Open Access
Add to BibSonomyAbstract
On the substrate carrying a sub-wavelength grating covered with a thin metal layer, a fluorescent dye-labeled cell was observed by fluorescence microscope. The fluorescence intensity was more than 20 times greater than that on an optically flat glass substrate. Such a great fluorescence enhancement from labeled cells bound to the grating substrate was due to the excitation by grating coupled surface plasmon resonance. The application of a grating substrate to two-dimensional detection and fluorescence microscopy appears to offer a promising method of taking highly sensitive fluorescence images.
©2008 Optical Society of America
1. Introduction
Surface plasmon resonance (SPR) modes [1–3] are produced by the resonance between surface plasmon polaritons (SPP) at a metal-dielectric interface and p-polarized incident light at a required angle and enhance the electric field intensity to about 20–100 times the level of the incident field [1, 3]. Highly sensitive detection by SPR has therefore been applied to various kinds of biosensor [4, 5]. The electric field of SPR can be applied to excite fluorescent molecules, which is known as surface plasmon field-enhanced fluorescence spectroscopy (SPFS) [6–8]. The fluorescence observed in SPFS is also enhanced due to excitation by the enhanced surface plasmon-field. In addition to the advantage of high sensitivity, the merit of selective detection specific to the surface makes application to bio-chips [7, 8] possible. The present paper does not however aim to provide a complete account of the advantages of SPR for microanalysis.
In propagated SPR (cf. localized SPR [9, 10]), the prism-coupled SPR (by attenuated total internal reflection setup) of Kretschmann geometry has been widely used to study specific binding on a biosensor and to develop multi-array systems [4, 11]. However, the incident angle is inevitably greater (around 60 degree at 630nm-wavelength) in water or buffer solution even if a substrate and prism with high refractive index (1.85) are used. Such a severe restriction of a wide incident angle makes it difficult to apply SPFS to a standard fluorescence microscopy. Grating-coupled SPR (GC-SPR) [1, 3, 12, 13], another form of propagated SPR, makes direct coupling between SPP and the radiation modes possible. As is well known, the SPP dispersion relation for a periodic metal layer with a grating pitch of Λ is described in Eq. (1).
in which k SPP, k 0, and k g are wave vectors of SPP, incident light, and grating corresponding to 2π/Λ, respectively, and εd, εm, and θ are complex dielectric constants for dielectric and metal and the incident angle, respectively. The use of metal grating with the required grating pitch, depth, layer structure, etc., enhances the fluorescence due to excitation by GC-SPR.
Some studies of GC-SPR excited fluorescence were already reported [14–16]. Barnes et al. [14] and Lin et al. [15] fabricated a thick emissive layer of Alq3 coated with a grating-structured metal film. Such an emission of light through a metal film can provide an application to the organic light-emitting diodes (OLEDs) and biosensor. Smolyaninov et al. [16] fabricated a fluorescent dye layer on a dielectric grating deposited on the top of a thin metal film and the observed enhanced fluorescence was considered to be promising in sensing application. On the other hand, the goal of this study using the glass plate with a grating and the layer configuration on a grating is totally different from those of the other earlier studies. We used the grating-glass plate coated with metal and SiO2 films as a substrate for fluorescence microscopic observation of a labeled cell using GC-SPR excited fluorescence. The configuration of layers prepared on the grating was decided as Cr/Ag/Cr/SiO2, i.e., 200nm-thick silver layer, 20nm-thick SiO2 overlayer, and adhesion layer of Cr (less than 1 nm), so that an enhanced fluorescence could efficiently be detected without a fluorescence quench. By controlling the pitch of grating, the required incident angle, i.e. the resonance angle, can be adjusted to below 10 degrees unlike in prism-coupling SPR. A small incident angle provides advantages: (1) an objective lens with a small numerical aperture (NA) without oil-immersion can be used even in a large magnification, (2) in an objective with a large NA, the incident angle range can be adjusted to be small and that of fluorescence to be still wide for highly S/N detection by minimizing an aperture stop (AS). GC-SPR thus provides a promising excitation field in applying SPFS to microscopy.
