- Email: [email protected]
Large-scale arrays of nanomechanical sensors for biomolecular fingerprinting
Sensors and Actuators B 187 (2013) 111–117
Contents lists available at SciVerse ScienceDirect
Sensors and Actuators B: Chemical journal homepage: www.elsevier.com/locate/snb
Large-scale arrays of nanomechanical sensors for biomolecular fingerprinting C. Guthy a , M. Belov a , A. Janzen a , N.J. Quitoriano b , A. Singh a , V.A. Wright c , E. Finley d , T.I. Kamins e , S. Evoy a,∗ a
Department of Electrical and Computer Engineering, University of Alberta, 9107 - 116th St., Edmonton, Alberta, T6G 2V4, Canada Department of Mining and Materials Engineering, McGill University, Montreal, Quebec, H3A 2B2, Canada Department of Chemistry, University of Alberta, Edmonton, Alberta, T6G 2G2, Canada d National Institute for Nanotechnology, 11421 Saskatchewan Drive, Edmonton, Alberta, T6G 2M9, Canada e Department of Electrical Engineering, Stanford University, 350 Serra Mall, Stanford, CA, 94305, USA b c
a r t i c l e
i n f o
Article history: Received 14 July 2012 Received in revised form 12 September 2012 Accepted 21 September 2012 Available online 28 September 2012 Keywords: NEMS Nanoresonators Large-scale integration Electronic noses
a b s t r a c t A review of activities involving the development of large arrays of nanomechanical resonators is presented. This review includes demonstration of the use of these arrays for the detection of biological targets. Both top-down and bottom-up approaches to the realization of such arrays were developed. Using a top-down approach, a nanomachining method for the fabrication of large arrays of doublyclamped silicon carbonitride (SiCN) resonators with width as narrow as 16 nm and a yield approaching 100% was developed. The specific detection of protein-A using such resonator arrays functionalized with single domain antibody fragments (sdAb) was also demonstrated with femtogram-level mass sensitivity. A nano-imprinting based fabrication of these resonator arrays was also realized, opening up their potential for cost-effective manufacturing. On a bottom-up approach, resonant silicon nanowires were also produced using directed chemical vapor deposition methods. These bottom-up resonant nanowires were in turn successfully employed for the specific detection of streptavidin with attogram-level mass sensitivity. © 2012 Elsevier B.V. All rights reserved.
1. Introduction Detection of biological and chemical agents is critical to many areas of the life sciences including: disease diagnosis, drug molecule screening, and rapid analysis of various molecular systems. Technologies currently used for such assaying include mass spectrometry (MS), nuclear magnetic resonance spectroscopy (NMR), and enzyme-linked immunosorbent assay (ELISA). ELISA [1] is a widely employed, array-based, analytical technique for the parallel analysis of antigens or antibodies. This technique however requires fluorescent tagging, which may disrupt the biochemical properties being investigated. Other platforms, such as quartz crystal microbalance (QCM), surface acoustic wave sensors (SAW), and surface plasmon resonance sensors (SPR) offer tagless alternatives for the analysis of molecular mixtures. Neither of these techniques is however well-suited for the detection of a large number of analytes [2]. Micro- and nanoresonators have been shown to be promising platforms for the tagless array-based detection of molecular systems. The binding of the analyte onto the sensor surface is detected through a shift of resonant frequency induced by its added mass.
∗ Corresponding author. E-mail address: [email protected] (S. Evoy). 0925-4005/$ – see front matter © 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.snb.2012.09.070
The mass sensitivity of mechanical resonators scales favorably as their mass is reduced, offering a compelling path for the development of large arrays of sensors of exceptional sensitivities. Sub 100 nm-wide nanoelectromechanical systems (NEMS) were first reported by Carr et al. [3]. The properties of silicon in addition to stiction issues inherent to this process however limited the fabrication yield to less than 25% for widths below 50 nm. Sacrificial layer processes also typically require critical point drying (CPD) following immersion of the resonator in order to prevent stiction. Such critical point drying is known to leave contamination on the sensor surface that may interfere with the detection of biological analytes [4]. While Dalton range (∼10−24 g) mass sensitivity has been proposed as the ultimate limit of nanoresonator-based detection [5], femtogram (1 fg = 10−15 g) and attogram (1 ag = 10−18 g) level sensitivities have been widely reported [6–11]. For instance, Waggoner et al. demonstrated the detection of prostate specific antigen (PSA) with sensitivities down to 50 fg/mL using trampoline-like nanomechanical resonators in conjuction with a secondary mass labeling technique [12]. More recently, zeptogram (1 zg = 10−21 g) level mass resolutions have been demonstrated [13,14]. However, these state-of-the-art mass sensitivities were achieved only at cryogenic temperatures, and the employed fabrication techniques are not readily scalable for large array-based detection.
112
C. Guthy et al. / Sensors and Actuators B 187 (2013) 111–117
A review of activities involving the development of alternatives to traditional sacrificial layer nanomachining process for the realization of large arrays of nanomechanical resonators is here presented. These development efforts include both top-down and bottom-approach to the realization of such arrays. Using a topdown approach, we developed a novel nanomachining method for the fabrication of large arrays of doubly-clamped SiCN resonators with width as narrow as 16 nm with a yield approaching 100% [15]. This novel process combines the surface machining of a SiCN glassy layer with bulk machining techniques for the release of the device [15]. This process allows the realization of sub-20 nm wide and tens of micron long suspended structures with a yield approaching 100% without the need of critical point drying [15]. This process has also more recently produced the first sub-10 nm wide suspended structures ever realized by surface machining [16]. The specific detection of protein-A using these resonator arrays is now here reported. In addition, a nanoimprinting based approach to the fabrication of these devices was also realized, opening up the potential of this technology to cost-effective manufacturing [17]. A bilayer resist consisting of PMMA 495/LOR 3A allowed high fabrication yields for resonators of widths ranging from 120 nm to 300 nm, thicknesses of 40 nm and 70 nm, and a length of 14 m. To our knowledge, these 120 nm resonators are the narrowest suspended structures ever fabricated via nanoimprinting. The last few years have also seen the development of alternate “bottom-up” techniques for the fabrication of nanodevices. Bottom-up approaches offer the advantages of higher throughput, and potentially lower manufacturing costs. For example, silicon nanowires have been used to detect biological and chemical species [18]. These sensors operate by monitoring changes of electrical conductivity associated with the binding of the analyte to the nanowire. A single-walled carbon nanotube (SWNT) resonator has been described by Sazonova et al. [19]. Also, mechanical resonators have been produced from 43 nm diameter platinum [20] and 75 nm diameter silicon nanowires [14]. Both these structures were tested using magnetomotive actuation, which requires ultra-high vacuum and cryogenic cooling. We have developed cantilevered nanomechanical silicon resonators using a directed chemical vapor deposition method. We have recently reported the analysis of such wires with diameters as small as 40 nm using a simpler and cost-effective room-temperature interferometry technique [21]. Given that those dimensions compare to the mean free path of air molecules at ambient pressure, we specifically observed exceptionally high quality factors, as high as 7000, at atmospheric pressure, which is only a factor of ∼3 lower than that in vacuum [21]. We here report the successful use of these nanowire resonators for the specific detection of proteins with attogram-level mass sensitivity.
2. Experimental 2.1. Fabrication of nanoresonator arrays The SiCN nanoresonators were initially realized via a combination of electron beam lithography and surface nanomachining. A 50 nm thick SiCN layer was deposited by plasma-enhanced chemical vapor deposition (PECVD) onto single-crystal, (1 0 0), silicon wafers (500 m-thick, 100 mm-diameter) [15]. The SiCN-coated wafers were then annealed in a tube furnace at 500 ◦ C for 6 h which resulted in a tensile stress of ∼200 MPa. Next, the resonator beams and the supporting pads were patterned using electron beam lithography (EBL). A 30 nm thick Cr film, deposited by thermal evaporation and subsequently lifted off in acetone, was used as a mask for reactive ion etching. Finally, the resonators were released by anisotropic etching in KOH solution (35%) saturated with IPA [15,22].
