Electro-optically modulated lossy-mode resonance

2.1 Numerical analysis

In this work, double-domain (optical and EC) approach that utilizes both, the optical and electrical advantages of ITO thin films is assumed. Figure 1 shows the schematic of an interface between the domains for optical fiber structure (Figure 1A) and its equivalent electrical model (Figure 1B). Considered simplified electrical model of the sensor corresponds to a metal–insulator–semiconductor (MIS) capacitor with a single Cu electrode and concentration of free charge carriers (electrons) of ITO (N) ranging from 3e19 cm⁻³ to 3e20 cm⁻³. This range is in agreement with experimental results presented elsewhere [31, 32]. The thickness of ITO film was set to 250 nm which corresponds to microscopic analysis of the ITO films on the surface of an optical fiber allowing for receiving LMR [33]. The ITO thickness was optimized using in situ spectrum monitoring approach during the deposition to attain 1st order LMR in the spectral range of operation. Moreover, a monolayer (0.2 nm) of SiO₂ representing oxidized surface of the material, was introduced in the electrical model at the ITO-metal interface to make the MIS capacitor assumption valid. Presence of this layer in the model originates from the fact that ITO is often suboxidized and its surface becomes fully stoichiometric when exposed to standard environmental conditions [17]. In an experimental setup shown schematically in Figure 1A, the ITO-LMR sensor is immersed in an electrolyte (represented by metal in the electrical model), and the ITO film plays a role of a working electrode biased by the potential E. When E changes, as it happens in EC setups, the charge carrier density at the ITO surface follows. In this work the mean density of accumulated charge carriers N _acc(E) has been calculated according to the procedure reported in [34]. Next, optical properties at the surface of ITO, namely n _acc and k _acc have been calculated using well-established Drude model of scattering of electrons by relatively immobile ions [35] according to Eq. (1), where ω is the field angular frequency, ε _∞ = 3.9 is the high-frequency permittivity, Γ = 2.9e15 s⁻¹ is the electron collision frequency [36], and ω _p represents the plasma oscillation resonance. The ω _p is determined by the ITO doping and defined by Eq. (2), where N _acc is the charge carrier density, e is the elementary charge, ε ₀ is the vacuum permittivity and m _eff = 0.35m ₀ [37] is the effective mass of electron.

Figure 1:

Schematic representation of (A) experimental setup for combined optical and EC analysis with the investigated ITO-coated optical fiber structure and part of its cross-section. In (B) electrically equivalent model of the optical structure is schematically shown, where a monolayer of SiO₂ was introduced on ITO to represent oxidation of the film surface.

Next, the calculated n _acc and k _acc were used to describe a layer with a thickness corresponding to accumulation depth (d _acc). For calculation of the spectral response of the ITO-LMR structure, whole ITO layer, i.e., bulk and the accumulation layer, has been considered as a single effective layer of 250 nm and described by effective RI, i.e., n _eff and k _eff. For this purpose, an optimization-based approach for extraction of optical constants has been formulated inspired by Male method [38]. The approach is based on a transfer matrix model suitable for description of electromagnetic response of plane–parallel media. Finally, spectral response of the ITO-LMR structure for different E was calculated using a numerical model described in [39], where silica glass and water were assumed as substrate and external medium, respectively.

2.2 ITO-coated optical fiber fabrication

Reactive magnetron sputtering of ITO was carried out in a high-vacuum chamber capable to attain the back-down pressure of 1e−6 Pa and assure the high purity of the deposition. The configuration of the deposition system is given in detail in [17]. A commercial planar 3-inch magnetron equipped with a compound ITO target (In₂O₃/SnO₂ of composition 90/10 wt% and a purity of 99.99%) was installed in the stainless-steel vacuum chamber. Samples were placed 15 cm away from the sputtering target. For the fabrication of the samples approx. 15 cm long multimode polymer-clad silica fibers with 400/730 μm (core/cladding) diameter were used. For each fiber its central part (ca. 25 mm) of the cladding was removed and after ITO deposition it served as an active sensing area. The samples were rotated during the deposition to receive a uniform ITO coating around them. The sputtering was driven using RF COMET Cito 1310 supply with frequency 13.56 MHz and transmitted power of 150 W. Argon (purity of 99.999%) flow was kept constant at 100 sccm. Pressure (p) and deposition time (t) were tuned in range 0.1–1 Pa and 100 to 29 min, respectively, to tailor the highest optical sensitivity corresponding to the first attenuation band of LMR [27]. The range of p was found during initial experiments and recognized as appropriate for maintaining effective sputtering processes. It is worth noting that the deposition parameters were optimized to tailor ITO properties for dual-domain interrogation with respect to the system geometry and conditions of sputtered target. The in situ and online optical spectrum monitoring during the deposition was employed to prepare sensitivity-wise optimized structures, i.e., 1st order LMR was sought.

