Influence of Annealing Temperature on Structural Compositions and pH Sensing Properties of Sol-Gel Derived YTi_xO_y Electroceramic Sensing Membranes

All the material and precursor solutions used in this research were of analytical grade. In order to prepare the YTi_xO_y based sol-gel, yttrium oxide (Y₂O₃ granule, Admat Midas Inc., purity 99.99%) was dissolved in 2 N HCl solution to achieve a final concentration of 0.1 mol/L. Further, clear and transparent Y₂O₃ sol-gel were placed in water bath at 80°C and stirred constantly for 3 h. Subsequently, titanium precursor solution (titanium isopropoxide, [(CH₃)₂CHO]₄Ti, Sigma Aldrich, purity 97%) was added in the Y₂O₃ sol-gel by dropwise under vigorous stirring. After the addition of titanium isopropoxide, the reaction is allowed to proceed up to 30 min. To obtain a stable homogeneous solution, lactic acid (0.5 ml) was added into the solution acting as a sol stabilizer. However, glycerol (1 ml) also was added into the solution to ensure the good spin condition like film thickness and proper adhesion on the Si substrate. This sol-gel solution was kept at ambient temperature for 48 h for aging so as to acquire the optimal viscosity during the deposition of the film.

P-type (<100>) Si wafer with a resistivity of 1–10 Ω-cm was used for the fabrication of YTi_xO_y sensing film based EIS sensor. Prior to the deposition of sensing film, the wafer was cleaned by a standard RCA cleaning procedure and further submerged into 1% HF solution to remove the native oxide. A ∼ 40 nm YTi_xO_y film was deposited on the cleaned Si wafer by a spin coating technique with the spin rate of 3000 rpm for 30 s under suction. To enhance the adhesion between the YTi_xO_y thin film and Si substrate as well as the evaporation of organic solvent from YTi_xO_y films, the deposited film was dried at 300°C for 5 min. Further, the YTi_xO_y films were then underwent for annealing in a furnace with various annealing temperatures from 700 to 900°C in the presence of O₂ atmosphere for 45 min. In order to establish ohmic contact, an aluminum film (∼100 nm) was deposited on the back side of silicon wafer by a thermal evaporation method. The sensing region of the YTi_xO_y film was defined by using epoxy gel, which acts as an isolation layer. Finally, EIS structure was assembled on the copper line of a printed circuit board by the silver gel. To prevent the leakage of electrolyte solution, epoxy was used to distinguish the EIS structure and Cu line. Three devices on each annealing temperature were measured to evaluate sensing performance of the YTi_xO_y EIS sensors. The schematic structure of YTi_xO_y based EIS sensor is illustrated in Fig. 1.

Zoom In Zoom Out Reset image size

The electrical performance of YTi_xO_y sensing membrane was evaluated by measuring capacitance-voltage (C-V) measurement of the prepared EIS devices in different buffer solutions with a Ag/AgCl reference electrode and HP4284 LCR meter (ac signal frequency: 500 Hz). All electrical experiments were performed in a Faraday cage to avoid the light and noise interference. The crystal structure and orientation of the YTi_xO_y films were carried out by a Rigaku D/MAX2000 XRD with Cu K_α (λ = 1.54176 Å) line. The surface morphology and roughness of the YTi_xO_y films were observed using an NT-MDT Solver P47 AFM with a silicon probe operated in tapping mode. The root-mean-square (R_rms) variation of the height values was calculated from 3 × 3 μm² scan areas. The chemical bonding of the dielectric film was studied using a Physical Electronics Quantum 2000 XPS instrument equipped with a monochromatic Al K_α X-ray (1486.7 eV) source. The symmetric Gaussian–Lorentzian product functions were used to approximate the line shapes of the fitting components. Peak fitting was performed after Shirley background subtraction.

In order to understand the effect of annealing on the orientation and crystalline nature of YTi_xO_y sensing films, XRD measurements were performed as shown in Fig. 2a. The sample performed at annealing temperatures of 700, 800 and 900°C, no apparent crystalline phase was observed and the pattern showed a large diffuse background, suggesting that all annealed YTi_xO_y films exist in the amorphous structure.

