1 23
Applied Physics A
Materials Science & Processing
ISSN 0947-8396
Volume 108
Number 4
Appl. Phys. A (2012) 108:801-807
DOI 10.1007/s00339-012-6972-9
Wavelength and coherence effects on the
growth mechanism of silicon nanopillars
and their use in the modification of
spontaneous lifetime emission of BODIPY
dye molecules
Sabriye Acikgoz, Bukem Bilen, Asli
C. Saygili, Gulen Aktas, Amitav Sanyal &
Mehmet Naci Inci
1 23
Your article is protected by copyright and
all rights are held exclusively by
Springer-Verlag. This e-offprint is for personal use only
and shall not be self-archived in electronic
repositories. If you wish to self-archive your
work, please use the accepted author’s
version for posting to your own website or
your institution’s repository. You may further
deposit the accepted author’s version on a
funder’s repository at a funder’s request,
provided it is not made publicly available until
12 months after publication.
Appl Phys A (2012) 108:801–807 DOI 10.1007/s00339-012-6972-9
Wavelength and coherence effects on the growth mechanism
of silicon nanopillars and their use in the modification
of spontaneous lifetime emission of BODIPY dye molecules
Sabriye Acikgoz· Bukem Bilen · Asli C. Saygili · Gulen Aktas· Amitav Sanyal · Mehmet Naci Inci
Received: 29 February 2012 / Accepted: 3 May 2012 / Published online: 23 May 2012 © Springer-Verlag 2012
Abstract Silicon nanopillars are grown by an
electrochemi-cal anodization of p-type silicon wafers at low current densi-ties in a hydrofluoric acid solution. CW, white light, and var-ious UV pulsed lasers are employed as illumination sources in sample preparation to study wavelength and coherence effects on the growth mechanism of the nanopillars. Coher-ence is observed to be the foundation of regularity in obtain-ing conical shapes. The pillar size is found to be almost lin-early proportional to the employed illumination wavelength during their growth. BODIPY dye molecules are chemically attached to these silicon nanopillars and the radiative decay rates are investigated by means of a time-resolved fluores-cence experiment. The decay rate of the dye molecules em-bedded in the vicinity of various size pillar tips is signifi-cantly affected due to different apex angles of the conical nature. It is demonstrated that the pillar size and the sepa-ration between pillars can be adjusted if one uses a coher-ent light source with an appropriate wavelength during the course of fabrication process. Since change in the decay rate is due to tips of the pillars only, separation of a few microme-ters between pillar tips allows one to directly monitor a dye, which is embedded to the tip of a single nanopillar, via a confocal microscopic method for the spontaneous lifetime
S. Acikgoz· B. Bilen · A.C. Saygili · G. Aktas · M.N. Inci (
) Department of Physics, Bogazici University, Bebek,34342 Istanbul, Turkey e-mail:naci.inci@boun.edu.tr
A. Sanyal
Department of Chemistry, Bogazici University, Bebek, 34342 Istanbul, Turkey
S. Acikgoz
Department of Material Science and Engineering, Karamano˘glu Mehmetbey University, 70100 Karaman, Turkey
measurements, without having needed to any extra efforts for an in situ imaging process. It is observed that as the pillar size gets smaller, the inhibition in the spontaneous lifetime of BODIPY is more pronounced. In addition, a more regu-lar pilregu-lar structure yields nonvarying decay rates of the dye molecules throughout the silicon sample.
Keywords Silicon· Nanopillar · Laser · Wavelength ·
Coherence· Dye · Radiative decay rate
1 Introduction
Diodes, transistors, and various kinds of integrated circuit components are the first device applications of the semicon-ductor technology. With their nanoscale dimensions, semi-conductors have become indispensible materials for many optoelectronics and lightwave applications, such as solar cells, nanoscale electronic devices, new lasers, waveguides, chemical, and biosensors [1–3]. Silicon has a very special place amongst all in semiconductor technology due to its unique intrinsic properties for device processing as well as being an excellent substrate for silicon and nonsilicon appli-ances.
