Infrared Temperature Control System for a Completely Noncontact

A completely noncontact temperature system is described for amplification of ..... Comparative Medicine, University of Virginia Health System). PCR pr...
0 downloads 0 Views 287KB Size
Download PDF
Anal. Chem. 2007, 79, 1294-1300

Infrared Temperature Control System for a Completely Noncontact Polymerase Chain Reaction in Microfluidic Chips Michael G. Roper,†,‡ Christopher J. Easley,† Lindsay A. Legendre,† Joseph A. C. Humphrey,§ and James P. Landers*,†,§,|

Department of Chemistry and Department of Mechanical and Aerospace Engineering, University of Virginia, Charlottesville, Virginia 22904, and Department of Pathology, University of Virginia Health Science Center, Charlottesville, Virginia 22908

A completely noncontact temperature system is described for amplification of DNA via the polymerase chain reaction (PCR) in glass microfluidic chips. An infrared (IR)sensitive pyrometer was calibrated against a thermocouple inserted into a 550-nL PCR chamber and used to monitor the temperature of the glass surface above the PCR chamber during heating and cooling induced by a tungsten lamp and convective air source, respectively. A time lag of less than 1 s was observed between maximum heating rates of the solution and surface, indicating that thermal equilibrium was attained rapidly. Moreover, the time lag was corroborated using a one-dimensional heattransfer model, which provided insight into the characteristics of the device and environment that caused the time lag. This knowledge will, in turn, allow for future tailoring of the devices to specific applications. To alleviate the need for calibrating the pyrometer with a thermocouple, the on-chip calibration of pyrometer was accomplished by sensing the boiling of two solutions, water and an azeotrope, and comparing the pyrometer output voltage against the known boiling points of these solutions. The “boiling point calibration” was successful as indicated by the subsequent chip-based IR-PCR amplification of a 211-bp fragment of the B. anthracis genome in a chamber reduced beyond the dimensions of a thermocouple. To improve the heating rates, a parabolic gold mirror was positioned above the microfluidic chip, which expedited PCR amplification to 18.8 min for a 30-cycle, three-temperature protocol. Amplification of specific DNA sequences by the polymerase chain reaction (PCR) is used in a myriad of settings, including forensic DNA analysis,1 molecular diagnostics,2 and molecular * To whom correspondence should be addressed. Phone: (434)-243-8658. Fax: (434)-924-3048. E-mail: [email protected]. † Department of Chemistry, University of Virginia. ‡ Current address: Department of Chemistry and Biochemistry, Florida State University, Tallahassee, FL 32306. § Department of Mechanical and Aerospace Engineering, University of Virginia. | University of Virginia Health Science Center. (1) Budowle, B.; Johnson, M. D.; Fraser, C. M.; Leighton, T. J.; Murch, R. S.; Chakraborty, R. Crit. Rev. Microbiol. 2005, 31, 233-254. (2) Song, Y. Anaerobe 2005, 11, 79-91.