2.3 Measurements

Optical transmission of the ITO-coated fiber structure was interrogated in the range λ = 350–1050 nm using Ocean Optics HL-2000 white light source and Ocean Optics USB4000 spectrometer. Although the spectral response of the LMR structure is polarization dependent, the setup was kept simple, without any polarization managing devices. In general, the magnitude and location of the minimum in the spectral response of the sensor differs for each polarization. However, with the unpolarized light source and the multimode optical fiber the more pronounced resonance dip may be observed and traced, since contributions of the polarizations superimpose with one another [40]. Therefore, the setup was used as a simple, robust, and experimentally proved concept. First, the sensitivity of the samples to changes of external RI (n _ext) was determined by immersion in water/glycerin solutions of different concentrations n _D = 1.3330–1.4200 RIU. Rudolph J57 automatic refractometer was used to verify n _ext of the solutions. Next, each sample was installed in the EC cell, as shown in Figure 1A, where it acted as a working electrode. A platinum wire and an Ag/AgCl 0.1 M KCl were used as a counter and a reference electrode, respectively. The experimental setup has been described in details in [33]. The EC performance of the films was tested in cyclic voltammetry (CV) configuration using a PalmSens Emstat3+ potentiostat/galvanostat controlled by a PSTrace 5.4 software. All the measurements were performed at room temperature in 0.1 M KCl (POCh, Poland) solution containing 1 mM 1,1′-ferrocenedimethanol (Acros Organics) as a redox system. All CVs were obtained at E reaching from −1 to 1 V versus Ag/AgCl electrode at a scan rate of 20 mV/s. Optical transmission (T) was traced in parallel to EC measurements. In-house developed software based on Matlab was used to synchronize optical and EC data.

3.1 Influence of charge carrier density on the modulation of ITO’s optical properties

ITO is an n-type semiconductor in which due to changes in composition and molecular structure the density of free charge carriers (N) can be tuned in a wide range, typically from 1e19 to 1e20 cm⁻³ [41]. Additionally, N can be locally modified by electric field E [42]. Negative and positive E induces, respectively, accumulation and depletion of majority carriers at the ITO surface. The more negative E becomes the higher the density of accumulated carriers N _acc and the deeper (higher d _acc) into the surface the accumulation spreads. Figure 2A shows the dependence of mean N _acc(E) for different N. It can be seen that ability to form the accumulation layer is determined by N, and for its lower values (order of 1e19 cm⁻³) the d _acc can reach tens of nanometers (Figure 2B) with an average increase of N _acc by over an order of magnitude (Figure 2A). For EC applications of ITO, N plays a crucial role. The Fermi level occurring in the forbidden energy gap of a semiconductor is influenced by the doping. Enhanced N brings the Fermi level up, what may encourage part of the electrons to contribute to the EC charge exchange between the electrolyte and the electrode.

Figure 2:

E-induced evolution of accumulation layer where (A) and (B) shows mean charge carrier density and thickness of the layer, respectively, for initial bulk carrier density N reaching 3e19, 8e19, and 3e20 cm⁻³.

Modulation of N _acc is followed not only by changes of electrical, but also optical properties of ITO [21]. Figure 3 shows analytical estimation of how significant the changes of RI of ITO in the accumulation layer can be. For N = 3e19 cm⁻³ the changes are effective when E < −0.5 V. In the analyzed case, in UV–Vis spectral range n _acc may decrease and k _acc may increase by over 0.5 RIU and 1 RIU, respectively, and reach over d _acc = 20 nm in depth of the material. Further decrease in E induces rise of N _acc and shift of the Fermi level toward metallic-like character of the ITO layer. This is in particular indicated by the turn and rise of n _acc(λ) as E becomes more negative < −0.5 V (Figure 3A).