In 1990, electrochemical anodization process of silicon wafer in HF based solution is discovered by Canham [4]. It is demonstrated that both photoluminescence and struc-tural properties of silicon-made nanostructures are dramati-cally affected by various factors of the fabrication parame-ters, such as HF concentration, current density, etching time, and the light source used for illumination [5–7]. Altering the growth parameters, especially changing the properties of the illumination light source, one could manufacture dif-ferent nanostructures on a silicon wafer such as nanopores,
802 S. Acikgoz et al.
nanospheres, nanorods, conical nanopillars, etc. In the tradi-tional electrochemical etching process, white light is used as an illumination source and nanopores of random sizes are formed on a silicon wafer [8]. These porous silicon (PSi) samples show an intense red photoluminescence due to quantum confinement when excited with an ultraviolet op-tical source. Moreover, Naddaf et al. studied the role of the illumination wavelength during the etching of the porous sil-icon formation using a halogen lamp. It is found that the PL intensity of PSi is up to 100 times brighter if the PSi sub-strate is irradiated with a 450 nm light source; and the PL spectrum is blue-shifted by some 80 nm [9]. On the other hand, it is observed that 700 nm light illumination used in the fabrication process has no effects on subsequent PL spectrum. Only short-wavelength irradiations have a signifi-cant effect on the photoluminescence and structural proper-ties of PSi due to their short penetration depth [10].
Employing a coherent light source, whether it is contin-uous wave (CW) or pulsed, instead of a white light one, re-sults in formation of conical silicon pillars. In a CW laser-induced etching process, surface formation depends on the laser power density. While pore-like structures are formed at low laser power densities, irregular pillar-like structures are obtained at high power densities [11]. In order to ob-tain regular and sharp conical silicon pillars, the front size of silicon wafer must be illuminated using a coherent pulsed source. Crouch et al. observed that laser pulse rate is an ef-fective parameter on the size of silicon pillars. For example, a femtosecond pulsed laser yields formation of smaller sili-con pillars than a nanosesili-cond laser [12].
Although conventional fabrication of silicon nanostruc-tures is electrochemical anodization of silicon in HF acid, formations of the complex surface morphologies are ob-served as a result of the laser-induced etching (LIE) tech-nique [13]. In the LIE process, the surface morphology de-pends upon laser power density, laser wavelength, etching time and the etchant solution only, despite having no electric bias applied [14,15]. Most of the previous studies, involv-ing the femtosecond laser induced nanostructure formation on silicon, are performed by irradiating samples in HF acid solution. However, Radu et al. report that two different liq-uid solutions, such as chlorine and fluorine precursors, can be used as etchant solutions for micrometer and submicrom-eter scale structuring of silicon [16].
In this article, effects of coherence and wavelength on the growth mechanism of silicon nanopillars are investigated. Our samples are obtained by an electrochemical anodiza-tion of p-type silicon wafers under illuminaanodiza-tion of CW and pulsed lasers, white light, and blue light emitting diodes. A coherent light source alone is not a sufficient agent to be effective in achieving regular and sharp conical silicon nanopillars. On the other hand, wavelength of a pulsed co-herent UV light source is very effective on the regularity and
structural size of silicon nanopillars. It is observed that UV pulsed diode lasers of 337, 405, and 467 nm give the best regular nanopillars. It is demonstrated that when the light source is not pulsed, the coherence becomes ineffective at longer wavelengths.
The easy fabrication of the pillars and control of the structure with wavelength makes silicon an attractive mate-rial, since the samples with different pillar structures have different optoelectronic properties. Besides, the open net-work of the samples with pillars on the surface aligned along the laser beam direction can be used to modify the prop-erties of emitters; for example, filling the pores with dyes for sensing applications. As the fluorescence dynamics of dye-nanobody is concerned, the rate of spontaneous emis-sion for an atom in the vicinity of nanobodies is either en-hanced or inhibited depending on the nanobody structure and the distance of the atom to the nanobody [17]. This is mainly due to the modification of the local electromag-netic field leading to changes in the optical properties of the adsorbed molecule and, therefore, enhancement or inhibi-tion in the decay rate. Here in this work, Boradiazaindacene (BODIPY) dye molecules are attached to our silicon pillars and the inhibition of spontaneous emission is observed by time-resolved lifetime measurements. Our experimental re-sults show that one can control inhibition in the spontaneous emission rate by adjusting the nanopillar size through the fabrication parameters. In addition, no in situ imaging mech-anism is needed to monitor a single tip of the nanostruc-ture since the separation distance between the nanopillars is much greater than the spot-size of the excitation source, which is interrogated in a confocal time-resolved setup.