1294 Analytical Chemistry, Vol. 79, No. 4, February 15, 2007

biology.3 Advancements in PCR have been facilitated by the discovery of more thermostable polymerases,4 real-time detection of PCR products enabling a quantitative assessment of the reaction during the amplification,5 and the use of miniaturized reaction systems to reduce reagent use and speed temperature transitioning.6 Miniaturized systems, such as microfluidic chips or capillary systems, offer other advantages beyond reducing reagent consumptionsthese include faster cycle times facilitated by the reduced sample volume (reducing the mass to be thermal cycled) and allowing for facile integration of both pre- and post-PCR processing steps.7One method for fast thermal cycling is noncontact heating via infrared (IR) excitation of the vibrational bands of water. IR-mediated PCR, pioneered by our group and demonstrated in capillaries,10 and both polymeric11 and glass12 microdevices, has been shown to be capable of fast thermocycling. The use of a broadband tungsten lamp/convective fan allows for remote heating and cooling and facilitates building the temperature control hardware into the instrumentation and not into the device, allowing for simple and cost-effective microdevice fabrication. While not yet demonstrated in a microfluidic device, a commercially available system is available for fast, noncontact thermal cycling by blowing preheated air over glass capillaries.13,14 For a noncontact method to be effective for selective heating of microdomains on a fluidic chip, it must be coupled with a (3) Peano, C.; Severgnini, M.; Cifola, I.; DeBellis, G.; Battaglia, C. Expert Rev. Mol. Diagn. 2006, 6, 465-480. (4) McDonald, J. P.; Hall, A.; Gasparutto, D.; Cadet, J.; Ballantyne, J.; Woodgate, R. Nucleic Acids Res. 2006, 34, 1102-1111. (5) Watzinger, F.; Ebner, K.; Lion, T. Mol. Aspects Med. 2006, 27, 254-298. (6) Roper, M. G.; Easley, C. J.; Landers, J. P. Anal. Chem. 2005, 77, 38873893. (7) Easley, C. J.; Karlinsey, J. M.; Bienvenue, J. M.; Legendre, L. A.; Roper, M. G.; Feldman, S. H.; Hughes, M. A.; Hewlett, E. L.; Merkel, T. J.; Ferrance, J. P.; Landers, J. P. Proc. Natl. Acad. Sci. U.S.A. 2006, 103, 19272-19277. (8) Paegel, B. M.; Blazej, R. G.; Mathies, R. A. Curr. Opin. Biotechnol. 2003, 14, 42-50. (9) Kricka, L. J.; Wilding, P. Anal. Bioanal. Chem. 2003, 377, 820-825. (10) Oda, R. P.; Strausbach, M. A.; Huhmer, A. F. R.; Borson, N.; Jurrens, S. R.; Craighead, J.; Wettstein, P. J.; Eckloff, B.; Kline, B.; Landers, J. P. Anal. Chem. 1998, 70, 4361-4368. (11) Giordano, B. C.; Ferrance, J.; Swedberg, S.; Huhmer, A. F. R.; Landers, J. P. Anal. Biochem. 2001, 291, 124-132. (12) Easley, C. J.; Karlinsey, J. M.; Landers, J. P. Lab Chip 2006, 6, 601-610. (13) Wittwer, C. T.; Fillmore, G. C.; Hillyard, D. R. Nucleic Acids Res. 1989, 17, 4353-4357. (14) Wittwer, C. T.; Fillmore, G. C.; Garling, D. Anal. Biochem. 1990, 186, 328331. 10.1021/ac0613277 CCC: $37.00

© 2007 American Chemical Society Published on Web 01/17/2007

comparably simple but sensitive method for temperature sensing. While both the IR-mediated and the forced-air system utilize manually inserted thermocouples for direct sensing of solution temperature, these contact-based sensors require surface passivation to avoid inhibition of PCR (as would a thermocouple fabricated into the chamber) or placement into “dummy” chambers adjacent to the PCR chamber or capillary.11-15 However, a more viable solution involves a method that, like the heating, senses temperature through a noncontact process, again minimizing the cost of microchip fabrication. In this report, a simple, effective, and robust temperature sensing method is described which, together with IR-mediated heating, enables completely noncontact temperature control for performing PCR in microfluidic devices. This involves an IR pyrometer sensing the surface temperature above a PCR chamber with a design that assures the rapid equilibration between the PCR solution and the chamber surface. Thermal modeling is used to define physical properties of the device and environment needed to achieve this rapid equilibration, and the model verifies an experimentally observed thermal equilibration of the device during the initial PCR thermal cycles. In line with a completely noncontact method of temperature control, calibrating the surface temperature relative to the PCR solution temperature is accomplished using the boiling point of water and an azeotrope within the chip. The effectiveness of the method is illustrated with the successful PCR of a fragment of Bacillus anthracis gene in a fluidic channel too small for conventional thermocouple-based sensing. Using a goldcoated parabolic mirror to focus stray IR radiation back onto the device allows for 30 cycles of PCR amplification to be complete in 19 min. EXPERIMENTAL SECTION Microfluidic devices were made as previously described.16 Glass wafers (Telic Co., Valencia, CA) precoated with a layer of chrome and positive photoresist were exposed to UV radiation through a film mask. Irradiated photoresist and the underlying chrome were then removed, and the exposed glass was etched using hydrofluoric acid. Diamond-tipped drill bits (Tripple Ripple, 1.1-mm diameter, Crystalite Corp., Lewis Center, OH) were used to drill fluidic access holes, and after drilling, microdevices were cleaned with a 3:1 (v/v) solution of H2SO4/H2O2 followed by a 5:1:1 (v/v/v) solution of H2O/NH4OH/H2O2. Clean bottom plates, 200 µm thick, were then placed into contact with the etched devices and bonded together at 640 °C for 8 h. The photomask design consisted of two ellipses, each 0.75 mm wide and 3 mm long. The two ellipses were separated by 1.5 mm (from one center of the ellipse to the other center). Two channels, each impinging on the end of each ellipse, were used for filling the ellipses with PCR solution and insertion of the thermocouple. Etch depths were either 200 or 100 µm deep producing PCR reaction volumes of 550 and 230 nL, respectively. Thermocouple and Pyrometer PCR Setups. The PCR setup was similar to systems described elsewhere.12,15 Briefly, the microdevice was mounted on a Plexiglas stage that allowed access to the infrared radiation and convective cooling from the bottom (15) Legendre, L. A.; Bienvenue, J. M.; Roper, M. G.; Ferrance, J. P.; Landers, J. P. Anal. Chem. 2006, 78, 1444-1451. (16) Roper, M. G.; Frisk, M. L.; Oberlander, J. P.; Ferrance, J. P.; McGrory, B. J.; Landers, J. P. Anal. Chim. Acta 2006, 569, 195-202.

