APPLIED PHYSICS LETTERS
VOLUME 77, NUMBER 3
17 JULY 2000
Mechanical resonant immunospecific biological detector B. Ilic, D. Czaplewski, and H. G. Craigheada) School of Applied and Engineering Physics and the Nanobiotechnology Center, Cornell University, Ithaca, New York 14853
P. Neuzil Institute of Microelectronics, Singapore Science Park II, Singapore 117685
C. Campagnolo and C. Batt Department of Food Science and the Nanobiotechnology Center, Cornell University, Ithaca, New York 14853
共Received 11 February 2000; accepted for publication 24 May 2000兲 We have demonstrated high-sensitivity detection of bacteria using an array of bulk micromachined resonant cantilevers. The biological sensor is a micromechanical oscillator that consists of an array of silicon-nitride cantilevers with an immobilized antibody layer on the surface of the resonator. Measured resonant frequency shift as a function of the additional cell loading was observed and correlated to the mass of the specifically bound Escherichia coli O157:H7 cells. Deposition and subsequent detection of E. coli cells was achieved under ambient conditions. © 2000 American Institute of Physics. 关S0003-6951共00兲03429-X兴
flexural rigidity EI along the beam is constant, where E is the modulus of elasticity and I⫽wt 3 /12 the area moment of inertia of the rectangular beam, the resonant frequency of the oscillator can be approximated from the following general equation of the transverse mechanical vibrations:
There is widespread technological and scientific interest in the use of micro- and nanoelectromechanical systems as miniaturized biological sensors and actuators. In recent years, micro-1,2 and nanomechanical3 oscillators, in particular, have been suggested as a new class of resonant chemical and biological sensors. For instance, very delicate mechanical oscillators combined with sensitive displacement detection schemes4–9 have resulted in a number of remarkably powerful experimental techniques. These include various scanning force microscopy10 techniques, such as detection of specific molecular interactions,11–13 cell adhesion,14,15 and specific binding,16 and studies of both single-molecule dynamics17 and temporal changes of mechanical properties of enzymes,18 proteins,19 and living cells.20 In this work, detection of bacteria using a resonantfrequency-based mass detection biological sensor has been accomplished. The biological sensor is composed of an array of low-stress silicon-nitride cantilever beams that are fabricated using bulk micromachining techniques. Signal transduction of the resonant structures was accomplished by measuring the out-of-plane vibrational resonant mode using an optical deflection system. In order to test the feasibility of the process, measurements were made using the Dimension 3000 Digital Instruments 共DI兲 atomic-force microscope. In this setup, the reflected laser beam from the cantilever’s apex is sensed by a split photodiode, used as a position-sensitive detector 共PSD兲. The A – B signal from the PSD was extracted directly from the DI signal access module using a HP3562A dynamic signal analyzer 共Fig. 1兲. The measured operational mode of the cantilever is an out-of-plane translational vibration due to thermal mechanical noise and ambient vibrations in air. Therefore, the necessity to externally drive the mechanical oscillator was eliminated, simplifying the test apparatus. In general, neglecting damping and assuming that the
EI
共1兲
where z denotes the transverse displacement of the beam, (⬃3100 kg/m3) the mass density of the cantilever, w is the width, and t the thickness of the cantilever, which are perpendicular and parallel to the direction of bending, respectively. Using appropriate boundary conditions for a rectangular beam with one end fixed and the other free, Eq. 共1兲 yields the corresponding resonant frequency f⫽
共 1.875兲 2 t 2 l2
冑
E , 12
共2兲
where l denotes the length of the cantilever. From Eq. 共2兲 the approximate resonant frequency shift due to additional mass loading, assuming the cell mass is much less than the mass of the oscillator, is given by
FIG. 1. Schematic of the laser deflection setup used to measure the transverse vibrations of the mechanical oscillators.
a兲
Electronic mail:
[email protected]
0003-6951/2000/77(3)/450/3/$17.00
4z 2z ⫽0, 4 ⫹ wt y t2
450
© 2000 American Institute of Physics
Ilic et al.
