pubs.acs.org/NanoLett
Infrared Perfect Absorber and Its Application As Plasmonic Sensor Na Liu, Martin Mesch, Thomas Weiss, Mario Hentschel, and Harald Giessen* 4. Physikalisches Institut, Universita¨t Stuttgart, D-70569 Stuttgart, Germany ABSTRACT We experimentally demonstrate a perfect plasmonic absorber at λ ) 1.6 µm. Its polarization-independent absorbance is 99% at normal incidence and remains very high over a wide angular range of incidence around (80°. We introduce a novel concept to utilize this perfect absorber as plasmonic sensor for refractive index sensing. This sensing strategy offers great potential to maintain the performance of localized surface plasmon sensors even in nonlaboratory environments due to its simple and robust measurement scheme. KEYWORDS Perfect absorbers, LSPR resonances, infrared sensors, glucose sensing, plasmonics
I
metamaterial losses.18 There are also theoretical proposals of wide-angle perfect absorbers in the gigahertz frequency regime using microstructures in combination with thick metallic films.27 Alternatively, perfect absorption can also be achieved with structured metallic surfaces,28 microcavities,29 and subwavelength hole arrays.30 In general, it is desirable to have perfect absorbers which are insensitive on the incident angle and the polarization of light for practical applications. In this letter, we introduce a novel plasmonic device which combines the concepts of a perfect absorber and an LSPR sensor. We demonstrate experimentally for the first time a narrow-band perfect absorber working as plasmonic sensor in the near-infrared regime. We show that this plasmonic device yields ∼99% absorbance in the experiment and remains highly absorptive over a wide range of incident angles for both transverse electric (TE) and transverse magnetic (TM) configurations. We also show that careful consideration of damping in the metal film is of particular importance for realizing perfect absorbance in experiment. Furthermore, we demonstrate that our plasmonic device can work as LSPR sensor, where mode volumes in the attoliter range are common. Different from existing LSPR sensors which measure the spectral shift of a resonance upon a refractive index change of the surrounding medium,1 our plasmonic absorber sensor detects rather a relative intensity change dI(λ)/I(λ) at a fixed wavelength λo induced by a refractive index change dn. A figure of merit FOM*, which is introduced by J. Becker et al., is defined as FOM* ) max|[dI(λ)/dn(λ)]/(I(λ)|. λo is chosen where FOM* has a maximum value.31 We have achieved a FOM* around 87 in the experiment by measuring the intensity changes with different local dielectric materials (air and water) at the sample surface. Our FOM* is nearly four times larger than that of plasmonic gold nanorod sensors.31 Our absorber sensor concept offers substantial advantages over classical
n plasmonics, the optical properties of metallic nanoparticles are the basis of many fascinating applications such as chemical and biomedical sensing,1 surfaceenhanced spectroscopy,2 and near-field scanning optical microscopy.3 The collective excitation of conduction electrons in gold or silver nanoparticles can lead to a localized surface plasmon resonance (LSPR) when interacting with an incident light field.4 Such LSPR strongly depends on the size, shape, and surrounding dielectric environment of nanostructures.5 In particular, the latter dependence opens a route toward refractive index sensing in which small concentrations of target molecules can be detected.5 So far, different plasmonic nanostructures have been used to optimize LSPR sensors with a large spectral shift for a given change in refractive index, including nanospheres, nanoshells, nanorice, nanostars, and etc.1,6-10 More complex plasmonic sensors based on specific physical mechanisms, for example, on the classical analog of electromagnetically induced transparency11-15 have also been explored. In practical applications, losses are inevitable in plasmonic metallic nanostructures. Significant effort has been paid to achieve low-loss devices, for instance by optimizing structural geometries16 and by using gain materials.17 Recently, the concept of perfect metamaterial absorbers initiated a new research area in which losses are actually put to the advantage.18-22 Metamaterials are resonant metallic nanostructures with unit cells much smaller than the operating wavelength of light.23-26 It has been shown that perfect absorption is possible with metamaterials by properly engineering of electric and magnetic responses in the gigahertz frequency regime.18-21 The basic idea is to minimize the reflectance through impedance matching and simultaneously eliminate the transmittance by maximizing the
* To whom correspondence should be addressed. E-mail:
[email protected]. Received for review: 12/11/2009 Published on Web: 06/18/2010 © 2010 American Chemical Society
2342
DOI: 10.1021/nl9041033 | Nano Lett. 2010, 10, 2342–2348
