift of 580 Hz. In addition, the Q of the cantilever was determined to be approximately 40. The resonant frequency shift for devices operated in solutions with a concentration less than 0.25 mM was below detectable limits given the integration time of the device used and the sensitivity limitations imposed by the design. In solutions above 1 mM, the metal deposited at the cantilever tip became too rough for optical resonance measurements. This limitation of sensor dynamic range can be improved by redesigning the sensor and separating the electrodeposition and optical measurement areas. Finally, Fig. 11 shows electrodeposition at the tip of an unreleased cantilever beam. This deposition was driven by a PV cell array under microscope level illumination.
SUPINO AND TALGHADER: METAL-ION CONCENTRATION IN FLUIDS
Fig. 11. Scanning electron micrograph of an unreleased cantilever beam with copper electrodeposited at tip.
V. DISCUSSION A power supply consisting of three series connected PV cells provides ample current and sufficient voltage to drive an electrochemical reaction. The PV cell array was capable of producing of current with a maximum power output of 4.02 0.4 under light intensities of about 80 . This level of current is more than was needed to drive an electrochemical reaction in a solution with a 1 mM ion concentration which required a steady-state current of about 35 nA and average power of 50.2 nW. To provide a comparison between the light intensities used in PV cell evaluation, the intensity of direct sunlight is on the . It should be noted that this intensity also order of 1000 includes IR radiation which lies outside the absorption spectrum of the PV cell array. A total efficiency of optical power to electrical power conversion can be determined to be about 15% for the wavelength of light used in this characterization. This does not claim to be the solar cell efficiency as the tests were performed using a 660-nm laser source as opposed to a calibrated broadband illumination source. Due to variations in efficiency with wavelength, the overall array efficiency under broadband illumination conditions will be lower than 15%. Fig. 5 provides a comparison between illuminating the PV array with monochromatic laser illumination and a broadband microscope light source. Rough calculations treating the microscope light source as a black-body radiator and integrating over the wavelength sensitivity of a silicon photodetector gives an intensity of . The short circuit current under these conditions 4000 was approximately the same as the short circuit current for of laser light, or a PV cell array illuminated at 2760 approximately 50% lower efficiency. A better comment on efficiency can be found by comparing the external quantum efficiency of the PV cell array with a standard silicon photodiode. The total external quantum efficiency of the PV cell array can be calculated to be about 31%. This compares to typical quantum efficiencies of between 80–90% for standard photodiode devices [11], [27]. It is likely that most of this difference is due to the short absorption length of our cells. A secondary cause of the lower PV array efficiency can be attributed to a non-optimum anti-reflection coating. Thin film calculations show that the reflectivity of 660nm light off of the
1305
interface to be about 8.4%. Typical optimized AR coatSi/ ings for modern photodiodes often achieve values below 1%. Even with reduced efficiency, the current PV cell design is sufficient for use in an electrochemical based sensor without further optimization. The mass detection experiment produced frequency shifts of 2.88 kHz and 580 Hz for deposition in 1 mM and 0.25 mM solutions, respectively. Substituting the shifts in frequency due to mass deposition into (7), a deposited mass of 1.4 ng is determined for the 1 mM case and 0.28 ng for the 0.25 mM case. These calculations assume a Young’s modulus of 100 GPa and for low-stress [28]. In coma density of 2850 parison, integration of the curves in Fig. 8 corresponding to the two measured concentrations gives a total charge transferred of and 6.43 for operation in 1 and 0.25 2.01 mM concentrations, respectively. From Faraday’s Law, the total electroplated metal mass should have been 5.0 and 1.6 ng given the amount of charge transferred during electrodeposition. Differences between the deposited mass as calculated using the electrodeposition current curves and the measured frequency shifts could be due to a parallel electrochemical process such as the electrolysis of water during electrodeposition or due to a current leakage path. If there is a breach in the PECVD silicon nitride layer insulating the PV cells and interconnect metal, a leakage path will exist taking away from current participating in electrodeposition at the cantilever tip. Furthermore, using (8) with the measured frequency shifts, ions are deterconcentrations of 0.26 and 0.052 mM of mined for the 1 and 0.25 mM solutions. This calculation as. sumes a somewhat difficult to define electrode radius of 18 The empirical constant in (8) was determined to be 4.0 by performing electrodeposition current measurements in a solution of known concentration and fitting the theoretical diffusion current (4) to the experimentally measured diffusion current. Equation (8), derived using the diffusion current (4), also assumes a circular microelectrode, which differs from the cantilever’s rectangular electrode geometry. The differences between the detected concentrations are metal-ion concentration to the actual again theorized to be due to a parallel electrochemical reaction or leakage path reducing the total current participating in metal electrodeposition. If the response of the metal-ion sensor is assumed to be linear with concentration and the results are normalized to the 1 mM concentration solution, the measured result for the 0.25 mM concentration would be 0.19 mM, a 24% error. Additionally, it should be mentioned that the current geometry and resonance measurement technique is suboptimal. The measured frequency shift of 580 Hz was larger than the minimum discernable frequency shift of approximately 200–400 Hz noticed during experimentation. From this observation, it is noted that the minimum detectable mass for this resonant mass sensor is on the order of about 0.1–0.2 ng. These resonant frequency measurements were carried out in air where the cantilever Q was approximately 40. Moving such experiments to vacuum should allow an increase in cantilever Q to at least 1000 if not orders of magnitude higher. Such an improvement would lead to an increase in sensitivity of about 25 times at minimum and possibly over 250 times depending on the Q of the cantilever in vacuum. In addition, the large size of these cantilever
1306
JOURNAL OF MICROELECTROMECHANICAL SYSTEMS, VOL. 15, NO. 5, OCTOBER 2006
TABLE I HALF-CELL REDUCTION POTENTIALS FOR COMMON METAL IONS. ADAPTED FROM [20]