2 Experimental
P-Si samples are prepared by electrochemical anodiza-tion of p-type silicon wafers at low current densities in HF:C2H5OH (1:1) solutions under illumination of white light, pulsed hydrogen laser (Spectra Physics, 337 nm), He-Ne laser (Thorlabs, 633 nm), and blue and violet pulsed diode laser head with wavelength 467 nm and 405 nm (Pi-coquant, LDH-C-D-470 and LDH-P-C-405B). Laser power densities of pulsed hydrogen, diode violet, diode blue, and CW HeNe lasers are 93.8 mW/cm2,62.5 mW/cm2, 50.0 mW/cm2, and 31.3 mW/cm2, respectively. The illumi-nation area of the laser beam is different for each laser used but it is approximately 11.8–12.6 mm2, which is sufficient to capture the general trend in the nanostructure size gra-dient. Laser sources with different wavelengths are used to investigate the effects of the wavelength on the PSi growth mechanism. The clean silicon wafers are first cut into pieces and aluminum contacts are coated as thin films at the back of the samples by evaporation in Edwards Coating Sys-tem, E306A. Copper wires are attached to the aluminum
Wavelength and coherence effects on the growth mechanism of silicon nanopillars and their use 803
films at the back of the silicon samples by silver paste. The siliconsamples then are immersed into HF:C2H5OH solu-tion. The copper wire is connected to the positive terminal of the power supply and the stainless steel is connected to the negative terminal as in Fig.1. The current is kept con-stant during anodization. The current density and also the time it takes for creating the best samples with the most intense photoluminescence is noted and all samples are pre-pared accordingly. The current density passing through the
Fig. 1 The anodization setup
Fig. 2 Photoluminescence spectrum of a PSi sample
silicon sample and the etching time are 100 mA/cm2and 30 min, respectively.
Photoluminescence experiments are performed for all PSi samples in a dark environment and at the room tem-perature using a fiber optic spectrometer (Ocean Optics, USB4000-VIS-NIR). It is observed that PSi substrate ex-hibits an emission spectrum with a double peak at around 620 and 670 nm due to the quantum confinement ef-fect of silicon nanostructures. The normalized PL spec-tra of all samples are observed to be almost the same as seen in Fig. 2. After PSi samples are analyzed by spec-troscopy, they are impregnated by BODIPY. Dye attach-ment to the PSi surfaces were carried out as follows. First, 4, 4-difluoro-1,3,5,7-tetramethyl-8-[(10-bromo)]-4-bora-3a,4a-diaza-s-indacene, a bromo derivative of the BODIPY dye was synthesized according to previously reported litera-ture [18]. Thereafter, this dye was covalently immobilized onto the surface by first treating the surface with a so-lution of 3-aminopropyltri(methoxy)silane in dry toluene (0.2 mL/10 mL) at room temperature, followed by heat-ing thus modified surface in a solution of BODIPY-Br in N-methylpyrrolidone (2 mg/10 mL) at 60 °C for 20 h. Sur-faces were extensively washed with dichloromethane to re-move any residual unbound dye molecules. The penetra-tion of the dye molecules is studied by spectroscopy and the results show that the dried dye uniformly covers the pore walls.
Time resolved fluorescence lifetime measurements are performed using a TimeHarp 200 PC-Board system (Pico-quant, GmbH). Figure3shows the optic experimental setup. The excitation source used in the experiment is an ultraviolet pulsed diode laser head with a wavelength of 405 nm (LDH-C-D-470 Picoquant, GmbH). In order to obtain a Gaussian beam illumination, a single mode optical fiber is used as a waveguide (Thorlabs, S405-HP). The separation of the flu-orescence emission and the excitation occurs at a dichroic mirror. The excitation light is focused onto the sample us-ing a microscope objective of 0.55 numerical apertures with a working distance of 10.1 mm (Nikon, ELWD 100X). A pinhole, which has a diameter 75 μm, is placed in the focal
Fig. 3 Optical setup
804 S. Acikgoz et al.
Fig. 4 SEM pictures of P-Si
samples produced using various illumination of light sources. (a) White light, (b) blue LED (450 nm), (c) CW HeNe laser (633 nm), (d) pulsed hydrogen laser (337 nm), (e) pulsed diode laser (405 nm), and (f) pulsed diode laser (467 nm)
plane, in order to eliminate out of focus excitation of fluo-rescence.
