One-Step Selective Chemistry for Silicon-on-Insulator Sensor

LETTER pubs.acs.org/Langmuir

One-Step Selective Chemistry for Silicon-on-Insulator Sensor Geometries Oliver Seitz,† Poornika G. Fernandes,† Gazi A. Mahmud,† Huang-Chun Wen,‡ Harvey J. Stiegler,† Richard A. Chapman,† Eric M. Vogel,† and Yves J. Chabal*,† † ‡

University of Texas at Dallas, Richardson, Texas, United States Texas Instrument, Dallas, Texas, United States

bS Supporting Information ABSTRACT: A one-step functionalization process has been developed for oxide-free channels of field effect transistor structures, enabling a self-selective grafting of receptor molecules on the device active area, while protecting the nonactive part from nonspecific attachment of target molecules. Characterization of the self-organized chemical process is performed on both Si(100) and SiO2 surfaces by infrared and X-ray photoelectron spectroscopy, atomic force microscopy, and electrical measurements. This selective functionalization leads to structures with better chemical stability, reproducibility, and reliability than current SiO2-based devices using silane molecules.

F

ield-effect transistor (FET)-based biosensors require grafting of self-assembled monolayers that are functionalized to attach appropriate probe molecules.1
5 A chemically self-selective functionalization of active and inactive regions of electronic sensors may be a means to enhance their performance, especially for detection of ultralow concentration analytes when very low aspect ratio (e.g., nanowire) devices are present. The ability to graft chemically active molecules only6 on the channel region of a FET biosensor7 that responds to a change in surface electric field8,9 is therefore useful because it could reduce the loss of analytes attached to the nonactive part of the device and, therefore, increase the probability of any single analyte to be attached on the active part. Most importantly, this functionalization adapted to the standard structure could be useful to develop some fundamental understanding of commonly used FETsensor structures. An example would be to understand the effect on the electrical response induced by target molecules attached at the vicinity of the active part of the device. Another one, illustrated below, would be the effect on the response of a pH sensor when hydroxyl groups are not present at the interface. Overall, such selectivity would result in full protection of the area surrounding the channel by minimizing the probability of unwanted adsorption and reducing parasitic effects that could result from ion penetration10 and in active functionalization of the channel (Figure 1). In view of the difficulties involved in photolithographic-based processes (spatial inaccuracy and chemical damage to SAM), a self-organized and selective chemical process is highly desirable and effective to avoid these problems. The main difficulty in achieving such selectivity in siliconbased devices has been associated with the belief that a thin r 2011 American Chemical Society

thermal oxide layer is absolutely necessary to minimize interface states in the active region of the device. Although the electronic quality of Si/SiO2 interfaces is unrivaled, wet chemical functionalization of silicon oxide has been problematic and precluded the needed selectivity. For instance, grafting on silicon oxide surfaces has been achieved using primarily silane chemistry (Si
O
Si interfacial bond formation), which is hindered by polymerization, uncontrolled self-assembled monolayer (SAM) formation, and degradation in aqueous media.11 Although improvements in processing reproducibility and layer stability have been achieved by using longer chain molecules,11 the lack of selectivity, due to the fact that both the channel and the surrounding areas are both oxidized, prevents the efficient collection of analytes as suggested in Figure 1. Recently, the success in functionalizing oxide-free Si directly has opened new possibilities for device improvements and is the basis of this work.12,13 We demonstrate here that active SAMs, known to have much higher stability, reproducibility, and reliability than currently used silane molecules on oxidized Si, can be attached directly on oxide-free (100)-Si surfaces. Moreover, we show that oxidefree nanoribbons can be selectively functionalized by active SAMs, using a one-step process, leading to enhanced passivation of the other parts of the device with inactive (methyl terminated) SAMs, as evidenced by IR spectroscopy (IRS), X-ray photoelectron spectroscopy (XPS), atomic force microscopy (AFM), and electrical measurements performed directly on chips. This Received: February 4, 2011 Revised: May 25, 2011 Published: May 31, 2011 7337

dx.doi.org/10.1021/la200471b | Langmuir 2011, 27, 7337–7340

Langmuir

LETTER

Figure 1. Schematic representation of one-step functionalization with the mixture
COOEth alkyl chains and OTS molecules leading to selective functionalization of the active part of the device and BOX protection from ion penetration.