that different from many plasmonic applications where losses deteriorate the performance of potential devices for the realization of perfect absorbers the substantial losses in the metal are desirable and can be exploited as well. In the case of five times damping constant of bulk gold, the reflectance minimum slightly deviates from perfect absorbance. The losses in gold with a damping constant equal to three times that of bulk gold are sufficient to yield a strong narrow-band resonance. In fact, the structural parameters described in Figure 1 were particularly optimized for the gold film (i.e., perfectly impedance matched18 to the air on top of the sample), which has a damping constant three times that of bulk gold because it provides the best match to our experimental gold film. To better understand the nature of our perfect absorber, the current distribution at resonance was simulated and is depicted in Figure 2b. It is evident that antiparallel currents are excited in the gold disk and the bottom gold layer.36,37 Actually, this is often called a magnetic resonance because the circulating currents result in a magnetic moment which can strongly interact with the magnetic field of the incident light.36,37 At resonance, a strong enhancement of the localized electromagnetic field is established between the two layers. Consequently, electromagnetic energy can be efficiently confined in the intermediate MgF2 spacer and therefore no light is reflected back. This gives rise to a pronounced reflectance dip in the spectrum with nearly zero intensity, therefore leading to ∼100% absorbance. In fact, our device can work as a perfect absorber over a wide range of incident angles. Figure 2c shows the angular dispersions of the absorbance peak at various angles of incidence for both TE and TM configurations. For the TM polarization, the absorbance peak is nearly independent of the incident angle and it is 96% even at 80°. This is because the direction of the magnetic field of the incident light remains unchanged with various incident angles and it can efficiently drive the circulating currents at all angles of incidence. Conversely, for the TE polarization, the magnetic field cannot drive the circulating currents efficiently at large angles.19 Nevertheless, the absorbance still remains at 50% at 80°. The designed structure is compatible with nanofabrication techniques. The thick gold film and the MgF2 dielectric layer were subsequently deposited onto a glass substrate using electron-gun evaporation. The disk array was defined in positive resist (polymethylmethacrylate, PMMA) using standard electron beam lithography followed by a gold liftoff procedure. The sample has a structure area of 100 µm × 100 µm. The top left picture in Figure 3a shows the normal view of the sample obtained by a scanning electron microscopy. To demonstrate the feasibility of our absorber working as a LSPR sensor, Figure 3a shows the results of a proof-ofprinciple experiment, displaying the measured reflectance spectra with air (n ) 1) and water (n ) 1.312),38 on the sample surface. The measurements were performed by a Fourier-transform infrared spectrometer with electric field
FIGURE 1. Schematic of the perfect absorber structure and the incident light polarization configuration. The diameter and thickness of the gold disks are 352 and 20 nm, respectively. The periods in both x and y-directions are 600 nm. The thickness of the MgF2 spacer is 30 nm and the thickness of the gold mirror is 200 nm. The whole structure resides on a glass substrate.
sensing methods and is of special interest in numerous applications due to its low-background detection scheme. Figure 1 illustrates the geometry of the absorber sensor structure. It consists of two functional layers. The top layer is a two-dimensional gold disk array and the bottom layer is a gold mirror. The two layers are separated by an MgF2 dielectric spacer. Because of the presence of the gold mirror, the transmittance of the structure is totally eliminated across the entire near-infrared frequency regime (i.e., T ) 0). The structure is designed to be polarization independent in xand y-directions at normal incidence. To investigate the resonant behavior of the structure, numerical simulations were performed by employing a Finite Integration Time Domain algorithm.32 For excitation of the structure, we use normally incident light with its polarization along the xdirection as shown in Figure 1. The permittivity of the MgF2 spacer33 is taken as 1.9, and the permittivity of bulk gold in the near-infrared34 is described by the Drude model with the plasma frequency ωpl ) 1.37 × 1016 s-1 and the damping constant ωc ) 4.08 × 1013 s-1. Owing to the surface scattering and grain boundary effects in thin films, the damping constant of the gold film in the real system is likely higher than that of bulk gold.35 To elucidate the influence of the damping constant of the gold film on the minimum reflectance value at resonance, Figure 2a presents the simulated reflectance spectra for damping constants of one, three, and five times that of bulk gold. As shown in Figure 2a, reflectance dips with different amplitudes are observable. In particular, a strong resonance with nearly zero reflectance (R ) 0.28%) is achieved for three times damping constant of bulk gold (see the red curve in Figure 2a). Consequently, a perfect absorber is obtained (A ) 1 - T R). Nevertheless, for a damping constant equal to one time that of bulk gold, the reflectance minimum is around 30% and it leads to only 70% absorbance. Notably, this illustrates © 2010 American Chemical Society
2343
DOI: 10.1021/nl9041033 | Nano Lett. 2010, 10, 2342-–2348
FIGURE 2. (a) Simulated reflectance spectra in dependence on the damping constant of the gold film. Reflectance with zero intensity is achieved using a damping constant that is equal to three times that of bulk gold. (b) Calculated current distribution at resonance where perfect absorbance occurs. Antiparallel currents are excited in the gold disk and the gold film. (c) The simulated angular dispersions of the absorbance peak for TE and TM configurations.
