1. Introduction

Plasmonic sensors based on LSPR allow rapid detection of biomolecular interactions in real time, which is particularly valuable for disease diagnosis and routine point-of-care (POC) clinical assessments [1–6]. The LSPR sensing follows the basic principle of the excitation of charge density oscillations (localized surface plasmons) propagating along the metal nanostructure/dielectric interface when the wave vector of incident light meets the resonance conditions [7,8]. The electromagnetic fields correlated with the oscillations above are highly sensitive to changes in the refractive index of the surroundings [9]. Compared with conventional LSPR sensors, optical fiber-based LSPR sensors have been of particular interest by the advantages of facile integration, small size, long-distance sensing and anti-electromagnetic interference [10–12]. Moreover, plastic optical fibers, compared with traditional silica fibers, are of high significance because of their high machinability, easy handling and low cost [13,14]. In addition, various fiber geometries have been proposed, including straight decladded [15], partially decladded [16], tapered [17,18], U-bent fiber [19], D-shaped [20], etc. To be specific, D-shaped optical fibers with large polished surfaces can more effectively preserve the integrity of the plasmonic structure and prevent them from being wrinkled. Furthermore, D-shaped optical fibers allow easy access to a large evanescent wave for efficient sensing, promoting the interaction with analyte and providing a flat detection plane during the detection process, which makes it attracted much attention [21].

LSPR optical fiber sensors are more significantly influenced by wrapping material, surface charge and inter-particle interactions than NPs size [22]. Recent progress in nanofabrication has encouraged the development of LSPR optical fiber sensor, particularly nanomaterials, which can break through the limitations of conventional plasmonic sensor. Monolayer of noble metal NPs as sensing structures was first applied in optical fibers sensing. Camara et al. fabricated a fiber-based LSPR sensor using Au NPs for dengue diagnosis [23]. Based on Au NPs, a tapered optical fiber sensor was developed by Lokendra for the detection of uric acid in human serum [24]. However, for conventional optical fiber LSPR sensors, biomolecules are poorly adsorbed on pure NPs surface, making the refractive index changes on the sensor surface cannot be fully identified, which limits the sensitivity of LSPR optical fiber sensors [25]. To solve this problem, graphene with molecular adsorption property is introduced. Moreover, graphene is a plasmon material with stable chemical properties that prevents oxidation of metal nanostructures, which can extend the lifespan of the sensors. Based on Gr/Ag NPs hybrid structure, Jiang et al. proposed an LSPR optical fiber sensor to detect ethanol and glucose concentration, achieving a sensitivity of 700.3 nm/RIU [26]. Nancy et.al designed hybrid graphene/gold plasmonic sensor to ssDNA detection [27]. Jeeban et al. developed a fiber-based LSPR sensor using GO/Ag NPs hybrid structure for biomolecule detection [28]. The sensing effect of the above hybrid structures arises primarily from the electromagnetic enhancement of metal NPs gap and the plasmonic resonance which is confined to the 2D plane. In contrast, bulk plasmon resonance based on 3D nanostructure has become an effective method to achieve better detection performance because the strong plasmonic hybridization response can be achieved both the parallel and vertically stacked plasmonic structure [29,30]. Combing multilayer Ag NPs with graphene oxide as spacer, Li et al. achieved SERS substrate exhibiting excellent detection capability, which provides a valid idea for the development of LSPR optical fiber sensors [31]. Based on Au NPs and multilayer graphene film 3D composite structure, Li et al. developed an LSPR optical fiber sensor that achieved 1251.44 nm/RIU [32]. Here, graphene, as a spacer for 3D hybrid structures, can transfer electrons to the Au NPs, thereby enhancing the plasmon excitations in the metal structure and further improved the strength of the local electromagnetic fields [33]. However, chemical synthesis methods make it difficult to form metal NPs with high uniformity and small gap over large areas. And optical absorption of graphene increases linearly with layers, limiting the performance of the LSPR optical fiber sensor. In order to improve the sensitivity and reproducibility of LSPR optical fiber sensors, new structures and fabrication methods are required to be developed.

