Optoplasmonic nanodevices are the key components to bridge the gap of integrated electronic circuits and optical chips for ultrafast optical information storage and processing [1]. One of the key aspects for their functionality is the facile and fast switchability of plasmon resonances to accommodate the demands of ultrasmall size and high operating speed [2], which, however, is nontrivial for conventional strategies.

One of the key obstacles is the compatibility issue of the current tunable plasmonic systems with the existing optoelectronic systems, which rules out most of the plasmon tuning systems based on wet (electro)chemistry [3–10]. An electric tunable plasmonic system appears more compatible but it adds up the complication of circuit designs and mostly operates in the infrared (IR) and terahertz regions [11,12]. An optically tunable plasmonic system is much superior in terms of simplicity, modulation speed, and ultrasmall sizes, making it a promising strategy towards the photonic chips. Although several optically addressable plasmonic systems have been proposed, the tunability of plasmons via this approach appears either irreversible [13–15] or in the infrared (IR)/terahertz regions [16–18]. For some systems, it has to be implemented in a liquid environment [19–22], which makes them far from practical applications.

Phase change materials such as germanium antimony telluride (GST) and vanadium dioxide (VO2) are commonly applied for nonvolatile memory devices [23,24]. However, due to their high phase transition temperature (∼250°C for GST, ∼80°C for VO2), it is a great challenge to integrate them into optoplasmonic nanosystems as such high local temperature may deform the plasmonic nanocomponents and surrounding dielectrics after cycles of operation [25,26], making their switchability vulnerable. Thus, most of their applications are limited to metasurfaces for propagating surface plasmons [27,28], whereas modulation of gap plasmons is rare [29,30].

In this paper, we use vanadyl oxalate hydrates (VOC2O4·xH2O) films as the spacer in Au NP on mirror (NPoM) structure, which are capped by titanium dioxide (TiO2) to maintain its stability in air [Fig. 1(a)]. The Au NPoM structure presents a dipolar mode of coupled plasmons due to its image charge [Fig. 1(b)] [31], which results in a coupled plasmon resonance in the scattering spectra [Fig. 1(c)]. Since this peak is very sensitive to the refractive index (RI) of the gap medium and the size of the nanocavity [32], we can modulate either of these two parameters to tune the plasmon resonances. As the VOC2O4·xH2O films contain water molecules in a form of sol-gel matrix which can be dehydrated over time, the RI of such composite films can be modulated at ambient temperature. This RI tuning mechanism is totally different from the phase change material of VO2 which largely depends on the change of lattice symmetry at high temperature (∼80°C) [33,34]. Another unique advantage of using VOC2O4·xH2O films is that their (de)hydration is reversible even at small thickness, which is critical to support the coupled gap plasmons of NPoM. The TiO2 overlayer functionalizes as a water reservoir [35], which not only protects the VOC2O4 layer from cracking but also supplies water molecules for the rehydration of the VOC2O4 films when the temperature cools down, making plasmon shift fully reversible [Fig. 1(a)]. Such reversible shift of the plasmon resonances can also be optically controlled via photothermal effect, which unlocks a new type of thermo-responsive optical material that can be well integrated with optoplasmonic switches and optical chips of the next generation [36–38].

The product of VOC2O4 is obtained via the reduction of V2O5 with H2C2O4 [39]. The ethanol not only serves as solvent for the reaction system, but also facilitates the processing of ultrathin films via spin-coating (see Appendix A for experimental details). The reaction was monitored with UV-vis spectra, which suggests the reduction completes within 10 h [Figs. 2(a) and 2(b)]. The energy dispersive X-ray spectroscopy (EDS) of the VOC2O4 films [Appendix B, Fig. 6(a)] indicates the presence of vanadium (V) with significant amounts of carbon (C) and oxygen (O), which were mainly originated from the oxalic acid. The valence state of V in the films is characterized by the X-ray photoemission spectroscopy (XPS), where the peaks of V 2p3/2 and V 2p1/2 exactly match the position of tetravalent vanadium [Fig. 2(c)] [40].

The naturally dried VOC2O4 powders are hydrated (VOC2O4·5H2O) as confirmed with the X-ray diffraction (XRD) patterns at room temperature [red line in Fig. 2(d)]. However, when the temperature rises to 80°C, the VOC2O4 hydrates are reduced with only two water molecules per VOC2O4 [blue line in Fig. 2(d)].