REFERENCES [1] B. A. Warneke and K. S. Pister, “MEMS for distributed wireless sensor networks,” in Proc. ICECS 2002. 9th IEEE Int. Conf. Electron. Circuits Syst., Sep. 2002, vol. 1, pp. 291–294. [2] B. Warneke, B. Atwood, and K. S. Pister, “Smart dust mote forerunners,” in Tech. Dig. MEMS 2001. 14th IEEE Int. Conf. Micro Electro Mech. Syst., Jan. 2001, pp. 357–360. [3] B. A. Warneke, M. D. Scott, B. S. Leibowitz, L. Zhou, C. L. Bellew, J. A. Chediak, J. M. Kahn, K. S. P. Bernhard, and E. Boser, “An autonomous 16 solar-powered node for distributed wireless sensor networks,” in Proc. IEEE Sens. 2002: First IEEE Int. Conf. Sens., Jun. 2002, vol. 2, pp. 1510–1515. [4] B. Warneke, M. Last, B. Liebowitz, and K. S. Pister, “Smart dust: Communicating with a cubic-millimeter computer,” Computer, vol. 34, no. 1, pp. 44–51, Jan. 2001. [5] A. Mason, N. Yazdi, A. Chavan, K. Najafi, and K. D. Wise, “A generic multielement microsystem for portable wireless applications,” Proc. IEEE, vol. 86, pp. 1733–1746, Aug. 1998. [6] L. Zhou, J. M. Kahn, and K. S. Pister, “Corner-cube restroreflectors based on structure-assisted assembly for free-space optical communication,” J. Microelectromech. Syst., vol. 12, pp. 233–242, Jun. 2003. [7] M. V. Kruger, M. H. Guddal, R. Belikov, A. Bhatnagar, O. Solgaard, C. Spanos, and K. Poola, “Low power wireless reasout of autonomous sensor wafer using MEMS grating light modulator,” in Proc. IEEE/ LEOS Int. Conf. Opt. MEMS, Aug. 2000, pp. 67–68. [8] M. Suster, D. J. Young, and W. H. Ko, “Micro-power wireless transmitter for MEMS sensing and communication applications,” in Microelectromech. Syst. Conf., Aug. 2001, pp. 25–28. [9] M. Suster, W. K. Ko, and D. J. Young, “An optically powered wireless telemetry module for high-temperature MEMS sensing and communication,” J. Microelectromech. Syst., vol. 13, pp. 536–541, Jun. 2004. [10] G. Asada, A. Burstein, D. Chang, M. Dong, M. fielding, E. Kruglick, J. Ho, F. Lin, T. Lin, H. Marcy, R. Mukai, P. Nelson, F. Newberg, K. S. Pister, G. Pottie, H. Sanchez, O. M. Stafsudd, S. Valoff, G. Yung, and W. J. Kaiser, “Low power wireless communication and signal processing circuits for distributed microsensors,” in Proc. 1997 IEEE Int. Symp. Circuits Syst. Circuits and Systems in the Information Age, Jun. 1997, vol. 4, pp. 2817–2820. [11] C. L. Bellew, S. Hollar, and K. S. Pister, “An soi process for fabrication of solar cells, transistors and electrostatic actuators,” in The 12th Int. Conf. Solid State Sens. Actuators and Microsyst., 2003, vol. 2. [12] J. B. Lee, Z. Chen, M. G. Allen, A. Rohatgi, and R. Arya, “A high voltage solar cell array as an electrostatic MEMS power supply,” in IEEE Proc., ser. Micro Electro Mech. Syst.. : IEEE, 1994, pp. 331–336. [13] D. M. Ryan, R. M. LaFollette, and L. Salmon, “Microscopic batteries for micro electromechanical systems (MEMS),” in Proc. Intersoc. Energy Conv. Eng. Conf., Jul. 1997, vol. 1, pp. 77–82. [14] M. Chiao, K. B. Lam, and L. Lin, “Micromachined microbial fuel cells,” in Proc. IEEE Micro Electro Mech. Syst. (MEMS), Jan. 2003, pp. 383–6. [15] H. Kulah and K. Najafi, “An electromagnetic micro power generator for low-frequency environmental vibrations,” in Proc. IEEE Int. Conf. Micro Electro Mech. Syst. (MEMS), Jan. 2004, pp. 237–240. [16] B. H. Starj, P. D. Mitcheson, P. Miao, T. C. Green, E. M. Yeatman, and A. S. Holmes, “Power processing issues for micro-power electrostatic generators,” in Proc. IEEE 35th Annu. Power Electron. Special. Conf., Jun. 2004, vol. 6, pp. 4156–4162. [17] P. B. Koeneman, I. J. Busch-Vishniac, and K. L. Wood, “Feasibility of micro power supplies for MEMS,” J. Microelectromech. Syst., vol. 6, pp. 355–362, Dec. 1997. [18] J. Santini, M. Cima, and R. Langer, “A controlled-release microchip,” Nature, vol. 397, no. 6717, pp. 335–8, Jan. 1999. [19] G. T. Kovacs, C. W. Storment, and S. P. Kounaves, “Microfabricated heavy metal ion sensor,” Sens. Actuators B, vol. 23, pp. 41–47, 1995. [20] P. H. Rieger, Electrochemistry, 2nd ed. New York: Chapman and Hall, 1994. [21] K. B. Oldham and J. C. Myland, Fundamentals of Electrochemical Science. New York: Academic, 1994. [22] T. Drews, S. Krishnan, J. Alameda, D. Gannon, R. Braatz, and R. Alkire, “Multiscale simulations of copper electrodeposition onto a resistive substrate,” IBM J. Res. Devel., vol. 49, no. 1, pp. 49–63, Jan. 2005. [23] S. Timoshenko, D. H. Young, and W. Weaver, Vibration Problems in Engineering, 4th ed. New York: Wiley, 1974.
mm
structures leads to an increase in the mass of the cantilever and a subsequent decrease in sensitivity. Reduction in cantilever size would not only reduce physical dimensions of the sensor, but would also lead to an increase in sensitivity. Finally, a few notes regarding sensor selectivity are in order. From well tabulated half-cell potential tables [20], it can be seen that a number of different metals can be electrodeposited at voltages near the electrodeposition potential of copper, as shown , , and , in Table I. These metal ions include among others. When the metal-ion sensor is exposed to an environment with multiple ion species present, more than one type of metal ion will be deposited, and selectivity is compromised. Therefore, the metal-ion sensor as presented is better suited to general heavy metal-ion detection than as a sensor used to discern individual species. With little added complexity to the sensor architecture, the cantilever resonant mass detection can be coupled with anodic stripping voltammetry (ASV) techniques [19] to enable specific ion selectivity. If the sensor were connected to an external voltage source, the metal at the tip of the cantilever could be oxidized in an electrolyte solution. By scanning through a range of potentials, current peaks occur at various potentials indicating the presence of different ions. The magnitude of the current peaks allow the determination of the ratio of metal ions present on the electrode. Typical microfabricated ASV devices employ an array of microelectrodes to increase the total amount of metal deposited and therefore increase the sensitivity of the device. Incorporating ASV with a cantilever resonant mass detection device would allow a single electrode to perform sensitive mass measurements to measure small metal-ion concentrations with the added ability of detecting multiple species. VI. CONCLUSION An extremely low-power electrochemical sensor was demonstrated. The device scavenges optical power from its surroundings in order to power an electrochemical reaction. The electrochemical reaction is used to sense the metal-ion concentration in a liquid in which the sensor is immersed. The integrated . The metal-ion sensor had a total volume below 0.046 PV cells driving the electrochemical process could provide over of current under laser excitation, but the electrochem10 ical system operated at power levels as low as 50.2 nW. This power level is easily supplied by scavenged room light. Individual sensors were capable of sensing metal-ion concentration with a factor of two, making them suitable for spot checks of environmental and industrial processes.