In this paper, LSPR optical fiber sensor based on 3D Au NPs with Gr_m as spacer of two different stacking form: n*(Au NPs/Gr_m)/D-POF and n*(Gr_m/Au NPs)/D-POF (n-1-4) were developed using a facile two-step fabrication method. Both sensing structures improved the performance of the sensor since the multilayer hybrid structures excited stronger plasmonic hybridization response. Moreover, compared with n*(Au NPs/Gr_m), n*(Gr_m/Au NPs) excited stronger plasmonic hybridization coupling, results from the more upward-pointing tip created a stronger hot spot and promoted the coupling of plasmon excitations, which facilitated the generation of homogeneous plasmon hybridization. The performance of n*(Gr_m/Au NPs)/D-POF was evaluated applying the finite element method, the strongest 3D hot spot occurring when n = 3, which was consistent with our experimental results. The structure was also proven with a remarkably high sensitivity up to 2160 nm/RIU, with excellent linearity, reproducibility and response properties. Moreover, the proposed LSPR optical fiber sensor was successfully used to detect glucose concentrations showing excellent detection performance, which indicated that it has great potential in clinical medicine.

2. Materials and methods

2.1 Materials

Au target materials were purchased from Fuzhou Invention photoelectrical Tech Co., Ltd. Ethanol and Iron trichloride hexahydrate (FeCl₃) were provided by Sinopharm Chemical Reagent Co., Ltd. Glucose was offered by Tianjin Dingshengxin Chemical Industry Co., Ltd.

2.2 Methods

2.2.1 Fabrication of the n*(Gr_m/Au NPs)/D-POF sensor

The n*(Gr_m/AuNPs)/D-POF sensors were developed using a flexible two-step fabrication method. (Fig. 1). Firstly, Gr_m was grown on the copper foil by CVD method and then Au film with 8 nm thickness was deposited directly onto the Gr_m/copper foil substrate via thermal evaporation method (deposition rate was about 0.8 Å/s). Subsequently, the 1×2 cm² Au NPs/Gr_m/copper foil was placed in FeCl₃ solution (∼1M) to corrode copper foils [34]. Then the remaining Au NPs/Gr_m units were transferred to deionized (DI) water and washed three times to remove the etchant involved. Secondly, the AuNPs/Gr_m was inverted transferred to the fiber. On this basis, the two-layer structure was obtained by inverted transferring other Au NPs/Gr_m units to the previous one-layer plasmonic structure fiber. By constant repetition of the above transfer process, the multilayer plasmonic structure would be fabricated. To make the Au NPs/Gr_m layer contact closely with optical fiber, the probes were placed on the heating table at 40 ℃ for 10 min [21].

 

Fig. 1. Schematic of the preparation procedure of n*(Gr_m/Au NPs)/D-POF

Download Full Size | PDF

2.2.2 Experimental setup

Figure 2 illustrates an experimental setup for analyzing the performance of the designed n*(Gr_m/Au NPs)/D-POF LSPR sensor. A tungsten lamp (Ocean Optics HL-2000) was used as an excitation light source. An optical fiber spectrometer (Ideaoptics Instruments, PG2000) was connected to the computer to record the LSPR spectra. Besides, a dedicated microfluidic chip was used for analytes test, which was fixed to the D-POF sensor. The surface morphology of 3D nanostructures was characterized under a Zeiss Gemini Ultra-55 scanning electron microscope (SEM).

 

Fig. 2. Schematic of an experimental setup based on the n*(Gr_m/Au NPs)/D-POF sensor.