Figure 2. FTIR absorption spectra of (a) hydrogenated Si (100) surface, (b) COOEth terminated alkyl chains SAM, and (c) COOH alkyl chains terminated SAM. (a) is referenced to the surface with its cleaned thermal oxide; (b) and (c) are referenced to the Si
H surface. No SiO2 is observed after all the SAM preparation process. Inset: Full SiO2 vibration modes after etching.

selective chemistry is applied to a working FET sensor geometry to assess its usefulness for more advanced sensor devices and to investigate fundamental issues associated with their response to pH changes. The FET devices are fabricated on p-type (100) silicon-oninsulator (SOI) wafers with a buried oxide thickness of 145 nm. Lithographically prepared Si nanoribbon channels (30 nm thick, 1
5 μm wide, 50 μm length, covered with 3 nm thermal oxide) are connected between the source/drain gold pads. Each chip (with many devices) includes two larger areas representing the BOX oxide and the oxidized SOI channel region to permit XPS characterization. These areas undergo all the treatments that the channel and the box SiO2 do during the sensor fabrication. Alkyl monolayers are formed via thermal hydrosilylation of alkenes.13 The freshly etched Si sample is immersed, for 4 h at 200 °C, in deoxygenated ethyl undecylenate (CH2dCH
(CH2)8
COOEth) molecules containing 1% octadecyltrichlorosilane (OTS), chosen for their protective and nonreactive methyl head groups. The devices are then cleaned with solvent, dried, and activated from
COOEth head groups to
COOH and then further to NH2 for gold nanoparticle (AuNp) attachment and pH sensing device testing. A detailed description of the chemical process is given in the Supporting Information. The above mixed chemistry, involving (
COOEth) SAM attachment in the presence of OTS molecules and subsequent

(
COOH) activation, is first investigated with IRS on a blanket Si(100) surface. Figure 2 shows the differential IR absorption spectra at three critical stages: (a) a fully H-terminated surface, characterized by three Si
H stretch bands at ∼2100 cm
1, and the loss of SiO2 in the 900
1300 cm
1 range; (b) a surface terminated by a COOEth-terminated alkyl chain obtained after grafting; and (c) a surface terminated by COOH after the activation step. The spectral assignment is straightforward13 and confirms that the desired chemistry is achieved without any oxidation of the initial H-terminated surface (no absorption band in the 950
1250 cm
1 region). The CdO vibration mode (1740 cm
1) and the CH2 stretching modes (especially the antisymetric (asym.) mode at 2922 cm
1) in (b) indicate that a high density SAM layer has been grafted.13
16 The red shift of the CdO mode (1714 cm
1) as well as the loss of methyl group at 2983 cm
1 in (c) confirm the removal of the ethyl head group and the presence of closely packed carboxylic acid head groups. Importantly, an upper limit of 3% of a monolayer of Si
O is established for the potential presence of Si
O after the two-step process leading to the COOH-terminated surfaces (see the Supporting Information). Despite the presence of OTS molecules, the surface is fully functionalized with a dense functional SAM layer13
16 and an amount of remaining Si
H termination (see Supporting Information) similar to what is observed during standard hydrosilylation. Similar experiments on SiO2 surfaces confirm that only OTS molecules are grafted, despite the presence of ethyl undecylenate molecules and the subsequent activation steps (IRS data not shown, see XPS below). To verify that selective chemistry is also achievable on the actual device, XPS spectra are recorded in the chip probe areas equivalent to the channel (Figure 3a) and to the nonactive part surroundings (Figure 3b). In the channel area (Figure 3a), the Si spectrum shows no suboxidation of the silicon surface (100
102 eV range), in agreement with the IRS results on the blanket surface (Figure 2). The presence of a weak band at 104 eV (high quality SiO2)17 is due to the X-ray beam size (larger than the probed area).The C1s spectrum confirms the presence of
COOEth molecules identified by the high binding energy peaks at 289 and 286 eV, corresponding to OdC
O and O
CH2
CH3, respectively. In the probed area corresponding to the BOX (SiO2 part, Figure 3b), there is only a strong band at 104 eV and only one type of C1s is observed at 285 eV, confirming that only aliphatic carbons from the OTS alkyl chain are attached. An important aspect of this work is the one-step method for achieving high-quality selective functionalization. This method is effective because the channel is fully hydrogenated after mild HF 7338