Figure 1. Instrumental layout of pyrometer-controlled PCR. The pyrometer was placed off-axis from the tungsten lamp to ensure the pyrometer was not heated during experiments. During and after calibration of the pyrometer, the microdevice was secured to the setup to ensure the same location above the PCR chamber was measured for accurate temperature sensing. The digital control lines from the DAQ card to the relays are shown as dotted lines.

of the stage. Two solid-state relays (CMX60D10, Crydom Corp., San Diego, CA) were placed in series with a 12-V power supply to power both the fan and the tungsten lamp (CXR, 8 V, 50 W, General Electric, Cleveland, OH). The pyrometer (MI-N5, Mikron Infrared, Inc., Oakland, NJ) was oriented 45° from vertical above the microdevice, and a red diode laser within the pyrometer was used to align the sensing area of the pyrometer onto the surface of the device above the PCR chamber (Figure 1). When the microdevice was placed on the stage and aligned with both the pyrometer sensing area and the focal spot of the tungsten lamp, the physical location of the microfluidic chip was recorded using double-sided tape. In some experiments stated in the text, a goldcoated parabolic mirror (Edmund Optics, Barrington, NJ) was placed 50 mm above the microdevice to focus the stray radiation from the lamp onto the top of the microfluidic chip. A thermocouple (T-240C, Physitemp Instruments, Inc., Clifton, NJ) was used to initially calibrate the pyrometer. The thermocouple voltage was digitized and amplified using a commercially available circuit (TAC-386-T, Omega Engineering, Stamford, CT) and a homemade circuit for a total amplification factor of 1875 (75 mV/°C). This voltage was recorded using a data acquisition (DAQ) card (PCI-6014, National Instruments, Austin, TX). The pyrometer voltage output was fed into the same DAQ card using a 1-kΩ resistor across the current output of the pyrometer. To control the output of the IR lamp, a proportional integral derivative feedback algorithm (written in LabView, National Instruments) using the thermocouple or pyrometer input was used to control the duty cycle of a 1000-Hz TTL output to the solidstate relay in series with the lamp. The control to the relay in series with the fan was either on (TTL high) or off (TTL low). The thermocouple was calibrated by filling the 550 nL microfluidic device with PCR buffer, inserting the thermocouple, and placing mineral oil above the fluidic access holes. The microdevice and Analytical Chemistry, Vol. 79, No. 4, February 15, 2007