Download Full Size | PDF

3. Results and discussion

3.1 Theoretical analysis

Compared with conventional evanescent field detection, LSPR greatly enhances the energy density of the local electromagnetic fields on the surface of metal nanostructures, thus improving its sensitivity to the change in the refractive index of the surrounding medium. The finite element method was employed to demonstrate the different electromagnetic fields enhancement effects on the 3D plasmonic hybrid structure with different layers. The structure schematics of two types of LSPR optical fiber sensors were shown in Fig. 3(a) and 3(b). The diameter and nanogap of gold hemispheres were set to 60 nm and 10 nm respectively. Considering the conditions of excitation wave generation, we used P polarized light (TM mode light) with an incident angle of 80 degree as the incident light. According to Fig. 3(c)–3(j), the strong electromagnetic fields existed not only among the in-plane Au NPs but also on the sub-nanometer graphene gap regions, which was due to the vertical electron resonance and electron transforms facilitated by the monolayer graphene. Moreover, the intensity and density of the “hot spot” varied significantly with the number of layers of the 3D plasmonic hybrid structure. This was due to the more stacked plasmonic layers would stimulate more plasmonic couplings on the vertical plane, in which the multilayer plasmonic couplings could further enhance the intensity of the local electromagnetic fields. It was also revealed that there was a synergistic effect of the local electromagnetic fields that would induce strong plasmonic coupling along the vertical direction. Figure 3(k) showed the variation of the electric fields enhancement (E/E₀) as the plasmonic structure layers increasing. As indicated by this simulation curve, the intensity and density of local electromagnetic fields could be effectively manipulated by the number of plasmonic structure layers. Furthermore, the electric fields were gradually strengthened with the increase in the number of layers. However, once the number was higher than three, the electric field enhancement effect showed a progressively decreasing trend since the evanescent wave propagation distance was limited by the thickness of the number of plasmonic layers [20,35].

 

Fig. 3. (a) Schematics of n*(Gr_m/Au NPs)/D-POF and (b) n*(Au NPs/Gr_m)/D-POF (n = 3). (c)-(f) Electric-field distribution for the structure n*(Gr_m/Au NPs)/D-POF (n = 1-4). (g-j) Electric-field distribution for the structure n*(Au NPs/Gr_m)/D-POF (n = 1-4). (k) Electric-field enhancement (E/E₀) for the two types of 3D hybrid structure with different plasmonic layers.

Download Full Size | PDF

On the other hand, the local electromagnetic fields of n*(Gr_m/Au NPs) were stronger than n*(Au NPs/Gr_m) as illustrated in Fig. 3(c)-(j) and Fig. 3(k). Figure 3(k) showed that respective layer of n*(Gr_m/Au NPs)/D-POF had a more significant electric field enhancement variation (E/E₀) than n*(Au NPs/Gr_m)/D-POF, and the maximum value of electric field enhancement was achieved in 3*(Gr_m/Au NPs)/D-POF. As Fig. 3(c) and Fig. 3(g) shown, hot spots exist at both the curved end and the sharp tips of the hemisphere in n*(Gr_m/Au NPs), while hot spots exist only at the tip of the hemisphere in n*(Au NPs/Gr_m). Therefore, for n*(Gr_m/Au NPs) it is a multilayer strong electric field coupling, while n*(Au NPs/Gr_m) has a relatively weak electric field coupling effect. Besides, n*(Gr_m/Au NPs) transmits more electromagnetic energy to better enhance the electric field, n*(Au NPs/Gr_m) scattering makes the electric field relatively weak. Moreover, the maximum electromagnetic fields for both structures occurred at the tip of the first layer of nanoparticles closest to the fiber. For n*(Au NPs/Gr_m)/D-POF, the light firstly touched the tip forming strong hot spot, and the light intensity decreased with the increase in the distance. In the other words, the light was largely absorbed in the first layer NPs tips, thus resulting in the relatively weak resonance intensity. In contrast, with the tip on top, the light intensity had a relatively slow decrease, the aggregation of the tip with the NPs further strengthened the electromagnetic field. Compared with n*(Au NPs/Gr_m), n*(Gr_m/Au NPs) excited stronger plasmon hybridization response, due to the fact that the “hot spot” was closer to the upper surface which promoted stronger coupling of the hybridization response between layers. As a result, stronger electromagnetic fields were achieved. Besides, the multilayer hybrid structure of graphene coated Au NPs achieved the bigger specific surface areas compared with 2D graphene film decorated with Au NPs, which also made them a flat surface more suitable for biomolecule adsorption. Since the change in effective refractive index depended on the number of molecules adsorbed on the nanostructure, the uppermost layer of graphene could adsorb more molecules to facilitate the change in effective refractive index thus increasing its sensitivity. For 3*(Au NPs/Gr_m)/D-POF, the biomolecules adsorption was limited on the surface of the Au NPs, thereby causing low sensitivity. Accordingly, to improve the performance of sensors, biocompatible 2D materials can be used because of its plasmonic property and the ability to provide a flat surface for the adsorption of molecules.