dx.doi.org/10.1021/la200471b |Langmuir 2011, 27, 7337–7340

Langmuir

LETTER

Figure 3. XPS spectra of the Si2p and C1s (inset) regions after functionalization in the mixture ethyl undecylenate/OTS molecules. (a) Spectra are recorded on the area equivalent to the channels; (b) spectra are recorded on the area equivalent to the surrounding SiO2.

Figure 5. Real-time response to changes in pH of NH2-terminated Si surface after full modification of the COOH-terminated alkyl chains SAM. pH solutions are composed of 0.1 M KCl and 10 mM phosphoric acid salts and then adjusted with KOH solution. Figure 4. AFM image (5 μm  5 μm) recorded after attachment of AuNps. The selectivity of the two different functional groups (NH2 vs CH3) favors the attachment of the AuNps on the channel part of the device (center part) as compared to the SiO2 part (above and below the channel). Left side: Schematic showing the AFM area measured on the device.

etching while the BOX SiO2 is chemically clean (hydrocarbonfree) and remains hydroxyl-terminated. Immersion in the mixture leads to efficient functionalization, with selective hydrosilylation on H/Si and silanization on SiO2. If the process is done in two separate steps, then the second step (silanization) is hindered by the long immersion in ethyl undecylenate molecules, resulting in poor protection of the oxide surface due to the loss of hydroxyl groups. Several attempts to perform the selective functionalization in two separate steps have consistently resulted in poor OTS layers. A visual confirmation of the selective functionalization can be established by further transforming the
COOH termination of the channel into NH2 termination, as described in the Supporting Information, to permit selective attachment of Au nanoparticles (AuNps). The AFM image (Figure 4) shows a high concentration of AuNps in the channel region compared to the nonselective attachment in the oxide region (ratio ∼ 10:1) due to their higher affinity for gold. The electrical quality of the Si/SAM interface on oxide-free Si, determined by photoluminescence measurements is found to be comparable to that of a Si/thermal SiO2 and better than Si
H

interfaces (see the Supporting Information). It is therefore expected that devices prepared accordingly can perform well as sensors. Figure 5 shows such a test for pH sensing of a NH2-terminated Si/SAM system in a microfluidic configuration,5,8 which is a precursor to actual biosensing. The device reproducibly responds to pH changes over the whole pH = 2
10 range. At low pH (2
4
6), a ΔI of 10 nA/pH is observed, while at higher pH (6
8
10) a lower sensitivity is recorded (∼5 nA/pH). This behavior is in good agreement with a pH response due to the presence of
NH2 groups that are transformed to NH3þ at low pH. However, the response at high pH (∼10) is more surprising because the well-accepted mechanisms invokes charging of hydroxyl groups on the SiO2 layer.8 Since the interface is oxide-free, the response at high pH suggests that hydroxyl groups are not necessary for a high-pH response, and that the changes in the equilibrium concentrations of NH2 and NH3þ at the surface is sufficient to generate changes in pH (see the Supporting Information for a discussion and a comparison with bare SiO2 and fully OTS-covered devices). This conclusion underscores the usefulness of the selective functionalization to test sensor response mechanisms. In summary, active functionalization via robust Si
C bonds is selectively achieved on the channel area of sensor devices, and thorough passivation of the surrounding oxidized areas of chips by a methyl-terminated SAM has been achieved by a one-step chemistry. Such selective chemistry can keep the device 7339

dx.doi.org/10.1021/la200471b |Langmuir 2011, 27, 7337–7340

Langmuir

LETTER

responsive to analytes while reducing nonspecific attachment (physisorption). This method is highly reproducible and results in devices that are much more stable in aggressive solutions (due to Si
C bonds), feature negligible nonradiative recombination, and have good electrical properties as illustrated by the reproducible pH sensing performance. Such selective FET-sensor structures can therefore be used to address fundamental issues associated with the commonly used FET-based biosensors.