SUPINO AND TALGHADER: METAL-ION CONCENTRATION IN FLUIDS
[24] C. Harris and A. Piersol, Eds., Harris’ Shock and Vibration Handbook, 5th ed. New York: McGraw-Hill, 2002. [25] R. Buser, B. Stauffer, and N. de Rooij, “Realization of mesa array in (001) oriented silicon wafers for tactile sensing applications,” in The Electrochemical Society, Fall Meeting, Oct. 1986, vol. 86, p. 879. [26] G. Meyer and N. Amer, “Novel optical approach to atomic force microscopy,” Appl. Phys. Lett. , vol. 53, no. 12, pp. 1045–1047, 1988. [27] M. Fukuda, Optical Semiconductor Devices. New York: Wiley, 1999. [28] L. Kiesewetter, J. Zhang, D. Houdeau, and A. Steckenborn, “Determination of young’s moduli of micromechanical thin films using the resonance method,” Sens. Actuators A, vol. 35, no. 2, pp. 153–159, Dec. 1992. Ryan Supino (S’03) received the B.S. degree in electrical engineering (summa cum laude) from the University of Minnesota, Twin Cities, in 2000. He was awarded an NSF Graduate Fellowship and received the M.S. and Ph.D. degrees in electrical engineering also from the University of Minnesota in 2003 and 2006, respectively. His graduate research was focused on MEMS technology, where he researched a variety of topics ranging from thermal management in micromechanical devices to the development of active microsensors for remote environments. Currently, he is a Research Scientist at Honeywell International, Plymouth, MN, where he is involved in the research and development of MEMS inertial sensor products.
1307
Joseph J. Talghader (S’93–M’95) received the B.S. degree in electrical engineering from Rice University, Houston, TX, in 1988. He was awarded an NSF Graduate Fellowship and attended the University of California at Berkeley, where he received the M.S. and Ph.D. degrees in 1993 and 1995, respectively. From 1992 to 1993, he worked at Texas Instruments as a Process Development Engineer. After graduating from Berkeley in 1995, he joined Waferscale Integration. In 1997, he joined the faculty at the University of Minnesota, Minneapolis, where he is now an Associate Professor. Dr. Talghader has received 3M Nontenured Faculty Awards on three occasions. He has served on various program committees and reviews, including service as Program chair of the 2003 IEEE/LEOS Optical MEMS Conference and as Guest Editor of the IEEE JOURNAL OF SELECTED TOPICS IN QUANTUM ELECTRONICS 2004 Special Issue on Optical Microsystems. He will be Conference Chair of the 2006 IEEE/LEOS Optical MEMS Conference at Yellowstone/Big Sky.
1299
Micromachined Particles for Detecting Metal-Ion Concentration in Fluids Ryan Supino, Student Member, IEEE, and Joseph J. Talghader, Member, IEEE
Abstract—A remote microfluidic metal-ion sensor is developed using an electrochemical system integrated with a compact photovoltaic cell power supply. The sensor is designed to detect the sum of metal ions in a remote environment. The sensor uses electrodeposition to remove ions from the fluid around the sensor and deposit them on an electrode at the tip of a cantilever. The electrodeposited mass changes the resonant frequency of the cantilever, which can be determined upon read-out. The sensor is designed to be dropped in liquids or flow through microfluidic systems and can be used in parallel with many other similar sensors. The photovoltaic cells are directly integrated on the device and are capable of producing tens of microwatts of power at about 15% efficiency with laser excitation. However, the sensor operates at power levels of 50 nW with small voltages and currents using only scavenged daylight or room light. The complete device is integrated into a total volume below 3 , which is more than two orders of magnitude smaller 0.046 than other remote electrically-powered sensors reported to date. Although it is expected that multiple devices will be used in parallel to gain statistical data, individual particles detect metal-ion concentration within 24% of the actual concentration, making them suitable for safety testing and endpoint monitoring among other applications. [1714]
mm
Index Terms—Electrochemical, MEMS power supply, power scavenging, remote sensor.
I. INTRODUCTION XTENSIVE work has been performed in an effort to develop remote, self-powered microdevices [1]–[5]. Many of these devices have achieved impressive complexity, integrating a power source with sensing, communications, and even processing capabilities. Due to this level of complexity, especially the addition of communications and processing capabilities, the power requirements of such devices are often relatively substantial and involve dedication of large portions of device area to power supplies and communication, which places strict lower limits on device volume. As such, the smallest reported electriin cally powered remote device is only approximately 16 volume [3]. The primary difficulty associated with microscopic remote sensors is the inherent difficulty communicating with such a device. A number of different wireless communication schemes
E
Manuscript received October 25, 2005; revised March 16, 2006 Subject Editor M. Mehregany. R. Supino was with the Department of Electrical and Computer Engineering, University of Minnesota, Minneapolis, MN 55455 USA. He is now with Advanced Sensors and Microsystems, Honeywell International, Plymouth, MN 55411 USA (e-mail: [email protected]) The authors are with the Department of Electrical and Computer Engineering, University of Minnesota, Minneapolis, MN 55455 USA (e-mail: [email protected]). Digital Object Identifier 10.1109/JMEMS.2006.880246
have been developed for remote sensors typically based on either optical [6], [7] or RF [8]–[10] designs. Both types of communication schemes have drawbacks. Wireless RF communications tend to require a fairly large area antenna, on the order of , and typically have short data transmission distances on a the order of a few meters. Optical communication technologies are often designed to be passive (i.e., without an onboard light source). While they typically have longer data transmission distances, they require a large amount of sensor area to overcome diffraction limits on coupling and can often be fairly complex in design. Both types of communication schemes can achieve relatively low power consumption, on the order of less than 100 , however, this level of power usage is still difficult to pro. vide onboard a device with a total volume less than 1 Remote microdevices not only need a means of communicating data from the sensor to a receiving station, but also need a means of powering these operations. A variety of work has been performed on micropower generation for remote microsensors. Such work has included microsolar power supplies [11], [12], microbatteries [13], microfuel cells [14], low-frequency vibration energy scavenging [15], [16], in addition to others [17]. Many of these micro power supply devices are capable of pro. ducing modest amounts of power, on the order of 1–100 Unfortunately, most of these power sources are quite large, often or more. consuming volumes of 10 For a variety of applications including environmental sensing, drug or chemical delivery, industrial process flow monitoring, and microfluidic device evaluation, the development of extremely low-power fluidic sensors would be desirable. This paper introduces an electrochemical device architecture to detect metal-ion concentrations in electrolytes. The advantage of this type of architecture is that electrolytic reactions operate at low-voltages and require extremely low currents. The use of electrolysis allows the creation of metal-ion sensors using electrodeposition mechanisms or the creation of drug and chemical delivery systems using the electrochemical dissolution of thin film membranes [18]. In this paper, a fluidic metal-ion sensor is developed using electrodeposition as a sensing mechanism [19]. Since communication requires large amounts of sensor area, a means of storing data onboard the sensor could lead to large reductions in sensor volume. The use of electrodeposition allows the sensor device to store metal-ion concentration information on board. The electrodeposition sensing mechanism requires very low voltage and current operation allowing power to be delivered by three onboard photovoltaic (PV) cells. The sensor itself is comprised of a cantilever acting as a resonant mass sensor with an electrode at the tip where metal from the surrounding environment is electrodeposited as driven by the PV cell array. The total mass de-
1057-7157/$20.00 © 2006 IEEE
1300
JOURNAL OF MICROELECTROMECHANICAL SYSTEMS, VOL. 15, NO. 5, OCTOBER 2006
tabulated half-cell reactions. Summing these half-cell reactions can be deteran equilibrium cell potential of mined for the deposition of copper ions [20]. Under non-equilibrium conditions, or more specifically, during current flow, other modifications to the potential are necessary. These additional modifications, or overpotentials, can be added to the equilibrium cell potential giving [21]
(2)
Fig. 1. Illustration of copper electrodeposition. Cu is reduced at the cathode plating copper on the electrode surface and H O is oxidized at the anode giving off O gas.