Zoom In Zoom Out Reset image size

The surface morphology is an important characteristic of sensing membrane for assessment of its performance in the electrolyte solution. To obtain the surface roughness and grain size of sensing films, the annealed YTi_xO_y sensing membranes were explored from AFM images in Figs. 2b–2d. The surface roughness values of 0.46, 0.71 and 1 nm were obtained for the YTi_xO_y sensing membranes annealed at 700, 800 and 900°C, respectively. From the obtained morphology, we can clearly observed that the surface roughness increased with the degree of annealing temperature. This property ascribes that the native grains are coalesced and organized into larger grain due to the post-annealing treatment. This divergent nature of larger grain influences the surface morphology of the sensing membrane. Furthermore, the sample annealed at 900°C showed more rougher surface, indicating that yttrium and oxygen atoms diffuse from the film and relocate toward the interface where it forms yttrium silicate (YSi_xO_y) due to the interaction between the Y₂O₃ and SiO₂.³⁵

Impact of annealing temperature on the surface composition and chemical state of the YTi_xO_y sensing membrane was further investigated by using XPS technique. Figs. 3a–3c show the Y 3d, Ti 2p and O 1s spectra of the YTi_xO_y sensing films treated at different annealing temperatures, respectively. A signification variation in the peak position and intensity caused by a conventional furnace anneal is attributed to the surface modification in sensing membrane. As shown in Fig. 3a, the binding energy of Y 3d_5/2 and 3d_3/2 double peaks for the sample after annealing at 700°C located at 157.2 and 159.1 eV, suggesting weak (YTi)O₂ structure formed on the surface of sample due to the subjection to air.³⁶ The YTi_xO_y sensing film annealed at 800°C, the binding energy of Y 3d_5/2 and 3d_3/2 double peaks shifted toward lower binding energies, appeared at 157 and 158.9 eV, respectively.^27,28 Moreover, it was found that the increasing of annealing temperature at 900°C, Y 3d doublet peak shifted toward more lower binding energies by 0.3 eV. The position of Y 3d_5/2 and 3d_3/2 double peaks was obtained at 156.7 and 158.6 eV, respectively. It is found that the increase in annealing temperature have adverse effect on the surface. Thus, this finding suggests that the Y and O atom diffuse toward the interface and further intermixed with Si atoms, which facilitate the formation of an Y-silicate layer at high temperature annealing. The XPS spectra of Ti 2p_3/2 and 2p_1/2 double peaks for the YTi_xO_y films annealed at different temperatures were presented in Fig. 3b. The peak position of Ti 2p_3/2 and 2p_1/2 at 458.3 and 464 eV, respectively, for sample annealed at 700°C is assigned for Ti⁴⁺ in the YTi_xO_y film.³⁷ It was also observed that the binding energy of Ti 2p_3/2 peak increased from 458.3 to 458.4 eV with increasing in the annealing temperature from 700 to 800°C, respectively, arising from Y-O-Ti bond. This shift also evidenced the similar phenomenon as described above, the Y atom moved toward interface to make silicates, while the incorporation of Ti leads to the formation of Y-O-Ti bond which is more mechanically stable. The O 1s spectra for the YTi_xO_y films as a function of annealing temperature are shown in Fig. 3c, with three peak curve fitting lines. In the three peak sets, one peak at 529.1 eV represented the O atom in Y₂O₃,³⁸ another peak at 529.9 eV corresponding to a weak Y-O-Ti bond was observed for the 700°C-annealed sample, and the other peak at 531.7 eV can be regarded carbon contamination in sample. In contrast, the increase in annealing temperature at 800°C the binding energy of O 1s peak shifted toward a lower binding energy by 0.1 eV (529.8 eV), indicating that Ti replaces the oxygen and forms stoichiometric Y-O-Ti bond. Additionally, further increment in annealing temperature at 900°C, the binding energy decreased (529.7 eV), there is the diffusion of Y and O atom toward the interface to induce the formation of an interfacial layer. This observation suggests that the YTi_xO_y sensing membrane annealed at 800°C suppresses the formation of an interfacial yttrium silicate, thereby leading to a stronger stoichiometric YTi_xO_y structure.

