Anal. Chem. 2002, 74, 3592-3598
A Versatile Planar QCM-Based Sensor Design for Nonlabeling Biomolecule Detection Hiroyuki Sota,* Hiroshi Yoshimine, Robert F. Whittier, Masanori Gotoh, Yasuro Shinohara, and Yukio Hasegawa
Department of Research and Development, Amersham Biosciences K.K., 3-25-1, Hyakunincho, Shinjuku-ku, Tokyo 169-0073, Japan Yoshio Okahata
Department of Biomolecular Engineering, Tokyo Institute of Technology, 4259 Nagatsuda, Midori-ku, Yokohama 226-8501, Japan
Despite high theoretical sensitivity, low-cost manufacture, and compactness potentially amenable to lab-on-a-chip use, practical hurdles have stymied the application of the quartz crystal microbalance (QCM) for aqueous applications such as detection of biomolecular interactions. The chief difficulty lies in achieving a sufficiently stable resonance signal in the presence of even minute fluctuations in hydrostatic pressure. In this work, we present a novel versatile planar sensor chip design (QCM chip) for a microliter-scale on-line biosensor. By sealing the quartz resonator along its edges to a flat, solid support, we provide uniform support for the crystal face not exposed to solvent, greatly decreasing deformation of the crystal resonator under hydrostatic pressure. Furthermore, this cassette design obviates the need for direct handling when exchanging the delicate quartz crystal in the flow cell. A prototype 27-MHz sensor signal exhibited very low noise over a range of flow rates up to 100 µL/min. In contrast, signals obtained from a conventional QCM sensor employing an O-ring-based holder were less stable and deteriorated even further with increasing flow rate. Additional control designs with intermediate amounts of unsupported undersurface yielded intermediate levels of stability, consistent with the interpretation that deformation of the crystal resonator under fluctuating hydraulic pressure is the chief source of noise. As a practical demonstration of the design’s high effective sensitivity, we readily detected interaction between myoglobin and surfacebound antibody. Although great strides have been made in the development of microfluidic devices intended for state-of-the-art biological sample analysis, development of suitable liquid-phase sensors has lagged. Ideally, such sensors should possess high sensitivity but be miniaturizable and inexpensive to manufacture in quantity. In theory, quartz crystal microbalance (QCM) technology not only * To whom correspondence should be addressed: (fax) +81-3-5331-9361; (e-mail)
[email protected].
3592 Analytical Chemistry, Vol. 74, No. 15, August 1, 2002
meets these requirements but in addition can provide real-time nonlabeling detection of macromolecular interactions.1-3 Such QCM sensors are based upon quartz crystal resonators whose oscillations involve the interconversion of electrical and mechanical energy. The most sensitive sensors use AT-cut quartz crystals, whose physical oscillation occurs parallel to the crystal thickness (thickness shear mode). The metal electrodes deposited on opposite faces of the crystal wafers are amenable to chemical modification techniques for immobilization of macromolecules. To date, however, the difficulty of achieving high practical sensitivity in a solvent-immersed quartz crystal resonator, particularly one subject to solvent flow, has stymied efforts to create widely applicable QCM-based biosensors. Under any given set of conditions, a quartz crystal resonator will have an optimal resonance frequency. The theoretical mass sensitivity, i.e., the change in optimal frequency per change in unit mass, is proportional to the square of the fundamental frequency, as expressed by Sauerbrey’s equation.4