posited can be determined by making resonant frequency measurements of the cantilever beam after the sensor is removed from its liquid and correlating this information to the concentration of ions in the environment. Collecting the sensors involves taking a small volume of the electrolyte. Since a large number of sensors can be dispersed at once, statistically a small sample will have at least a few sensors. The sensors are dried onto a convenient surface (silicon and lab paper were used in this work) where they could be scanned by a laser beam to determine added mass. II. BACKGROUND The microparticle metal-ion sensor will be composed of three important elements, the electrochemical based sensor, a mass sensitive resonant cantilever structure, and a PV cell array power supply. This section will discuss the theoretical background associated with the electrochemical sensing system and the mass sensitive resonant cantilever. A. Electrochemistry Electrodeposition is the sensing mechanism employed by the metal-ion sensor. When driven by a power source, the device electrodeposits metal ions from the solution onto the tip of a cantilever beam. Such a system represents an electrochemical cell. This cell is composed of two electronic conductors, the anode and cathode, and an ionic conductor, the fluidic medium. When a voltage is applied across the anode and cathode, a system of oxidation-reduction reactions occurs. As an example, let two electrodes be immersed in an electrolyte containing copper ions as shown in Fig. 1. The reduction of copper occurs at the cathode and the oxidation of water takes place at the anode. The total chemical reaction is given as follows: (1) To determine the voltage necessary to drive the electroplating of copper, the chemical reaction can be broken down into well
is the equilibrium cell potential, is the ideal gas where constant, is absolute temperature, is the number of moles of electrons transferred in the reaction as given by (1), and is , Faraday’s constant. The overpotential terms are given by , and , while is the concentration of copper ions in solution. The concentration is in reference to a standard is a dimensionless term. concentration of 1 M, therefore , originates from the reThe first overpotential term, sistivity of the fluidic medium and the external electrical circuit. The ohmic overpotential is a geometrically dependent term and must be calculated over the complex geometry of the elec, detrochemical device. The second over-potential term, scribes the electron transfer kinetics associated with a reaction. In an electrochemical reaction, electrons are transferred to the reactant in the case of a reduction reaction and away from the reactant in the case of an oxidation reaction. The rate at which electrons are transferred, and thus the rate at which species are reduced or oxidized in an electrochemical reaction is dependent on the potential dropped across the electrode/electrolyte interface. , is a concentration dependent overpotential that Finally, is due to the consumption of metal ions at the cathode. As metal ions are plated at the surface, the solution adjacent to the electrode becomes depleted of metal ions and a concentration gradient is formed. The establishment of a concentration gradient sets up a diffusion limitation to the deposition reaction possibly limiting the rate of metal deposition. This deposition limitation will be discussed as the mechanism governing the operation of the metal-ion sensor device. and are considFor the purposes of this work, ered negligible. Consequently, the current limitations imposed by ohmic loss or reaction rate kinetics are expected to be large compared to the limitation on deposition current imposed by metal-ion diffusion. is exFirst, let us consider the ohmic overpotential. pected to be small due to the low-current operation of the sensor. For extremely low concentration solutions and low-current operation, rough calculations for the sensor geometry show that the expected overpotential due to ohmic loss to be less than 0.1V. Variations in the conductivity of the solution should not have a major impact on cell potential for the current levels anticipated. , is related to reaction Since the activation overpotential, rate constants, we expect it to be very dependent on temperature. However, it can be shown that the electron transfer kinetics for copper electrodeposition are very fast and therefore the limitation to current imposed by the reaction rate is much larger than
SUPINO AND TALGHADER: METAL-ION CONCENTRATION IN FLUIDS
1301
the current limit imposed by the concentration gradient. Specifically, the current limitation due to electron kinetics is predicted using the Butler-Volmer relationship [21]
The final important electrochemical theory for the sensors is a means of relating the total amount of deposited metal to the current flowing in the system. The relationship between reduced or oxidized metal and total charge transferred is given as
(3)
(5)
where is known as the exchange current and represents the anodic and cathodic currents flowing at equilibrium, is the are constants total overpotential applied to the cell, and that describe the symmetry of the forward and reverse reactions. Using (3) in conjunction with typical copper electrodeposition is capable of flowing constants predicts that a current of 1.6 with an applied over potential of only 200 mV [22]. This current is much higher than is observed in devices used in this study, is not the rate limiting mechanism. indicating that is the basis for the metal-ion The third overpotential sensor operation. This overpotential is formed due to the depletion of reactant species in the proximity of the working electrode, which leads to a diffusion-based limitation on the transport of metal ions from the bulk solution to the cathode surface. Instead of viewing this mechanism as an overpotential, it can also be expressed as a diffusion limited current. For microelectrodes, the diffusion limited current is given as [19]
is the atomic mass, is current flow, is total rewhere is Faraday’s constant, and is the number of action time, electrons transferred per unit reaction. This equation is known as Faraday’s Law of Electrolysis. This relationship assumes that there are no parallel reactions occurring, such as the electrolysis of water, during the electrochemical reaction.