characterized by the black curve in Figure 3a, the experimental reflectance reaches a minimum of 1% at 185.6 THz (1.6 µm) in air, which corresponds to an experimental absorbance of 99%. When water (red curve) is applied onto the sample surface, a clear increase of the reflectance intensity from 1% to 28.7% at 185.6 THz is visible, resulting from the refractive index change of the local dielectric environment. The correspondingly simulated reflectance
polarization as illustrated in Figure 1 at normal incidence. The sensing principle relies on the fact that zero reflectance (i.e., perfect impedance matching) occurs only for a certain refractive index of the surrounding medium. The variation of the refractive index of the surrounding medium gives rise to nonzero reflectance (i.e., nonperfect absorbance) and therefore allows for the extremely sensitive detection of the intensity change in reflectance at a fixed frequency. As © 2010 American Chemical Society
2344
DOI: 10.1021/nl9041033 | Nano Lett. 2010, 10, 2342-–2348
FIGURE 3. (a) Top: Experimental demonstration of perfect absorbance in air. Experimental tuning of the reflectance and absorbance spectra by changing the dielectric environment which is adjacent to the gold disks from air to water is shown. The SEM image of the sample is presented in the left column. Bottom: The correspondingly simulated reflectance and absorbance spectra with different dielectric materials (air and water) on the structure surface. In simulations, a damping constant equal to three times that of bulk gold was utilized. The agreement between experimental and simulated results is nearly perfect. (b) Experimental FOM* as a function of frequency. The highest value of FOM* is 87, which is reached at a frequency slightly away from the minimum reflectance. In the discrete calculation of the experimental FOM*, the reflectance difference between water (Iwater (λ)) and air (Iair (λ)) was taken for the derivative in eq (1). Iair (λ) was taken for the denominator I (λ) in eq 1. The experimental reflectance spectrum with air on the structure surface is replotted with a black curve. © 2010 American Chemical Society
2345
DOI: 10.1021/nl9041033 | Nano Lett. 2010, 10, 2342-–2348
and absorbance spectra characterized by dashed lines are also shown in Figure 3a. In simulations, a damping constant equal to three times that of bulk gold for the gold film was used. The agreement between experimental and simulated results is nearly perfect. Subsequently, the FOM* can be calculated according to
FOM* ) max
|
dI(λ)/dn(λ) I(λ)
|
(1)
where dI(λ)/I(λ) is the relative intensity change at a fixed wavelength induced by a refractive index change dn.31 I (λο) corresponds to the intensity where FOM* reaches a maximum value. Figure 3b presents the experimental FOM* for water (with a dn of 0.312) as a function of frequency. The maximum value of FOM* is around 87 and it occurs at the position where the slope of dI(λ)/I(λ) is highest right next to the perfect absorption. Compared to the experimental FOM* of plasmonic gold nanorods31 which is around 24, our FOM* is much higher. The fundamental advantage of our absorber sensor lies in the fact that it allows detection of photons that are reflected by the nonperfect absorber upon refractive index change versus a nearly dark reference measurement where only few photons are reflected from the perfect absorber. For optimum operation, specific care has to be given to reduce the background intensity due to scattering as well as due to detector noise. Our experimentally determined sensitivity in terms of wavelength shift per refractive index unit is around 400 nm/RIU, which is compatible with excellent current LSPR sensor data.15 It is also noteworthy that our sensor design is highly scalable and can be tuned to the visible or gigahertz frequency regimes by changing the sizes appropriately. In addition, for specific applications perfect absorbance can also be designed for surrounding dielectrics other than air, such as water or glucose solution with a certain refractive index. As an example, Figure 4a presents the simulated spectral tuning of an absorber sensor which exhibits perfect absorbance for water as the surrounding medium. The reflectance spectrum, which is characterized by the black curve in Figure 4a, reaches an extremely low intensity at 0.04% at 175.5 THz in water. For clarity, the resonance position is highlighted by a pink dashed line as shown in Figure 4a. It is