This SiCN technology was subsequently employed in conjunction with nanoimprint lithography using a bilayer of PMMA 495/LOR as imprint resist. Two bilayer thicknesses were investigated. A 150 nm/150 nm bilayer was initially employed, and reproducibly yielded suspended beams as narrow as 300 nm. A 100 nm/100 nm bilayer was then employed, which yielded released devices as narrow as 120 nm. The imprint itself was performed using a Nanonex NX-2500 system at a temperature of 190 ◦ C, a pressure of 200 psi, and a hold time of 2 min. An oxygen plasma was then employed to remove any resist residue at the bottom of the pattern. This oxygen plasma cleaning was performed in a Trion reactive-ion etch (RIE) system, and at low pressures to ensure etch anisotropy (10 mT, 7 sccm O2 , 60 W). The etch time employed varied from 60 s to 120 s given that the residual layer height was slightly different for each imprint. Bottom-up cantilevered silicon nanowires were grown onto a silicon-on-insulator (SOI) wafer that consisted of a 7 m thick (1 1 0)-oriented device layer, and a 100 nm thick buried oxide layer [23]. Vertical {1 1 1} sidewalls were formed by patterning and etching trenches in the (1 1 0) top silicon device layer using KOH. Gold catalyst particles were then deposited from colloidal suspension using a dip-and-dry process. The CVD growth was performed using silane, HCl, and B2 H6 (the boron dopant source) as precursors. Silane preferentially decomposes on the gold surface at a growth temperature of T = 680 ◦ C. The deposited silicon dissolves in the gold, eventually forming a super-saturated Au–Si liquid alloy. Silicon atoms then precipitate from this alloy to form a nanowire whose diameter is about that of the gold particle. The nanowires are epitaxially grown horizontally from the vertical sidewall surface, and are thus rigidly anchored (covalently bonded) at their base. 2.2. Interferometric setup The experimental optical interferometry setup used for assaying of the resonance frequencies of the nanoresonators is shown in Fig. 1 [15,22]. The resonator arrays were mounted onto a piezoelectric element which was actuated by the tracking output of a spectrum analyzer (Agilent model 4411B). The beam of a laser diode ( = 655 nm) was directed through a beamsplitter and focused onto the substrate using a NA = 0.45 microscope objective. At resonance, motion of the nanoresonators relative to the substrate created a moving fringe pattern that was reflected back through the microscope objective, was redirected by the beamsplitter, and impinged on an AC-coupled photodetector (New Focus model 1601). 2.3. Biofunctionalization In order to utilize such SiCN nanoresonators for the specific detection of biological targets, analyte-specific functional layers need to be immobilized onto their surface. The binding affinity of these layers is then assessed by monitoring the resonant frequency shifts related to any added mass on each individual device. The specific detection of protein-A using single domain antibody fragments (sdAb) was here chosen as model system. Protein-A is a 56 kDa surface protein originally found in the cell wall of Staphylococcus aureus. The protein has found widespread use in biochemical research because of its ability to bind immunoglobulins. The resonator chips were cleaned in cold piranha (3:1 H2 SO4 :H2 O2 ) just prior to the experiments and stored in DI water. The resonance frequency of all devices was assessed prior to the start of the functionalization procedure. The samples were then transferred into 100% ethanol for 2 min in a petri dish. This was followed by silanization in 2% 3-aminopropyl(triethoxysilane) (APTES) in ethanol (pH 5 adjusted using glacial acetic acid for 2 min). The samples were washed again in 100% ethanol for 5 min followed by blow drying and curing in an incubator @ 80 ◦ C for
C. Guthy et al. / Sensors and Actuators B 187 (2013) 111–117
113
Fig. 1. Diagram of the optical interferometric setup employed for the resonant assaying of the nanoresonators.
30 min. After cooling, the cured samples were washed for 2 min in phosphate-buffered saline (PBS) buffer. The silanized samples were activated in freshly prepared 2% solution of gluteraldehyde in water for 1 h followed by washing in PBS buffer for 5 min. The samples were then incubated in 1 g/mL solution of sdAb for 1 h at 37 ◦ C in the incubator at a speed of 180,000 rpm. This was followed by washing in 10% ethanolamine for 10 min, then in PBS-tween (0.05%) for 5 min, one wash in PBS for 5 min and two washes in DI water for 2 min each. Ethanolamine treatment ensures the blocking of all unreacted aldehyde groups on resonator surface and prevents any non-specific attachment of analyte. The samples are blow-dried, and the resonance frequency of each device was measured. After the resonance characterization, the samples were washed in PBS buffer for 2 min. They were finally exposed to the target Protein-A (5 g/mL) for 1 h at 37 ◦ C in a rotating incubator at a speed of 180,000 rpm. The samples were then washed in PBS-tween (0.05%) for 5 min followed by one wash in PBS for 5 min and two washes in DI water for 2 min each. The resonance frequencies of the devices was assessed again immediately after exposure to the target protein.
In the case of the bottom-up silicon nanowires, the detection of proteins was rather demonstrated using streptavidin as target. Nanowires were first exposed to an O2 plasma (Harrick Plasma) for 90 s at 400–500 mTorr and a power of 18 W. This initial step cleaned their surfaces and generated the hydroxyl groups (OH) required for the bonding of methoxysilane. Another wet cleaning and hydroxylation process followed, using hot piranha for 120 s, an H2 O rinse, a 100% anhydrous ethanol wash, and drying in argon. This cleaning and hydroxylation process also allowed previously functionalized nanowires to be cleaned and reused. The resonant frequencies of the nanowires were assessed a first time immediately after this cleaning step (Fig. 2(a)). A monolayer of 3mercaptopropyl trimethoxysilane (C6 H16 O3 SiS) was deposited on Si surfaces immediately after the plasma cleaning by exposure to C6 H16 O3 SiS vapor at approximately 5 Torr for 18 h. This deposition was followed by a wash in 100% anhydrous ethanol to remove physisorbed C6 H16 O3 SiS molecules. The samples were then dried in argon, and the resonant frequencies measured again to determine any shifts induced by the silane deposition (Fig. 2(b)). Substrates of porous silicon were exposed to the C6 H16 O3 SiS alongside the
Fig. 2. Resonant response of two individual bottom-up Si nanowires. (a) Immediately after cleaning, (b) immediately after silanization, (c) immediately after biotinylation, and (d) immediately after specific capture of streptavidin.
114
C. Guthy et al. / Sensors and Actuators B 187 (2013) 111–117
nanowires, and measured using IR-spectroscopy to confirm the successful deposition of the linker molecule. The 3-mercaptopropyl trimethoxysilane served as a linker to the bind biotin when the samples were then immersed in a biotin solution consisting of 1 mg of Pierce AZ-Link Biotin HPDP 21341, C24 H37 N5 O3 S3 , 0.5 mL dimethyl-sulfoxide (DMSO) and 0.5 mL phosphate buffered saline (PBS, 20 mM, 7.4 pH) [24] for 1 h. The DMSO was replaced gradually by toluene, followed by a wash in 100% anhydrous ethanol, and drying in argon. The resonance of the nanowires was once again assessed to measure the frequency shift associated to the attachment of the biotin (Fig. 2(c)). Biotin has a well-known affinity and selectivity to streptavidin [25], and the attachment of streptavidin was performed by immersing the samples in a mixture of 50 L streptavidin (of concentration 10 mg/mL) and 50 L of PBS (10 mM with 0.1% nonionic surfactant Triton X) for 1 h. Following this immersion, the buffered solution was gradually replaced by Millipore-purified H2 O. The samples were then washed in chloroform, 100% anhydrous ethanol, and dried with argon. The resonant frequency was measured one last time following this specific capture of the streptavidin (Fig. 2(d)). Negative control experiments were also performed in which the biotinylated nanowires (Fig. 2(c)) were exposed to a solution of streptavidin that was already saturated with biotin. This presaturation was performed by mixing 0.5 mg biotin, 0.5 mL DMSO and 0.5 mL PBS with the streptavidin solution. The binding sites of this streptavidin being already filled with biotin, it was not expected to attach to the biotin located on the nanowires.
3. Results and discussion 3.1. SiCN NEMS A typical (5 × 5) array of of doubly-clamped SiCN nanomechanical resonators produced using “top down” nanomachining that contains 24 nominally identical devices is shown in Fig. 3(a). A close-up image of an individual 16 nm wide, 14 m long resonator is shown in Fig. 3(b). Resonators as narrow as 8 nm have been obtained using electron beam lithography exposure [15,16]. Fig. 3(c) shows a typical resonant frequency spectrum obtained from a SiCN device. The experimental data obtained for the specific detection of protein-A using the SiCN nanoarrays are summarized in Table 1. Immobilization of single domain antibody fragments (sdAb) onto the nanoresonator arrays resulted in consistent resonant frequency downshifts of ∼341 kHz, corresponding to an attached mass of 12.3 fg or 58,000 sdAb molecules. The frequencies further downshift by ∼216 kHz due to protein A attachment (Fig. 4), corresponding to the capture of 7.7 fg or 11,000 molecules of Protein A. Another “negative control” array II (Fig. 5) was processed identically except Protein-A was not added to the solution. The frequency shifts due to this “solvent only” step are much smaller (∼54 kHz or 1.9 fg), further supporting the assertion that the significant frequency downshifts in array I are indeed due to protein-A capture. Fig. 6 shows a typical 300 nm-wide resonator structure realized via the imprinting of a 300 nm thick PMMA 495/LOR 3A resist bilayer [17]. Imprint temperatures of at least 70 ◦ C above the glass transition temperature Tg = 105 ◦ C of the PMMA allowed an optimal fabrication yield of 97%. Line-widening effects related to the oxygen plasma etch resulted in beams that were 20–30 nm wider than the design dimensions. The minimum resonator width obtained using a 200 nm-thick resist bilayer was approximately 120 nm. The fabrication yield however rapidly dropped to below 50% for narrower structures. These 120 nm-wide structures are however the narrowest suspended structures realized via nanoimprinting.