Zoom In Zoom Out Reset image size

To understand the sensing mechanism of an ISFET or EIS sensor, it is important to consider the presence of electrolyte between an insulator film and reference electrode. Moreover, Equations 1 describes the flatband voltage (V_FB) of an EIS sensor which is defined as:³⁹

where ϕ_Si is the work function of silicon, C_ox and Q_ox are the oxide capacitance and accumulated charge in the oxide, respectively, q is an elementary charge, and Q_ss is the surface charges and interface states in the oxide. The presence of electrolyte solution generates three different kinds of potential termed as E_ref is the constant reference electrode potential, χ^sol is the surface dipole potential of the solution, and ψ_o is the surface potential associated with the function of solution pH, representing the flatband voltage of an EIS sensor. The site binding theory model has been proposed to understand the change in the surface charge of an oxide insulator in a liquid electrolyte. The difference in the size and charge of ion elicits the formation of the electrical double layer at the interface of an oxide membrane and electrolyte solution. As per this model, amphoteric insulating surface exhibited the hydroxyl group which can be protonated (Si-OH₂⁺) or deprotonated (Si-O⁻) relying on the proton concentration in the buffer solution. Both protonation and deprotonation are responsible for the change in the surface potential of an oxide insulator. For this phenomenon, H⁺ and OH⁻ ions are called potentially determining ions. This model was widely accepted for an EIS structure, however, the protonation and deprotonation are described as follow:

where [H⁺_s] ion is the concentration of proton, which exists at the surface of an oxide insulator, and [SiOH₂⁺], [SiO⁻] and [SiOH] are the binding site for proton concentration near the interface.

According to Boltzmann distribution, [α] is a concentration of ion species which exist at location i in the electrical double layer. [α]_i compared to bulk concentration [α]_b is described as follow:

where k_B is Boltzmann constant and T is the absolute temperature. In chemical reactions, the surface potential of the electrolyte is related to the charge density accumulated on insulator's surface. The net surface charge density (σ₀) is given by

While the total density of binding sites is

The obtained voltage (ψ₀) dependent on the sensing membrane material and the pH value of an electrolyte solution is explained by using SiO₂ and Al₂O₃ two surface models proposed by bousse et al., hence combining Equations 2–6, the calculated surface potential is:⁴⁰

where pH_pzc is the pH value with zero charge and β is the dimensionless pH sensitivity parameter which determines the chemical sensitivity of sensing membrane and depends upon the density of surface binding site as showing in the Equation 8:

where C_DL is the double layer capacitance (based on Gouy-Chapman-stern model which the value is 20 μF/cm²). According to the theoratical model if β >> 1, the maximum sensitivity of an EIS sensor is 59.2 mV per unit of pH (25°C), which termed as Nernstian response. Apparently, from Equation 8, it is theorized that the pH sensitive performance and linearity of the sensing membrane are proportional to the number of the surface site per unit area. However, the density of surface site is firmly associated with a surface roughness of the film. Change in the H⁺ ion concentration results in the V_FB shift in the capacitance-voltage (C-V) curve.

In order to investigate the effect of different annealing temperatures (700, 800 and 900°C) on the sensing characteristics of YTi_xO_y sensing membranes, we quantified C-V curves at pH ranges from pH 2 to 12, as shown in Figs. 4a–4c. From the obtained C-V curve, the V_FB shifted toward the negative side as the pH of buffer solution increased due to the lowering of surface potential as described in site binding model previously. To measure the pH sensitivity of YTi_xO_y based EIS sensor, the significant change in the voltage region at 50% of maximum capacitance at each buffer solution compared to detect the corresponding hydrogen ions concentration. However, to calculate the linearity of YTi_xO_y EIS devices which fabricate with different annealing temperatures, reference voltage (V_REF) as a function of pH was plotted in the inset of Figs. 4a–4c_. The pH sensitivities of YTi_xO_y sensing membranes after annealing at 700, 800 and 900°C were determined to be 56.16 ± 0.41, 58.52 ± 0.17 and 55.91 ± 0.11 mV per unit of pH, respectively. Among these tested devices, the YTi_xO_y EIS device annealed at 800°C exhibited the highest sensitivity. The possible mechanism of the high sensitivity could be associated with the surface roughness of YTi_xO_y sensing membrane as observed in previous AFM analysis (see in Fig. 2c). With increment in the surface roughness of YTi_xO_y sensing membrane, it effectively increases the total surface area at insulator/electrolyte interface, which results in sensing membrane accommodating more charge (H⁺ or OH⁻) on its surface, hence pH sensitivity was improved at 800°C devices as compared to other samples. Moreover, 900°C-annealed device exhibted a lower sensitivity compared to 800°C in spite of having higher surface roughness, because at 900°C a solid state reaction takes place in which yttrium and silicon atom intermixed with each other, hence leading to the formation of a thicker silicate layer with a lower dielectric constant.⁴¹