3 Results and discussion
Porous silicon samples are formed under the illumination of various light sources; such as pulsed diode lasers of 405 nm and 467 nm, pulsed hydrogen laser of 337 nm, CW HeNe laser of 633 nm, blue LED of 440 nm, and white light. Sur-face roughness of the etched areas and nanoscale morphol-ogy are studied using an environmental scanning electron microscope (ESEM). Figure4 shows the surface morphol-ogy of the porous layers prepared with the light sources mentioned above. When images are compared, the effect of the laser light on sample formation can be seen directly. Nanopores are observed in silicon wafers under illumination of white light, blue LED, and the HeNe laser. On the other hand, silicon nanopillars are formed on silicon wafers due to coherence and wavelength effects of different UV pulsed sources.
When the silicon wafer is illuminated with white light source during the electrochemical anodization process, nanopores with different sizes are distributed all over the entire etched area as shown in Fig.4a. Non-monochromatic and incoherent properties of white light source leads to a sur-face disorder. When the silicon wafer is etched under illumi-nation of a CW blue LED of 440 nm in wavelength, the re-sultant structure is more regular than the white light one but randomness in the surface morphology of the sample is still the dominant appearance as shown in Fig.4b. This indicates that relatively more regular nanostructures can be obtained
Fig. 5 A dye molecule attached to a silicon nanopillar
using a single wavelength light source instead of using a white light one. In addition to monochromatic property of the light source, if one adds the coherence property, it is ob-served that the quality of the discreteness of the etched sur-face morphology is enhanced. For example, Fig.4c shows the surface morphology of a silicon wafer etched under co-herent 633 nm HeNe laser. It is seen that the individual ele-ments of the nanostructure of the porous silicon surface are more regularly and discretely pronounced but their sizes are being transformed from nano to microstructure due to the relatively longer wavelength of the HeNe laser. However, these discrete microstructures are not quite regular either, due to the longer wavelength and CW properties of HeNe laser. Moreover, using an ultraviolet, monochromatic, co-herent and pulsed illumination light source allows obtaining quite regular and nice conical pillars. Figures4e,4d, and4f show the conical pillars produced with pulsed hydrogen and pulsed diode lasers. It is observed that the size of the conical
Wavelength and coherence effects on the growth mechanism of silicon nanopillars and their use 805
Fig. 6 The change in the tip size and in vertex of a dye molecule as a
function of illumination wavelength
pillars depends on the wavelengths of the pulsed and coher-ent light sources. As short wavelengths give smaller pillars, longer wavelengths give bigger pillars.
To calculate the size of the silicon pillars, a model shown in Fig.5is used: silicon pillars have conical shapes of radius rat depth t from the tip and vertex angle of α. At t= 100 nm depth, a relation between α and wavelength of the source used is found for each sample. Figure6shows the changes in the tip size and α as a function of illumination wavelength. The radius, therefore, the vertex of the nanopillars changes as a result of illumination wavelength. While the average radius of the nanopillars is about 200 nm, which are formed under the illumination of a hydrogen pulsed diode laser, the nanopillars, which are fabricated by blue laser irradiation, are greater than 260 nm.
Silicon samples shown in Fig. 4 are impregnated by BODIPY dye molecules and the spontaneous emission rate of the dye molecules is studied in the optical setup shown in Fig.3. The attached dye molecule is at a distance of d, which is a fixed distance in our experiments. Pillars grown are seen to be away from each other in the range of 1–2 μm for var-ious wavelengths we used during the production. Since our laser excitation spot-size is about 0.8 micron, one can attach dyes or quantum dots only to the tips of the pillars and the inhibition will be due to the tip only without in situ mon-itoring. Apart from the conical tip of the pillars, the spon-taneous emission rate (or lifetime τ ) is mainly due to the tip and does not vary from its value in the atmosphere for other parts of the pillars. In other words, as one goes from the tip to the bottom of the pillar, the structure gets much thicker, which does not cause any inhibition in BODIPY’s fluorescence lifetime. The optical system used in our exper-imental work shown in Fig.3 is based on a confocal light detection scheme via a 75 μm pinhole in the setup, which allows monitoring the reflected light coming from the very center of the small focused area only. In other words, the possibility of getting illuminations apart from the focal cen-ter is eliminated by this pinhole. If, for example, the illu-mination is due to tip, no reflections other than the tip are
allowed to enter detector since the confocal setup allows fo-cusing this part only, and the lifetime value would be smaller than that of the BODIPY on a normal ordinary substrate or in the atmosphere. If the focused area is not due to the tip but somewhere down the bottom of the pillar, the measured lifetime would not be different than its natural value in the atmosphere since the thick bottom part of the pillar would act as an infinitely large optical environment compared to the radiation wavelength of the BODIPY dye molecules.