Fig. 3. (a) Scanning electron micrograph (SEM) of a 5 × 5 array of doubly-clamped 16 nm wide nanomechanical resonators fabricated by electron-beam lithography with a yield of 100%. (b) Close-up SEM of a 16 nm wide and 14 m long resonator. (c) Typical resonance curve for a 16 nm wide, 50 nm thick and 10 m long resonator. From the FWHM of the peak a Q-factor of ∼1000 is obtained.
Given the high-tensile stress present in the device layer, resist imperfections such as nicks along the device edge were found to be highly detrimental to the successful realization of sub-100 nm wide suspended structures. Imperfections in the resist undercut profile are likely to cause the observed edge roughness. Arrays of nanoresonators with widths in the 120–300 nm range, thicknesses of 40 nm and 70 nm, and a length of 14 m were realized. The 40 nm thick and 70 nm thick devices showed average resonant frequencies of ∼16.6 MHz and 21.7 MHz, respectively [17] and quality factors in the 6000–8000 range. We also demonstrated the specific detection of streptavidin with mass sensitivities of ∼3.5 fg using such nanoresonator arrays functionalized with biotin [17].
C. Guthy et al. / Sensors and Actuators B 187 (2013) 111–117
115
Table 1 Summary of the average resonant frequency shifts and corresponding added masses of the two SiCN NEMS arrays (I and II) due to capture of single domain antibody fragments (sdAb), protein-A and “solvent only” step. Attach-ment
F (kHz)
Error (kHz)
m (fg)
m/S (fg/m2 )
Sample
sdAb Prot A
341 216
±5 ±5
12.3 7.7
1.4 0.9
58,000 11,000
Control
sdAb Sol.
248 54
±5 ±5
8.9 1.9
1.0 0.2
42,000 3,000
Fig. 4. Resonant frequencies of the 24 individual resonators of Array I. The blue diamonds, green squares and red circles designate the individual device frequencies of the bare resonators, resonators with immobilized antibodies and resonators with captured protein-A, respectively. Immobilization of antibodies results in frequency downshifts of ∼341 kHz. Capture of protein-A further down-shifts the frequencies by ∼216 kHz. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.)
3.2. Si nanowires The nanowire dimensions were measured in a high-resolution Hitachi FE 4800 scanning electron microscope (SEM). Typical nanowire lengths and diameters ranged from 3 to 4 m and from 100 to 150 nm, respectively. The height of the wires above the substrate ranged from 2 to 7 m. Fig. 7 shows arrays of nanomechanical resonators produced by the bottom-up CVD method [20]. Fig. 2 schematically shows resonant response of two nanowires prior to any chemical attachment (Fig. 2(a)), after silanization (Fig. 2(b)), after biotin attachment (Fig. 2(c)), and after the specific
Fig. 5. Resonant frequencies of the 24 individual resonators of Array II. This “negative control” array was processed identically except protein-A was not added to the solution in the detection step. Similarly to Array I, immobilization of antibodies results in frequency downshifts of ∼248 kHz. The second “solvent only” step results in much smaller frequency downshifts of ∼54 kHz.
Total # of molecules
Fig. 6. Scanning electron micrograph of a 300 nm-wide SiCN resonator element fabricated by imprint lithography.
capture of the target streptavidin (Fig. 2(d)). The silanization step caused a net downward shift of resonant frequency by f = 33 ± 1 kHz. The attachment of biotin onto the nanowires caused a further downward frequency shift of f ∼ 25 kHz, corresponding to an added mass of 63 ag. The molecular weight of a biotin molecule is 540 Da, or 0.9 zg. The experimentally observed added mass therefore corresponded to a biotin surface density of 6.5 molecules per 100 nm2 . Given that a biotin molecule can occupy [26] an area 8.5 nm2 , this experimental finding indicates that the biotin covers almost 55% of the nanowire surface. The specific capture of streptavidin caused an additional frequency shift of 120 kHz, corresponding to an attached mass of 300 ag. The molecular weight of a single streptavidin molecule is 60 kDa, or 99.6 zg. This added mass corresponded to a
Fig. 7. 40 nm-wide bottom-up silicon nanowire resonators produced by chemical vapor deposition.
116
C. Guthy et al. / Sensors and Actuators B 187 (2013) 111–117
streptavidin surface density of 0.3 molecules per 100 nm2 . Given that a streptavidin molecule occupies [27] an area of 100 nm2 , this experimental finding indicates that the streptavidin covers approx. 30% of the nanowire surface. Finally, exposure of similarly biotinylated nanowires to a solution of pre-saturated (by the biotin) streptavidin showed negligible shifts, demonstrating the specificity of the streptavidin capture. 4. Conclusions Large arrays of nanomechanical resonators as narrow as 8 nm have been produced using a novel SiCN surface machining method. These arrays were used to detect antigen–antibody binding events without the need of extrinsic markers. The specific detection of protein-A using single domain antibody fragments (sdAb) was demonstrated as proof of concept. Although resonator arrays produced by electron beam lithography (EBL) show excellent uniformity and reproducibility, the serial nature of the EBL technique is not suitable for low cost, large-scale manufacturing However, this SiCN fabrication technology was recently employed in conjunction with nanoimprint lithography, demonstrating its amenability to cost-efficient manufacturing. Using a bottom-up CVD approach we also produced arrays of resonant nanowires that are promising for high sensitivity applications. This technique, similarly to other bottom-up approaches, is however less suitable for the realization of very large arrays. The specific detection of streptavidin was accomplished with these devices. In both cases an optical interferometric method was used to monitor the resonant frequency changes of the devices induced by the attachment of molecular systems. The readout was performed with a simple and cost-effective optical technique, and is therefore conducive to the development of large arrays of individual sensors. Akin to the concept of chemoresistive electronic noses, large arrays of nanomechanical resonators would enable the development of nanoresonator-based ‘electronic ears’ for the fingerprinting of metabolomics systems. Acknowledgments This work was supported by the National Research Council of Canada, the Province of Alberta, and the Natural Science and Engineering Research Council of Canada. Fabrication of the SiCN resonators was performed in the University of Alberta Nanofab facilities. The bottom-up Si nanowires were fabricated at HewlettPackard Laboratories, Palo Alto, CA References [1] R.M. Lequin, Enzyme immunoassay (EIA)/enzyme-linked immunosorbent assay (ELISA), Clinical Chemistry 51 (2005) 2415–2418. [2] H.J. Lee, A.W. Wark, R.M. Corn, Microarray methods for protein biomarker detection, Analyst 133 (2008) 975–983. [3] D.W. Carr, S. Evoy, L. Sekaric, J.M. Parpia, H.G. Craighead, Measurement of mechanical resonance and losses in nanometer scale silicon wires, Applied Physics Letters 75 (1999) 920–922. [4] Y. Wang, J.A. Henry, A.T. Zehnder, M.A. Hines, Surface chemical control of mechanical energy losses in micromachined silicon structures, Journal of Physical Chemistry B 107 (2003) 14270–14277. [5] K.L. Ekinci, Y.T. Yang, M.L. Roukes, Ultimate limits to inertial mass sensing based upon nanoelectromechanical systems, Journal of Applied Physics 95 (2004) 2682–2689. [6] N.V. Lavrik, P.G. Datskos, Femtogram mass detection using photothermally actuated nanomechanical resonators, Applied Physics Letters 82 (2003) 2697–2699. [7] B. Ilic, Y. Yanh, H.G. Craighead, Virus detection using nanoelectromechanical devices, Applied Physics Letters 85 (2004) 2604–2606. [8] L.M. Fischer, V.A. Wright, C. Guthy, N. Yang, M.T. McDermott, J.M. Buriak, S. Evoy, Specific detection of proteins using nanomechanical resonators, Sensors and Actuators B 134 (2008) 613–617.