Zoom In Zoom Out Reset image size

For the development of solid-state EIS sensor devices, reliability and stability are the most important factor have to be considered. In order to evaluate the stability of an EIS sensor, hysteresis phenomena have been performed. The hysteresis effect is also known as a non-ideal effect, which is mainly due to the slow interaction between the interior surface sites and ions in electrolyte solution. These interior surface sites are presented beneath the sensing surface, which mostly are defects. The hysteresis voltage described as the voltage difference between the initial and terminal voltage average, when YTi_xO_y sensing membrane was submerged in an alternative cycle of pH loop 7→ 4→ 7→10→7 over a period of 25 min in each buffer solution. The hysteresis voltages of YTi_xO_y based EIS sensors annealed at 700, 800 and 900°C are 27.6, 2.6 and 10.4 mV, respectively. Thus, the sample annealed at 800°C was shown a small hysteresis deviation (2.6 mV) in Fig. 5. Since the oxygen vacancies and various defects in sensing membrane are responsible for binding of ion on surface site of the membrane, which ultimately detains the reference voltage. Annealing at 800°C in the presence of oxygen ambient condition possibly saturates the oxygen vacancies and several defects, while increasing in annealing temperature 900°C evokes the diffusion yttrium and oxygen ion at their respective atomic positions which generate dangling bond and interfacial trap, this self-diffusion process leads to the formation of non-stoichiometric YTi_xO_y, thus is responsible for a higher hysteresis voltage compared with 800°C-annealed film.

Zoom In Zoom Out Reset image size

For the long-term measurement process, drift rate is a very essential parameter to examine the stability of EIS sensor devices. When EIS sensor is used for the detection of pH or biological molecule in the electrolyte solution, a comparatively slow chemical alteration occurred on the sensing membrane surface due to its vulnerability in the electrolyte solution. These chemical alterations occurred in the sensing membrane surface are responsible for the modification in the density of surface binding site after a period of time. This complete phenomenon has a deleterious impact on the overall capacitance of sensing membrane, which resembles the relatively slow, monotonic and temporal change in measurement circuit and the reference voltage. Impact of annealing temperature (700, 800 and 900°C) on the drift characteristics of YTi_xO_y based EIS sensors depicts in Fig. 6. The samples were submerged in pH 7 solution for 12 h. Change in the reference voltage (ΔV_REF) is described as:

The drift rates of YTi_xO_y EIS devices after annealing at 700, 800 and 900°C were found to be 1.98, 0.10 and 0.54 mV/h, respectively. This obtained result indicates that the sample annealed at 800°C had a smaller drift as compared to other annealed samples. The smaller drift rate at 800°C is possibly attributed to the presence of a lower crystal defect, oxygen vacancy, and dangling bond, accountable for trapping of H⁺ and OH⁻ ion on the surface in the electrolyte solution, thus this optimal temperature prevents the clustering of ions. However, a larger drift occurred at high annealing temperature (700°C) rate may be due to more C-O bonds in the film to enhance the formation of crystal defect and buried surface sites, which are also known as trap centers or trap sites. This trapping causes the reference voltage shift over a period of time, which may worsen the drift response. In this investigation, we found that the sol-gel derived YTi_xO_y EIS sensor displayed better pH sensitivity performances (sensitivity, hysteresis voltage and drift rate) as compared to other ISFET or EIS sensors composing HfO₂ and ZrO₂ (55 mV per unit of pH, NA and NA),⁴² Ta₂O₅ (56.19 mV per unit of pH, NA and NA),¹⁵ WO₃ (50.2 mV per unit of pH),⁴³ IGZO (53.3 mV per unit of pH, NA and NA),⁴⁴ SiO₂ (52.4 mV per unit of pH, 2.3 mV and 3 mV/h),⁴⁵ Al₂O₃/SiNw Nano-ribbon (53.3 mV per unit of pH, NA and NA),⁴⁶ IrO_x (51.7 mV per unit of pH, 16.5 mV and 0.3 mV/h),⁴⁷ and SnO₂ (57.36 mV per unit of pH, 2.9 mV and 6.73 mV/h)⁴⁸ sensing membranes.

Zoom In Zoom Out Reset image size