Meanwhile, the enhancement factor of electric-field intensity, Ef, on metal surfaces with a sinusoidally modulated height profile can be described [1] as follows:
in which εm′ and εm″ are the real part and imaginary parts, respectively, of a complex dielectric constant, and θ and R are the incident angle and reflectivity. A greater fluorescence enhancement is therefore expected on a grating surface with the appropriate metal layer, a small incident angle, and use of a substrate that shows a sharp SPR dip with low reflectivity. The use of a silver layer as a metal grating is promising because the real and imaginary parts have a high negative and a low positive value, respectively, across a wide visible spectrum; for example, the values of εm′ and εm″ correspond to -17 and 0.4, respectively, at a wavelength of 633nm. These conditions in silver suggest the possibility of a greater enhancement than in gold. In the results of our initial experiment, though the surface profile of grating pattern fabricated in this study is not sinusoidal, more than 24-times enhanced fluorescence image of cells was observed on the grating substrate by an optical microscope compared with that on a general slide glass.
2. Experiment
2.1 Fabrication of grating on SiO2 substrate surface
SiO2 substrates were cleaned twice by sonicating in ethanol for 15 min. A sub-wavelength structure was fabricated on the surface of a SiO2 plate under precise regulation of its shape and surface morphology. The periodic grating pattern of a positive resist was formed on the SiO2 by two-beam interference using a He-Cd laser with a wavelength of 325 nm. The subwavelength structure of SiO2 was etched using inductive-coupled plasma (ICP) with C3F8 gas [17]. The periodicity of the grating and the depth and duty ratio (ratio of convex surface) were 480nm, 31nm, and 0.40, respectively; the wall was almost vertical, as shown in scanning probe microscopy (Fig. 1).
Fig. 1. Scanning probe microscope image of grating fabricated on SiO2 surface with 480-nm pitch and 31-nm depth.
2.2 Coating with metal layers
Slide glass plates and grating substrates were cleaned by sonicating in a 1% Hellmanex (Hellma, Müllheim, Germany) solution for 20 min and rinsing extensively with fresh Milli-Q water, after which they were dried completely under air flow.
Thin metal layers covered with a SiO2 overlayer (with an adhesion layer of Cr) were prepared on both substrates via the following process. First, a Cr layer of less than 1-nm thickness was deposited using an rf sputter setup (Rikensya, specially made) at a power of 20W at room temperature in an Ar gas flow of 20cc/min. A 200nm silver layer was then deposited onto the Cr layer by sputtering at a power of 30W at room temperature in an Ar gas flow, after which Cr sputtering onto the Ag layer was repeated without opening the chamber. Finally, a 20 nm-thick SiO2 layer was deposited onto the Cr layer by rf-magnetron sputtering at a power of 40W at room temperature in an Ar gas flow of 20cc/min. 20nm-thick SiO2 layer can suppress the energy transfer (quench) of fluorescence excited by GC-SPR field to a metal layer [8] and keep a metal layer stable against water. The surface profile of the substrate with the grating after coating was checked by scanning probe microscope. The vertical wall had changed to a slope with an incline of about 25 degrees (Figure 2).
Three different substrates, i.e. a grating substrate with coating, a slide glass plate with coating, and a slide glass without coating, were used in the study. All the substrates were modified with various chemicals after preparation of the metal and SiO2 films. A 3-aminopropyltriethoxysilane (APTES, Sigma-Aldrich) aqueous solution (1 vol %) was poured onto the substrates, which were then kept at 40 °C for 1 hr. After reaction, the APTES solution was removed and the substrates rinsed by immersion in milli-Q water (18.2 MΩ) and ethanol and dried. For measurement of the substrates by angle-scan SPFS, poly(ethylene glycol)-α-biotin-ϖ-NHS ester [Biotin-PEG-NHS3400 (Nekter)] prepared in aqueous solution (2mM) was poured onto the surface of the substrates, which were rinsed with milli-Q water at 1hr after reaction between the NHS group and the amino group of the substrate surface.
2.3 SPR-SPFS measurement
A He-Ne laser beam of 632.8 nm wavelength passes an optical chopper (used also as the reference for the lock-in amplifier) and two polarizers for intensity and polarization control. Using a θ-2θ goniometer, the light reflected at the substrate at an incident angle (θ) of 5–30 degrees is monitored by photo diode (Figure 3). Here, the diameter of the laser spot was about 1 mm. The optical field, which is strongly enhanced at the surface plasmon resonance angle, can be used to excite fluorescence molecules attached to the surface-bound analyte. The emission is monitored by a photomultiplier (after passing through an appropriate lens, 10%-ND filter, and a narrow band interference filter, λ=670±5 nm) mounted on the goniometer unit at a fixed angle of 65 degrees.