evident that at 175.5 THz the reflectance increases with increasing the refractive index of the glucose solution, which is applied on the structure surface. The perfect absorber can therefore be a highly promising device for detecting refractive index changes of a sensing agent. FOM* calculations based on eq 1 by extracting the data from Figure 4a show that the highest value of FOM* is achieved not exactly at the frequency of perfect absorption, but rather slightly detuned (see Figure 4b). Such detuning is also demonstrated by J. Becker et al. using their nanorod plasmonic sensor. Subsequently, the careful optimization of the structural parameters for achiev© 2010 American Chemical Society
FIGURE 4. (a) Simulated reflectance spectra of an absorber sensor designed for water as reference medium. Spectral tuning occurs with glucose solutions varying from 0 to 25% that have different refractive indices. The resonance position of the perfect absorber in an aqueous environment at 175.5 THz is highlighted by a dashed pink line. The diameter and thickness of the gold disks are 330 and 19 nm, respectively. The periods in both x- and y-directions are 600 nm. The thickness of the MgF2 spacer is 24 nm and the thickness of the gold mirror is 200 nm. The whole structure resides on a glass substrate. In simulations, a damping constant equal to three times that of bulk gold was utilized. (b) Calculated FOM* as a function of frequency: In the calculation of the numerical FOM*, we used a finer differential quotient with n ) 1.322 and 1.302. The average intensity value of the cases n ) 1.322 and 1.302 was then taken for the denominator I (λ) in eq 1. The maximum value is 94, and it is reached slightly away from the reflectance minimum. The simulated reflectance spectrum with water on the structure surface is replotted with a black curve.
ing drastically low reflectance near resonance is the key issue to obtain a very high FOM*. 2346
DOI: 10.1021/nl9041033 | Nano Lett. 2010, 10, 2342-–2348
(8)
In essence, our sensor scheme suggests potential refractive index sensor platforms in which LSPR sensing is based on straightforward reflectance measurements simply using a single-wavelength light source and the exact incidence angle does not matter as much. Such advantage will benefit the development of simple and cheap sensors for chemical sensing and biomedical diagnostics. So far, the best plasmonic sensors have been mostly synthesized by chemical methods1 because gold or silver nanostructures obtained from chemical synthesis can be single crystalline. The introduction of the perfect absorber sensors discards this restriction since the intrinsic losses in the metal are essential to achieve perfect absorption. As a result, many nanofabrication technologies such as electron beam lithography, mask colloidal lithography,10 interference lithography,39 nanosphere lithography,40 nanoimprint lithography,41 and focusedion beam writing15 can be widely applied to manufacture ultrasensitive plasmonic sensors. In conclusion, we have experimentally demonstrated a plasmonic perfect absorber in the near-infrared, which is independent of polarization. Specifically, this wide-angle absorber can work as an LSPR sensor with high FOM* close to triple digit values. The results agree well with theoretical predictions. Also, our absorber sensor is very flexible and can be designed for achieving perfect absorbance in specific surrounding materials by impedance matching, even when lossy metals are being employed in the nanostructure. The combination of basic sensing principles and particular physical phenomena opens up a new pathway for plasmonic sensing and it will facilitate a new class of nanoscopic bio/ chemosensors. The presented sensing strategy has general applicability for a large variety of nanofabrication technologies and it has profound significance for developing cheap sensing equipment in which only simple measurement technologies are needed.
(9) (10) (11) (12) (13) (14) (15) (16) (17) (18) (19) (20)
(21) (22) (23) (24) (25) (26) (27)
Acknowledgment. We thank Professor M. Dressel for useful discussions and comments. We acknowledge S. Hein for his metamaterial visualizations. We thank Patrick Mai and Andreas Tittl for extensive help. We acknowledge H. Gra¨beldinger and M. Ubl for technical assistance. This work was financially supported by Deutsche Forschungsgemeinschaft (SPP1391 and FOR557), by Landesstiftung BW, and by BMBF (13N9048 and 13N10146).
