Fluorescence manipulation from graphene and hydrogen-terminated graphene covered Au nanoparticles

A great deal of fundamental investigation and technological innovation have arisen based on the Purcell effect, which stated that fluorescence emission could be altered by resonant coupling to the external electromagnetic (EM) environment.¹ Enhanced and suppressed fluorescence emission have been observed from luminescent materials coupled to various entities possessing a strong EM field, such as optical cavities,^2,3 photonic crystals,^4,5 and metal nanostructures.^6–10 Numerous devices based on altered fluorescence emission have been demonstrated, including surface plasmon enhanced LEDs,^11,12 single-photon sources,¹³ and bio-sensors and chemical sensors.^7,14 Enhancement of fluorescence is of great interest because it can provide a higher detection sensitivity and signal-to-noise ratio for fluorescence analysis and imaging.^7,15,16 On the other hand, conditional quenching of fluorescence can be effectively applied for other forms of sensing, such as selective quenching or negative sensing.¹⁴ In this sense, the ability of artificially enhancing or quenching the fluorescence signals is greatly beneficial for specific fluorescence-based applications.

In the above-mentioned plasmonic entities, metal nanostructures [especially nanoparticles (NPs)] are the most commonly used plasmonic substrates due to their strongly enhanced optical properties through localized surface plasmon resonance (LSPR).¹⁷ The interparticle coupling of LSPR would result in generation of hot-spots with greatly enhanced local EM fields, producing systems that can be applied in metal-enhanced fluorescence (MEF). According to the well-known MEF mechanism, the nature and level of change in the fluorescence depend upon the field strength, while the field strength critically depends on the metal type,¹⁸ the separation distance,¹⁹ the distance between the metal surface and fluorophore,²⁰ and the size and morphology of metal nanostructures.^7,21 Experiments performed by controlling the above-stated factors did show an MEF enhancement factor (EF) from 2–3 to a scale of several hundreds, as typically summarized in Table S1 (see the supplementary material).^{7,9,18,19,21–26} Reviews of the experimental activity can be found in Refs. 27–29.

Graphene is chemically inert and stable single atom thick and has proven to be the best candidate for hybridizing with metal nanostructures.³⁰ In such a metal–graphene hybrid system, graphene can be considered the thinnest protective film for metal plasmonic substrates, especially for Ag-, Cu-, and Al-NPs, as these are easily oxidized in air (impose negative effects on fluorescence signals).³¹ Besides, there exist strong interactions between graphene and metal nanostructures through a charge transfer mechanism or refractive index effect. Specifically, graphene can tune LSPR spectral properties, i.e., the resonance frequency and quality factor.^30,32 On the other hand, the light absorption efficiency and Raman spectrum of graphene themselves would also be modified by metal nanostructures.^33,34 Applications based on these effects have been successfully developed, including photodetectors, photovoltaic devices, SERS, and plasmon sensors.^31,35–37 However, the influence of graphene materials on the fluorescence emission in an MEF system has not been investigated. In particular, it should be reminded that graphene shows an alterable EM shielding effect due to its tunable electrical conductivity via defect engineering,³⁸ which is expected to bring a different impact on the local EM fields around metal nanostructures. In addition, graphene itself has proven to be an efficient quencher of fluorescence, and its quenching efficiency can also be easily controlled by defect engineering;³⁹ it is thereby possible to extend the range of fluorescence manipulation in a metal–graphene hybrid system.

For this purpose, here, we employ pristine graphene and hydrogen-terminated graphene to modify the surface of Au-NPs deposited on top of quartz substrates and successfully manipulate the fluorescence of Rhodamine-6G (R6G) from enhancement (∼fourfold) to quenching (∼7.6-fold). To investigate the underlying mechanisms, controlled experiments and simulations of the local EM fields are carried out. The enhancement of fluorescence on bare Au-NPs is due to plasmon resonance energy transfer from Au-NPs to nearby R6G molecules. In Au-graphene hybrid nanostructures, the overall fluorescence levels are possibly ascribed to the combined effect of the (i) local field enhancement by Au-NPs and (ii) quenching effect by graphene. This work may be helpful in understanding the effect of graphene materials on the local EM fields generated at the edges and corners of metal NPs and provide a simple pathway for fluorescence manipulation.