[9] M. Li, H.X. Tang, M.L. Roukes, Ultra-sensitive NEMS-based cantilevers for sensing, scanned probe and very high-frequency applications, Nature Nanotechnology 2 (2007) 114–120. [10] K.L. Ekinci, X.M. Huang, M.L. Roukes, Ultrasensitive nanoelectromechanical mass detection, Applied Physics Letters 84 (2004) 4469–4471. [11] B. Ilic, H.G. Craighead, S. Krylov, W. Senaratne, C. Ober, P. Neuzil, Attogram detection using nanoelectromechanical oscillators, Journal of Applied Physics 95 (2004) 3694–3713. [12] P.S. Wagoneer, M. Varshney, H.G. Craighead, Detection of prostate specific antigen with nanomechanical resonators, Lab Chip 9 (2009) 3095–3099. [13] Y.T. Yang, C. Callegari, X.L. Feng, K.L. Ekinci, M.L. Roukes, Zeptogram-scale nanomechanical mass sensing, Nano Letters 6 (2006) 583–586. [14] X.L. Feng, R. He, P. Yang, M.L. Roukes, Very high frequency silicon nanowire electromechanical resonators, Nano Letters 7 (2007) 1953–1959. [15] C. Guthy, R. Das, B. Drobot, S. Evoy, Resonant characteristics of ultranarrow SiCN nanomechanical resonators, Journal of Applied Physics 108 (2010) 014306. [16] M.A. Mohammad, S.K. Dew, S. Evoy, M. Stepanova, Fabrication of sub-10 nm silicon carbon nitride resonators using a hydrogen silsesquioxane mask patterned by electron beam lithography, Microelectronic Engineering 88 (2011) 2338–2341. [17] A. Janzen, S. Poshtiban, A. Singh, S. Evoy, Fabrication of nanoresonator biosensing arrays using nanoimprint lithography, Journal of Micro/Nanolithography, MEMS, and MOEMS 11 (2012) 023013. [18] Y. Cui, Q. Wei, H.Ch. Park, M. Lieber, Nanowire nanosensors for highly sensitive and selective detection of biological and chemical species, Science 293 (2001) 1289–1292. [19] V. Sazonova, Y. Yaish, H. Ustunel, D. Roundy, T.A. Arias, P.L. McEuen, A tunable carbon nanotube electromechanical oscillator, Nature 431 (2004) 284–287. [20] A. Husain, J. Hone, W. Henk, Ch. Postma, X.M.H. Huang, T. Drake, M. Barbic, A. Scherer, M.L. Roukes, Nanowire-based very-high-frequency electromechanical resonator, Applied Physics Letters 83 (2003) 1240–1242. [21] M. Belov, N.J. Quitoriano, S. Sharma, W.K. Hiebert, T.I. Kamins, S. Evoy, Mechanical resonance of clamped silicon nanowires measured by optical interferometry, Journal of Applied Physics 103 (2008) 074304. [22] L.M. Fischer, N. Wilding, M. Gel, S. Evoy, Low-stress silicon carbonitride for the machining of high-frequency nanomechanical resonators, Journal of Vacuum Science and Technology B 25 (2007) 33–37. [23] M.S. Islam, S. Sharma, T.I. Kamins, R.S. Williams, Ultrahigh-density silicon nano-bridges formed between two vertical silicon surfaces, Nanotechnology 15 (2004) L5–L8. [24] A. Janshoff, K.S. Dancil, C. Steinem, D.P. Greiner, V.S.-Y. Lin, C. Gurtner, K. Motesharei, M.J. Sailor, M.R. Ghadiri, Macroporous p-type silicon Fabry–Perot layers. Fabrication, characterization, and applications in biosensing, Journal of the American Chemical Society 120 (1998) 12108–12116. [25] J. Dechancie, K.N. Houk, The origins of femtomolar protein–ligand binding: hydrogen bond cooperativity and desolvation energetics in the biotin–(strept)avidin binding site, Journal of the American Chemical Society 129 (2007) 5419–5429. [26] Instructions EZ-Link Biotin-HPDP 21341, Pierce, www.piercenet.com [27] W.A. Hendrickson, A. Pahler, J. Smith, Y. Satow, E.A. Merritt, R.P. Phizackerley, Crystal structure of core streptavidin determined from multiwavelength anomalous diffraction of synchrotron radiation, Proceedings of the National Academy of Sciences of the United States of America 86 (1989) 2190–2194.
Biographies Csaba Guthy is a Research Associate at the Department of Electrical and Computer Engineering of the University of Alberta in Edmonton, Canada. He received B.Sc. and M.Sc. degrees in physics from the University of Debrecen, Hungary in 2001 and M.Sc. and Ph.D. in materials science and engineering from the University of Pennsylvania (Philadelphia, PA, USA) in 2007. He has conducted extensive research in diverse areas such as thermal transport in carbon nanotubes, polymer composites, resonance behavior of nanoelectromechanical systems (NEMS), diffusion processes in intermetallic compounds, and development of a thermoacoustic thermal-to-electrical power converter. His research interests include the development of complex, highlyreliable, robust and unique technological solutions, R&D of advanced technologies for rapid detection and identification of chemical, biological and explosive threats, advanced non-destructive testing technologies, development of integrated sensors, system integration, innovative medical diagnostics technologies, novel energy storage technologies for automotive and stationary applications, modeling and simulation. Miro Belov received physics degrees, B.Sc. and M.Sc., from the Comenius University, Bratislava, Ph.D. from the University of Alberta, Edmonton, and have worked in both academic and industrial R&D environments. His education and research experience are rooted in condensed matter physics, NEMS, optics and magnetics. His research interests include high-resolution (ultrafast) measurements of (bio-) nano-mechanical, optical and magnetic structures, micro-, nano-fabrication for biomedical applications. Alex Janzen received a B.Sc. in engineering physics in 2009 and an M.Sc. in microelectromechanical systems in 2011 from the University of Alberta in Edmonton Canada. His M.Sc. work involved fabricating nanomechanical resonator devices
C. Guthy et al. / Sensors and Actuators B 187 (2013) 111–117 using nanoimprint lithography, and the testing of these devices for the specific detection of proteins. He currently works as a Research Associate at the University of Alberta involving work with boron-doped polysilicon in collaboration with Micralyne. Nathaniel Quitoriano is an Assistant Professor at the Department of Mining and Materials Engineering of McGill University in Montreal Canada. Dr. Quitoriano heads the Semiconductor Nanostructures Lab (SNL) which researches device applications of metal-catalyzed semiconductor growth, including semiconductor nanowires. Applications of particular interest include: solar cells, lasers, photo-diodes, and sensors. In 2000, Nate Quitoriano received Bachelor’s degrees in both electrical engineering and computer science and materials science and engineering from the University of California, Berkeley. As an undergraduate, he worked with Professor Tim Sands on establishing the kinetics of an ohmic, transient-liquid-phase bond for semiconductors. Nate received his Ph.D. in materials science engineering in 2006 at the Massachusetts Institute of Technology under the supervision of Gene Fitzgerald. While at MIT, he worked on III–V, lattice-mismatched semiconductors and grew high-quality InP on GaAs using graded, compositional buffers InGaAs and InGaP to slowly increase the lattice constant. Following MIT, he worked in Stan William’s group at Hewlett-Packard Laboratories under the direction of Ted Kamins where he studied Si and Ge nanowires for use as sensors and electrical devices and successfully demonstrated Si nanotube resonators and guided Si nanowire growth. Amit Singh is currently an Associate Research Scientist at the Department of Pharmaceutical Sciences, Northeastern University, Boston, USA. He received his M.Sc. and Ph.D. in biotechnology from the University of Pune, India. He worked as a postdoctoral fellow in Dr. Evoy’s laboratory at the University of Alberta and the National Institute for Nanotechnology. His research interests include multifunctional nanomaterial synthesis, characterization materials and their biomedical applications, biosensors, drug-delivery systems and cancer therapeutics. Vincent A. Wright received a Ph.D. degree in chemistry at the University of British Columbia, Vancouver in 2005. He worked as a post-doctoral fellow at the Univerisity of Alberta in Jillian Buriak’s research group investigating the functionalization of silicon, silica, and stainless steel surfaces. He currently works as a Research Scientist at Gilead Alberta ULC.
117
Eric Finley received a B.Sc. in physics from the University of Alberta, Edmonton in 1998. He is currently serving as Technical Officer at the National Institute of Nanotechnology in Edmonton Canada. Ted Kamins is a Consulting Professor in the Electrical Engineering Department at Stanford University. He received his B.S., M.S. and Ph.D. degrees from the University of California, Berkeley. He then joined the Research and Development Laboratory of Fairchild Semiconductor, where he performed early work contributing to the understanding of the then-emerging field of polycrystalline silicon and also contributed to device applications of epitaxial silicon. After moving to Hewlett-Packard, he worked in a number of materials and device-related areas, beginning with the development of UV-sensitive photodiodes, which are enabling devices for spectrophotometers – instruments in production for more than 20 years, first by HP and now by Agilent Technologies. He then contributed to the emerging areas of silicon-on-insulator and rapid thermal processing. Subsequent work dealt with advanced epitaxy and device technology for the silicon-germanium, heterojunction bipolar transistor, which is currently becoming widely used for wireless communications. Most recently Ted was a Principal Scientist in the Information and Quantum Systems Laboratory at Hewlett-Packard Laboratories in Palo Alto, California, where he was focusing on advanced nanostructured electronic materials and devices. Stephane Evoy received a B.Eng. and M.Sc.A. in engineering physics from the Ecole Polytechnique de Montréal, in 1992 and 1994, respectively. He also received Ph.D. in Applied Physics from Cornell University in 1998. He is currently Associate Professor in the Department of Electrical and Computer Engineering at the University of Alberta. Dr. Evoy has co-authored over 60 papers in the area of biosensors, nanoelectromechanical systems (NEMS), nanofabrication, and scanning probe microscopy. He co-edited the 2004 textbook: “Introduction to Nanoscale Science and Engineering.” His research currently focuses on cantilever- and SPR-based detection of metabolites and bacterial pathogens, the development of sample preparation and detection technologies leveraging the natural affinities of bacteriophage, as well as the development of metabolomics fingerprinting platforms based on large arrays of nanomechanical resonators.