Fig. 4. Optical arrangement of grating and incident light. When the grating vector is parallel to a plane including propagating p-polarized light, ψ is defined as 0 degrees.
A substrate mounted on a rotational plate (Fig. 4; rotational angle, ψ) was attached to our SPR-SPFS instrument. As shown in Fig. 4, the rotational angle, ψ, was defined as 0 degrees where the incident plane was parallel to the grating vector. All measurements were performed at room temperature (22–23 °C).
2.4 Fluorescence microscopy
For fluorescence microscopy, an upright microscope (BX51WI, Olympus, Tokyo, Japan) equipped with a halogen lamp (12V100WHAL-L, Olympus) and a 40× objective (NA 0.75) was used as shown in Fig. 5. The samples were observed using a Cy5 filter cube (excitation wavelength 630±15nm; emission wavelength 670±20 nm, Omega Optical, Brattleboro, VT, USA, dichroic mirror). Fluorescence microscopy images were obtained with a CCD camera (ORCA-ER, HAMAMATSU).
The illumination intensity (lamp power: 20µW with ND-25 filters), open-aperture diaphragms, exposure time of 1 second, and CCD camera gain were kept constant during the series of measurements. Fluorescence images were analyzed using the software NIH Image J. The samples observed by fluorescence microscopy were transfected cells modified with Cy5-labeled antibody to enhanced green fluorescent protein (Cy5-labeled anti-GFP antibody). A cover glass was attached to the top of all substrates and a 20µL-phosphate buffer solution or sample solution was injected into the substrate by pipette as in the SPR-SPFS measurement. The substrate was then mounted on the rotational stage and turned quarter circle. When the substrate was set as a grating vector parallel to the direction of s-polarized light at the dichroic mirror reflecting the incident light from the lamp, the position was called lateral, and when set perpendicular, the position was called longitudinal. All measurements were performed at room temperature (22–23 °C).
2.5 Samples
A labeled protein [streptavidin labeled with Cy5 (Amersham)] was prepared as a 1.6µM phosphate buffer saline solution for measurement of angle-scan GC-SPR and -SPFS. A cover glass was attached to the top of all substrates and a 20µL solution was injected into the substrate by pipette. After rinsing, the proteins were homogeneously bound to the surface of the substrates.
For measurement by fluorescence microscopy, cells expressing a protein tag recognized by fluorescently labeled antibody were prepared. Enhanced green fluorescent protein (GFP, Clontech, USA) was used as a cell-surface displaying protein tag. The GFP coding sequence was amplified by polymerase chain reaction (PCR) with a BglII restriction site and inserted into the restriction site of pDisplay (Invitrogen, USA). The plasmid was transfected to a COS cell (a cell line derived from kidney cells of the African green monkey) with lipofectamin 2000 (Invitrogen). After two days, the transfected cells were collected and fixed with 4% paraformaldehyde in phosphate buffer saline. The transfection efficiency, as checked by confocal microscopy (FV-1000, Olympus, Japan), was approximately 50%.
The anti-GFP antibody was modified with the fluorescent molecule Cy5 using a Cy5 labeling kit (Amersham). The ratio of modification of Cy5 to antibody was 2–3 mole/mole. After reaction, the antibody was passed through a column for refining. The transfected cells were mixed with labeled antibody and incubated and cells binding antibody were collected after rinsing by repeated rounds of centrifugation, discarding of supernatant, and resuspension.
3. Results
3.1 Grating-coupled surface plasmon resonance (GC-SPR)
Figure 6 shows the plot of reflection spectra against incident angle for Cy5-streptavidin bound to substrates in phosphate buffer. Three different substrates were used for comparison: (1) substrates coated with a silver film and SiO2 with grating of ψ=0, 45, and 90 degrees, (2) slide glass coated with silver and SiO2, and (3) slide glass without coating. For the plates in (1), SPR dips were clearly observed at θ=9 and 11 degrees with ψ of 0 and 45 degrees, respectively, but not with ψ of 90 degrees. The result at 90 degrees coincides with that for the slide glass with coating; reflectivity was constant at ca. 90% during measurement. The high reflectivity at 90 degrees was due to the metal layer without resonance to plasmon polaritons. For the slide glass, reflectivity was constant at below 10% and most of the light was able to pass through the substrate.