The loss of water molecules also changes its conductivity which shows temperature dependence [Figs. 2(e) and 2(f)]. The conductivity decreases dramatically when the temperature increases from 10°C to 30°C [Fig. 2(e)] but increases slightly as the temperature further increases over 40°C [Fig. 2(f)]. The first decrease of the conductivity is mainly due to the loss of surface absorbed water which significantly reduces the mobility of oxalic ions, while the slight increase at higher temperature is mainly due to the release of water molecules from the VOC2O4·xH2O crystallites in the sol-gel films, which promotes the mobility of the oxalic ions again. When the temperature cools from 80°C to 10°C, its conductivity switches back with strong hysteresis as it takes time for reabsorption and diffusion of the water molecules.

Unlike other inorganic hydrates, the VOC2O4·xH2O can be spin-coated on a flat substrate to form continuous thin films, which is due to the formation of O-V-O networks as evidenced by the weaker Raman mode of V=O stretching compared to that of the crystallites in the powders [Appendix B, Fig. 6(b)]. Such a smooth thin film is temporarily sustained by the sol-gel matrix with free water molecules as the hydration layer to protect them from cracking. Longer time of aging in air can lead to dehydration and oxidation of the films, which can be avoided by spin-coating a protection layer of TiO2 on top. Such composite films of VO2C2O4/TiO2 have negligible strains at the interface as the TiO2 sol-gel is highly reactive with VOC2O4, forming -Ti-O-V- networks at the interface between VOC2O4 and TiO2 via condensation [41]. This protective TiO2 layer not only maintains the mechanical and chemical stability of the VOC2O4 films but also stores the water molecules for VOC2O4 films with humidity-responsive electric and optical properties.

For a proof-of-concept demonstration, we use VOC2O4 as the spacer in the plasmonic nanostructures such as the Au NPoMs, which are protected by the TiO2 films [Fig. 1(a)]. The atomic force microscope (AFM) images [Appendix B, inset of Fig. 6(c)] suggest the VOC2O4 spacer is smooth and uniform with a thickness of ∼14nm as determined by ellipsometry [Appendix B, Fig. 6(c)]. A visible surface bump of only ∼10nm can be found with AFM imaging [Fig. 3(a)], suggesting successful overcoating of TiO2 films on the Au NPoMs. The SEM image reveals the Au NPs are separated with a distance larger than 200 nm [Fig. 3(b)], excluding the possibility of interparticle coupling. This large separation also ensures the scattering spectra collected are from single Au NPoMs.

The plasmon resonance of the Au NPoM was monitored by dark field (DF) scattering spectra with a heating stage, which shows temperature-dependent property [Fig. 3(c)]. As the temperature increases from 15°C to 80°C, the plasmon resonance of the Au NPoM blueshifts from 730 to 670 nm, which can switch back to 730 nm after the temperature cools down to 15°C again [Fig. 3(d)]. This plasmon resonance is the dipolar mode of gap plasmons in Au NPoM as no obvious shift of the transverse mode (∼620nm) is observed if the Au mirror is replaced by the Si substrate (Appendix C, Fig. 7). Such a reversible switching process is very robust and can be cycled for over 10 times [Figs. 3(e) and 3(f)]. Moreover, the switching performance of this thermo-responsive plasmonic system can be maintained as long as proper sealing technique is applied, which can be nicely reproducible after 10 days (Appendix C, Fig. 8).

This reversible plasmon switching is mainly due to the change of RI as the thickness of VOC2O4 films remains almost constant (∼14nm) throughout the thermal cycling from 15°C to 80°C [Appendix B, Fig. 6(d)]. The RI change of the VOC2O4 films is mainly due to the loss of water molecules with rising temperature especially in the range from 10°C to 40°C (Fig. 4). Thermogravimetric analysis (TGA) of the VOC2O4 powders shows the weight decreases linearly with increasing temperature but such a decrease clearly slows down in the temperature range from 45°C to 80°C [Fig. 4(a)]. The weight loss in the first phase (15°C to 45°C) is mainly due to the loss of absorbed free water molecules in VOC2O4·5H2O powders, which correlates to the huge decrease of its conductivity [Fig. 2(e)]. After the temperature rises above 45°C, the VOC2O4·5H2O almost all coverts to VOC2O4·2H2O, which is very stable until the temperature reaches above 100°C [42]. By fitting of the curves in Fig. 4(a) we obtain the semi-empirical correlation between the weight of VOC2O4·xH2O powders and the temperature (T) as W(T)%=113.26−0.781T(15°C<T≤40°C),W(T)%=74+144e−0.073T(40°C<T≤80°C).

Thus, the average RI change with temperature can be calculated with the formula navg=W(T)·nVO+[1−W(T)]·nair,where nVO and nair are the RI of VOC2O4 and H2O, respectively. Thus, we can correlate the temperature with the RI of VOC2O4·xH2O, which is presented as the dashed line in Fig. 4(d).

