1. Introduction

Phase-change materials revolutionized dissemination of digital data since 1990s by enabling the rewritable optical data storage [1]. As the use of optical data storage is surpassed by the all-electronic memory devices, the phase-change materials find their way into other optical devices. The phase-change materials can be categorized as non-volatile and volatile. The non-volatile phase-change materials are mostly based on chalcogenides, such as Ge₂Sb₂Te₅, SbTe. These materials exhibit stable amorphous and crystalline phases at room temperature [2]. A typical amorphous to crystalline transition requires annealing the materials below the melting temperature, whereas the crystalline to amorphous transition is achieved by melting and quenching. The need for high temperatures and quenching for amorphous to crystalline transition restricts the application of non-volatile phase-change materials. So far, non-volatile phase-change materials are used for tunable optical applications like enhanced data storage [3], changeable surface colors [4–6], switching waveguides [7–9], tuning near infrared transmission [10–12], tunable infrared emitters [13,14], tunable infrared absorbers [15], and beam steering [16]. The volatile phase-change materials like VO₂ exhibit dynamic phase-changing that can be induced thermally [17]. The materials are in insulator phase at lower temperatures and transition to metallic phase by elevating the temperature. The phase-change is reversed as the material cools down following a hysteresis curve. The transition temperature is given as 340 K for VO₂. The list of optical applications based on VO₂ includes uncooled bolometers [18], controlling near infrared transmission [19], dynamic control over infrared absorption [20–24] and emission [25,26]. Here, we demonstrate dynamic tuning of enhanced infrared absorption spectroscopy surfaces using VO₂ thin films.

Infrared absorption spectroscopy is a common material characterization technique that relies on identification of vibrational absorptions in the infrared regime specific to molecular bonds. For characterization, materials are typically deposited on an infrared transparent substrate such as a CaF₂ window, if the amount of the material is sufficient. The absorbance is then measured as the difference between the incident and the transmitted infrared light. Alternatively, KBr powder is used to mix with the material and pulverized to form pellets for transmission measurements [27]. The absorbance is proportional to the electric field intensity, |E|², following the Beer’s Law. For the CaF₂ substrate, the electric field intensity on the surface of the substrate is ~0.7|E_o|² in the absence of any material, where |E_o|² is the electric field intensity of the incident light [Fig. 1]. A lower field intensity on the surface is a direct result of the partial destructive interference of the incident light and the reflected light from the surface. If the amount of material is low, then the absorbance of the infrared light must be enhanced to the detectable levels. Plasmonic field enhancement is demonstrated to be a promising method for detection of ultralow amounts of materials owing to very high field enhancements on nanostructured metal features [28,29]. Their use as a universal enhanced infrared absorption spectroscopy surfaces, however, is limited due to the spatially localized field and narrow spectral bandwidth. For instance, a metal antenna array designed to detect the Amide I and II absorption bands of protein molecules exhibits a maximum field intensity enhancement factor of ~500 at the antenna edges, however the peak value of the average field enhancement on the antennas is only ~3.3 [Fig. 1]. Moreover, the field intensity is only enhanced in a narrow bandwidth.

 

Fig. 1 (a) Illustration of infrared absorption spectroscopy surfaces: (i) Cross-section of a semi-infinite CaF₂ substrate, (ii) top-down illustration of a single Ag nanoantenna on Si, and (iii) cross-section of a thin CaF₂ film on optically-thick Al. (b) The field-intensity enhancement spectra for the surfaces in (a). The field-intensity on the antennas is averaged for 10-nm thick volume on the surface.

Download Full Size | PDF

Recently, we have demonstrated that simply enhancing the electric field uniformly on the entire surface exhibits a better performance compared to plasmonic-field enhancement for the purpose of universal infrared absorption spectroscopy of ultrathin materials like monolayers of protein molecules [30]. The surfaces consist of an infrared-transparent dielectric film on a metal mirror [Fig. 1]. The field enhancement is achieved by constructive interference of the incident light and the reflected light from the surface. Using a low refractive index material like CaF₂ and a highly reflective mirror layer like Al maximizes the peak enhancement and the enhancement bandwidth. The peak field enhancement wavelength is approximately t/4n, where t and n are the thickness and the refractive index of the CaF₂ layer, respectively. Hence, the peak enhancement wavelength and the enhancement spectrum can be tuned by changing the CaF₂ thickness that is the motivation for using a VO₂ thin film. It is worth noting that Al can be replaced with other metals such as Ag and Au that provide high reflectance in the infrared, should Al poses a challenge to be used together with VO₂.