1295

thermocouple were then placed on a conventional PCR thermal cycler (GeneAmp 2400 PCR System, Perkin-Elmer, Wellesley, MA), and the temperature of the conventional cycler was increased in increments of 5 °C and held for 30 s at each increment. The average thermocouple voltage versus temperature of the conventional cycler was used to calibrate the thermocouple. The pyrometer was calibrated by placing the thermocouple within the microdevice and the IR lamp used to perform five mock PCR cycles with 15-s hold times at 95, 60, and 72 °C. The average pyrometer output voltage at the average temperature recorded by the thermocouple was used for calibration. When reporting the data for the pyrometer or thermocouple, all values are average ( one standard deviation. Calibration of the Pyrometer by Measurement of Boiling Points. A 230-nL microfluidic device was filled with water or a 13.1/38.2/48.7 (v/v/v) mixture of water/2-propanol/toluene and positioned in the focal spot of the tungsten lamp. The lamp was subsequently turned on at full power and the pyrometer output voltage recorded as a function of time. This procedure was repeated three times with each solution and the derivatives of these traces were used to determine the pyrometer output voltage that correlated with the known boiling points of the solutions (see Results and Discussion for more details). PCR Procedure. Microdevices were rinsed with acetone, dried, and passivated with 10 µL of SigmaCote (Sigma-Aldrich, St. Louis, MO) prior to making PCR mastermix solutions. All PCR reagents were from Sigma-Aldrich, and primers were from MWG Biotech, Inc. (High Point, NC). For PCR amplification, a 10-µL PCR solution was made, which was composed of 10 mM TrisHCl, 50 mM KCl, pH 8.3, 3 mM MgCl2, 0.2 mM each dNTP, 0.4 µM forward and reverse primer, 0.1 units/µL Taq polymerase, and either 7 ng of λ-phage DNA or purified DNA from 3000 colony forming units of B. anthracis spores (spores and anthrax primers kindly donated by Dr. Sanford H. Feldman, Department of Comparative Medicine, University of Virginia Health System). PCR protocol for a 500-bp fragment of λ-phage DNA consisted of an initial denaturation at 95 °C for 60 s, followed by 30 cycles of 95 °C at 15 s, 68 °C for 15 s, and a final extension at 72 °C for 120 s.17 Protocol for amplification of a 211-bp fragment of the virulence B gene on pXO1 of the B. anthracis genome consisted of a 60-s initial denaturation at 95 °C, followed by 30 cycles of 95 °C for 5 s, 62 °C for 5 s, and 72 °C for 5 s. A final extension at 72 °C was performed for 30 s. Control amplifications for each of the thermal cycling protocols were performed without template DNA and are shown in the appropriate figures. After PCR was complete, amplified product was removed from the device, diluted with 24 µL of water and 0.8 µL of a PCR marker (N3234S, New England BioLabs, Inc., Ipswich, MA). This mixture was then separated on a commercial capillary electrophoresis instrument using laser-induced fluorescence detection as described previously.15 Sizing results of amplified PCR products are given as average ( standard deviation as determined by comparison of the migration time of the amplified product to the migration time of the sizing ladder. RESULTS AND DISCUSSION The composition of an optical pyrometer is relatively simple containing an optical filter and a detector, typically a thermopile (a collection of thermocouples in series). The basis for using a 1296