Contents lists available at SciVerse ScienceDirect
Sensors and Actuators B: Chemical journal homepage: www.elsevier.com/locate/snb
Large-scale arrays of nanomechanical sensors for biomolecular fingerprinting C. Guthy a , M. Belov a , A. Janzen a , N.J. Quitoriano b , A. Singh a , V.A. Wright c , E. Finley d , T.I. Kamins e , S. Evoy a,∗ a
Department of Electrical and Computer Engineering, University of Alberta, 9107 - 116th St., Edmonton, Alberta, T6G 2V4, Canada Department of Mining and Materials Engineering, McGill University, Montreal, Quebec, H3A 2B2, Canada Department of Chemistry, University of Alberta, Edmonton, Alberta, T6G 2G2, Canada d National Institute for Nanotechnology, 11421 Saskatchewan Drive, Edmonton, Alberta, T6G 2M9, Canada e Department of Electrical Engineering, Stanford University, 350 Serra Mall, Stanford, CA, 94305, USA b c
a r t i c l e
i n f o
Article history: Received 14 July 2012 Received in revised form 12 September 2012 Accepted 21 September 2012 Available online 28 September 2012 Keywords: NEMS Nanoresonators Large-scale integration Electronic noses
a b s t r a c t A review of activities involving the development of large arrays of nanomechanical resonators is presented. This review includes demonstration of the use of these arrays for the detection of biological targets. Both top-down and bottom-up approaches to the realization of such arrays were developed. Using a top-down approach, a nanomachining method for the fabrication of large arrays of doublyclamped silicon carbonitride (SiCN) resonators with width as narrow as 16 nm and a yield approaching 100% was developed. The specific detection of protein-A using such resonator arrays functionalized with single domain antibody fragments (sdAb) was also demonstrated with femtogram-level mass sensitivity. A nano-imprinting based fabrication of these resonator arrays was also realized, opening up their potential for cost-effective manufacturing. On a bottom-up approach, resonant silicon nanowires were also produced using directed chemical vapor deposition methods. These bottom-up resonant nanowires were in turn successfully employed for the specific detection of streptavidin with attogram-level mass sensitivity. © 2012 Elsevier B.V. All rights reserved.
1. Introduction Detection of biological and chemical agents is critical to many areas of the life sciences including: disease diagnosis, drug molecule screening, and rapid analysis of various molecular systems. Technologies currently used for such assaying include mass spectrometry (MS), nuclear magnetic resonance spectroscopy (NMR), and enzyme-linked immunosorbent assay (ELISA). ELISA [1] is a widely employed, array-based, analytical technique for the parallel analysis of antigens or antibodies. This technique however requires fluorescent tagging, which may disrupt the biochemical properties being investigated. Other platforms, such as quartz crystal microbalance (QCM), surface acoustic wave sensors (SAW), and surface plasmon resonance sensors (SPR) offer tagless alternatives for the analysis of molecular mixtures. Neither of these techniques is however well-suited for the detection of a large number of analytes [2]. Micro- and nanoresonators have been shown to be promising platforms for the tagless array-based detection of molecular systems. The binding of the analyte onto the sensor surface is detected through a shift of resonant frequency induced by its added mass.
∗ Corresponding author. E-mail address: [email protected] (S. Evoy). 0925-4005/$ – see front matter © 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.snb.2012.09.070
The mass sensitivity of mechanical resonators scales favorably as their mass is reduced, offering a compelling path for the development of large arrays of sensors of exceptional sensitivities. Sub 100 nm-wide nanoelectromechanical systems (NEMS) were first reported by Carr et al. [3]. The properties of silicon in addition to stiction issues inherent to this process however limited the fabrication yield to less than 25% for widths below 50 nm. Sacrificial layer processes also typically require critical point drying (CPD) following immersion of the resonator in order to prevent stiction. Such critical point drying is known to leave contamination on the sensor surface that may interfere with the detection of biological analytes [4]. While Dalton range (∼10−24 g) mass sensitivity has been proposed as the ultimate limit of nanoresonator-based detection [5], femtogram (1 fg = 10−15 g) and attogram (1 ag = 10−18 g) level sensitivities have been widely reported [6–11]. For instance, Waggoner et al. demonstrated the detection of prostate specific antigen (PSA) with sensitivities down to 50 fg/mL using trampoline-like nanomechanical resonators in conjuction with a secondary mass labeling technique [12]. More recently, zeptogram (1 zg = 10−21 g) level mass resolutions have been demonstrated [13,14]. However, these state-of-the-art mass sensitivities were achieved only at cryogenic temperatures, and the employed fabrication techniques are not readily scalable for large array-based detection.
112
C. Guthy et al. / Sensors and Actuators B 187 (2013) 111–117
A review of activities involving the development of alternatives to traditional sacrificial layer nanomachining process for the realization of large arrays of nanomechanical resonators is here presented. These development efforts include both top-down and bottom-approach to the realization of such arrays. Using a topdown approach, we developed a novel nanomachining method for the fabrication of large arrays of doubly-clamped SiCN resonators with width as narrow as 16 nm with a yield approaching 100% [15]. This novel process combines the surface machining of a SiCN glassy layer with bulk machining techniques for the release of the device [15]. This process allows the realization of sub-20 nm wide and tens of micron long suspended structures with a yield approaching 100% without the need of critical point drying [15]. This process has also more recently produced the first sub-10 nm wide suspended structures ever realized by surface machining [16]. The specific detection of protein-A using these resonator arrays is now here reported. In addition, a nanoimprinting based approach to the fabrication of these devices was also realized, opening up the potential of this technology to cost-effective manufacturing [17]. A bilayer resist consisting of PMMA 495/LOR 3A allowed high fabrication yields for resonators of widths ranging from 120 nm to 300 nm, thicknesses of 40 nm and 70 nm, and a length of 14 m. To our knowledge, these 120 nm resonators are the narrowest suspended structures ever fabricated via nanoimprinting. The last few years have also seen the development of alternate “bottom-up” techniques for the fabrication of nanodevices. Bottom-up approaches offer the advantages of higher throughput, and potentially lower manufacturing costs. For example, silicon nanowires have been used to detect biological and chemical species [18]. These sensors operate by monitoring changes of electrical conductivity associated with the binding of the analyte to the nanowire. A single-walled carbon nanotube (SWNT) resonator has been described by Sazonova et al. [19]. Also, mechanical resonators have been produced from 43 nm diameter platinum [20] and 75 nm diameter silicon nanowires [14]. Both these structures were tested using magnetomotive actuation, which requires ultra-high vacuum and cryogenic cooling. We have developed cantilevered nanomechanical silicon resonators using a directed chemical vapor deposition method. We have recently reported the analysis of such wires with diameters as small as 40 nm using a simpler and cost-effective room-temperature interferometry technique [21]. Given that those dimensions compare to the mean free path of air molecules at ambient pressure, we specifically observed exceptionally high quality factors, as high as 7000, at atmospheric pressure, which is only a factor of ∼3 lower than that in vacuum [21]. We here report the successful use of these nanowire resonators for the specific detection of proteins with attogram-level mass sensitivity.
2. Experimental 2.1. Fabrication of nanoresonator arrays The SiCN nanoresonators were initially realized via a combination of electron beam lithography and surface nanomachining. A 50 nm thick SiCN layer was deposited by plasma-enhanced chemical vapor deposition (PECVD) onto single-crystal, (1 0 0), silicon wafers (500 m-thick, 100 mm-diameter) [15]. The SiCN-coated wafers were then annealed in a tube furnace at 500 ◦ C for 6 h which resulted in a tensile stress of ∼200 MPa. Next, the resonator beams and the supporting pads were patterned using electron beam lithography (EBL). A 30 nm thick Cr film, deposited by thermal evaporation and subsequently lifted off in acetone, was used as a mask for reactive ion etching. Finally, the resonators were released by anisotropic etching in KOH solution (35%) saturated with IPA [15,22].