Au-NPs were obtained by vacuum thermal evaporation (Technol ZHD-400 system) of a 5 nm gold film on a quartz substrate, followed by annealing at 500 °C for 30 min in a vacuum tube furnace with rapid cooling. Graphene was synthesized on copper foils (a thickness of 25 µm) by chemical vapor deposition (CVD) at 1045 °C with a mixture flow of H₂ (60 sccm) and CH₄ (60 sccm). The growth pressure was 300 Pa, and the growth time was 10 min. Hydrogenation of graphene was performed by a planar low frequency (13.56 MHz, 15 W, 20 Pa) inductively-coupled plasma source, and the hydrogenation degree was tuned by varying the plasma irradiation time. Graphene covered Au-NPs on quartz substrates were achieved by transferring graphene on Au-NPs using a standard PMMA transfer procedure.⁴⁰ After that, all as-prepared substrates were soaked in R6G solution (with a very low concentration of ∼10⁻⁷ M in water to avoid aggregation) for about 30 min. De-ionized water was used to flush away unadsorbed molecules. The samples were then dried in air. The device for resistance (R_s) measurements was fabricated using UV lithography, followed by O₂ plasma etching. Finally, metal contacts (Ni/Au, 5 nm/50 nm) were thermally evaporated.

Scanning electron microscopy (SEM, FEI Inspect F50) was used to investigate the morphology of Au-NPs with and without coverage of graphene film. Atomic force microscopy (Asylum Research MFP-3D AFM) was employed to investigate the height profile. The optical absorption was measured by a Hitachi UV-visible spectrophotometer. Fluorescence and Raman spectrum measurements were carried out using a JY HR800 micro-Raman system. The excitation wavelength was 514 nm (100× objective, NA = 0.9). Fluorescence intensity mapping was carried out with a Witec Alpha 300R confocal Raman microscope. A laser power of ∼0.02 mW was chosen to avoid laser-induced heating as well as photobleaching of R6G. Time-resolved fluorescence spectra were conducted on a Fluorolog3-TCSPC spectrofluorometer equipped with a xenon lamp and a NanoLED-460LH (pulse duration: 750 ps). All the fluorescence spectra were measured under the same conditions for valid comparisons. Resistance measurements were measured using a Keithley 2612 semiconductor analyzer.

The optical properties of both Au-NPs and the local EM fields were calculated using the finite difference time domain (FDTD) software method with the periodic boundary conditions in plane. The mesh (x × y × z) size was set to 0.25 nm × 1 nm × 0.25 nm for time-saving under reliable accuracy. The dielectric constant function of Au-NPs was adopted from Ref. 41. The relative dielectric constants of pristine graphene and H-terminated graphene were expressed as $εr=1+σdωε0$ after ignoring the imagine part of conductivity (σ), where d = 0.35 nm is the thickness of pristine graphene or H-terminated graphene, ω is the interested frequency, and ɛ₀ is the permittivity of vacuum. The conductivity σ = 1/(R_sd) was calculated to be ∼2.8 × 10⁶ S m⁻¹ and ∼2.9 × 10⁴ S m⁻¹ for pristine graphene and H-terminated graphene, respectively, which were extracted from the device measurements.

In order to assure the sample preparation appropriately, SEM and AFM micrographs of the prepared samples were captured and are illustrated in Fig. 1. A distinct morphology of Au nanostructures with relatively uniform size and distribution was identified by SEM and AFM. Figure 1(a) shows the typical large scale and small scale (inset) SEM images of Au-NPs prepared by 5 nm thick gold film deposition. The size distribution of Au-NPs on the quartz substrate from AFM results is shown in the inset of Fig. 1(b). The average diameter of the prepared Au-NPs is ∼30 nm. As mentioned before, a standard PMMA transfer procedure for graphene was performed to achieve graphene/Au-NP hybrid nanostructures. Figure 1(c) shows a typical SEM micrograph of a hybrid structure of Au-NPs covered with a single-layer CVD-grown graphene film, indicating the clean surface and the mostly wrinkle-free transferred graphene. In addition, the morphology of the underlying Au-NPs has no obvious change, even if soaked in solutions during the removal of PMMA. Subsequently, we performed optical absorbance measurements for bare Au-NPs and Au-NPs covered with pristine graphene film and H-terminated graphene film on top of quartz substrates (see Fig. S1 in the supplementary material). The absorption spectrum of bare Au-NPs shows one peak centered at ∼545 nm originating from LSPR. With the coverage of pristine graphene and H-terminated graphene, the absorption peak shows a redshift, which can be explained by a change in the physical environment of Au-NPs.³⁰ Figure 1(d) demonstrates the optical absorption spectrum calculated by the FDTD method for 30 nm Au-NPs with an interparticle separation of ∼2 nm, which is in good agreement with the experimental results. The blueshift and redshift of the absorption peak are also obtained (not shown here) when reducing and increasing the interparticle separation, respectively. The full width at half maximum of the peak from the experimental results is much larger than that from calculation, indicating that the dominating contribution of light absorption is from 30 nm Au-NPs with an interparticle separation of ∼2 nm. Importantly, the emission spectrum of R6G significantly overlaps with the absorption spectrum of Au-NPs, thus guaranteeing efficient energy transfer between the Au-NPs and the vicinal R6G molecules and finally enhancing the fluorescence.