(4)
where is the beam thickness, is the length, is Young’s modulus, and is the material density of the cantilever. This resonant frequency equation is derived by solving the wave partial differential equation using boundary conditions of a beam with rectangular cross section and uniform thickness and width. As mass is added to the cantilever, a shift in the resonant frequency of the structure occurs. The change in resonant frequency can be approximated using Rayleigh’s Method [24] and is given by
where is the number of electrons transferred during the electrochemcial reaction, two in the case of copper ion reduction, is Faraday’s constant, is the electrode area, is the metal-ion diffusivity, is the concentration of ions in the bulk fluid is the electrode radius, and is an empirical conmedium, stant that represents geometrical deviations from an ideal circular microelectrode. Using typical copper electrodeposition parameters with a copper ion concentration of 1 mM and an electrode size of 30 it can be shown that the diffusion limited current 30 is on the order of 7 nA [22], which is almost three orders of magnitude lower than the current limit due to electron kinetics. Therefore, the device can be assumed to be diffusion limited at concentrations at or below 1 mM. The diffusion current expressed in (4) is directly dependent on the concentration of metal ions in solution. If an electrochemical device is operated with a power source that can handle the voltage requirements in addition to current levels greater than the diffusion limit, then the current will be limited by the diffusion process and will be proportional to the metal-ion concentration. In the case of a metal-ion sensor, this allows the integration of concentration data over a long period of time in order to measure small concentrations of metal ions. It is also important to notice that there is an initial time dependent transient in the diffusion current. This transient is due to the expansion of the depletion region and reaches a constant level as a stable depletion hemisphere forms. As long as the integration time of the sensor is much greater than this transient time, the error associated with the initial increased deposition rate will be negligible. For shorter integration times, the transient can be included directly since it has a well-quantified form.
B. Resonant Cantilever Mass Sensing The previous discussion of electrochemical theory is applicable to any type of electrochemical microparticle. This section considers the specifics of a sensor using cantilever mass detection. To begin, the fundamental resonant frequency of an undamped cantilever beam with one end clamped and the other end free is given by [23] (6)
(7) where is the shift in resonant frequency of the cantilever is the added mass, is the mass of the cantilever bebeam, fore loading, and is the cantilever resonant frequency before loading. The change in resonant frequency given by (7) assumes that the mass is added in a point at the tip of the cantilever. If the added mass is spread out along the cantilever length, a reduction in the sensitivity of the frequency shift due to mass loading occurs. Therefore, it is important to ensure that the mass loading is confined to a region at the tip of the cantilever beam. The resonant frequency shift can be measured in order to determine the total amount of mass added to the beam. In the case of a metal-ion sensor, this would be the total mass of metal electrodeposited at the cantilever’s tip. With an understanding of the frequency sensitivity of a cantilever beam to mass loading, the expected relationship between the resonant frequency shift of a metal-ion sensor and the concentration of ions in the fluidic medium can be determined. Combining (4), (5), and (7) the following relationship is obtained. (8)
1302
JOURNAL OF MICROELECTROMECHANICAL SYSTEMS, VOL. 15, NO. 5, OCTOBER 2006
Fig. 2. Drawing of complete sensor device. (a) PV cell mesa. (b) Cantilever with electrode at tip connected to PV cell array. (c) Counter electrode connected to PV cell array.
In (8), it is assumed that for typical sampling times, the diffusion transient contributes a negligible amount to the overall charge transfer during metal deposition. Therefore, the transient portion of (4) is ignored. This is a reasonable assumption as long as sampling times are on the order of minutes. Typical diffusion , coefficients for metal ions in solution are around leading to a transient decay time on the order of 10 s. For increased accuracy during shorter sampling times, it is possible to include the transient portion of the diffusion curve. Additionally, it is assumed that sufficient light is present such that electrodeposition occurs the entire time the device is in solution such that any etching of the metal that may occur is negligible. This is a reasonable assumption as long as the device is not exposed to extremely acidic environments and is operating for the majority of the exposure time. III. DESIGN AND FABRICATION Some of the basic mechanisms of an electrochemical based microsensor were discussed in the previous section in addition to some specifics of the sensing mechanism used by the metal-ion sensor device. This section will present greater detail of the sensor architecture and discuss the fabrication process. The metal-ion sensor is an integrated device composed of a power supply combined with a cantilever resonant mass sensor. The cantilever has a Pt working electrode at the tip. Surrounding the cantilever is a Pt counter electrode. The cantilever tip is connected to the cathode side of a photovoltaic (PV) cell array and the counter electrode is connected to the PV cell anode. The PV cell array drives the electrodeposition process, gathering metal ions from solution and plating them on the electrode at the tip of the cantilever. Fig. 2 gives an illustration of the basic architecture of the sensor structure. The power supply is created using a unique SOI process. Individual PV cells are created using a simple - junction design. The PV cells are isolated from one another using anisotropic etching of the silicon while stopping on the SOI buried oxide layer. Corner compensation techniques are used to create square shaped mesas for the PV cells [25]. The 54.7 sloped side walls of the PV cell mesas allow easy metal coverage between the separate mesas in order to create a power supply capable of producing about 1.5 V. The metal interconnect lines are insulated from the PV cell using a thermal oxide layer and a silicon nitride layer. The thermal oxide layer provides two additional benefits. First, it helps to tie up surface states reducing the recombination
Fig. 3. Process flow of complete integrated metal-ion sensor device.
of photogenerated electron-hole pairs at the surface of the diffusion. Secondly, it provides a simple antireflection coating increasing the overall light gathering efficiency of the PV cells. thick with a (100) The p-type device layer of the SOI is 20 is deposited using orientation. A thin layer of low-stress low pressure chemical vapor deposition (LPCVD) and serves as a diffusion mask for the PV cell array. This film is patterned, regions of the PV cell where subsequently a defining the thick phosphosilicate glass (PSG) film is deposited to serve as the diffusion dopant source. The wafer is annealed at 1000 for a period of one hour. Subsequently, the films are stripped and layer is deposited and patterned another thin low-stress for PV cell mesa definition. Separate PV cells are defined using a KOH etch to remove the silicon between cells. The KOH etch is terminated on the buried oxide of the SOI wafer and uses 45% to etch through the 20- -thick device KOH solution at 80 etch mask is stripped. layer. Subsequently, the Next, a 1000 thick thermal oxide is grown at 1050 in a dry oxygen ambient. The oxide film is used to tie up surface region of the PV cell which increases efficiency. states in the The cantilever structural layer is next added by depositing and . Alupatterning another 6000 thick layer of low-stress minum interconnect metal is deposited by e-beam evaporation to form an ohmic contact between PV and is annealed at 450 cell devices. A 100- layer of Ti is followed by a 300- layer of Pt for both the counter and working electrodes. Fig. 3 gives a depiction of the process flow. The Pt metal serves as the deposition area on the cantilever beam, therefore it is necessary to insulate the metal connecting
SUPINO AND TALGHADER: METAL-ION CONCENTRATION IN FLUIDS
1303
Fig. 4. Photomicrograph of completed metal-ion sensor.