Analytical Chemistry, Vol. 79, No. 4, February 15, 2007

pyrometer as a temperature sensor is that, as an object is heated, the blackbody radiation from the object shifts to shorter wavelengths, which fall into the range that can be detected by the pyrometer. In this report, the pyrometer is composed of a thermopile situated behind a Ge lens, which is transparent to wavelengths between 8 and14 µm. As the solution in the PCR chamber is heated by the IR radiation from the tungsten lamp, the glass above the chamber is also heated producing blackbody radiation that is detected by the pyrometer and recorded as a change in the output voltage of the thermopile. Output voltages of the thermopile were calibrated to temperatures using either a thermocouple or the boiling points of reference solutions and used to successfully control PCR amplification in microdevices. Pyrometer Calibration and Temperature Sensing Time Lag. The pyrometer was placed off the vertical axis of the lamp to ensure that the pyrometer was not heated during experiments (Figure 1). Since the emissivity of glass is known, the surface temperature of the top glass plate can be deduced directly with the pyrometer; however, due to difficulties in maintaining the exact optical properties of glass pre- and post-thermal bonding, a calibration that defined the pyrometer output voltage (from the surface) relative to the PCR solution temperature was performed initially with a thermocouple. One consequence of correlating the surface temperature with solution temperature is the lag time associated with temperature equilibration between these two regions. The optimal design of a microdevice would have a negligible time lag, allowing the surface temperature to accurately reflect the solution temperature; alternatively, a suboptimal design would render the surface temperature completely insensitive to changes in the corresponding solution temperature. The pyrometer was calibrated by performing five mock PCR cycles between 60, 95, and 72 °C with a dwell time of 15 s at each and comparing the temperature in the PCR chamber (inserted thermocouple) with that on the surface (pyrometer output voltage). As shown in Figure 2A, the pyrometer signal was increasing and decreasing in proportion to the solution temperature and was linear. However, further analysis of the traces indicated that the temperature measured by the thermocouple increased prior to an increase in the pyrometer output, indicative of a lag associated with the equilibration of the two regions. Quantitative information on this time lag was made by calculating the difference between the heating rate maximums sensed by the thermocouple and that of the pyrometer (obtained by taking the derivative of the traces with respect to time); the time lag for cooling was measured in a similar manner with the minimums of the cooling rates (Figure 2B). Heating occurred in two steps, the first from 60 to 72 °C and the second from 72 to 95 °C, leading to time lags of 1.2 ( 0.8 and 0.3 ( 0.2 s, respectively, whereas cooling involved transitioning from 95 to 60 °C and corresponded to a time lag of 0.4 ( 0.2 s. These positive values indicated that the surface temperature always lagged the solution temperature; however, since the absolute numbers were small, the surface temperature provided an acceptable method for indirect control of the solution temperature. The large error in the measurements likely resulted from the incidental heating of the entire device (not just the PCR domain) over the course of the experiment (discussed in detail below). Ideally, the device could be designed and fabricated such (17) Huhmer, A. F. R.; Landers, J. P. Anal. Chem. 2000, 72, 5507-5512.

Figure 2. Calibration of the pyrometer to a thermocouple. (A) Five mock PCR cycles of a three-temperature thermal cycling program controlled by a thermocouple (black line, left Y-axis) were used to calibrate the pyrometer output voltage (red line, right Y-axis). Data values from each hold step were averaged and used in attaining the calibration constants. It is apparent that the time lag between the two temperatures was larger in the initial cycles compared to the later cycles and this effect was likely due to the entire device heating as the cycles progressed (explained in more detail at the end of the text). (B) The lag time between the maximum heating rates measured by the thermocouple and pyrometer are plotted for the various transition periods of the cycling shown in (A). Heating to denature entailed the temperature change from 72 to 95 °C, heating to extension was 6072 °C, and cooling to anneal was 95-60 °C. While it appears that the lag time was largest during the heating to extension, the error, due to the changing overall device temperature, was large enough that the results were not statistically different.

that this time lag would be zero, which would allow the surface temperature to correspond exactly with the solution temperature. To this end, a one-dimensional heat-transfer model was used to investigate the properties of the device and environment that contributed to this time lag.