3.2 Characterization of the n*(Gr_m/Au NPs)/D-POF LSPR sensor

The obtained n*(Gr_m/Au NPs)/POF (n = 1-4) with sensing length of 1.5 cm was exhibited in Fig. 4(a). The morphologies of the 1-4 layers Gr_m/Au NPs hybrid nanostructures were demonstrated in Fig. 4(b)-(e). Figure 4(c)-(e) clearly show layered AuNPs, and the homogeneity and affinity of the plasmonic structure ensured high coverage of the fiber surface, which enhanced the plasmonic coupling in the vertical direction. Besides, monolayer graphene can be observed obviously in Fig. 4(d). In addition, the quality and thickness of the monolayer graphene were measured using Raman spectroscopy (Fig. 4(f)). The main Raman peak of graphene included D band (∼1350 cm^-1), G band (∼1580 cm^-1), and 2D band (∼2670 cm^-1). The I_2D/I_G in the spectrum was nearly 2.25, thus indicating a typical spectrum for monolayer graphene. Moreover, the insert was the Raman mapping for 2D peak, which indicted high uniformity of the monolayer graphene. The nucleation, growth and distribution of AuNPs were impacted by the surface diffusion coefficient, which were correlated with the number of layers of graphene [36]. Based on homogeneous monolayer graphene films, PVD method was used to efficiently form sufficiently small and uniform gaps among Au NPs. Furthermore, the aggregation of Au NPs was avoided, which facilitated the effective penetration of the evanescent wave. Moreover, monolayer graphene, as a spacer, avoided the vertical Au NPs contacting, which produced almost optimal electromagnetic fields enhancement effect. Lastly, since the active metal was buried within the graphene without being exposed to the atmosphere, it was expected the n*(Gr_m/Au NPs)/POF could have a longer lifespan.

 

Fig. 4. (a) Photo of the sensors with n*(Gr_m/Au NPs)/D-POF. (b)-(e) SEM images of n*(Gr_m/Au NPs)/D-POF (n = 1-4). (f) Raman spectrum of monolayer graphene. Insert is Raman mapping at 2D peak with an area of 20×20 µm².

Download Full Size | PDF

3.3. Sensing performance measurement

3.3.1 Performance of the n*(Gr_m/Au NPs)/D-POF

To measure the performance of n*(Gr_m/Au NPs)/D-POF (n = 1-4) LSPR sensor, the normalized transmission spectra were collected and then compared in the ethanol solutions with RI ranging from 1.340 to 1.352, as presented in Fig. 5(a)–5(d). In general, the experimental results were generally consistent with our predicted results. Moreover, the variation of red shift and resonance wavelength were summarized in Fig. 5(e). The redshift increased from 5.97 nm to 25.92 nm and the resonance wavelength shifted from 494.16 nm to 685.13 nm. To be specific, the 3*(Gr_m/Au NPs)/D-POF had the most significant red shift corresponding an extremely high sensitivity up to 2160 nm/RIU. In addition, figure of merit (FOM) is another main parameter that describe the sensing performance of an optical sensor. The FOM is defined as FOM = Sensitivity/FWHM, where FWHM is the full width at half maximum [37,38]. And the 3*(Gr_m/Au NPs)/D-POF obtained a FOM of 17.28. Notably, the red shift increased obviously with the plasmonic layers stacking, and this could be attributed to the enhancing multiple plasmonic couplings on the vertical plane. However, when stacking more than three layers, the red shift slightly decreased, due to the high thickness of the 3D overall nanostructure that made the evanescent wave difficult to pass through. Besides, the SPR dips decreased with the increase in the alcohol refractive index, which could be attributed to the larger penetration depth and the higher energy consumption of evanescent field introduced by the increase of the refractive index [39]. The results above indicated excellent sensitivity of the 3*(Gr_m/Au NPs)/D-POF, which could be attributed to the following factors below. Firstly, periodic stacking Gr_m/Au NPs plasmonic structure produced strong plasmonic hybridization response in vertical direction, which effectively improved the LSPR effect; Secondly, graphene film could provide a flat surface for testing and effectively promote the molecular adsorption ability, which would significantly contribute to the sensitivity. Importantly, the evanescent wave could penetrate the numerous stacking layers for the atomically thin monolayer graphene film and 8 nm Au film without sharp attenuation. The dip resonance depth represents the resonance intensity [40]. As shown in Fig. 5(f), the dip resonance depth reaches a maximum at n = 3, due to the fact that the electric field also reaches a maximum resulting in a strong plasmonic hybridization response. But the plasmonic structure of one-layer is too thin to produce a good resonance response [20].