’ ASSOCIATED CONTENT

bS

Supporting Information. Device structure, fabrication, and functionalization; hydrogen estimation left on the surface; upper limit of SiO2 on the surface, photoluminescence measurements, and complementary pH sensing measurements using different surfaces. This material is available free of charge via the Internet at http://pubs.acs.org.

’ AUTHOR INFORMATION Corresponding Author

*E-mail: [email protected].

’ ACKNOWLEDGMENT This work is supported by the Texas Higher Education Coordinating Board (NHAR Program), Texas Instrument Inc., and the National Science Foundation (Grant CHE-0911197). ’ REFERENCES (1) Chaki, N. K.; Vijayamohanan, K. Biosens. Bioelectron. 2002, 17, 1–12. (2) DiBenedetto, S. A.; Facchetti, A.; Ratner, M. A.; Marks, T. J. Adv. Mater. 2009, 21, 1407–1433. (3) Faucheux, N.; Schweiss, R.; Lutzow, K.; Werner, C.; Groth, T. Biomaterials 2004, 25, 2721–2730. (4) Patolsky, F.; Zheng, G. F.; Lieber, C. M. Nat. Protoc. 2006, 1, 1711–1724. (5) Stern, E.; Klemic, J. F.; Routenberg, D. A.; Wyrembak, P. N.; Turner-Evans, D. B.; Hamilton, A. D.; LaVan, D. A.; Fahmy, T. M.; Reed, M. A. Nature 2007, 445, 519–522. (6) Masood, M. N.; Chen, S.; Carlen, E. T.; Berg, A. v. d. ACS Appl. Mater. Interfaces 2010, 2, 3422–3428. (7) Cui, Y.; Zhong, Z. H.; Wang, D. L.; Wang, W. U.; Lieber, C. M. Nano Lett. 2003, 3, 149–152. (8) Cui, Y.; Wei, Q. Q.; Park, H. K.; Lieber, C. M. Science 2001, 293, 1289–1292. (9) Patolsky, F.; Lieber, C. M. Mater. Today 2005, 8, 20–28. (10) Fernandes, P. G.; Seitz, O.; Chapman, R. A.; Stiegler, H. J.; Wen, H.-C.; Chabal, Y. J.; Vogel, E. M. Appl. Phys. Lett. 2010, 97, 034103. (11) Seitz, O.; Fernandes, P. G.; Tian, R.; Karnik, N.; Wen, H.-C.; Stiegler, H. J.; Chapman, R. A.; Vogel, E. M.; Chabal, Y. J. J. Mater. Chem. 2011, 21, 4384. (12) Aureau, D.; Varin, Y.; Roodenko, K.; Seitz, O.; Pluchery, O.; Chabal, Y. J. J. Phys. Chem. C 2010, 114, 14180–14186. (13) Seitz, O.; Dai, M.; Aguirre-Tostado, F. S.; Wallace, R. M.; Chabal, Y. J. J. Am. Chem. Soc. 2009, 131, 18159–18167. (14) Porter, M. D.; Bright, T. B.; Allara, D. L.; Chidsey, C. E. D. J. Am. Chem. Soc. 1987, 109, 3559–3568. (15) Seitz, O.; Bocking, T.; Salomon, A.; Gooding, J. J.; Cahen, D. Langmuir 2006, 22, 6915–6922. (16) Snyder, R. G.; Strauss, H. L.; Elliger, C. A. J. Phys. Chem. 1982, 86, 5145–5150. (17) Himpsel, F. J.; McFeely, F. R.; Talebibrahimi, A.; Yarmoff, J. A.; Hollinger, G. Phys. Rev. B 1988, 38, 6084–6096. 7340

dx.doi.org/10.1021/la200471b |Langmuir 2011, 27, 7337–7340