Hydrogen termination of graphene was carried out in a plasma reactor, aiming to achieve high resistivity of graphene. To monitor the alteration in the electrical property, a pristine graphene device on a 285 nm SiO₂/Si substrate [the SEM image is presented in the inset of Fig. 2(a)] was fabricated for H⁺ plasma treatment. The electrical resistance of graphene as a function of hydrogen plasma irradiation time is given in Fig. 2(a). The electrical resistance was monitored after each plasma treatment. Initially, the resistance increases gradually with the irradiation time up to ∼16 min, and after that, the increase accelerated at a faster pace. For short irradiation time, hydrogen is rarely distributed on the graphene surface, which impedes the electrical conduction slightly. With a further increase in irradiation time, a higher percentage of the graphene surface is terminated by hydrogen atoms, and conductible electrons in the π band decrease sharply, leading to a rapid increase in electrical resistance. Notice that more defects are probably induced with the irradiation and also restrain the electrical conduction by scattering. It has been observed that graphene is expected to become insulating from a semi-metallic nature if its surfaces are fully terminated by hydrogen atoms.³⁸ In our case, after subjected to plasma hydrogenation for more than 40 min, the electrical resistance sharply increases to ∼110 k$Ω$⁠, which is about 100 times larger than that of graphene film before hydrogenation. Figure 2(b) presents the Raman spectra excited by 514 nm light for pristine graphene and H-terminated graphene. The characteristic graphene peaks such as the G peak (∼1580 cm⁻¹) and 2D peak (∼2680 cm⁻¹) are shown in these two spectra. The intensity ratio of 2D/G in the spectrum for pristine graphene is about 2, signifying a single-layer of graphene. Two new peaks at ∼1340 cm⁻¹ and ∼1620 cm⁻¹ are noticeably observable in the spectrum of H-terminated graphene. The strong peak at ∼1340 cm⁻¹ is assigned to the D peak, which is attributed to the breaking of the translational symmetry of the sp² C–C bond by hydrogenation. The peak at ∼1620 cm⁻¹ is termed the D′ peak, which results due to the intravalley double resonance process that only occurs in the presence of defects.³⁰ In addition, slightly broadened and decreased intensities of both G and 2D peaks for H-terminated graphene are observed. All these features indicate that the electronic structure of pristine graphene has been modified by defect engineering via hydrogen plasma treatment and agree well with the resistance measurements. Notably, graphene film of the same batch before and after hydrogen termination was used to cover the Au-NP surface on top of quartz substrates for fluorescence measurements.

Fluorescence signals of R6G on four different substrates were measured under the same conditions with an excitation laser of 514 nm. The left-hand panel of Fig. 3(a) schematically displays R6G molecules on (i) bare quartz, (ii) Au-NPs/quartz, (iii) Gr/Au-NPs/quartz, and (iv) H-Gr/Au-NPs/quartz. The corresponding mappings of fluorescence intensity measured in arbitrarily selected areas (5 × 2.5 µm²) are demonstrated in the right-hand panel of Fig. 3(a). Compared to the bare quartz substrate, the observed fluorescence intensities of R6G on the other substrates exhibit a range of response from enhancement to quenching. The comparison with a clear contrast can also be verified from the fluorescence spectra, as shown in Fig. 3(b). The degree of enhancement is evaluated by the ratio of integrated fluorescence intensity to that on the bare quartz substrate. Notice that the quenching factor is the inverse of the enhancement factor. Data shown in the inset of (b) are the relative fluorescence intensities of each sample collected from 10 different locations. It can be seen that the fluorescence signals are significantly enhanced on Au-NPs/quartz with an enhancement factor of ∼4 times, and the spectrum peak is slightly red-shifted by ∼5 nm as compared to that on bare quartz while for the H-Gr/Au-NP/quartz substrate, the enhancement decreases to ∼1.8 times. Contrary to both the above-mentioned cases, the relative fluorescence signals are strongly quenched by a factor of ∼7.6 times on the Gr/quartz substrate. These results demonstrated that the fluorescence signal of R6G can be effectively manipulated by these pre-designed substrates from enhancement to quenching.

As reported previously, the phenomenon of MEF is attributed to two origins.^8,19,42 The first is surface plasmons excited by the light transferring part of its energy to the fluorophore (R6G), and the fluorophore can then radiate, resulting in an increase in fluorescence intensity. The second is the strong coupling between the excited state of the fluorophore and plasmons. The excited fluorophore can interact with surface plasmons, converting part of the fluorophore’s non-radiative near-field emission to be radiated by surface plasmons as far-field emission (observed enhanced fluorescence). For the latter, the lifetime of the fluorophore is generally expected to be reduced due to a strong interaction between the surface plasmon and the excited state of the fluorophore.⁴² To verify this, we performed time-resolved fluorescence spectra of R6G on bare quartz and Au-NP/quartz substrates (see Fig. S2 in the supplementary material). Notably, almost no change in lifetime of R6G was found in the presence of Au-NPs. Therefore, we conclude that plasmon resonance energy transfer is predominantly responsible for the enhancement of fluorescence emission.