∆F/∆m ) -2F02A-1(Fqµq)-1/2 where ∆F and F0 are the shift in frequency and fundamental frequency expressed in hertz and A is the electrode area, Fq the density of quartz, and µq the shear modulus of quartz. Since the fundamental frequency is inversely proportional to the thickness of the quartz crystal, thinner crystals theoretically offer higher sensitivity. Commonly used quartz crystals (AT-cut) typically resonate at 5-10 MHz,5 requiring approximate thicknesses of 160-330 µm.1 In a few cases, thinner, more sensitive crystals have also been used. Lin et al. used a thicker quartz crystal chemically milled at its resonating center to 55 µm and fixed this into a holder using epoxy sealant instead of O-rings.6 (1) Janshoff, A.; Galla, H.-J.; Steinem, C. Angew. Chem., Int. Ed. 2000, 39, 40044032. (2) Okahata, Y.; Kawase, M.; Niikura, K.; Ohtake, F.; Furusawa, H.; Ebara, Y. Anal. Chem. 1998, 70, 1288-1296. (3) Caruso, F.; Rodda, E.; Furlong, D. N.; Niikura, K.; Okahata, Y. Anal. Chem. 1997, 69, 2043-2049. (4) Sauerbrey, G. Z. Phys. 1959, 155, 206-222. (5) Handley, J. Anal. Chem. 2001, 73, 225A-229A. 10.1021/ac025526b CCC: $22.00
© 2002 American Chemical Society Published on Web 06/22/2002
More important than the theoretical sensitivity discussed above is the practical sensitivity as determined by the stability of the optimum oscillation frequency. This stability is expressed as the quality factor Q, reflecting the sharpness of the energetic optimum. The following equation defines the relationship of Q to other parameters.7
Q ) 2πfsL1/R1 where fs is the mechanical resonance frequency, L1 is the motional inductance, and R1 is the motional resistance in the Butterworthvan Dyke equivalent circuit for a quartz resonator. This equation shows that Q is inversely proportional to motional resistance, which can also be described as energy dissipation. Such dissipation is greater in solvent than in air for two major reasons. The first problem is unavoidable for biosensor applications; aqueous solvents are denser and more viscous than air, leading to a greater transfer of acoustic energy from the crystal to the surrounding medium. A second problem is that the quartz crystal resonator is itself an electric device, requiring measures to prevent electrical shortcircuiting between the electrodes deposited on its opposite surfaces. While some form of sealing is needed, any physical constraint imposed upon the free, uniform mechanical oscillation of the crystal will lead to further dissipation of energy and lowering of the Q factor. In a typical arrangement, the naked resonator (AT-cut quartz crystal with metal electrode strips) is sealed and held within a small liquid chamber with O-rings, with one resonator face exposed to the solvent and the other face exposed to air.8-12 While most of the acoustic energy is concentrated between the opposing electrodes in the center, and can be further concentrated by asymmetric resonator designs,13 acoustic energy propagated laterally through the crystal wafer will be absorbed by the O-rings. Thus, sealing methods that reduce physical constraint have the potential to improve practical sensitivity. The considerations above are relevant to QCM biosensors operated under ideal hydrostatic conditions in batch mode. Most practical applications for monitoring separations or determining kinetic parameters require a biosensor that can operate under flow. Solvent flow adds a dynamic component in that the optimal resonance frequency is also exquisitely sensitive to fluctuations in hydraulic pressure.14 The more sensitive (i.e., thinner) the quartz crystal, the more susceptible it is to surface stress (deformation) from both hydrostatic and hydrodynamic pressure. Most flow-type QCM biosensors reported to date have utilized relatively thick, low-frequency resonators (9 MHz) coupled with relatively low turnover (e1 vol/min) and large flow cell volumes (6) Lin, Z.; Yip, C. M.; Joseph, I. S.; Ward, M. D. Anal. Chem. 1993, 65, 15461551. (7) Buttry, D. A.; Ward, M. D. Chem. Rev. 1992, 92, 1355-1379. (8) Fawcett, N. C.; Craven, R. D.; Zhang, P.; Evans, J. A. Anal. Chem. 1998, 70, 2876-2880. (9) Storri, S.; Santoni, T.; Minunni, M.; Mascini, M. Biosens. Bioelectron. 1998, 13, 347-357. (10) Lin, Z.; Ward, M. D. Anal. Chem. 1996, 68, 1285-1291. (11) Ma´sson, M.; Yun, K.; Haruyama, T.; Kobatake, E.; Aizawa, M. Anal. Chem. 1995, 67, 2212-2215. (12) Schneider, T. W.; Martin, S. J. Anal. Chem. 1995, 67, 3324-3335. (13) Hillier, A. C.; Ward, M. D. Anal. Chem. 1992, 64, 2539-2554. (14) Heusler, K. E.; Grzegorzewski, A.; Ja¨ckel, L.; Pietrucha, J. Ber. Bunsen-Ges. Phys. Chem. 1988, 92, 1218-1225.