This SiCN technology was subsequently employed in conjunction with nanoimprint lithography using a bilayer of PMMA 495/LOR as imprint resist. Two bilayer thicknesses were investigated. A 150 nm/150 nm bilayer was initially employed, and reproducibly yielded suspended beams as narrow as 300 nm. A 100 nm/100 nm bilayer was then employed, which yielded released devices as narrow as 120 nm. The imprint itself was performed using a Nanonex NX-2500 system at a temperature of 190 ◦ C, a pressure of 200 psi, and a hold time of 2 min. An oxygen plasma was then employed to remove any resist residue at the bottom of the pattern. This oxygen plasma cleaning was performed in a Trion reactive-ion etch (RIE) system, and at low pressures to ensure etch anisotropy (10 mT, 7 sccm O2 , 60 W). The etch time employed varied from 60 s to 120 s given that the residual layer height was slightly different for each imprint. Bottom-up cantilevered silicon nanowires were grown onto a silicon-on-insulator (SOI) wafer that consisted of a 7 m thick (1 1 0)-oriented device layer, and a 100 nm thick buried oxide layer [23]. Vertical {1 1 1} sidewalls were formed by patterning and etching trenches in the (1 1 0) top silicon device layer using KOH. Gold catalyst particles were then deposited from colloidal suspension using a dip-and-dry process. The CVD growth was performed using silane, HCl, and B2 H6 (the boron dopant source) as precursors. Silane preferentially decomposes on the gold surface at a growth temperature of T = 680 ◦ C. The deposited silicon dissolves in the gold, eventually forming a super-saturated Au–Si liquid alloy. Silicon atoms then precipitate from this alloy to form a nanowire whose diameter is about that of the gold particle. The nanowires are epitaxially grown horizontally from the vertical sidewall surface, and are thus rigidly anchored (covalently bonded) at their base. 2.2. Interferometric setup The experimental optical interferometry setup used for assaying of the resonance frequencies of the nanoresonators is shown in Fig. 1 [15,22]. The resonator arrays were mounted onto a piezoelectric element which was actuated by the tracking output of a spectrum analyzer (Agilent model 4411B). The beam of a laser diode ( = 655 nm) was directed through a beamsplitter and focused onto the substrate using a NA = 0.45 microscope objective. At resonance, motion of the nanoresonators relative to the substrate created a moving fringe pattern that was reflected back through the microscope objective, was redirected by the beamsplitter, and impinged on an AC-coupled photodetector (New Focus model 1601). 2.3. Biofunctionalization In order to utilize such SiCN nanoresonators for the specific detection of biological targets, analyte-specific functional layers need to be immobilized onto their surface. The binding affinity of these layers is then assessed by monitoring the resonant frequency shifts related to any added mass on each individual device. The specific detection of protein-A using single domain antibody fragments (sdAb) was here chosen as model system. Protein-A is a 56 kDa surface protein originally found in the cell wall of Staphylococcus aureus. The protein has found widespread use in biochemical research because of its ability to bind immunoglobulins. The resonator chips were cleaned in cold piranha (3:1 H2 SO4 :H2 O2 ) just prior to the experiments and stored in DI water. The resonance frequency of all devices was assessed prior to the start of the functionalization procedure. The samples were then transferred into 100% ethanol for 2 min in a petri dish. This was followed by silanization in 2% 3-aminopropyl(triethoxysilane) (APTES) in ethanol (pH 5 adjusted using glacial acetic acid for 2 min). The samples were washed again in 100% ethanol for 5 min followed by blow drying and curing in an incubator @ 80 ◦ C for
C. Guthy et al. / Sensors and Actuators B 187 (2013) 111–117
113
Fig. 1. Diagram of the optical interferometric setup employed for the resonant assaying of the nanoresonators.
30 min. After cooling, the cured samples were washed for 2 min in phosphate-buffered saline (PBS) buffer. The silanized samples were activated in freshly prepared 2% solution of gluteraldehyde in water for 1 h followed by washing in PBS buffer for 5 min. The samples were then incubated in 1 g/mL solution of sdAb for 1 h at 37 ◦ C in the incubator at a speed of 180,000 rpm. This was followed by washing in 10% ethanolamine for 10 min, then in PBS-tween (0.05%) for 5 min, one wash in PBS for 5 min and two washes in DI water for 2 min each. Ethanolamine treatment ensures the blocking of all unreacted aldehyde groups on resonator surface and prevents any non-specific attachment of analyte. The samples are blow-dried, and the resonance frequency of each device was measured. After the resonance characterization, the samples were washed in PBS buffer for 2 min. They were finally exposed to the target Protein-A (5 g/mL) for 1 h at 37 ◦ C in a rotating incubator at a speed of 180,000 rpm. The samples were then washed in PBS-tween (0.05%) for 5 min followed by one wash in PBS for 5 min and two washes in DI water for 2 min each. The resonance frequencies of the devices was assessed again immediately after exposure to the target protein.
In the case of the bottom-up silicon nanowires, the detection of proteins was rather demonstrated using streptavidin as target. Nanowires were first exposed to an O2 plasma (Harrick Plasma) for 90 s at 400–500 mTorr and a power of 18 W. This initial step cleaned their surfaces and generated the hydroxyl groups (OH) required for the bonding of methoxysilane. Another wet cleaning and hydroxylation process followed, using hot piranha for 120 s, an H2 O rinse, a 100% anhydrous ethanol wash, and drying in argon. This cleaning and hydroxylation process also allowed previously functionalized nanowires to be cleaned and reused. The resonant frequencies of the nanowires were assessed a first time immediately after this cleaning step (Fig. 2(a)). A monolayer of 3mercaptopropyl trimethoxysilane (C6 H16 O3 SiS) was deposited on Si surfaces immediately after the plasma cleaning by exposure to C6 H16 O3 SiS vapor at approximately 5 Torr for 18 h. This deposition was followed by a wash in 100% anhydrous ethanol to remove physisorbed C6 H16 O3 SiS molecules. The samples were then dried in argon, and the resonant frequencies measured again to determine any shifts induced by the silane deposition (Fig. 2(b)). Substrates of porous silicon were exposed to the C6 H16 O3 SiS alongside the
Fig. 2. Resonant response of two individual bottom-up Si nanowires. (a) Immediately after cleaning, (b) immediately after silanization, (c) immediately after biotinylation, and (d) immediately after specific capture of streptavidin.
114
C. Guthy et al. / Sensors and Actuators B 187 (2013) 111–117
nanowires, and measured using IR-spectroscopy to confirm the successful deposition of the linker molecule. The 3-mercaptopropyl trimethoxysilane served as a linker to the bind biotin when the samples were then immersed in a biotin solution consisting of 1 mg of Pierce AZ-Link Biotin HPDP 21341, C24 H37 N5 O3 S3 , 0.5 mL dimethyl-sulfoxide (DMSO) and 0.5 mL phosphate buffered saline (PBS, 20 mM, 7.4 pH) [24] for 1 h. The DMSO was replaced gradually by toluene, followed by a wash in 100% anhydrous ethanol, and drying in argon. The resonance of the nanowires was once again assessed to measure the frequency shift associated to the attachment of the biotin (Fig. 2(c)). Biotin has a well-known affinity and selectivity to streptavidin [25], and the attachment of streptavidin was performed by immersing the samples in a mixture of 50 L streptavidin (of concentration 10 mg/mL) and 50 L of PBS (10 mM with 0.1% nonionic surfactant Triton X) for 1 h. Following this immersion, the buffered solution was gradually replaced by Millipore-purified H2 O. The samples were then washed in chloroform, 100% anhydrous ethanol, and dried with argon. The resonant frequency was measured one last time following this specific capture of the streptavidin (Fig. 2(d)). Negative control experiments were also performed in which the biotinylated nanowires (Fig. 2(c)) were exposed to a solution of streptavidin that was already saturated with biotin. This presaturation was performed by mixing 0.5 mg biotin, 0.5 mL DMSO and 0.5 mL PBS with the streptavidin solution. The binding sites of this streptavidin being already filled with biotin, it was not expected to attach to the biotin located on the nanowires.
3. Results and discussion 3.1. SiCN NEMS A typical (5 × 5) array of of doubly-clamped SiCN nanomechanical resonators produced using “top down” nanomachining that contains 24 nominally identical devices is shown in Fig. 3(a). A close-up image of an individual 16 nm wide, 14 m long resonator is shown in Fig. 3(b). Resonators as narrow as 8 nm have been obtained using electron beam lithography exposure [15,16]. Fig. 3(c) shows a typical resonant frequency spectrum obtained from a SiCN device. The experimental data obtained for the specific detection of protein-A using the SiCN nanoarrays are summarized in Table 1. Immobilization of single domain antibody fragments (sdAb) onto the nanoresonator arrays resulted in consistent resonant frequency downshifts of ∼341 kHz, corresponding to an attached mass of 12.3 fg or 58,000 sdAb molecules. The frequencies further downshift by ∼216 kHz due to protein A attachment (Fig. 4), corresponding to the capture of 7.7 fg or 11,000 molecules of Protein A. Another “negative control” array II (Fig. 5) was processed identically except Protein-A was not added to the solution. The frequency shifts due to this “solvent only” step are much smaller (∼54 kHz or 1.9 fg), further supporting the assertion that the significant frequency downshifts in array I are indeed due to protein-A capture. Fig. 6 shows a typical 300 nm-wide resonator structure realized via the imprinting of a 300 nm thick PMMA 495/LOR 3A resist bilayer [17]. Imprint temperatures of at least 70 ◦ C above the glass transition temperature Tg = 105 ◦ C of the PMMA allowed an optimal fabrication yield of 97%. Line-widening effects related to the oxygen plasma etch resulted in beams that were 20–30 nm wider than the design dimensions. The minimum resonator width obtained using a 200 nm-thick resist bilayer was approximately 120 nm. The fabrication yield however rapidly dropped to below 50% for narrower structures. These 120 nm-wide structures are however the narrowest suspended structures realized via nanoimprinting.