(28) (29) (30) (31) (32) (33) (34)
REFERENCES AND NOTES (1) (2) (3) (4) (5) (6) (7)
Lal, S.; Link, S.; Halas, N. J. Nat. Photonics 2007, 1, 641–648. Moskovits, M. Rev. Mod. Phys. 1985, 57, 783–826. Paesler, M. A.; Moyer, P. J. Near-field Optics: TheoryInstrumentation, and Applications; Wiley: New York, 1996. Kreibig, U.; Vollmer, M. Optical Properties of Metal Clusters; Springer: Berlin, 1995. Stewart, M. E.; Anderton, C. R.; Thompson, L. B.; Maria, J.; Gray, S. K.; Rogers, J. A.; Nuzzo, R. G. Chem. Rev. 2008, 108, 494–521. Soennichsen, C.; Alivisatos, A. P. Nano Lett. 2005, 5, 301–304. Park, T. H.; Mirin, N.; Lassiter, J. B.; Nehl, C. L.; Halas, N.; Nordlander, P. ACS Nano 2008, 2, 25–32. © 2010 American Chemical Society
(35) (36)
(37)
2347
Lassiter, J. B.; Aizpurua, J.; Hernandez, L. I.; Brandl, D. W.; Romero, L.; Lal, S.; Hafner, J. H.; Nordlander, P.; Halas, N. J. Nano Lett. 2008, 8, 1212–1218. Dmitriev, A.; Haegglund, C.; Chen, S.; Fredriksson, H.; Pakizeh, T.; Kaell, M.; Sutherland, D. S. Nano Lett. 2008, 8, 3893–3898. Larsson, E. M.; Alegret, J.; Kaell, M.; Sutherland, D. S. Nano Lett. 2007, 7, 1256–1263. Hao, F.; Nordlander, P.; Sonnerfraud, Y.; Van Dorpe, P.; Maier, S. A. ACS Nano 2009, 3, 643–652. Hao, F.; Sonnefraud, Y.; Van Dorpe, P.; Maier, S. A.; Halas, N. J.; Nordlander, P. Nano Lett. 2008, 8, 3983–3988. Zhang, S.; Genov, D.; Wang, Y.; Liu, M.; Zhang, X. Phys. Rev. Lett. 2008, 101, No. 047401. Liu, N.; Langguth, L.; Weiss, T.; Ka¨stel, J.; Fleischhauer, M.; Pfau, T.; Giessen, H. Nat. Mater. 2009, 8, 758–762. Liu, N.; Weiss, T.; Mesch, M.; Langguth, L.; Eigenthaler, U.; Hirschner, M.; So¨nnichsen, C.; Giessen, H. Nano Lett. 2009, 10, 1103–1107. Zhou, J.; Koschny, T.; Soukoulis, C. M. Opt. Express 2008, 16, 11147–11152. Plum, E.; Fedotov, V. A.; Kuo, P.; Tsai, D. P.; Zheludev, N. I. Opt. Express 2009, 17, 8548–8551. Landy, N. I.; Sajuyigbe, S.; Mock, J. J.; Smith, D. R.; Padilla, W. J. Phys. Rev. Lett. 2008, 100, 207402. Tao, H.; Bingham, C. M.; Strikwerda, A. C.; Pilon, D.; Shrekenhamer, D.; Landy, N. I.; Fan, K.; Zhang, X.; Padilla, W. J.; Averitt, R. D. Phys. Rev. B 2008, 78, 241103. Avitzour, Y.; Urzhumov, Y. A.; Shvets, G. Phys. Rev. B 2009, 79, No. 045131. Mosallaei, H.; Sarabandi, K. A one-layer ultra-thin meta-surface absorber. In Antennas Propag. Soc. Int. Symp., 2005 IEEE; IEEE: Los Alamitos, CA, 2005; Vol. 1B, pp 615–618. Wu, C.; Avitzour, Y.; Shvets, G. Ultra-thin, wide-angle perfect absorber for infrared frequencies. In Proc. SPIE, Proceedings of Metamaterials: Fundamentals and Applications, San Diego, CA, August 10–14, 2008; Noginov, M. A., Zheludev, N. I., Boardman, A. D., Engheta, N., Eds.; SPIE: Bellingham WA; 