Appl. Phys. Lett., Vol. 77, No. 3, 17 July 2000
⌬f⫽
共 1.875兲 2 4
冑
EI m load , 12 m 1.5 c
451
共3兲
where m load and m c are the added cell and unloaded oscillator mass, respectively. In order to simplify the system, Eq. 共3兲 further assumed that the flexural rigidity remained constant following the binding of the cells. The cantilever fabrication starts with 具 100典 silicon wafers covered with either 320 nm of low-pressure or 600 nm of plasma-enhanced chemical-vapor deposition 共LPCVD or PECVD兲 low-stress silicon nitride. Cantilever thickness was verified using an ellipsometer 共Rudolph Research model 43603兲. To define the resonating beam and cantilever substrate, photolithography is carried out on the frontside and the exposed silicon nitride is etched in a reactive ion etch 共RIE兲 chamber using CF4. Next, backside alignment is carried out and the back of the cantilever substrate is also defined by RIE. To prevent roughening of the frontside silicon nitride during subsequent KOH etching, 2 m of PECVD oxide was deposited. When backside etching neared the front, the PECVD oxide was removed via 共6:1兲 buffered oxide etch solution and the etch continued from both sides until the cantilevers were released. Cantilever lengths varied from 100 to 500 m. Removing the longer (⬎300 m) cantilevers intact from the subsequent rinse solution is made difficult by the surface tension of that solution, which tends to cause both stiction and breaking of the cantilevers. This problem was circumvented by the use of a high-pressure CO2 critical-point dryer. However, stiction was not a problem for the shorter (⬍300 m) cantilevers, which were used for cell detection. From the measured frequency versus length curve, the Youngs modulus for the PECVD and LPCVD low-stress silicon-nitride films was determined to be 93 and 110 GPa, respectively. After releasing, cantilevers were immersed into a solution of E. coli serotype O157:H7 antibodies 共Kirkegaard & Perry Laboratories Inc., Gaithersburg, MD兲 for 5 min. They were then rinsed in deionized water and nitrogen dried. Tapping-mode atomic-force topographs, in conjunction with an already developed surface scratching scanning probe microscope technique using a diamond-coated tapping-modeetched silicon probe, revealed an average antibody thickness of 40 nm across ten substrates. Cantilevers were then immersed into a buffer solution containing different concentrations of E. coli cells ranging from 106 to 109 E. coli cells/ml. The cells were incubated at room temperature for 15 min. Devices were first rinsed in a solution of 0.05% tween to remove any loosely bound cells, then in deionized water and nitrogen dried. Subsequent washing in tween did not affect the bound cells. The Leo 982 scanning electron microscope 共SEM兲, operating at 5 keV, was used to evaluate the topography of the resonator following cell binding. In Fig. 2, scanning electron micrographs show a random distribution of bound cells immobilized on the surface of various cantilevers. In order to determine the mass bound to each of the cantilevers in the array, frequency spectra were taken before and after binding of the cells to the antibodies. When a binding event occurred, the additional cell mass loading causes a shift in the resonant frequency of the micromechanical oscil-
FIG. 2. Scanning electron micrographs of cells bound to the immobilized antibody layer on the surface of various cantilevers. In order to reduce charging effects during SEM imaging, samples were prepared by evaporating a thin (⬍10 nm) layer of Au/Pd. Samples 共a兲, 共b兲, 共c兲, and 共d兲 were prepared in a similar manner and show a random distribution of the cells.
lator 共Fig. 3兲. In accord with Eq. 共3兲, we observed a linear dependence of the frequency shift in the regime where the total cell mass is much smaller than the mass of the oscillator. Slight deviations from linearity are attributed to nonuniform loading of the cells, due to their random distribution on the surface of the cantilever, causing variations in the flexural rigidity along the beam. To confirm the specificity of the binding events, cantilevers without the presence of immobi-
FIG. 3. Measured frequency shift vs the number of bound E. coli cells for a cantilever with 共a兲 l⫽100 m, w⫽20 m, and t⫽320 nm and 共b兲 l ⫽200 m, w⫽10 m, and t⫽600 nm. The dashed line represents a linear regression fit to our data. Data point ⌬ f ⫽0 was a consequence of several cantilevers that were measured after being treated with a buffer solution containing no cells, rinsed with deionized water 共18.2 M⍀ cm兲, and nitrogen dried. Cells were counted manually from an optical micrograph recorded using a Zeiss Axiotron AF optical microscope.
452
Ilic et al.
Appl. Phys. Lett., Vol. 77, No. 3, 17 July 2000
FIG. 4. The measured thermal noise spectra, corresponding to Fig. 3共a兲, due to the transverse vibrations of the cantilevers before (——) and after (——) cell attachment. The observed small signal-to-noise ratio is caused by the low reflectivity of the silicon nitride.