Fig. 3. (a) Scanning electron micrograph (SEM) of a 5 × 5 array of doubly-clamped 16 nm wide nanomechanical resonators fabricated by electron-beam lithography with a yield of 100%. (b) Close-up SEM of a 16 nm wide and 14 m long resonator. (c) Typical resonance curve for a 16 nm wide, 50 nm thick and 10 m long resonator. From the FWHM of the peak a Q-factor of ∼1000 is obtained.
Given the high-tensile stress present in the device layer, resist imperfections such as nicks along the device edge were found to be highly detrimental to the successful realization of sub-100 nm wide suspended structures. Imperfections in the resist undercut profile are likely to cause the observed edge roughness. Arrays of nanoresonators with widths in the 120–300 nm range, thicknesses of 40 nm and 70 nm, and a length of 14 m were realized. The 40 nm thick and 70 nm thick devices showed average resonant frequencies of ∼16.6 MHz and 21.7 MHz, respectively [17] and quality factors in the 6000–8000 range. We also demonstrated the specific detection of streptavidin with mass sensitivities of ∼3.5 fg using such nanoresonator arrays functionalized with biotin [17].
C. Guthy et al. / Sensors and Actuators B 187 (2013) 111–117
115
Table 1 Summary of the average resonant frequency shifts and corresponding added masses of the two SiCN NEMS arrays (I and II) due to capture of single domain antibody fragments (sdAb), protein-A and “solvent only” step. Attach-ment
F (kHz)
Error (kHz)
m (fg)
m/S (fg/m2 )
Sample
sdAb Prot A
341 216
±5 ±5
12.3 7.7
1.4 0.9
58,000 11,000
Control
sdAb Sol.
248 54
±5 ±5
8.9 1.9
1.0 0.2
42,000 3,000
Fig. 4. Resonant frequencies of the 24 individual resonators of Array I. The blue diamonds, green squares and red circles designate the individual device frequencies of the bare resonators, resonators with immobilized antibodies and resonators with captured protein-A, respectively. Immobilization of antibodies results in frequency downshifts of ∼341 kHz. Capture of protein-A further down-shifts the frequencies by ∼216 kHz. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.)
3.2. Si nanowires The nanowire dimensions were measured in a high-resolution Hitachi FE 4800 scanning electron microscope (SEM). Typical nanowire lengths and diameters ranged from 3 to 4 m and from 100 to 150 nm, respectively. The height of the wires above the substrate ranged from 2 to 7 m. Fig. 7 shows arrays of nanomechanical resonators produced by the bottom-up CVD method [20]. Fig. 2 schematically shows resonant response of two nanowires prior to any chemical attachment (Fig. 2(a)), after silanization (Fig. 2(b)), after biotin attachment (Fig. 2(c)), and after the specific
Fig. 5. Resonant frequencies of the 24 individual resonators of Array II. This “negative control” array was processed identically except protein-A was not added to the solution in the detection step. Similarly to Array I, immobilization of antibodies results in frequency downshifts of ∼248 kHz. The second “solvent only” step results in much smaller frequency downshifts of ∼54 kHz.
Total # of molecules
Fig. 6. Scanning electron micrograph of a 300 nm-wide SiCN resonator element fabricated by imprint lithography.
capture of the target streptavidin (Fig. 2(d)). The silanization step caused a net downward shift of resonant frequency by f = 33 ± 1 kHz. The attachment of biotin onto the nanowires caused a further downward frequency shift of f ∼ 25 kHz, corresponding to an added mass of 63 ag. The molecular weight of a biotin molecule is 540 Da, or 0.9 zg. The experimentally observed added mass therefore corresponded to a biotin surface density of 6.5 molecules per 100 nm2 . Given that a biotin molecule can occupy [26] an area 8.5 nm2 , this experimental finding indicates that the biotin covers almost 55% of the nanowire surface. The specific capture of streptavidin caused an additional frequency shift of 120 kHz, corresponding to an attached mass of 300 ag. The molecular weight of a single streptavidin molecule is 60 kDa, or 99.6 zg. This added mass corresponded to a
Fig. 7. 40 nm-wide bottom-up silicon nanowire resonators produced by chemical vapor deposition.
116
C. Guthy et al. / Sensors and Actuators B 187 (2013) 111–117
streptavidin surface density of 0.3 molecules per 100 nm2 . Given that a streptavidin molecule occupies [27] an area of 100 nm2 , this experimental finding indicates that the streptavidin covers approx. 30% of the nanowire surface. Finally, exposure of similarly biotinylated nanowires to a solution of pre-saturated (by the biotin) streptavidin showed negligible shifts, demonstrating the specificity of the streptavidin capture. 4. Conclusions Large arrays of nanomechanical resonators as narrow as 8 nm have been produced using a novel SiCN surface machining method. These arrays were used to detect antigen–antibody binding events without the need of extrinsic markers. The specific detection of protein-A using single domain antibody fragments (sdAb) was demonstrated as proof of concept. Although resonator arrays produced by electron beam lithography (EBL) show excellent uniformity and reproducibility, the serial nature of the EBL technique is not suitable for low cost, large-scale manufacturing However, this SiCN fabrication technology was recently employed in conjunction with nanoimprint lithography, demonstrating its amenability to cost-efficient manufacturing. Using a bottom-up CVD approach we also produced arrays of resonant nanowires that are promising for high sensitivity applications. This technique, similarly to other bottom-up approaches, is however less suitable for the realization of very large arrays. The specific detection of streptavidin was accomplished with these devices. In both cases an optical interferometric method was used to monitor the resonant frequency changes of the devices induced by the attachment of molecular systems. The readout was performed with a simple and cost-effective optical technique, and is therefore conducive to the development of large arrays of individual sensors. Akin to the concept of chemoresistive electronic noses, large arrays of nanomechanical resonators would enable the development of nanoresonator-based ‘electronic ears’ for the fingerprinting of metabolomics systems. Acknowledgments This work was supported by the National Research Council of Canada, the Province of Alberta, and the Natural Science and Engineering Research Council of Canada. Fabrication of the SiCN resonators was performed in the University of Alberta Nanofab facilities. The bottom-up Si nanowires were fabricated at HewlettPackard Laboratories, Palo Alto, CA References [1] R.M. Lequin, Enzyme immunoassay (EIA)/enzyme-linked immunosorbent assay (ELISA), Clinical Chemistry 51 (2005) 2415–2418. [2] H.J. Lee, A.W. Wark, R.M. Corn, Microarray methods for protein biomarker detection, Analyst 133 (2008) 975–983. [3] D.W. Carr, S. Evoy, L. Sekaric, J.M. Parpia, H.G. Craighead, Measurement of mechanical resonance and losses in nanometer scale silicon wires, Applied Physics Letters 75 (1999) 920–922. [4] Y. Wang, J.A. Henry, A.T. Zehnder, M.A. Hines, Surface chemical control of mechanical energy losses in micromachined silicon structures, Journal of Physical Chemistry B 107 (2003) 14270–14277. [5] K.L. Ekinci, Y.T. Yang, M.L. Roukes, Ultimate limits to inertial mass sensing based upon nanoelectromechanical systems, Journal of Applied Physics 95 (2004) 2682–2689. [6] N.V. Lavrik, P.G. Datskos, Femtogram mass detection using photothermally actuated nanomechanical resonators, Applied Physics Letters 82 (2003) 2697–2699. [7] B. Ilic, Y. Yanh, H.G. Craighead, Virus detection using nanoelectromechanical devices, Applied Physics Letters 85 (2004) 2604–2606. [8] L.M. Fischer, V.A. Wright, C. Guthy, N. Yang, M.T. McDermott, J.M. Buriak, S. Evoy, Specific detection of proteins using nanomechanical resonators, Sensors and Actuators B 134 (2008) 613–617.