The time-resolved fluorescence lifetime of the BOD-IPY dye molecule is performed using the PCI-Board sys-tem (TimeHarp 200, PicoQuant). The measurement of the fluorescence lifetime is based on the time correlated sin-gle photon counting (TCSPC) method. In this method, the time between the detected single photon of the fluorescence (start signal) and the excitation laser pulse (stop signal) is measured. The measured data is plotted as a fluorescence lifetime histogram. The fluorescence lifetime of BODIPY dye molecules attached to silicon nanostructures and in free space are compared. Decay parameters are determined us-ing the double exponential tailfit model, and the best fits are obtained by minimizing χ2values as seen in Fig.7.
The spontaneous emission of an emitter is not an intrin-sic property of the emitter and it is strongly affected by the surrounding environment. Therefore, the decay lifetime of an emitter in the vicinity of a nanostructure is inhibited or enhanced. Such structure may be, for example, a flat surface [19], a nanosphere [20], a nanorod [20], or a nanoparticle [21]. Understanding and controlling the emission properties of molecules in nanostructured geometries has a great poten-tial for applications in the area of nanooptics, biochemistry, and molecular biology [22].
In this work, we have seen that the decay rates of the dye molecules interacting with their surroundings are substan-tially different than those of free dye molecules. For free space lifetime of dye molecule, a silicon wafer is impreg-nated by BODIPY dye molecule and its lifetime is measured as 4.5 ns. When BODIPY dye molecules are embedded in porous silicon, which is formed under white light illumina-tion, the lifetime of the molecules is inhibited, however, it varies between 2.55 and 3.70 ns due to an irregular surface morphology. On the other hand, the lifetime values in the case of pulsed diode laser assisted samples are also inhib-ited but the variation in found to be insignificant due to reg-ularly shaped nanopillars (see Table1). The decay lifetimes of BODIPY are calculated from five different positions in all samples. Calculations of decay fitting parameters for the first lifetime data points of the hydrogen laser and blue laser illu-minated PSi samples are shown in Fig.7. Decay parameters of other positions are also analyzed using the same multi-exponential fitting model and calculated lifetime values are summarized in Table1. It is observed that the decay rate of the dye molecules decreases as the illumination wavelength
806 S. Acikgoz et al.
Fig. 7 (a) Fitting and calculation of decay parameters of BODIPY
dye on (blue) silicon wafer, (red) blue laser illuminated PSi, (violet) hydrogen laser illuminated PSi. (—) indicates multi-exponential fitting
curve. Residuals for fittings on (b) silicon wafer, (c) blue laser illumi-nated PSi, (d) hydrogen laser illumiillumi-nated PSi
Table 1 The fluorescence
lifetimes of BODIPY dye molecule in all samples
White light Illuminated τ (ns) Hydrogen laser illuminated τ (ns) Blue laser illuminated τ (ns) 2.55 2.66 3.50 2.66 2.77 3.57 2.88 2.79 3.59 3.55 2.82 3.62 3.70 2.86 3.65
Wavelength and coherence effects on the growth mechanism of silicon nanopillars and their use 807
increases since the illumination wavelength is almost lin-early proportional to the size of the grown structure. While the average lifetime for hydrogen laser illuminated sample is 2.80 ns, this value is 3.61 ns for a pulsed blue laser illu-minated sample.