2. Results and discussion

Inserting a thin VO₂ layer between the Al and CaF₂ layers enables changing the effective dielectric thickness by controlling the VO₂ phase [Fig. 2(a)]. Insulating VO₂ is lossless in the infrared regime with a refractive index of ~3.7 [22]. In the insulator phase, the effective dielectric thickness is the sum of VO₂ and CaF₂ thicknesses resulting in a large peak enhancement wavelength and pulling the 2^nd order enhancement peak into the IR regime. For the metallic phase, the CaF₂ layer is the only dielectric layer, resulting in blueshifting in the peak enhancement wavelength [Fig. 2(b)]. The overall enhancement spectrum is determined by the maximum of the enhancement spectra for the insulator and metal phases (the curve enveloping the shaded in Fig. 2(b)). The target is maximizing the spectral average of the enhancement factor, in other words maximizing the shaded area in Fig. 2(b), in the wavelength range of 2.5 – 10 µm that is the typical range for molecular vibrations. The VO₂ and CaF₂ thicknesses, t₁ and t₂, that maximize the mean spectral enhancement factor to ~3.1 are found as 350 nm and 880 nm, respectively, by varying both thicknesses [Fig. 2(c)]. Without a VO₂ layer, the maximum mean enhancement factor is ~2.6 that is achieved for 1.2-µm-thick CaF₂. More importantly, the overall enhancement spectrum is always larger than unity [Fig. 2(b)], whereas the enhancement factor is below unity in a narrow spectral range and even zeros out at a certain wavelength without a VO₂ layer [Fig. 1(b)]. A thinner VO₂ film can be used by introducing another layer of CaF₂ in between Al and VO₂ [Fig. 2(d)]. If the VO₂ is chosen to be optically thick in metallic phase (> 100 nm), the bottom CaF₂ layer is invisible to the incident light as infrared light cannot penetrate metallic VO₂. The bottom CaF₂ contributes to the overall optical response for insulating VO₂. The mean field-intensity enhancement factor peaks at ~3.2 using 370-nm bottom CaF₂ and 850-nm top CaF₂ for 100-nm VO₂ [Fig. 2(e)]. Reducing the VO₂ layer thickness to 50 nm provides a similar performance using 480-nm bottom CaF₂ and 850-nm top CaF₂ layer.

 

Fig. 2 (a) Cross-section illustration of stacked CaF₂ and VO₂ films on optically-thick Al. (b) Field-intensity enhancement on the CaF₂ surface for t₁ = 350 nm, t₂ = 880 nm. (c) Field-intensity enhancement factor averaged over the wavelength range of 2.5 – 10 µm for varying CaF₂ and VO₂ thicknesses. (d) Cross-section illustration of 850-nm CaF₂/100-nm VO₂/370-nm CaF₂ stack on optically-thick Al. (e) Field-intensity enhancement on the surface for the stack shown in (d). The optical properties of the VO₂ layer are obtained from Ref [22].

Download Full Size | PDF

The designed surface is numerically tested with a 10-nm-thick poly(methyl methacrylate) (PMMA) layer that is used as a probe material [Fig. 3(a)]. PMMA exhibits a number of absorption bands in the wavelength range of 3 to 10 µm, making it a common material to test the performance of infrared absorption spectroscopy surfaces [30,31]. The absorbance spectrum for the PMMA layer is quantified as the difference between the reflectance spectra with and without the PMMA layer. The absorbance spectrum for the same PMMA layer residing on a CaF₂ substrate is calculated as the reference. The molecular vibration bands of PMMA exhibit larger absorption for the insulating VO₂, expect for the major absorption band at 1732 cm⁻¹, compared to the absorbance spectrum of PMMA on a CaF₂ substrate [Fig. 3(b)]. For the field-enhancement surface, transitioning to metallic phase enhances the absorbance at 1732 cm⁻¹ keeping the overall absorption enhancement 3 to 5 times larger than the reference signals.