As for the VOC2O4·xH2O films, the surface humidity can be roughly obtained with a humidimeter [Fig. 4(b)], which shows a similar trend to the powders [Fig. 4(a)], confirming the reliability of the TGA measurement. The correlation between the plasmon resonance and RI can be simulated with the finite element method (FEM) at a fixed gap size of ∼14nm [Fig. 4(c)]. By comparing the wavelength readings of the plasmon peak measured at different temperatures [Fig. 3(d)] to such correlation, we can obtain the relationship between the RI and temperature, which is shown as the red line in Fig. 4(d). We find that the RI obtained via surface plasmon resonance (SPR) method is close to the RI obtained by the TGA method. The slight deviation is mainly due to the differences on the packing and dimensions between powders and films. The excellent match of the results obtained from the TGA approach further manifests the reliability of SPR for RI sensing. This agreement also indicates the shift of plasmon is due to the change of RI induced by the change of water content in the gaps.

Normally, the shift of plasmon resonances induced by the RI change is small (<20nm); however, in this case, the shift of coupled plasmon resonances is large [up to 60 nm, Fig. 3(d)] since smaller nanogaps have higher sensitivity to the change of RI [Appendix C, Fig. 9(a)]. The current RI sensitivity of this hybrid NPoM system is 190 nm/RI unit with a figure of merit of 1.46.

Since the thermal effect can be locally delivered by plasmonic heating with a laser, such a reversible switching of coupled plasmons can be optically controlled with ultrafast response (∼nanoseconds) [43,44]. To demonstrate the optical tunability of the coupled plasmons, another sample of Au NPoM was fabricated and irradiated with a 532 nm continuous wave (CW) laser, which can increase the local temperature of Au NPoM up to 30°C with a power of ∼3mW [Appendix C, Fig. 9(b)]. As a result, a blueshift of the coupled plasmons is observed from 676 to 664 nm [Fig. 5(a)] due to photothermally induced change of the local RI. Thus, reversible switching of plasmon resonances can be optically controlled at high modulation frequency of laser power, which persists robustly for many cycles [Fig. 5(b)], making it a promising prototype of optoplasmonic switching devices. The response time of such optical modulation is ∼10μs [Fig. 5(c)]. Power-dependent plasmon shift suggests that a critical threshold of ∼3mW is required for noticeable shift of plasmon resonances.

Both the response time and tuning range are critical parameters of optoplasmonic devices. In addition, the compatibility with existing optoelectronic systems is also an important factor to consider. Here we make a brief comparison of the VOC2O4-based optoplasmonic system to other optically tunable plasmonic systems based on different types of materials (Table 1). Responsive polymers such as poly(N-isopropylacrylamide) (PNIPAM) are commonly applied for plasmon tuning which shows large tuning range and excellent reversibility [15,21]; however, it has to be implemented in an aqueous environment which deters their application in optoplasmonic chips. Other electric- and light-responsive polymers such as azobenzene-doped poly(methyl methacrylate) (AZO/PMMA) [45] and polyvinyldifloride (PVDF) [46] although can be tuned in the solid state, their reversibility is mediocre. Liquid crystals (LCs) [47] are a class of commercialized optical materials which have been widely applied to modulate the plasmon resonances but their tuning range is always small with relatively slow response time (∼seconds). Phase change materials such as GST [29,48] and VO2 [49,50] may be more integratable to the optoelectronic devices with features of high speed and large tuning range, but they mainly work in a spectrum over IR. Thus, this VOC2O4 material system we demonstrated here complements the traditional phase change materials in terms of working range without losing the typical features of large tunability and fast switching speed, making it a promising prototype of optoplasmonic sensing and nanophotonic devices.Table 1.

Comparison of This Work to Other Typical Tunable Nanoplasmonic Systems

In summary, we have established a temperature-responsive plasmonic system based on inorganic thin films of VOC2O4·xH2O. These inorganic thin films function as a spacer in the Au NPoMs, whose RI decreases as the temperature increases, resulting in a blueshift of plasmon resonance as supported by simulations. The VOC2O4·xH2O films can be rehydrated when the temperature drops, which redshifts the plasmon peak back again. This process is reversible for many thermal cycles, which can also be photothermally switched with laser illumination. Such temperature-responsive plasmons with solid-state thin films have great compatibility with optoelectronic devices, which can potentially be applied in optoplasmonic switches with ultrafast response.

Acknowledgment. We thank the support of the Center for Nanoscience and Nanotechnology at Wuhan University. We also thank Dr. Huatian Hu for his assistance on the simulation.