 

Fig. 5. (a-d) Transmission spectra of the n*(Gr_m/Au NPs)/D-POF (n = 1-4) in the ethanol solution. (e) The red shift and the resonance wavelength of n*(Gr_m/Au NPs)/D-POF, respectively. (f) Summary of the DRD for various sensor. (g) The red shift of n*(Gr_m/Au NPs)/D-POF as a function of RI. (h) Transmission spectra of 3*(Gr_m/Au NPs)/D-POF at the RI of 1.352 during seven cycles. (i) Typical response-recovery characteristic curves of 3*(Gr_m/Au NPs)/D-POF in ethanol solution with an RI of 1.352

Download Full Size | PDF

Linearity, reproducibility and response-recovery ability are also key indicators to evaluate the performance of sensors. The linearity relationship between the refractive index and the red shift was shown in Fig. 5(g). This figure clearly shown that the fitting coefficient were 0.993, 0.987, 0.996 and 0.986, corresponding to different layers, respectively. The fitting coefficient reached the maximum when the layers were equal to three, which indicated that the 3*(Gr_m/Au NPs)/D-POF had an excellent linear detection ability. To study the reproducibility of 3*(Gr_m/Au NPs)/D-POF, the transmission spectra of seven cycles were tested under the same conditions at a refractive index of 1.352. According to Fig. 5(h), the minimum value of transmission was close to a constant, which showed the outstanding reproducibility of 3*(Gr_m/Au NPs)/D-POF. Similarly, the response-recovery curve of the sensor with the refractive index of 1.352 at resonance wavelength of 603.84 nm was tested, and the transmission decreased to 0.86 in 2.5 s. In brief, the mentioned characteristics proved the excellent performance of the 3*(Gr_m/Au NPs)/D-POF LSPR sensor, which provides a strong guarantee for the detection of the surface refractive index.

3.3.2 Performance of the n*(Au NPs/Gr_m)/D-POF

Correspondingly, the performance of the n*(Au NPs/Gr_m)/D-POF was also tested as the steps mentioned above. The change in red shift and resonance wavelength with the number of plasmonic layers was illustrated in Fig. 6(a)-(d) and summarized in Fig. 6(e). The resonance peak shifted from 480.81 nm to 640.52 nm, and the red shift increased from 5.99 nm to 15.91 nm. And the maximum red shift occurred in 3*(AuNPs/Gr_m), with a relatively low refractive index sensitivity of 1317.61 nm/RIU. This is due to the relatively weak layer-to-layer plasmonic hybridization coupling generated by the plasmonic structure of multilayered AuNPs/Gr_m, which limited its sensitivity. In addition, the linearity, reproducibility and response-recovery characteristics were also investigated. According to Fig. 6(f), the values of R² were 0.982, 0.992, 0.997, 0.984 when n changed from 1 to 4, and 3*(Au NPs/Gr_m)/D-POF had a high linear fit coefficient R²= 0.997. As illustrated in Fig. 6(g)–6(i), the reproducibility and response-recovery characteristics of 3*(Au NPs/Gr_m)/D-POF were also tested, both with excellent performance. In addition, the linear response curve had the almost same response time corresponding to 3*(Gr_m/Au NPs)/D-POF, but took more than triple time to recover to the original level.