the tip of the cantilever to the PV cell so that no metal is deposited in this region. If metal is deposited along the entire length of the cantilever, the effect of mass loading is counteracted by an increase in the stiffness of the beam. This leads to a reduction in the resonant frequency shift, or sensitivity, due to the added mass of the deposited metal. To prevent this, a 3000 thick layer is deposited using a plasma vapor deposition system and is patterned to cover all metallization other than the counter and working electrodes. After completion of the front side of the wafer, a thick photoresist layer is deposited on the front in order to protect the sensors during backside processing. The backside is then patterned using a thick photoresist in order to define sensor separation and cantilever pit areas. This pattern is etched using a deep reactive ion etch (DRIE) through the full thickness of the SOI handle thick, and terminates on the buried layer, approximately 300 oxide layer. The buried oxide layer is then removed in the open areas using 10:1 buffered oxide etchant for a period of 20 min. Care is taken to ensure that the etch is stopped after the oxide is removed from the sensor separation areas and the cantilever pit areas while not completely undercutting the area between the PV cell mesas and the silicon handle layer. Finally, the sensors are released by immersing the wafer in acetone dissolving the photoresist on the front side holding the sensors together. Fig. 4 shows an optical micrograph of the completed device. IV. EXPERIMENTAL RESULTS Each of the various aspects of the metal-ion sensor were evaluated. This includes the PV cell characteristics, the current demands of an electrochemical sensing or delivery mechanism, and resonant mass sensing driven by electrodeposition. In analysis of the power supply, arrays consisting of three PV cells connected in series were tested. These arrays were separate test structures from the arrays integrated with the sensor and were about four times larger in area. Individual cells had an and the total PV cell mesa size was active area of 0.032 on a side. The PV cell array was characterized using 228 a laser diode with a wavelength of 660 nm. Fig. 5 shows the general I–V behavior of the PV cell array under varying laser illumination intensities. Also included in Fig. 5 is an I–V plot of
Fig. 5. I–V curve of PV cell array at three distinct laser intensity levels and one under white light illumination.
Fig. 6. PV cell array open circuit voltage versus light intensity.
a single photovoltaic cell under broadband microscope illumination. In addition to measuring the general I–V behavior of the PV cell arrays, measurements of open cell voltage and short circuit current were made over a range of light intensities. Fig. 6 shows that the open cell voltage varies from about 1 V with a light to 1.52 V at an intensity of 2780 . intensity of 79.8 This corresponds to a range of short circuit currents from 0.4 to 13.2 for the respective light intensities (Fig. 7). The diffusion limited current of the sensor was shown to be dependent on the metal-ion concentration in solution as given by (4). The current was measured as a function of time using a long cantilever beam with a 30 by 36 Pt 200 electrode at the tip. The experiment was performed by putting a solution on the canfew drops of various concentration tilever and then applying a deposition voltage of 1.5 V. Fig. 8 conshows the current–time (I–t) curves for a range of centration from 0.1 to 1 mM. The I–t curves exhibited the general inverse root time dependence predicted by (4) and show an increase in steady-state current with concentration. Fig. 9 is a
1304
JOURNAL OF MICROELECTROMECHANICAL SYSTEMS, VOL. 15, NO. 5, OCTOBER 2006
Fig. 7. PV cell short circuit current versus light intensity. Fig. 9. Plot of the steady-state diffusion current and total integrated charge versus concentration during electrodeposition in various CuSO solutions.
Fig. 8. I–t curves of Cu electrodeposition on 200 various CuSO solutions.
m cantilever structure in
plot of the steady-state electrodeposition current and the total integrated charge versus solution concentration. Both are linearly dependent on ion concentration as expected. To give an idea of the current requirement of the sensor, during electrodeposition solution, the average deposition current was in 1 mM 35 nA with a peak current of 0.16 . Mass detection was performed using a resonant cantilever structure. The resonant frequency was measured before and after electrodeposition in order to determine the amount of copper deposited at the tip of the beam. Measurement of the mechanical resonance was performed using an optical lever technique [26]. A laser was focused onto the cantilever electrode. The reflected light illuminated a position sensitive detector (PSD) which measured the cantilever displacement. The PSD signal was sent to a current amplifier then a spectrum analyzer, providing the frequency spectrum of the vibrating cantilever. Vibration of the cantilever was achieved due to thermal mechanical and ambient noise and therefore no active actuation mechanisms were used. Fig. 10 provides the results of the mass detection experiment solution for after electrodeposition in 0.25 and 1 mM 600s as previously shown in Fig. 8. In 1 mM solution,
Fig. 10. Frequency spectrum of 200 m long cantilevers before and after metal electrodeposition in 0.25 mM and 1 mM CuSO solution. A frequency shift of 580 Hz and 2.62 kHz were observed, respectively.
the resonant frequency of the cantilever was seen to shift from 31.848 kHz prior to deposition down to 28.968 kHz after metal deposition for a total freqency shift of 2.88 kHz. In 0.25 mM solution, the frequency shifted from 31.720 to 31.140 kHz for a total shift of 580 Hz. In addition, the Q of the cantilever was determined to be approximately 40. The resonant frequency shift for devices operated in solutions with a concentration less than 0.25 mM was below detectable limits given the integration time of the device used and the sensitivity limitations imposed by the design. In solutions above 1 mM, the metal deposited at the cantilever tip became too rough for optical resonance measurements. This limitation of sensor dynamic range can be improved by redesigning the sensor and separating the electrodeposition and optical measurement areas. Finally, Fig. 11 shows electrodeposition at the tip of an unreleased cantilever beam. This deposition was driven by a PV cell array under microscope level illumination.
SUPINO AND TALGHADER: METAL-ION CONCENTRATION IN FLUIDS
Fig. 11. Scanning electron micrograph of an unreleased cantilever beam with copper electrodeposited at tip.