To gain a deep understanding of the fluorescence manipulation based on Gr-Au-NP/quartz and H-Gr/Au-NP/quartz substrates, the FDTD method is adopted to calculate the local EM fields. The structure of Au-NPs was approximated by hemispheres using dimensional parameters equal to the mean size (30 nm) of the samples produced experimentally. The light with a wavelength of 514 nm irradiates the sample in the normal direction. Figure 4 shows the cross-sectional view of the simulated EM field $E2$ distribution maps around bare Au-NPs for the Au-NP/quartz, Gr/Au-NP/quartz, and H-Gr/Au-NP/quartz substrates. Clearly, the maximum EM field is located at the gap between the adjacent hemispheres. These so-called hot-spots generated by the localized surface plasmon resonance are responsible for fluorescence enhancement. Evidently, it is seen that the EM field strength decreases as Au-NPs are covered with pristine graphene film [Fig. 4(b)]. However, EM field strength is recovered with hydrogenation of graphene [Fig. 4(c)]. This is believed to be originated from the difference in electrical conductivity of pristine graphene and H-terminated graphene films covered on Au-NPs. Specifically, the EM field is greatly weakened outside the pristine graphene film due to the electrostatic shielding effect of conductive graphene itself while almost completely passing through the high resistance H-terminated graphene film.³¹ However, the observed fluorescence levels of R6G on these two substrates do not seem to be well consistent with the simulation results. The fluorescence quenching effect of pristine graphene or H-terminated graphene film itself must be taken into account in these hybrid systems. To clarify this, two controlled fluorescence measurements from R6G on Gr/quartz and H-Gr/quartz substrates were performed (see Fig. S3 in the supplementary material). It was found that the fluorescence intensities are, respectively, quenched off ∼10 and ∼2 times on pristine graphene and H-terminated graphene as compared to that of the pure quartz substrate. Therefore, taken together, the above-mentioned results and analyses indicate that resulting fluorescence intensities can be inferred to arise from the combined effect of near-field enhancement from Au-NPs and the quenching effect from graphene.

In summary, we manipulated the fluorescence emission of R6G by three different substrates including bare Au-NPs/quartz, Gr/Au-NPs/quartz, and H-Gr/Au-NPs/quartz. Au-NPs were fabricated by vacuum thermal evaporation. The observed fluorescence intensity of R6G exhibited a wide range of response from enhancement to quenching. A maximum of ∼fourfold increase was found on bare Au-NPs, with the enhancement decreasing to ∼1.8-fold by H-terminated graphene film covered Au-NPs, and finally, the fluorescence was significantly quenched off ∼7.6-fold by pristine graphene film covered Au-NPs. The physical mechanisms associated with fluorescence enhancement and quenching have been discussed in detail based on the controlled experimental and FDTD simulation results. Regarding Au-NPs on the quartz substrate, the enhancement can be attributed to the energy transfer through the overlap of the LSPR band and the emission spectrum of R6G. For the hydrogenated substrate, the high resistivity of graphene after subjecting it to plasma hydrogenation allows the strong local EM fields among Au-NPs to penetrate the graphene, thus enhancing the fluorescence of R6G adsorbed on the film. In contrast, local EM fields on and parallel to the pristine graphene surface are greatly weakened by its conductive surface, thereby lowering the efficiency of plasmon resonance energy transfer from Au-NPs to R6G. Meanwhile, the fluorescence quenching effect of graphene itself has been taken into account in this hybrid system. Therefore, we concluded that the overall fluorescence levels in metal–graphene nanostructures are attributable to the collective effect of EM field enhancement by Au-NPs and quenching by graphene. This work reveals the effect of graphene with different conductivities on the local EM fields, providing a simple approach to artificially manipulate fluorescence for specific molecular sensing, detecting, and imaging.

Details about the optical absorbance measurements and controlled experiments are provided in the supplementary material.

The data that support the findings of this study are available within the article.

This work was supported by the National Natural Science Foundation of China (NSFC) (Grant No. 61704092), the Natural Science Foundation of Jiangxi Province (Grant No. 20202BABL212012), the Natural Science Basic Research Program of Shaanxi (Program No. 2019JQ-059), and the Ph.D. Early Development Program of East China University of Technology (Grant No. DHBK2018077).