 

Fig. 3 (a) Cross-section illustration of 10-nm-thick PMMA on the sensor surface. (b) Simulated absorption spectra of the PMMA layer on the sensor for the insulating and metallic VO₂ and when the PMMA layer resides on a CaF₂ window.

Download Full Size | PDF

Thermal modelling of the surface is necessary as the insulator-to-metal transition is proposed to be achieved thermally. While elevating the surface temperature using a hot plate would be a common and convenient approach, we propose using the metal mirror layer as an electric heater to speed up the thermal response, reduce the energy consumption, and miniaturize the experimental setup. The first part of the thermal modelling is finding the temperature distribution on the surface’s cross-section [Fig. 4(a)]. One-dimensional heat transfer in the cross-plane direction is assumed considering the low thicknesses of the films (< 1 µm) and as well as the substrate (~mm) compared to the lateral dimensions (> cm). In the steady state, the heat conducted through the VO₂ (q_Cond1) and CaF₂ (q_Cond2) layers are equal to each other and also to the sum of the heat loss through radiation (q_Rad) and convection (q_Conv) from the surface [32]:

 

Fig. 4 (a) Heat flow directions on the cross-section of the surface. (b) 3D illustration of the surface with the metal layer patterned in a serpentine shape. The VO₂ and CaF₂ layers are not shown for clarity. Note that the gaps between the metal features are exaggerated for clarity. The gaps must be minimized for a better surface coverage.

Download Full Size | PDF

It is also necessary to calculate the power generated in the heater and the required voltage and electric current. The total heat loss is the sum of the radiative and convective heat loss from the surface and the heat conducted through the substrate ($q_{Cond3}/A=k_{subs}\left(T_{source}-T_3\right)/t_{subs}$). For a better thermal isolation, hence lower energy consumption, a low-thermal-conductivity material like glass (k ≈1 Wm⁻¹K⁻¹) must be used as the substrate instead of silicon (k ≈150 Wm⁻¹K⁻¹). Using a thick substrate, e.g., microscope slides, further reduces the heat conduction. q_Cond3 /A is calculated as ~2 × 10⁴ W/m² using t_subs = 2 mm and T₃ = 300 K. By accounting for the radiative and convective losses, the total heat loss, hence the required power input, amounts to ~6 × 10⁴ W/m². Assuming 1 cm² area, the required power is calculated as ~6 W that can be dissipated on a 6 Ω resistor carrying 1 A electric current. The dimensions of the heater is then determined assuming a uniform-cross section resistor: R = ρL/Wt, where ρ is the electrical resistivity, L is the length, W is the width, and t is Al film thickness [Fig. 4(b)]. The length to width ratio (L/W) of the heater is found as ~20 for ρ = 3 × 10⁻⁸ Ωm and t = 100 nm. Alternatively, the Al film thickness can be lowered down to 10 nm without sacrificing the opacity in the infrared and the electrical current can be increased to avoid patterning the Al layer. It must be noted, that patterning the Al layer into a serpentine shape with small gaps can be achieved using large-area fabrication techniques like the stencil method.

3. Summary

In conclusion, we propose thermally tuning the bandwidth of the universal infrared absorption spectroscopy surfaces based on a stack of CaF₂/VO₂/Al. The electric field intensity, hence, the absorbance spectrum on such surfaces primarily depends on the optical thickness of the dielectric layer. Thermally changing the VO₂ phase from insulator to metal decreases the effective dielectric thickness, hence blueshifting the field intensity enhancement spectrum. As a result, the overall field intensity enhancement can cover a larger wavelength range. The CaF₂ and VO₂ thicknesses are numerically found as 880 nm and 350 nm, respectively, to maximize bandwidth of the field enhancement. The designed surface can enhance the IR absorbance of ultrathin probe materials in a large bandwidth (2.5 – 10 µm) that is superior to the alternatives like CaF₂ windows and plasmonic field-enhancement surfaces. The thermal analysis of the surfaces reveals that the surface temperature stays close the VO₂ temperature at all times limiting the use of the surface with temperature-sensitive probe materials when the VO₂ layer is in metallic phase. Finally, the possibility of using the Al mirror layer as a heater is inspected by resistance and heat dissipation calculations.