V. DISCUSSION A power supply consisting of three series connected PV cells provides ample current and sufficient voltage to drive an electrochemical reaction. The PV cell array was capable of producing of current with a maximum power output of 4.02 0.4 under light intensities of about 80 . This level of current is more than was needed to drive an electrochemical reaction in a solution with a 1 mM ion concentration which required a steady-state current of about 35 nA and average power of 50.2 nW. To provide a comparison between the light intensities used in PV cell evaluation, the intensity of direct sunlight is on the . It should be noted that this intensity also order of 1000 includes IR radiation which lies outside the absorption spectrum of the PV cell array. A total efficiency of optical power to electrical power conversion can be determined to be about 15% for the wavelength of light used in this characterization. This does not claim to be the solar cell efficiency as the tests were performed using a 660-nm laser source as opposed to a calibrated broadband illumination source. Due to variations in efficiency with wavelength, the overall array efficiency under broadband illumination conditions will be lower than 15%. Fig. 5 provides a comparison between illuminating the PV array with monochromatic laser illumination and a broadband microscope light source. Rough calculations treating the microscope light source as a black-body radiator and integrating over the wavelength sensitivity of a silicon photodetector gives an intensity of . The short circuit current under these conditions 4000 was approximately the same as the short circuit current for of laser light, or a PV cell array illuminated at 2760 approximately 50% lower efficiency. A better comment on efficiency can be found by comparing the external quantum efficiency of the PV cell array with a standard silicon photodiode. The total external quantum efficiency of the PV cell array can be calculated to be about 31%. This compares to typical quantum efficiencies of between 80–90% for standard photodiode devices [11], [27]. It is likely that most of this difference is due to the short absorption length of our cells. A secondary cause of the lower PV array efficiency can be attributed to a non-optimum anti-reflection coating. Thin film calculations show that the reflectivity of 660nm light off of the
1305
interface to be about 8.4%. Typical optimized AR coatSi/ ings for modern photodiodes often achieve values below 1%. Even with reduced efficiency, the current PV cell design is sufficient for use in an electrochemical based sensor without further optimization. The mass detection experiment produced frequency shifts of 2.88 kHz and 580 Hz for deposition in 1 mM and 0.25 mM solutions, respectively. Substituting the shifts in frequency due to mass deposition into (7), a deposited mass of 1.4 ng is determined for the 1 mM case and 0.28 ng for the 0.25 mM case. These calculations assume a Young’s modulus of 100 GPa and for low-stress [28]. In coma density of 2850 parison, integration of the curves in Fig. 8 corresponding to the two measured concentrations gives a total charge transferred of and 6.43 for operation in 1 and 0.25 2.01 mM concentrations, respectively. From Faraday’s Law, the total electroplated metal mass should have been 5.0 and 1.6 ng given the amount of charge transferred during electrodeposition. Differences between the deposited mass as calculated using the electrodeposition current curves and the measured frequency shifts could be due to a parallel electrochemical process such as the electrolysis of water during electrodeposition or due to a current leakage path. If there is a breach in the PECVD silicon nitride layer insulating the PV cells and interconnect metal, a leakage path will exist taking away from current participating in electrodeposition at the cantilever tip. Furthermore, using (8) with the measured frequency shifts, ions are deterconcentrations of 0.26 and 0.052 mM of mined for the 1 and 0.25 mM solutions. This calculation as. sumes a somewhat difficult to define electrode radius of 18 The empirical constant in (8) was determined to be 4.0 by performing electrodeposition current measurements in a solution of known concentration and fitting the theoretical diffusion current (4) to the experimentally measured diffusion current. Equation (8), derived using the diffusion current (4), also assumes a circular microelectrode, which differs from the cantilever’s rectangular electrode geometry. The differences between the detected concentrations are metal-ion concentration to the actual again theorized to be due to a parallel electrochemical reaction or leakage path reducing the total current participating in metal electrodeposition. If the response of the metal-ion sensor is assumed to be linear with concentration and the results are normalized to the 1 mM concentration solution, the measured result for the 0.25 mM concentration would be 0.19 mM, a 24% error. Additionally, it should be mentioned that the current geometry and resonance measurement technique is suboptimal. The measured frequency shift of 580 Hz was larger than the minimum discernable frequency shift of approximately 200–400 Hz noticed during experimentation. From this observation, it is noted that the minimum detectable mass for this resonant mass sensor is on the order of about 0.1–0.2 ng. These resonant frequency measurements were carried out in air where the cantilever Q was approximately 40. Moving such experiments to vacuum should allow an increase in cantilever Q to at least 1000 if not orders of magnitude higher. Such an improvement would lead to an increase in sensitivity of about 25 times at minimum and possibly over 250 times depending on the Q of the cantilever in vacuum. In addition, the large size of these cantilever
1306
JOURNAL OF MICROELECTROMECHANICAL SYSTEMS, VOL. 15, NO. 5, OCTOBER 2006
TABLE I HALF-CELL REDUCTION POTENTIALS FOR COMMON METAL IONS. ADAPTED FROM [20]
REFERENCES [1] B. A. Warneke and K. S. Pister, “MEMS for distributed wireless sensor networks,” in Proc. ICECS 2002. 9th IEEE Int. Conf. Electron. Circuits Syst., Sep. 2002, vol. 1, pp. 291–294. [2] B. Warneke, B. Atwood, and K. S. Pister, “Smart dust mote forerunners,” in Tech. Dig. MEMS 2001. 14th IEEE Int. Conf. Micro Electro Mech. Syst., Jan. 2001, pp. 357–360. [3] B. A. Warneke, M. D. Scott, B. S. Leibowitz, L. Zhou, C. L. Bellew, J. A. Chediak, J. M. Kahn, K. S. P. Bernhard, and E. Boser, “An autonomous 16 solar-powered node for distributed wireless sensor networks,” in Proc. IEEE Sens. 2002: First IEEE Int. Conf. Sens., Jun. 2002, vol. 2, pp. 1510–1515. [4] B. Warneke, M. Last, B. Liebowitz, and K. S. Pister, “Smart dust: Communicating with a cubic-millimeter computer,” Computer, vol. 34, no. 1, pp. 44–51, Jan. 2001. [5] A. Mason, N. Yazdi, A. Chavan, K. Najafi, and K. D. Wise, “A generic multielement microsystem for portable wireless applications,” Proc. IEEE, vol. 86, pp. 1733–1746, Aug. 1998. [6] L. Zhou, J. M. Kahn, and K. S. Pister, “Corner-cube restroreflectors based on structure-assisted assembly for free-space optical communication,” J. Microelectromech. Syst., vol. 12, pp. 233–242, Jun. 2003. [7] M. V. Kruger, M. H. Guddal, R. Belikov, A. Bhatnagar, O. Solgaard, C. Spanos, and K. Poola, “Low power wireless reasout of autonomous sensor wafer using MEMS grating light modulator,” in Proc. IEEE/ LEOS Int. Conf. Opt. MEMS, Aug. 2000, pp. 67–68. [8] M. Suster, D. J. Young, and W. H. Ko, “Micro-power wireless transmitter for MEMS sensing and communication applications,” in Microelectromech. Syst. Conf., Aug. 2001, pp. 25–28. [9] M. Suster, W. K. Ko, and D. J. Young, “An optically powered wireless telemetry module for high-temperature MEMS sensing and communication,” J. Microelectromech. Syst., vol. 