We hypothesized that the underlying reason for the time lag was that the solution in the PCR chamber was heating faster than the glass above the PCR chamber; however, a more qualitative method of defining the reason for this time lag was obtained with a one-dimensional heat-transfer model (detailed in section one in the Supporting Information). Using device and environment specifications, the model predicted a time lag of 0.3 s between the solution and surface temperature (Figure S-2), correlating, within error, with the data obtained. The model also defined several parameters that could minimize this time lagsamong these were fabrication of a device with a thin cover plate (the thinner the top plate, the more accurately the two temperatures would correlate), a material with a large thermal diffusivity, and performing the experiments in an environment with a small convective heat-transfer coefficient (e.g., in a preheated chamber). The use of the pyrometer for these cases is currently under investigation. PCR Amplification of λ-Phage DNA. Having defined the parameters needed for accurate pyrometer-controlled temperature cycling, amplification of a 500-bp fragment of bacterial λ-phage DNA in a 550-nL volume was attempted. After calibration of the pyrometer, the thermocouple was removed and thermocycling carried out using 15-s dwell times at 95 and 68 °C for 30 cycles (Figure 3A). Following the final extension for 120 s at 72 °C, the PCR solution was removed from the microchip and analyzed by capillary electrophoresis. As shown in Figure 3B, amplified product was observed, indicating that the pyrometer provided an effective approach for controlling the temperature in low-volume chambers on microdevices for enzyme-mediated DNA amplification. While the cycling was relatively slow (∼30 min), this was due to not having a reflective surface above the device, as was reported in other IR-mediated PCR amplifications.12,15 Boiling Point Calibration. The pyrometer calibration remained valid only when the same microdevice was placed in the same location above the tungsten lamp. This result was due to different devices having slightly different glass thickness and resulting in different rates of heat transfer. Consequently, it became clear that a more simple method for calibrating the pyrometer without the need for a thermocouple or other external hardware was needed. Due to the complexity of using emissivity as a means for obtaining a direct temperature measurement, a method was used to calibrate the pyrometer by the boiling points of reference solutions. The criteria for choosing the solutions included boiling points that were within the range of temperatures normally used in PCR (∼60-95 °C) and the solutions needed to be composed of at least 10% water to approximate IR absorption in a manner similar to absorption by the PCR solution. Two solutions were defined as ideal for this purpose: pure water and an azeotrope consisting of water/2-propanol/toluene at a ratio of 13.1/38.2/48.7 (v/v/v). The rising temperature of the solution was measured through surface sensing (above the microchannel) with the pyrometer, which generated the pyrometer voltage versus time plot given in Figure 4. When the solution reached its boiling point, the temperature remained constant during the phase change, which was detected with the pyrometer as a change in slope (n ) 3, Figure 4). Using the derivative of these plots, the pyrometer voltage minimums were identified and used to calibrate the pyrometer against the known boiling point values for these Analytical Chemistry, Vol. 79, No. 4, February 15, 2007

1297

Figure 4. Calibration of the pyrometer using boiling points. Water (black line) or an azeotrope (gray line) was heated until the solution boiled (n ) 3). As more easily observed with water, the surface temperature of the glass still increased during the phase change, but the change in the heating rate could be quantitatively determined by taking the derivative of these traces with respect to time. The derivative of the azeotrope heating trace is shown in the inset with the unbroken arrow indicating the time point that boiling occurred. This time point was then used to determine at which pyrometer output voltage (broken line) corresponded to the boiling temperature.

Figure 3. Noncontact amplification of λ-phage DNA. (A) Thirty cycles of a two-temperature heating protocol using the pyrometer calibrated with the trace shown in Figure 2A. (B) After thermal cycling was complete, analysis of the product within the PCR chamber by capillary gel electrophoresis demonstrated that an amplified product was present, which migrated at the expected size (shown by /). The other peaks present at 3.5 and 4.0 min correspond to primer and dimer peaks, respectively. Amplification with no template DNA was subsequently performed and is shown as the bottom electropherogram.

reference solutions, 76.3 °C for the azeotrope and 99.6 °C for water.18 This analysis yielded two reference points with which to calibrate the pyrometer. While one could be skeptical about creating a calibration curve with only two points, it was known from the previous calibration that the pyrometer signal was linear in this temperature range (see Figure 2A), and it was difficult to find azeotrope solutions that met the aforementioned criteria. It is noteworthy that the microdevice used for all experiments in this section contained channels that were smaller (100 µm in depth) than the thermocouple (∼160 µm) to demonstrate the practicality of controlling temperature in low-volume conditions (230-nL ellipse). (18) CRC Handbook of Chemistry and Physics, Student Ed., 69th ed.; Weast, R. C., Ed.; CRC Press: Boca Raton, FL, 1988.