(>1 mL).15,16 Recent publications from research groups in Germany mention 20-MHz resonators, but details of the flow cell volumes, resonator mounting, and signal stability are not specified.17,18 Here we report the development of a novel versatile planar sensor chip design utilizing a 27-MHz resonator housed in a flow cell with minimal sample requirements (10-µL flow cell volume). The design enables robust measurements with high stability (1 kHz) changes in the operating frequency. However, resistance to the small frequency variation typical of noise and drift may well encompass real-world parameters apart from the Q factor, such as the ability of the planar crystal to withstand dynamic physical deformation. Two related rectangular QCM chip designs were also tested. These were modified to reduce substrate support (b, c) to test
Figure 5. Panel A: schematic top and side views of the four quartz wafer mounting designs whose resonance stabilities are shown monitored in panel B. (a) QCM chip with full substrate support, (b) QCM chip with perimeter support only, (c) QCM chip with bridge support only, (d) O-ring-sandwiched circular resonator. Shaded portions in (a-c) indicate the areas of nonadhesive but distributive contact between the resonator and the solid support. The crosshatched portion in (d) indicates the area of the circular quartz resonator clamped directly between the O-rings. The open and filled arrows also indicate the expected of downward hydraulic pressure and counteracting upward mechanical force. Panel B: Comparison of signal stability between oscillators using various holding methods. (a) QCM chip, (b) “Perimeter-contact” type QCM chip, (c) “Bridge-like-contact”-type QCM chip, (d) O-Ring-sandwiched circular resonator. Note that the hertz scale for measurements made with the O-ring design under flow does not match the uniform measurement scales used for the other 13 graphs.
how curtailing substrate contact area would affect noise levels. These modified QCM chips exhibited relatively low noise levels along with Q factors similar to the fully supported design (data not shown). However, both modified chips displayed a pattern of noise level increments consistent with their degree of similarity to the O-ring design; signal stability deteriorated at higher flow rates in inverse correlation with the extent of surface contact area. We attribute this flow rate-independent noise level of the new QCM design to its mounting, a nonadhesive but distributive contact with the substrate, which effectively prevents deformation upon exposure of the crystal wafer to increased hydraulic pressure (Figure 5, panel A). The accurate piezoelectric properties of the quartz crystal resonators rely upon their precise crystal structure, in which the internal atomic dipoles align precisely according to their cut angles and respond collectively to an external charge to
produce a stable oscillation.7 The uniform bond angle strain among the atoms in the crystal lattice leads to a uniform misalignment of dipoles that is crucial for maintaining the collective motion. The drumlike configuration formed between the crystal wafer and its support in the conventional O-ring design, and to a lesser extent in the two modified QCM chip designs, seems to engender instability. Our full substrate support QCM chip design not only reduces noise to a level well below any we have seen reported for a liquid-immersed QCM based sensor but does so with a highfrequency, high-sensitivity quartz resonator. With high practical sensitivity in a robust, miniaturizable design, the QCM chip holds promise for the development of biosensors suitable to microfluidic applications. Real-Time Detection of Biomolecular Interactions. In Figure 6, we demonstrate real-time detection of biomolecular Analytical Chemistry, Vol. 74, No. 15, August 1, 2002
3597
Figure 6. Demonstration of the specific detection of a proteinprotein interaction using the sensor system with a QCM chip. Fifty microliters of a 5.0 µg/mL (0.30 µM) sheep myoglobin solution was injected onto a mouse anti-myoglobin- or BSA-preimmobilized sensor surface constructed through an alkanethiol (DTDP) monolayer on the detection electrode of a 27-MHz QCM chip. Running buffer: 50 mM HEPES, pH 7.4, containing 150 mM NaCl and 0.005% (vol/vol) surfactant P20. Flow rate, 100 µL/min. Operation temperature, 25.00 °C.
interactions using myoglobin/anti-myoglobin binding as an example. Taking advantage of the QCM chip’s signal stability at high flow cell volume turnover rates, we injected myoglobin and achieved clear discrimination between specific (anti-myoglobin)
3598
Analytical Chemistry, Vol. 74, No. 15, August 1, 2002
and nonspecific (BSA) interactions over a short observation period (∼3 min). Comparison to Published QCM Biosensor Data. While scientific reports have recently appeared using 20-MHz quartz crystal resonators in flow cell configurations, the details of resonator mounting, flow cell volume, and resonance stability have not been specifically reported. A graph drawn only to rough scale in one of these reports suggests a signal stability in the range of (2-3 Hz (Figure 2C of ref 16). While excellent in comparison to other published flow cell data, this noise level in combination with the lower frequency of their quartz crystal resonator suggests that the design reported here still achieves at least 40-fold higher effective mass sensitivity. ACKNOWLEDGMENT H.S. thanks Ms. K. Kimura of Amersham Biosciences K.K. for her continuous encouragement and secretarial help. This work was partly supported by the Ministry of Agriculture, Forestry and Fishery of Japan (MAFF). Received for review January 17, 2002. Revised manuscript received April 1, 2002. Accepted May 9, 2002. AC025526B