lized antibodies were treated with a buffer solution containing cells. No cells were bound to the cantilevers, and in turn, the resonant frequency of the oscillators remained unchanged. Additional experiments were performed to test specific binding, where cantilevers with an immobilized layer of E. coli antibodies were exposed to Salmonella typhimurium cells and no frequency shifts were detected. The lack of binding events was further confirmed optically. The mass sensitivity of the cantilevers depends on the mechanical quality factor Q of the device, which can be determined from the bandwidth at resonance. In the viscous regime, where considerable air damping occurs, the Q of the oscillator is proportional to the inverse square root of the pressure. Therefore, for vibrations in air at atmospheric pressure and room temperature, the Q is small, ranging between 5 and 8 共Fig. 4兲. For measurements in solutions where damping is extreme the Q would, and sensitivity would, be significantly lower. However, in the molecular regime, the quality factor is inversely proportional to the pressure so that when placed in a 1 mTorr vacuum at room temperature, the Q of the oscillators increases to 104 . The resolution of the frequency spectrum is related to the width of the peak, and thus the Q, with values of approximately 0.1 Hz in vacuum and 10 Hz at atmospheric pressure. The sensitivity of the devices is determined from the slopes of the curves in Figs. 3共a兲 and 3共b兲, to be 6.81 and 5.115 Hz/pg, respectively. The minimum detectable mass of the cantilevers 关Fig. 3共a兲兴 in vacuum is 14.7⫻10⫺15 g, and at atmospheric pressure it is 100 times as large, thus making the mass of a single cell easily detectable under vacuum conditions but not detectable at atmospheric pressure with this cantilever geometry. Furthermore, the uncontrolled placement of the cell along the cantilever will slightly vary the shift in frequency as observed during our experiments.
In summary, we have demonstrated a sensor for the detection of specifically bound cells immobilized on antibodies using microfabricated resonant structures. The cantileverbased detection method utilized here enabled detection of 16 E. coli cells, which corresponds to a mass of ⬃6⫻10⫺12 g. Sixteen is the lowest number of cells that we deposited during our experiments. In order to maximize sensitivity, enhancement in the resolution or the mechanical quality factor, through either vacuum encapsulation or tailoring cantilever dimensions would be needed. In conjunction with already developed optical sequential position readout from arrays of oscillators,9 we can envision a multielement nanomechanical detector utilizing arrays of microfabricated cantilevers, each coated with a distinct antibody or selective surface, with each element offering the ability to discriminately measure either single antibody analyte or other binding events. The authors would like to thank F. M. Serry 共Digital Instruments兲, R. Hall 共FDA兲, E. Merschrod, M. Cabodi, R. Orth, and A. G. Olkhovets, for helpful discussions, and the staff at the Cornell Nanofabrication Facility for generous aid in fabrication. This work was supported in part by the Defense Advanced Research Projects Agency and the National Science Foundation.
W. E. Newell, Science 161, 1320 共1964兲. K. E. Petersen, Proc. IEEE 70, 420 共1982兲. 3 D. W. Carr, S. Evoy, L. Sekaric, H. G. Craighead, and J. M. Parpia, Appl. Phys. Lett. 75, 920 共1999兲. 4 M. Tortonese, R. C. Barrett, and C. F. Quate, Appl. Phys. Lett. 62, 834 共1993兲. 5 W. H. King, Anal. Chem. 36, 1735 共1964兲. 6 N. Blanc, J. Brugger, and N. F. de Rooij, J. Vac. Sci. Technol. B 14, 901 共1996兲. 7 G. Meyer and N. M. Amer, Appl. Phys. Lett. 53, 1045 共1988兲. 8 D. Rugar, H. J. Mamin, and P. Guethner, Appl. Phys. Lett. 55, 2588 共1989兲. 9 H. P. Lang, R. Berger, C. Andreoli, J. Brugger, M. Despont, P. Vettiger, Ch. Gerber, J. K. Gimzewski, J. P. Ramseyer, E. Meyer, and H.-J. Guntherodt, Appl. Phys. Lett. 72, 383 共1998兲. 10 G. Binnig, C. F. Quate, and C. Gerber, Phys. Rev. Lett. 56, 930 共1986兲. 11 E.-L. Florin, V. T. Moy, and H. Gaub, Science 264, 415 共1994兲. 12 V. T. Moy, E.-L. Florin, and H. Gaub, Science 266, 257 共1994兲. 13 G. U. Lee, L. A. Chrisey, and R. J. Colton, Science 266, 771 共1994兲. 14 G. Sagvolden, I. Giaever, E. O. Pettersen, and J. Feder, Proc. Natl. Acad. Sci. USA 96, 471 共1999兲. 15 U. Dammer, O. Popescu, P. Wagner, D. Anselmetti, H.-J. Guntherodt, and G. N. Misevic, Science 267, 1173 共1995兲. 16 D. Baselt, G. U. Lee, K. M. Hansen, L. A. Chrisey, and R. J. Colton, Proc. IEEE 85, 672 共1997兲. 17 X. S. Xie and R. C. Dunn, Science 265, 361 共1994兲. 18 M. Radmacher, M. Fritz, H. G. Hansma, and P. K. Hansma, Science 265, 1577 共1994兲. 19 M. Reif, M. Gautel, F. Oesterhelt, J. M. Fernandez, and H. E. Gaub, Science 276, 1109 共1997兲. 20 E. A-Hassan, W. F. Heinz, M. D. Antonik, N. P. D’Costa, S. Nageswaran, C.-A. Schoenenberger, and J. H. Hoh, Biophys. J. 74, 1564 共1998兲. 1 2