Fig. 6. Zero-order reflection spectra of Cy5-treptavidin bound to substrates in phosphate buffer: Substrates with grating of ψ=0 degrees (solid line), grating of ψ=45 degrees (dotted line), grating of ψ=90 degrees (broken line), slide glass covered (circles), and uncoated slide glass (triangles).
GC-SPR was found to be generated under the conditions applied here, and the GC-SPR field was found to be most enhanced when the plane propagating the light and the grating vector were parallel (i.e., ψ=0) as predicted from the polarization component of the incident light.
3.2 Fluorescence enhanced by GC-SPR
Figure 7 shows the fluorescence spectra for Cy5-streptavidin bound to substrates excited by a He-Ne laser. The interface was phosphate buffer solution. For grating plates with ψ of 0 and 45 degrees, the peaks of the surface plasmon field-enhanced fluorescence spectra were observed at the same angles as the individual SPR dips. The enhancement factor of fluorescence intensity was calculated by normalizing the maximum fluorescence intensity, with background noise subtracted, to the intensity on the slide glass without coating. Enhancement factor of fluorescence intensity for the 0 and 45 degree grating plates was 20 times and 16 times enhanced, respectively. The enhancement factor was found to depend on the reflectivity at the SPR dip, as suggested by Eq. (2). The enhancement factor for the coated glass was twice as large as for the uncoated plate, which is thought to be due to the mirror effect of the silver layer. The enhancement factor for the 90 degree grating plate was also twice as great as for the slide glass with the small incident angle. This means that the Plasmon polaritons were unable to couple to the grating mode at ψ of 90 degrees.
Fig. 7. SPFS spectra for Cy5-streptavidin bound to substrates in phosphate buffer; symbols as in Figure 5.
Fig. 8. Fluorescence microscope images of labeled cells on substrate: (a) lateral grating plate, (b) longitudinal grating plate, (c) plate with coating, and (d) uncoated slide glass. Bar corresponds to 10 µm.
3.3 GC-SPR enhanced fluorescence image observed by fluorescence microscopy
Transfected cells labeled with Cy5 were observed by fluorescence microscopy. Figure 8 (a)–(d) shows the fluorescence images detected on grating substrates set laterally and longitudinal, and slide glass with coating and without coating, respectively. The maximum fluorescence intensity with background subtracted was found to be 73.3, 64.6, 7.1, and 3.0 counts per unit pixel, respectively. Fluorescence enhanced by GC-SPR was also detected on the grating substrate by fluorescence microscope. The fluorescence intensity on the grating substrates positioned laterally [Figure 8 (a)] was found to be 24 times more enhanced than that on the slide glass without coating [Figure 8 (d)]. The intensity on the coated slide glass was enhanced approximately twice as much as that on the uncoated glass plate. These findings agree well with those of angle scan SPFS.
4. Discussion
In order to confirm that surface plasmon resonance was supported by grating coupling, the reflectivity on the grating was calculated by Rigorous Coupled Wave Analysis (RCWA) using the GSOLVER calculation software (Grating Solver Development Co.). The calculation conditions of the substrate were as shown in Figure 2, i.e. (1) SiO2 grating with 480-nm pitch, 31-nm depth, and vertical groove wall, (2) 200nm-thick Ag and 20nm-thick SiO2 layer coating, (3) groove with 25 degree incline on Ag/SiO2-coated surfaces, and interface with air. For the optical system, incident p-polarized light with wavelength of 632.8nm was used for the calculation. The reflection spectra obtained at various incident angles are shown in Fig. 9. The SPR dips calculated were found at 14 and 24 degrees at ψ of 0 and 45 degrees, respectively. The measurement curves depicted in Fig. 9 for ψ of 0 and 45 degrees showed dips at 13 and 26 degrees, respectively. The calculated curves are thus nearly coincident with the experimental results. Such an agreement between experimental result and theoretical calculation indicates the reliability of our conclusion.
Fig. 9. Results of RCWA calculation (solid lines) and measurement (broken lines) of reflectivity on grating of ψ=0 (with circles) and 45 degrees (with triangles). Interface was air.