 

Fig. 6. (a-d) Transmission spectra of the n*(Au NPs/Gr_m)/D-POF (n = 1-4). (e) The red shift and the resonance wavelength of n*(Au NPs/Gr_m)/D-POF. (f) The red shift of n*(Au NPs/Gr_m)/D-POF as a function of RI, respectively. (g) Transmission spectra at the RI of 1.352 during seven cycles, based on 3*(Au NPs/Gr_m)/D-POF. (h) Typical response-recovery characteristic curves of 3*(Au NPs/Gr_m)/D-POF. (i) Dynamic absorbance transmission response of 3*(Au NPs/Gr_m)/D-POF for 10 cycles.

Download Full Size | PDF

Table 1 summarizes the sensitivity of the SPR optical fiber sensors from the previous reports. The SPP excited under the special evanescent mechanism can significantly improve the performance of the sensors. The table indicates that the multilayer plasmonic structure had a higher sensitivity, which could be explained by the fact that the multilayer structure was able to couple plasmon response of each layer significantly increasing the energy density of the electromagnetic fields compared to the single layer plasmonic structure. In other words, the more stacked plasmonic layers would stimulate more plasmonic couplings on the vertical plane, where the multiple plasmonic couplings could further enhance the intensity of the local electromagnetic fields. However, the limited penetration depth of evanescent waves in multilayered nanostructures should be taken into account. In this work, combining atomically thin monolayer graphene film and 8 nm thickness Au NPs as plasmonic layer is extremely important for the increase of the stacked layers that the evanescent wave could penetrate.

Table 1. Comparison of D-shaped plastic optical fiber SPR sensor

View Table

3.3.3 Application of the n*(Gr_m/Au NPs)/D-POF

Assessment of glucose level is essential for the diagnosis and treatment of diseases that can lead to complications such as hypertension, heart disease and kidney failure. In comparison with electrochemistry, fluorescence, colorimetry and other detection techniques, LSPR-based sensing technology is a new, simple and cost-effective strategy for sensing changes in local refractive index before and after exposure to the analyte to detect glucose concentrations [18,43]. Herein, we have demonstrated the applications of the 3*(Gr_m/Au NPs)/D-POF as a signal amplification tag for highly sensitive recognition of glucose concentration. As shown in Fig. 7(a), with the increase in the glucose concentration from 2.5% to 20%, the corresponding resonance peak changed from 562.87 nm to 580.19 nm and the largest red shift was 20.95 nm, which indicated that our proposed LSPR optical fiber sensor had prominent detection capability and could be adapted successfully to the detection the concentration of glucose. Furthermore, the linearity of the glucose wavelength versus refractive index is exhibited in Fig. 7(b), where excellent linearity was strongly demonstrated with a linear fit coefficient of 0.993.

 

Fig. 7. (a) Normalized transmission spectra of 3*(Gr_m/Au NPs)/D-POF LSPR sensor with the aqueous glucose RIs from 1.3398 to 1.3557. (b) Resonance wavelength of the sensor as a function of aqueous glucose RIs.

Download Full Size | PDF

4. Conclusion

In summary, we have successfully fabricated the n*(Gr_m/Au NPs)/D-POF sensors using an effective and dexterous method. The proposed multilayer 3D hybrid structure can generate strong 3D hot spots, which contributes to achieve a higher sensitivity in the detection process. We also demonstrated experimentally that the LSPR resonance wavelength can be tuned as the increase of plasmonic layers in the visible wavelength. Moreover, the structure with 3*(Gr_m/Au NPs)/D-POF was found to be optimal in this paper, which had a high sensitivity (2160 nm/RIU), with excellent linearity (R²= 0.996), reproducibility and response-speed character (2.5 s). This study suggested that 3D bulk resonance was a practical option to improve the performance of the LSPR sensor. In applications, the LSPR sensor has been used to detect glucose concentrations and proven to have prominent detection performance which suggested 3*(Gr_m/Au NPs)/D-POF LSPR sensor has a promising prospect in medical diagnostics and clinical testing.