1298 Analytical Chemistry, Vol. 79, No. 4, February 15, 2007

Following this calibration, the pyrometer was tested for its effectiveness in controlling temperature cycling for the PCR amplification of a 211-bp gene fragment from B. anthracis. After completion of 30 cycles in 33.7 min, electrophoretic analysis of the solution in the PCR chamber indicated the presence of a 210 ( 3 bp DNA amplicon (Figure 5). This positive result indicated that with this type of calibration method IR-PCR chambers large enough for insertion of a thermocouple are no longer needed and lowers the volume limit of these chambers to subnanoliter. The use of the boiling point calibration takes advantage of the wellknown physical properties of reference solutions and provides a simple means for calibration and will allow for more portable and self-contained chip-based PCR devices in the future. Pyrometer Temperature Control with Increased Heating Rates. The cycling times for the amplifications shown in Figures 3 and 5 were slow compared to previously reported IR-mediated amplifications.12,15 Two factors contributed to the longer cycle times: the substantial mass of glass surrounding the PCR chamber and the lack of a reflective surface above the device to redirect stray radiation back onto the PCR chamber. A simple method for increasing the heating and cooling rates is by etching away a large portion of the mass around the PCR chamber.12 However, it is more difficult to increase the heating rates with a reflective surface while still allowing the pyrometer an unobstructed view of the PCR chamber (see Figure 1). It has been shown that reflection of IR can be accomplished using a goldcoated glass slide placed 1 mm above the PCR chamber,12,15 but this methodology was not compatible with the clearance needed for the pyrometer to probe the chamber surface. Alternatively, a 25-mm-diameter parabolic gold mirror with a 50-mm focal distance was positioned above the device, allowing ample room for the pyrometer to measure the surface. With the aim of improving

Figure 5. Control of PCR using boiling point calibration. The pyrometer was calibrated using the data in Figure 4 and used to control amplification of a gene fragment from B. anthracis in a 250nL PCR chamber. The product peak (shown by /) migrated at the expected size relative to a DNA ladder which consisted of 50, 150, 300, 500, and 766 bp fragments. The inset details the sizing information for the 211-bp product (open square). A control amplification showed no detectable amplified product at the expected size (bottom electropherogram).

heating rates, the pyrometer was calibrated by the boiling point method, and amplification of a fragment of the B. anthracis genome was attempted. As seen in Figure 6A, with the same thermal cycling parameters used in both Figures 3 and 5 to amplify the B. anthracis fragment, the heating rates were increased 44%, with completion of the PCR in 18.8 min. The reduced cycling time did not appear to adversely influence the amplification of the 211-bp amplicon, which was sized by capillary gel electrophoresis as a 214 ( 2 bp peak relative to DNA standards (Figure 6B), close to the expected 211-bp target. While an overall cycling time of 18.8 min was still slower than other IR-mediated PCR reactions,12,15 the goal of demonstrating remote pyrometer temperature sensing was accomplished with reasonably fast PCR. Cycling speed could be further enhanced by decreasing the hold time at each temperature and, as mentioned previously, etching away the glass around the PCR chambers for increased heating and cooling rates.12 Device Heating during Thermal Cycling. In addition to the ability to sense temperature in a noncontact fashion, the pyrometer also allowed for observation of a phenomenon that had not been observed with thermocouple-controlled IR-PCR. A qualitative evaluation of temperature cycles indicated that as thermocycling ensued successive cycle times decreased, leading to the theory that as the heat dissipated from the PCR chamber to the microchip, the ambient temperature of the entire device increased until the device came to equilibration. If the average temperature of the microdevice itself increased significantly over the course of the temperature cycling, this could be detrimental to the PCR amplification since recalibration of the pyrometer at the elevated temperatures would be required (i.e., the temperature gradient between the solution in the chamber and the surface above the chamber would be different from what was calibrated). While the reactions appeared to be successful, as shown in Figures 3B, 5,

Figure 6. Increased heating rates using a Au-coated mirror. (A) A gold-coated mirror was placed 5 cm above the PCR chamber and used to focus stray IR radiation back onto the device. With this mirror, the cycling time, relative to Figures 3A and 5, was reduced 44% to 18.8 min. (B) Amplification of a 211-bp gene fragment from B. anthracis was performed using this faster cycling method, and a 214 ( 2 bp peak was observed (peak shown by /, sizing data as the open square in the inset) when compared to the same DNA sizing ladder as shown in Figure 5. Amplification with no template DNA produced no detectable product (bottom electropherogram).