Figure 3:

E-induced evolution of an averaged (A) refractive index and (B) extinction coefficient in the accumulation layer. The results are shown for N reaching of 3e19 cm⁻³ where accumulation layer of ∼20 nm appears for E < −0.5 V (see Figure 2).

Since E influences n _acc and k _acc (Figure 3), the ITO-LMR structure undergoes the E-induced changes which manifest themselves in LMR response (Figure 4). The LMR follows the decrease of E, i.e., the lower E is applied the deeper LMR is observed and corresponding wavelength λ _R experiences a blueshift. This is clearly visible in Figure 4A for lower N, i.e., N = 3e19 cm⁻³, when E drops below −0.5 V. The spectral changes with E are significantly less effective for higher N (Figure 4B). These results qualitatively corresponds well with experimental studies reported in [16], and indicate that for formerly shown experimental results initial N may be below 3e19 cm⁻³. Following values of N of ITO films indicated in [43], as low as 1e19 cm⁻³ can be achieved for this material. Moreover, a significant difference between numerical and experimental analysis must be noted, i.e., numerical data are received at stationary conditions, while EC measurements are performed at dynamically changing (scanning) E. The lower is the scan rate, the higher LMR spectrum modulation is recorded [30]. It must be also noted that the λ _R decreases with N (Figure 4B), what is important for discussions on optical properties of ITO included in Section 3.2.

Figure 4:

(A) E-induced evolution of ITO-LMR spectral response when charge carrier density reaches 3e19 cm⁻³ and (B) LMR wavelength shift with E when N ranges from 3e19 cm⁻³ to 3e20 cm⁻³.

The results of numerical analysis (Figures 2 –4) indicate capability of electrical modulation of ITO-LMR properties. It is shown that charge carrier density, the depth of accumulation zone, and the LMR spectral shift are highly influenced by initial bulk density of charge carriers (free electrons in the case of ITO). Higher accumulation of carriers corresponds to higher electrical conductivity. In contrast, significant λ _R shifts with E can be achieved when N is low. Thus, in the next section, capability for tuning ITO properties at its deposition state will be discussed in detail.

3.2 Tuning ITO properties at the stage of its deposition

It has already been shown that properties of ITO film deposited by means of magnetron sputtering depend on many process parameters that include deposition power, used electric field (DC, RF, impulse DC), process gas composition, base and working pressure, distance between the sample and the target, as well as substrate temperature and deposition time [17, 29]. Since ITO is widely applied, an impact of these parameters on optical and electrical properties is crucial for certain application and the films have been intensively studied during the last two decades. In our earlier work devoted to ITO-LMR sensors [33], capability to change optical properties of ITO and corresponding sensors response by deposition pressure (p) has been shown. The encouraging results of that work allow us to set the p to be the tuning parameters also for the study presented here.

In Figure 5A and B spectral responses for ITO-LMR samples received at different set of p and t are shown. It has to be noted that the film thickness (d) is another variable parameter that was optimized by t, i.e., t was always adjusted to receive possibly similar spectral pattern (λ _R at ∼650 nm for n _ext = 1 RIU). The visible spectral range has been selected due to widely available instrumentation for interrogation in this range. In general, it is possible to receive similar spectral responses for different p, but t needs to be adjusted since the deposition rate (and subsequently d) follows the p. This effect is well-known and explained by a mean free path of sputtered atoms influenced by enhanced collisions in plasma at elevated p [44]. On top of the deposition rate, p has an impact on optical properties of the films, in particular its n and k [29, 33]. Investigations reveal that both the optical parameters are of crucial importance and strongly dependent on deposition process. In Figure 5C the spectral shifts, that corresponds to the RI sensitivity, are shown for different p. The λ _R shift follows the p – the RI sensitivity increases by over 150% when p increases from 0.1 to 1 Pa. Following in-detail theoretical analysis reported by Del Villar et al. [27], when n of the film increases, RI sensitivity of the LMR device follows. Since experimentally measured RI sensitivity increases with p in the discussed case (Figure 5C) one can expect, that n of ITO also increases with p. Thus, also following results shown in Figure 4B, when p increases the N drops. The N is a material property dependent on conditions of growth and the ITO film and its crystallinity. The crystallinity is strongly influenced by energy of impinging particles building the film [45, 46]. The energy of sputtered species is depleted in the plasma volume due to collisions with neutral gas atoms (Ar in our case). The energy depletion is proportional to the total number of gas atoms, and then can be considered as proportional to p [47]. The film prepared at higher p usually suffers from lower level of crystallinity, i.e., are typically amorphous [48]. Amorphous and polycrystalline structure of the film typically correspond to low mobility of charge carriers [49]. Hence, we can assume that rather amorphous ITO films prepared at higher p with lower N results in well-pronounced LMR and its high shift with applied E as shown in Figure 4B.