The following discussion concerns angle-dependent fluorescence. The angle of the fluorescence peak in Fig. 7 was found to be in good agreement with the resonance angle of the GC-SPR dip shown inFig. 6. This result suggests that the fluorescence from Cy5-labeled protein was excited by GC-SPR. The enhancement of fluorescence on the grating substrate shown in Fig. 7 was thought to be due to the enhanced electric field of GC-SPR.
As found from Eq. (2), however, the maximum value of Ef will ideally be more than 300 under a silver grating at 633nm and it deviated widely from the enhancement factor of fluorescence intensity obtained from our experiment. The reason can be considered as: 1) The Ef is maximum value at a metal surface. In this study, the 20nm-thick SiO2 overlayer was prepared on the metal film because of suppression of a quench. FDTD calculation (not shown here) showed that the Ef at the place of 20 nm from a metal surface is expected to be below half of maximum. 2) In the theoretical equation, a surface profile of metal grating was assumed to be sinusoidal. The surface profile of our substrate was not completely sinusoidal. The grating depth and the surface profile including duty ratio of the groove and slope of the groove walls were not considered in Eq. (2), but, they can contribute to the enhancement of the GC-SPR field. Therefore, the most appropriate condition of grating pattern must be found out and for the grating fabricated under such a condition 100 times-enhanced fluorescence can be expected. The determination of the factors controlling fluorescence enhancement is an important subject which we are now addressing.
Fig. 10. Cross section of fluorescence images corresponding to Figure 8(a) (solid line) and (c) (dotted line).
Fluorescence image enhanced by GC-SPR was also detected on the grating substrate by a microscope. The sub-wavelength order of the grating roughness presented no obstacle to taking fluorescence images of labeled cells, as shown in Fig. 8(a) and (b). All the images depicted in Fig. 8 were processed with the same threshold set for adjusting the contrast enhancement as appropriate for the image in Fig. 8(a). In Fig. 8a) and (b), the intensity of the cell background was large. This is because the scattering light (excitation light) enhanced by GC-SPR was not completely blocked by the filter used. Figure 10 is a cross section of the raw data (before processing) taken by CCD corresponding to Fig. 8(a) and (c). The arrows marked A and B indicate the enhanced fluorescence from the cells and the excitation light enhanced by GC-SPR, respectively. The enhanced background can be eliminated by using a more suitable filter.
The fluorescence enhancement factor for the lateral grating under the microscope was in good agreement with the result of angle-scan SPFS at ψ=0. The fluorescence was also found to be enhanced in the longitudinal grating [Fig. 8b)]. With the microscope, ψ cannot be defined in the same way as in SPR-SPFS measurement and therefore fluorescence intensity is basically independent of the orientation of the substrate (lateral or longitudinal). However, when the substrate was set in the lateral position, the incident light from the objective lens was richer in p-polarized component (for the substrate) than in the longitudinal position. This is due to the polarized property of the reflectivity in the dichroic mirror of the filter used. A small difference was thus observed between lateral and longitudinal positioning.
In the present study, a grating of 480-nm pitch was used for sensitive fluorescence microscopy. At this pitch, the use of a light beam with wavelength of 630nm provided a small resonance angle less than 10 degrees. If a light beam of 480 nm able to excite GFP is used, the resonance angle might be near to 30 degrees and a large enhancement factor could therefore not be expected. The combination of the grating pitch and the wavelength of the excitation light is considered to be important for application of GC-SPR to microscopy.
5. Conclusions
On the substrate with sub-wavelength periodic structure which we fabricated, an enhanced fluorescence image of the cell was observed with a general fluorescence microscope, in which the fluorescence intensity was enhanced 24 times more than on a slide glass plate without coating. In the measurement of reflectivity, a dip corresponding to GC-SPR was found at a small incident angle of 9 degrees, and an enhanced fluorescence excited by the GC-SPR-field (GC-SPFS) was also found at the same angle. GC-SPFS is thus an invaluable method of applying highly sensitive fluorescence microscopy. The method is very simple because complicated optical systems such as prisms and high-NA objective lenses are not necessary. The determination of the optimal conditions for the sub-wavelength grating on the substrate, e.g., pitch, depth, surface profile, and wavelength of incident light, is now the main task, and is expected to lead to fluorescence microscope observation with larger fluorescence enhancement.