and 6B, it was possible that amplification efficiency was not optimal due to inaccurate temperature sensing because of the increased ambient temperature at cycles later in the sequence. With the goal of determining the relative temperature change of the entire device during thermal cycling and what design/fabrication parameters could be altered to alleviate this effect, a more quantitative investigation of this phenomenon was attempted. Evidence to support the hypothesis that the entire microfluidic chip was warming over the course of the thermal cycling sequence was provided by the model outlined in section two of the Supporting Information. As shown in Figure S-4, the times for each successive temperature cycle decreased until the fifth cycle, indicating that the entire device had come to an equilibrium Analytical Chemistry, Vol. 79, No. 4, February 15, 2007

1299

the initial cycles until leveling off during the middle and final cyclessthis indicated that, while the solution temperature was constant during each hold step (as would be expected with temperature control by the thermocouple), the device temperature did increase until an equilibration was reached. The percentages that the signal changed over the 30 cycles shown in Figure 7 were 2.7 ( 1.0, 4.2 ( 1.3, and 3.6 ( 1.3% (for denature, anneal, and extension temperatures, respectively). While these numbers are not large, again, this effect may have lowered the amplification efficiency in this report. In the future, the calibration constants will be updated in a time-dependent or cycle number-dependent manner through the control program so that more accurate temperatures are measured with each cycle. Alternatively, a preheating step could be used to equilibrate the device during the first cycle, for example, by using a “hot-start” Taq polymerase.

Figure 7. Increased overall temperature of the microfluidic device. The average normalized outputs recorded by both the thermocouple and pyrometer at each denature, anneal, and extend step were plotted for each cycle in a 30-cycle, thermocouple-controlled mock PCR. As can be seen, the thermocouple traces are relatively constant since the temperature is being controlled with this sensor; however, the pyrometer signal increases over the first 15 cycles, indicating the temperature of the microdevice surface is increasing during this time. Plots are named by the method used to sense and the hold step, for example, “TC Denature” are values recorded from the thermocouple during the denature holds, and “Pyro Anneal” are values recorded from the pyrometer during the anneal holds. One standard deviation is shown for each data point for clarity.

temperature by this time. Due to thermal contact between the PCR chambers and the “frame” (the glass around the PCR chambers not involved in storing fluids), this excess glass was also heated at the same time as the PCR chambers (Figure S-5), serving as a heat “capacitor” in that heat could be stored during the initial cycles to increase the heating rates of later cycles. The model delineated two methods to reduce this effect: first, thermally isolate the PCR chambers from the frame and, second, implement flow conditions that produce low heat-transfer coefficients. While the former method is simply a matter of etching the glass to isolate the PCR chambers, the latter would be difficult to implement as a material of this type would take an excessively long time to cool from a heated state. Practical methods that circumvent (but do not alleviate) this phenomenon are given below. More quantitative evidence supporting the heating of the entire device was obtained when 30 cycles of mock PCR were controlled by an inserted thermocouple and the pyrometer used to probe the surface temperature (n ) 3). As shown in Figure 7, during each hold step, the normalized pyrometer output increased during

1300

Analytical Chemistry, Vol. 79, No. 4, February 15, 2007

CONCLUSIONS The first completely noncontact PCR device was described using a tungsten lamp and convective air source for heating and cooling, respectively, and a pyrometer for temperature sensing. In the future, we predict that the heat-transfer modeling will allow for optimization of microdevice design prior to experimentation instead of just providing a qualitative means by which to explain the phenomenon observed. Also, modeling will help dictate other materials from which to fabricate the microdevices so that the optimal heat-transfer properties are exploited, which may allow for direct temperature sensing of the PCR solution. Azeotropic calibration by boiling solutions of known composition that can be easily incorporated into the microfluidic architecture of the chip is ideal not only since it may allow for a completely enclosed system, but also allows for temperature control of solutions in smaller volumes than what can be currently performed using a thermocouple. While only DNA amplifications are shown in this report, this method of temperature control could easily be applied to other temperature-sensitive processes. No matter the application, use of a completely noncontact system will allow for simplification and cost reduction of microdevice fabrication as it alleviates the need for fabricating metal-based temperature sensors into the device. ACKNOWLEDGMENT The authors thank Micron Infrared, Inc. (Oakland, NJ) for loan of the initial pyrometer. SUPPORTING INFORMATION AVAILABLE Additional information as noted in text. This material is available free of charge via the Internet at http://pubs.acs.org. Received for review July 20, 2006. Accepted November 17, 2006. AC0613277