Figure 5:

Evolution of ITO-LMR sample properties with p. Examples of similar LMR spectral patterns in response to n _ext varying from 1 up to 1.42 RIU received at different ITO deposition p and t are shown in (A) and (B). In (C) is shown relative shift of the λ _R with n _ext for selected sets of p and t. The λ _R0 is the λ _R at n _ext = 1.3330 RIU for each set of p and t. In (D) is shown relation between resistance (R) of 25-mm-long active part of ITO-LMR structure and the p.

On top of optical properties, the p has also an impact on electrical properties of ITO. Tailoring of both electrical and the optical properties of the ITO film is important for two-domain simultaneous interrogation. Resistance (R) of the active part (25 mm) of the ITO-LMR has been measured directly on the optical fiber surface by an easy 2-pin method and the results for various p are shown in Figure 5D. It can be seen, that for p < 0.3 Pa the R does not exceed single kΩ, when for higher p (up to 1 Pa) it significantly increases. In general, R corresponds to N [21]. Furthermore, carrier mobility is typically higher in crystalline structures with larger grain domains [49] which preferential growth appears at lower p. It is documented in [29] that grain size and the level of crystallinity with preferential facet [h00] is strongly influenced by p. Vice versa for higher p smaller grains tend to growth, i.e., the deposited structure is expected to be polycrystalline or even amorphous, with reduced mobility of carriers that are scattered at boundaries of small grains. The results shown in Figure 5D fit well to the frame given above: the R increases significantly for p > 0.3 Pa, where lower level of crystallinity is expected. Thus, high n and high R of the ITO film correspond to low N that can be achieved for elevated p.

The effect of electro-optical modulation of the ITO-LMR is discussed next. In Figure 6A plots obtained during CV scans are shown for ITO-LMR samples received at different p. The E was applied between the working (ITO film) and reference electrode, and an electrolyte containing redox probe was used. The presence of the probe has made current peaks observable for certain E values, that correspond to reduction and oxidation reactions. These peaks can be clearly identified for ITO deposited at low pressure, e.g., p = 0.2 Pa. These are the films with higher crystallinity [29], and subsequently higher N, and expected higher charge carrier mobility. It should be reminded here, that N (in particular N _acc) may be further increased by the E applied to ITO (see Figure 2A). Due to enhanced N, the Fermi level of ITO rises and enables the exchange of the electrons with the electrolyte. In turn, when ITO films are prepared at higher pressures p > 0.2 Pa (Figure 6A), the redox current peaks tend to spread in the E domain and finally vanish, what can be clearly seen for p = 1 Pa. For the case of p > 0.3 Pa the scanning range of E has been increased, but even for these conditions the peaks can be hardly seen. It fits well with our previous statements that polycrystalline films with reduced mobility and N are preferentially deposited at higher p.

Figure 6:

(A) EC and (B) optical responses received for ITO-LMR used as a working electrode, when various ITO deposition p are considered. The analysis was performed in 0.1 M PBS and 1mM 1,1′-ferrocenedimethanol as a redox probe. Scan rate was set to 20 mV/s.

Optical responses of the ITO-LMR were recorded in parallel to CVs. Figure 6B shows relative shift of the λ _R with E. The λ _R was referred to initial measurements in the electrolyte when no E was applied (λ _R0). For electrochemically-active samples, i.e., those with appearing redox current peaks deposited at p < 0.3 Pa, only very limited modulation effect is observed. The LMR modulation becomes effective for negative E and p > 0.3 Pa. In the investigated p range the modulation effectiveness reaches its maximum for p = 1 Pa, where dλ _R/dE exceeds 40 nm/V. At these conditions the N seems to be the lowest in the investigated p range.