4 Conclusion
The illumination light source effects on the structural prop-erties of porous silicon are described for the samples pro-duced by an electrochemical anodization method. Conical regular nanopillar structures are formed under pulsed, coher-ent, and monochromatic UV illumination. While white light source irradiation, which contains multiple wavelengths, causes formation of irregular nanopores with various diam-eters, a blue LED irradiation gives relatively regular and greater nanopores. Our work shows that the porous layer structure can be controlled easily by suitable selection of a coherent illumination light source and better control can be achieved when a pulsed UV laser is used in the etching process. The wavelength effects on the pillar tip size and vertex angle are investigated for 337, 405, and 467 nm UV pulsed lasers. As the illumination wavelength increases, the nanopillar size increases proportionally. In addition to struc-tural analysis, the fluorescence dynamics due to the interac-tion of fluorescent BODIPY azide dye molecules with such silicon nanostructures is studied with a time resolved flu-orescence lifetime measurement system. The fluflu-orescence lifetime of a dye molecule in the close proximity of silicon nanopillar is inhibited, which means that the spontaneous emission rate depends on the pillar size. In other words, smaller nanopillars give rise to smaller lifetimes for BOD-IPY dye molecules. A pulsed UV tunable source can be con-veniently used to produce appropriate pillar sizes and hence the desired lifetimes. It might have been interesting to try whether it is possible to obtain a regular nanostructure with a laser illumination only, that is, similar to that of LIE, with-out any electrical current; however, this issue is left for a future project.
Acknowledgement This work was supported by TUBITAK (con-tract numbers: 106T011 and 107T206) and Bogazici University Re-search Fund (contract numbers: 05HB301 and 08HB301).
References
1. V. Sivakov, G. Andra, A. Gawlik, A. Berger, J. Plentz, F. Falk, S.H. Christiansen, Nano Lett. 9, 1549 (2009)
2. K.S. Dancil, D.P. Greiner, M.J. Sailor, J. Am. Chem. Soc. 121, 7925 (1999)
3. N.M. Park, T.S. Kim, S.J. Park, Appl. Phys. Lett. 78, 2575 (2001) 4. L.T. Canham, Appl. Phys. Lett. 57, 1046 (1990)
5. Z. Gaburro, H. You, D. Babic, J. Appl. Phys. 84, 6345 (1998) 6. M.C. Cheung, P.J.R. Roche, M.H. Hassan, A.G. Kirk, Z. Mi, V.P.
Chodavarapu, J. Nanophotonics 5, 53503 (2011)
7. N. Chiboub, N. Gabouze, J.N. Chazalviel, F. Ozanam, S. Moulay, A. Manseri, Appl. Surf. Sci. 256, 3826 (2010)
8. O. Erdamar, B. Bilen, Y. Skarlatos, G. Aktas, M.N. Inci, Phys. Status Solidi C 4, 601 (2007)
9. M. Naddaf, H. Hamadeh, Mater. Sci. Eng., C, Biomim. Mater., Sens. Syst. 29, 2092 (2009)
10. V.V. Doan, R.M. Penner, M.J. Sailor, J. Phys. Chem. 97, 4505 (1993)
11. R. Kumar, H.S. Mavi, A.K. Shukla, Micron 39, 287 (2008) 12. C.H. Crouch, J.E. Carey, J.M. Warrender, M.J. Aziz, E. Mazur,
Appl. Phys. Lett. 84, 2004 (1850)
13. N. Noguchi, I. Suemune, Appl. Phys. Lett. 62, 1429 (1993) 14. A. Ramizy, Z. Hassan, K. Omar, Sci. China, Technol. Sci. 54, 58
(2011)
15. R.K. Soni, G.R. Bassam, S.C. Abbi, Appl. Surf. Sci. 214, 151 (2003)
16. C. Radu, S. Simon, M. Zamfirescu, M. Ulmeanu, M. Enculescu, M. Radoiu, J. Appl. Phys. 110, 34901 (2011)
17. J.I. Gersten, A. Nitzan, J. Chem. Phys. 75, 1139 (1981)
18. J.L. Shepherd, A. Kell, E. Chung, C.W. Sinclar, M.S. Workentin, D. Bizzotto, J. Am. Chem. Soc. 126, 8329 (2004)
19. S. Astilean, S. Garrett, P. Andrew, W.L. Barnes, J. Mol. Struct.
651-653, 277 (2003)
20. R. Bardhan, N.K. Grady, J.R. Cole, A. Joshi, N.J. Halas, ACS Nano 3, 744 (2009)
21. M.R. Vasic, L.D. Cola, H. Zuilhof, J. Phys. Chem. C 113, 2235 (2009)
22. E.C. Wu, J.H. Park, J. Park, E. Segal, F. Cunin, M.J. Sailor, ACS Nano 2, 2401 (2008)