2008; Vol. 7029, p 70290W. Wang, B. N.; Koschny, T.; Soukoulis, C. M. Phys. Rev. B 2009, 80, No. 033108. Teperik, T. V.; Garcia de Abajo, F. J.; Borisov, A. G.; Abdelsalam, M.; Bartlett, P. N.; Sugawara, Y.; Baumberg, J. J. Nat. Photonics 2008, 2, 299–301. Smith, D. R.; Pendry, J. B.; Wiltshire, M. C. K. Science 2004, 305, 788–792. Shalaev, M. Nat. Photonics 2007, 1, 41–48. Liu, N.; Guo, H. C.; Fu, L. W.; Kaiser, S.; Schweizer, H.; Giessen, H. Nat. Mater. 2008, 7, 31–37. Liu, N.; Liu, H.; Zhu, S. N.; Giessen, H. Nat. Photonics 2009, 3, 157–162. Diem, M.; Koschny, T.; Soukoulis, C. M. Phys. Rev. B 2009, 79, No. 033101. Laroche, M.; Carminati, R.; Greffet, J. J. Phys. Rev. Lett. 2006, 96, 123903. Celanovic, I.; Perreault, D.; Kassakian, J. Phys. Rev. B 2005, 72, No. 075127. Hu, C. G.; Liu, L. Y.; Chen, X. N.; Luo, X. G. Opt. Express 2009, 17, 16745. Becker, J.; Truegler, A.; Jakab, A.; Hohenester, U.; Soennichsen, C. Plasmonics 2010, 5, 161. All the simulations in this paper are performed using the following software package: CST Microwave Studio; CST GmbH: Germany, 2010. Dodge, M. J. Appl. Opt. 1984, 23, 1980–1985. Ordal, M. A.; Long, L. L.; Bell, R. J.; Bell, S. E.; Bell, R. R.; Alexander, R. W., Jr.; Ward, C. A. Appl. Opt. 1983, 22, 1099–1119. Zhang, S.; Fan, W.; Malloy, K. J.; Brueck, S. R. J.; Panoiu, N. C.; Osgood, R. M. J. Opt. Soc. Am. B 2006, 23, 434–438. Liu, N.; Guo, H. C.; Fu, L. W.; Kaiser, S.; Schweizer, H.; Giessen, H. Adv. Mater. 2007, 19, 3628–3632. Cai, W. S.; Chettiar, U. K.; Yuan, H. K.; de Silva, V. C.; Kildishev, A. V.; Drachev, V. P.; Shalaev, V. M. Opt. Express 2007, 15, 3333–3341. Previous work on magnetic mirrors has demonstrated the ability of magnetic resonances in metamaterials for loss enhancement, see e.g., Fedotov, V. A.; Mladyonov, P. L.; Prosvirnin, S. L.; Zheludev, N. I. Phys. Rev. E 2005, 72, 056613. Schwanecke, A. S.; DOI: 10.1021/nl9041033 | Nano Lett. 2010, 10, 2342-–2348
Fedotov, V. A.; Khardikov, V. V.; Prosvirnin, S. L.; Chen, Y.; Zheludev, N. I. J. Opt. A: Pure Appl. Opt. 2007, 9, L1–L2. (38) Fernandez-Prini, R.; Harvey, A. H.,; Palmer, D. A. Aqueous Systems at Elevated Temperatures and pressures; Elsevier: New York, 2004. (39) Feth, N.; Enkrich, C.; Wegener, M.; Linden, S. Opt. Express 2007, 15, 501–507.
© 2010 American Chemical Society
(40) Gwinner, M.; Koroknay, E.; Fu, L.; Patoka, P.; Kandulski, W.; Giersig, M.; Giessen, H. Small 2009, 5, 400–406. (41) Wu, W.; Kim, E.; Ponizovskaya, E.; Liu, Y. M.; Yu, Z. N.; Fang, N.; Shen, Y. R.; Bratkovsky, A. M.; Tong, W.; Sun, C.; Zhang, X.; Wang, S.; Williams, R. S. Appl. Phys. A 2007, 87, 143–150.
2348
DOI: 10.1021/nl9041033 | Nano Lett. 2010, 10, 2342-–2348