13, pp. 536–541, Jun. 2004. [10] G. Asada, A. Burstein, D. Chang, M. Dong, M. fielding, E. Kruglick, J. Ho, F. Lin, T. Lin, H. Marcy, R. Mukai, P. Nelson, F. Newberg, K. S. Pister, G. Pottie, H. Sanchez, O. M. Stafsudd, S. Valoff, G. Yung, and W. J. Kaiser, “Low power wireless communication and signal processing circuits for distributed microsensors,” in Proc. 1997 IEEE Int. Symp. Circuits Syst. Circuits and Systems in the Information Age, Jun. 1997, vol. 4, pp. 2817–2820. [11] C. L. Bellew, S. Hollar, and K. S. Pister, “An soi process for fabrication of solar cells, transistors and electrostatic actuators,” in The 12th Int. Conf. Solid State Sens. Actuators and Microsyst., 2003, vol. 2. [12] J. B. Lee, Z. Chen, M. G. Allen, A. Rohatgi, and R. Arya, “A high voltage solar cell array as an electrostatic MEMS power supply,” in IEEE Proc., ser. Micro Electro Mech. Syst.. : IEEE, 1994, pp. 331–336. [13] D. M. Ryan, R. M. LaFollette, and L. Salmon, “Microscopic batteries for micro electromechanical systems (MEMS),” in Proc. Intersoc. Energy Conv. Eng. Conf., Jul. 1997, vol. 1, pp. 77–82. [14] M. Chiao, K. B. Lam, and L. Lin, “Micromachined microbial fuel cells,” in Proc. IEEE Micro Electro Mech. Syst. (MEMS), Jan. 2003, pp. 383–6. [15] H. Kulah and K. Najafi, “An electromagnetic micro power generator for low-frequency environmental vibrations,” in Proc. IEEE Int. Conf. Micro Electro Mech. Syst. (MEMS), Jan. 2004, pp. 237–240. [16] B. H. Starj, P. D. Mitcheson, P. Miao, T. C. Green, E. M. Yeatman, and A. S. Holmes, “Power processing issues for micro-power electrostatic generators,” in Proc. IEEE 35th Annu. Power Electron. Special. Conf., Jun. 2004, vol. 6, pp. 4156–4162. [17] P. B. Koeneman, I. J. Busch-Vishniac, and K. L. Wood, “Feasibility of micro power supplies for MEMS,” J. Microelectromech. Syst., vol. 6, pp. 355–362, Dec. 1997. [18] J. Santini, M. Cima, and R. Langer, “A controlled-release microchip,” Nature, vol. 397, no. 6717, pp. 335–8, Jan. 1999. [19] G. T. Kovacs, C. W. Storment, and S. P. Kounaves, “Microfabricated heavy metal ion sensor,” Sens. Actuators B, vol. 23, pp. 41–47, 1995. [20] P. H. Rieger, Electrochemistry, 2nd ed. New York: Chapman and Hall, 1994. [21] K. B. Oldham and J. C. Myland, Fundamentals of Electrochemical Science. New York: Academic, 1994. [22] T. Drews, S. Krishnan, J. Alameda, D. Gannon, R. Braatz, and R. Alkire, “Multiscale simulations of copper electrodeposition onto a resistive substrate,” IBM J. Res. Devel., vol. 49, no. 1, pp. 49–63, Jan. 2005. [23] S. Timoshenko, D. H. Young, and W. Weaver, Vibration Problems in Engineering, 4th ed. New York: Wiley, 1974.
mm
structures leads to an increase in the mass of the cantilever and a subsequent decrease in sensitivity. Reduction in cantilever size would not only reduce physical dimensions of the sensor, but would also lead to an increase in sensitivity. Finally, a few notes regarding sensor selectivity are in order. From well tabulated half-cell potential tables [20], it can be seen that a number of different metals can be electrodeposited at voltages near the electrodeposition potential of copper, as shown , , and , in Table I. These metal ions include among others. When the metal-ion sensor is exposed to an environment with multiple ion species present, more than one type of metal ion will be deposited, and selectivity is compromised. Therefore, the metal-ion sensor as presented is better suited to general heavy metal-ion detection than as a sensor used to discern individual species. With little added complexity to the sensor architecture, the cantilever resonant mass detection can be coupled with anodic stripping voltammetry (ASV) techniques [19] to enable specific ion selectivity. If the sensor were connected to an external voltage source, the metal at the tip of the cantilever could be oxidized in an electrolyte solution. By scanning through a range of potentials, current peaks occur at various potentials indicating the presence of different ions. The magnitude of the current peaks allow the determination of the ratio of metal ions present on the electrode. Typical microfabricated ASV devices employ an array of microelectrodes to increase the total amount of metal deposited and therefore increase the sensitivity of the device. Incorporating ASV with a cantilever resonant mass detection device would allow a single electrode to perform sensitive mass measurements to measure small metal-ion concentrations with the added ability of detecting multiple species. VI. CONCLUSION An extremely low-power electrochemical sensor was demonstrated. The device scavenges optical power from its surroundings in order to power an electrochemical reaction. The electrochemical reaction is used to sense the metal-ion concentration in a liquid in which the sensor is immersed. The integrated . The metal-ion sensor had a total volume below 0.046 PV cells driving the electrochemical process could provide over of current under laser excitation, but the electrochem10 ical system operated at power levels as low as 50.2 nW. This power level is easily supplied by scavenged room light. Individual sensors were capable of sensing metal-ion concentration with a factor of two, making them suitable for spot checks of environmental and industrial processes.
SUPINO AND TALGHADER: METAL-ION CONCENTRATION IN FLUIDS
[24] C. Harris and A. Piersol, Eds., Harris’ Shock and Vibration Handbook, 5th ed. New York: McGraw-Hill, 2002. [25] R. Buser, B. Stauffer, and N. de Rooij, “Realization of mesa array in (001) oriented silicon wafers for tactile sensing applications,” in The Electrochemical Society, Fall Meeting, Oct. 1986, vol. 86, p. 879. [26] G. Meyer and N. Amer, “Novel optical approach to atomic force microscopy,” Appl. Phys. Lett. , vol. 53, no. 12, pp. 1045–1047, 1988. [27] M. Fukuda, Optical Semiconductor Devices. New York: Wiley, 1999. [28] L. Kiesewetter, J. Zhang, D. Houdeau, and A. Steckenborn, “Determination of young’s moduli of micromechanical thin films using the resonance method,” Sens. Actuators A, vol. 35, no. 2, pp. 153–159, Dec. 1992. Ryan Supino (S’03) received the B.S. degree in electrical engineering (summa cum laude) from the University of Minnesota, Twin Cities, in 2000. He was awarded an NSF Graduate Fellowship and received the M.S. and Ph.D. degrees in electrical engineering also from the University of Minnesota in 2003 and 2006, respectively. His graduate research was focused on MEMS technology, where he researched a variety of topics ranging from thermal management in micromechanical devices to the development of active microsensors for remote environments. Currently, he is a Research Scientist at Honeywell International, Plymouth, MN, where he is involved in the research and development of MEMS inertial sensor products.
1307
Joseph J. Talghader (S’93–M’95) received the B.S. degree in electrical engineering from Rice University, Houston, TX, in 1988. He was awarded an NSF Graduate Fellowship and attended the University of California at Berkeley, where he received the M.S. and Ph.D. degrees in 1993 and 1995, respectively. From 1992 to 1993, he worked at Texas Instruments as a Process Development Engineer. After graduating from Berkeley in 1995, he joined Waferscale Integration. In 1997, he joined the faculty at the University of Minnesota, Minneapolis, where he is now an Associate Professor. Dr. Talghader has received 3M Nontenured Faculty Awards on three occasions. He has served on various program committees and reviews, including service as Program chair of the 2003 IEEE/LEOS Optical MEMS Conference and as Guest Editor of the IEEE JOURNAL OF SELECTED TOPICS IN QUANTUM ELECTRONICS 2004 Special Issue on Optical Microsystems. He will be Conference Chair of the 2006 IEEE/LEOS Optical MEMS Conference at Yellowstone/Big Sky.