Acknowledgment
KT, KK, and JN express thanks for financial support by KAKENHI (Grant-in-Aid for Scientific Research) No. 19049016 on Priority Area “Strong Photons-Molecules Coupling Fields (No. 470)” from the Ministry of Education, Culture, Sports, Science and Technology of Japan. Part of this research was also supported by Grants-in-Aid for AIST Upbringing of Talent in Nanobiotechnology Course from the Ministry of Education, Culture, Sports, Science, and Technology, Japan.
References and links
1. H. Raether: Surface plasmons on smooth and rough surfaces and on gratings (Springer-Verlag, Heidelberg1988)
2. A. Otto and B.O. Seraphin ed.: Optical properties of solids, new development (North-Holland Publishing Company1976) p. 677
3. W. Knoll, “Interfaces and thin films as seen by bound electromagnetic waves,” Annu. Rev. Phys. Chem. 49, 569–638 (1998). [CrossRef]
4. Z. Z. Wang, T. Wilkop, and Q. Cheng, “Characterization of micropatterned lipid membranes on a gold surface by surface plasmon resonance imaging and electrochemical signaling of a pore-forming protein,” Langmuir 21, 10292–10296 (2005). [CrossRef] [PubMed]
5. C.E. Jordan, A. G. Frutos, A. J. Thiel, and R. M. Corn, “Surface plasmon resonance imaging measurements of DNA hybridization adsorption and streptavidin/DNA multilayer formation at chemically modified gold surfaces,” Analytical Chemistry 69, 4939–4947 (1997). [CrossRef]
6. T. Liebermann and W. Knoll, “Surface-plasmon field-enhanced fluorescence spectroscopy,” Colloids Surf. A 177, 115–130(2000). [CrossRef]
7. K. Tawa and W. Knoll, “Mismatching base-pair dependence of the kinetics of DNA-DNA hybridization studied by surface plasmon fluorescence spectroscopy” Nucleic Acids Research 32, 2372–2377 (2004). [CrossRef] [PubMed]
8. K. Tawa and K. Morigaki, “Substrate supported phospholipid membranes studied by SPR and surface plasmon fluorescence spectroscopy (SPFS),” Biophyical Journal 89, 2750–2758 (2005). [CrossRef]
9. N. P. Prasad: In nanophotonics (John Wiley & Sons, New York2004) p.129 [CrossRef]
10. S. Link and M. A. El-sayed, “Spectral properties and relaxation dynamics of surface plasmon electronic oscillations in gold and silver nanodots and nanorods,” J. Phys. Chem. B 103, 8410–8426 (1999). [CrossRef]
11. H. J. Lee, A. W. Wark, and R. M. Corn, “Creating advanced multifunctional biosensors with surface enzymatic transformations,” Langmuir 22 (2006) 5241–5250. [CrossRef] [PubMed]
12. E. Popov, L. Tsonev, and D. Maystre, “Lamellar metallic grating anomalies,” Applied Optics 33, 5214– 5219 (1994). [CrossRef] [PubMed]
13. E. Popov, N. Bonod, and S. Enoch, “Comparison of plasmon surface waves on shallow and deep metallic 1D and 2D gratings,” Optics Express 15, 4224–4237 (2007). [CrossRef] [PubMed]
14. S. Wedge and W. L. Barnes, “Surface plasmon-polariton mediated light emission through thin metal films,” Optics Express 12, 3673–3685 (2004). [CrossRef] [PubMed]
15. N. -F. Chiu, C. Yu, S.-Y. Nien, J.-H. Lee, C.-H. Kuan, K.-C. Wu, C.-K. Lee, and C.-W Lin “Enhancement and tunability of active plasmonic by multilayer grating coupled emission,” Optics Express 15, 11608– 11615 (2007). [CrossRef] [PubMed]
16. Y.-J. Hung, I. I. Smolyaninov, C. C. Davis, and H.-C. Wu, “Fluorescence enhancement by surface gratings,” Optics Express 14, 10825–10830 (2006). [CrossRef] [PubMed]
17. K. Kintaka, J. Nishii, A. Mizutani, H. Kikuta, and H. Nakano, “Antireflection microstructures fabricated upon fluorine-doped SiO2 films,” Opt. Lett. 26, 1642–1644 (2001). [CrossRef]