[9] M. Li, H.X. Tang, M.L. Roukes, Ultra-sensitive NEMS-based cantilevers for sensing, scanned probe and very high-frequency applications, Nature Nanotechnology 2 (2007) 114–120. [10] K.L. Ekinci, X.M. Huang, M.L. Roukes, Ultrasensitive nanoelectromechanical mass detection, Applied Physics Letters 84 (2004) 4469–4471. [11] B. Ilic, H.G. Craighead, S. Krylov, W. Senaratne, C. Ober, P. Neuzil, Attogram detection using nanoelectromechanical oscillators, Journal of Applied Physics 95 (2004) 3694–3713. [12] P.S. Wagoneer, M. Varshney, H.G. Craighead, Detection of prostate specific antigen with nanomechanical resonators, Lab Chip 9 (2009) 3095–3099. [13] Y.T. Yang, C. Callegari, X.L. Feng, K.L. Ekinci, M.L. Roukes, Zeptogram-scale nanomechanical mass sensing, Nano Letters 6 (2006) 583–586. [14] X.L. Feng, R. He, P. Yang, M.L. Roukes, Very high frequency silicon nanowire electromechanical resonators, Nano Letters 7 (2007) 1953–1959. [15] C. Guthy, R. Das, B. Drobot, S. Evoy, Resonant characteristics of ultranarrow SiCN nanomechanical resonators, Journal of Applied Physics 108 (2010) 014306. [16] M.A. Mohammad, S.K. Dew, S. Evoy, M. Stepanova, Fabrication of sub-10 nm silicon carbon nitride resonators using a hydrogen silsesquioxane mask patterned by electron beam lithography, Microelectronic Engineering 88 (2011) 2338–2341. [17] A. Janzen, S. Poshtiban, A. Singh, S. Evoy, Fabrication of nanoresonator biosensing arrays using nanoimprint lithography, Journal of Micro/Nanolithography, MEMS, and MOEMS 11 (2012) 023013. [18] Y. Cui, Q. Wei, H.Ch. Park, M. Lieber, Nanowire nanosensors for highly sensitive and selective detection of biological and chemical species, Science 293 (2001) 1289–1292. [19] V. Sazonova, Y. Yaish, H. Ustunel, D. Roundy, T.A. Arias, P.L. McEuen, A tunable carbon nanotube electromechanical oscillator, Nature 431 (2004) 284–287. [20] A. Husain, J. Hone, W. Henk, Ch. Postma, X.M.H. Huang, T. Drake, M. Barbic, A. Scherer, M.L. Roukes, Nanowire-based very-high-frequency electromechanical resonator, Applied Physics Letters 83 (2003) 1240–1242. [21] M. Belov, N.J. Quitoriano, S. Sharma, W.K. Hiebert, T.I. Kamins, S. Evoy, Mechanical resonance of clamped silicon nanowires measured by optical interferometry, Journal of Applied Physics 103 (2008) 074304. [22] L.M. Fischer, N. Wilding, M. Gel, S. Evoy, Low-stress silicon carbonitride for the machining of high-frequency nanomechanical resonators, Journal of Vacuum Science and Technology B 25 (2007) 33–37. [23] M.S. Islam, S. Sharma, T.I. Kamins, R.S. Williams, Ultrahigh-density silicon nano-bridges formed between two vertical silicon surfaces, Nanotechnology 15 (2004) L5–L8. [24] A. Janshoff, K.S. Dancil, C. Steinem, D.P. Greiner, V.S.-Y. Lin, C. Gurtner, K. Motesharei, M.J. Sailor, M.R. Ghadiri, Macroporous p-type silicon Fabry–Perot layers. Fabrication, characterization, and applications in biosensing, Journal of the American Chemical Society 120 (1998) 12108–12116. [25] J. Dechancie, K.N. Houk, The origins of femtomolar protein–ligand binding: hydrogen bond cooperativity and desolvation energetics in the biotin–(strept)avidin binding site, Journal of the American Chemical Society 129 (2007) 5419–5429. [26] Instructions EZ-Link Biotin-HPDP 21341, Pierce, www.piercenet.com [27] W.A. Hendrickson, A. Pahler, J. Smith, Y. Satow, E.A. Merritt, R.P. Phizackerley, Crystal structure of core streptavidin determined from multiwavelength anomalous diffraction of synchrotron radiation, Proceedings of the National Academy of Sciences of the United States of America 86 (1989) 2190–2194.
Biographies Csaba Guthy is a Research Associate at the Department of Electrical and Computer Engineering of the University of Alberta in Edmonton, Canada. He received B.Sc. and M.Sc. degrees in physics from the University of Debrecen, Hungary in 2001 and M.Sc. and Ph.D. in materials science and engineering from the University of Pennsylvania (Philadelphia, PA, USA) in 2007. He has conducted extensive research in diverse areas such as thermal transport in carbon nanotubes, polymer composites, resonance behavior of nanoelectromechanical systems (NEMS), diffusion processes in intermetallic compounds, and development of a thermoacoustic thermal-to-electrical power converter. His research interests include the development of complex, highlyreliable, robust and unique technological solutions, R&D of advanced technologies for rapid detection and identification of chemical, biological and explosive threats, advanced non-destructive testing technologies, development of integrated sensors, system integration, innovative medical diagnostics technologies, novel energy storage technologies for automotive and stationary applications, modeling and simulation. Miro Belov received physics degrees, B.Sc. and M.Sc., from the Comenius University, Bratislava, Ph.D. from the University of Alberta, Edmonton, and have worked in both academic and industrial R&D environments. His education and research experience are rooted in condensed matter physics, NEMS, optics and magnetics. His research interests include high-resolution (ultrafast) measurements of (bio-) nano-mechanical, optical and magnetic structures, micro-, nano-fabrication for biomedical applications. Alex Janzen received a B.Sc. in engineering physics in 2009 and an M.Sc. in microelectromechanical systems in 2011 from the University of Alberta in Edmonton Canada. His M.Sc. work involved fabricating nanomechanical resonator devices
C. Guthy et al. / Sensors and Actuators B 187 (2013) 111–117 using nanoimprint lithography, and the testing of these devices for the specific detection of proteins. He currently works as a Research Associate at the University of Alberta involving work with boron-doped polysilicon in collaboration with Micralyne. Nathaniel Quitoriano is an Assistant Professor at the Department of Mining and Materials Engineering of McGill University in Montreal Canada. Dr. Quitoriano heads the Semiconductor Nanostructures Lab (SNL) which researches device applications of metal-catalyzed semiconductor growth, including semiconductor nanowires. Applications of particular interest include: solar cells, lasers, photo-diodes, and sensors. In 2000, Nate Quitoriano received Bachelor’s degrees in both electrical engineering and computer science and materials science and engineering from the University of California, Berkeley. As an undergraduate, he worked with Professor Tim Sands on establishing the kinetics of an ohmic, transient-liquid-phase bond for semiconductors. Nate received his Ph.D. in materials science engineering in 2006 at the Massachusetts Institute of Technology under the supervision of Gene Fitzgerald. While at MIT, he worked on III–V, lattice-mismatched semiconductors and grew high-quality InP on GaAs using graded, compositional buffers InGaAs and InGaP to slowly increase the lattice constant. Following MIT, he worked in Stan William’s group at Hewlett-Packard Laboratories under the direction of Ted Kamins where he studied Si and Ge nanowires for use as sensors and electrical devices and successfully demonstrated Si nanotube resonators and guided Si nanowire growth. Amit Singh is currently an Associate Research Scientist at the Department of Pharmaceutical Sciences, Northeastern University, Boston, USA. He received his M.Sc. and Ph.D. in biotechnology from the University of Pune, India. He worked as a postdoctoral fellow in Dr. Evoy’s laboratory at the University of Alberta and the National Institute for Nanotechnology. His research interests include multifunctional nanomaterial synthesis, characterization materials and their biomedical applications, biosensors, drug-delivery systems and cancer therapeutics. Vincent A. Wright received a Ph.D. degree in chemistry at the University of British Columbia, Vancouver in 2005. He worked as a post-doctoral fellow at the Univerisity of Alberta in Jillian Buriak’s research group investigating the functionalization of silicon, silica, and stainless steel surfaces. He currently works as a Research Scientist at Gilead Alberta ULC.
117
Eric Finley received a B.Sc. in physics from the University of Alberta, Edmonton in 1998. He is currently serving as Technical Officer at the National Institute of Nanotechnology in Edmonton Canada. Ted Kamins is a Consulting Professor in the Electrical Engineering Department at Stanford University. He received his B.S., M.S. and Ph.D. degrees from the University of California, Berkeley. He then joined the Research and Development Laboratory of Fairchild Semiconductor, where he performed early work contributing to the understanding of the then-emerging field of polycrystalline silicon and also contributed to device applications of epitaxial silicon. After moving to Hewlett-Packard, he worked in a number of materials and device-related areas, beginning with the development of UV-sensitive photodiodes, which are enabling devices for spectrophotometers – instruments in production for more than 20 years, first by HP and now by Agilent Technologies. He then contributed to the emerging areas of silicon-on-insulator and rapid thermal processing. Subsequent work dealt with advanced epitaxy and device technology for the silicon-germanium, heterojunction bipolar transistor, which is currently becoming widely used for wireless communications. Most recently Ted was a Principal Scientist in the Information and Quantum Systems Laboratory at Hewlett-Packard Laboratories in Palo Alto, California, where he was focusing on advanced nanostructured electronic materials and devices. Stephane Evoy received a B.Eng. and M.Sc.A. in engineering physics from the Ecole Polytechnique de Montréal, in 1992 and 1994, respectively. He also received Ph.D. in Applied Physics from Cornell University in 1998. He is currently Associate Professor in the Department of Electrical and Computer Engineering at the University of Alberta. Dr. Evoy has co-authored over 60 papers in the area of biosensors, nanoelectromechanical systems (NEMS), nanofabrication, and scanning probe microscopy. He co-edited the 2004 textbook: “Introduction to Nanoscale Science and Engineering.” His research currently focuses on cantilever- and SPR-based detection of metabolites and bacterial pathogens, the development of sample preparation and detection technologies leveraging the natural affinities of bacteriophage, as well as the development of metabolomics fingerprinting platforms based on large arrays of nanomechanical resonators.