Background

Recently, hydrogen peroxide (H2O2) is a major intermediate of biological cycles which has been used as a potential biomarker for oxidative stress diagnosis as well as a major catalyst for immune sensing [1, 2]. On the other hand, it is also an essential compound of bleach industries and waste water treatment. H2O2 has a major role in modulating mitochondrial function by inhibiting activities of the mitochondrial enzyme in a fully reversible fashion [3, 4]. The H2O2 sensing assay relies on the use of the enzyme horse radish peroxidase (HRP) to oxidize its substrates and detection using spectrophotometer [5]. H2O2 sensing in a simple way, with a short time detection with high specificity, is demanded for future disease diagnosis of the human body, and enzyme-free electro-catalytic methods have gained the attention for H2O2 sensing. Therefore, various catalysts such as metal, metal oxides, and redox polymers have been reported to detect H2O2 [612]. Huang et al. [13] have used the glassy carbon electrode modified by Si nanowire-dispersed CuO nanoparticle. Maji et al. [14] have demonstrated an amperometric H2O2 sensor based on reduced graphene oxide-coated silica modified with Au nanoparticles. Wang et al. [15] have developed a H2O2 sensor by using MoS2 nanoparticles. Sun et al. [16] have reported a dumbbell-like Pt-Pd-Fe3O4 nanoparticle-modified glassy carbon electrode which shows electro-catalytic reduction. Liu et al. [17] have reported an amperometric H2O2 sensor based on a Si substrate modified with carbon nanotube microelectrode coated by Pd nanoparticles. Kong et al. [18] have reported a non-enzymatic H2O2 sensor based on a Co3O4 nanowire grown over a reduced graphene oxide sheet. Hao et al. [19] have developed an amperometric H2O2 sensor based on Fe2O3 nanoparticles. Bai et al. [20] have reported a sensor based on carbon dot-decorated multi-walled carbon nano-composites. Silver (Ag) nanowire [21] and nanoparticle-decorated graphene [22] have been also reported for H2O2 sensing. Most of the above groups have used different materials using cyclic voltammetry/amperometric methods to sense H2O2 (ranging from few nanomolars to millimolars) due to different oxidation states in the presence of H2O2. On the other hand, high-k materials such as Al2O3 [23], Ta2O5 [24], and HfO2 [25] in an electrolyte-insulator-semiconductor (EIS) structure have been reported for pH sensing only; however, the Gd2O3 materials that have been reported are few [26, 27], and even then, there is no report for enzyme-free H2O2 sensing by using a GdO x (x < 1.5) material in a simple EIS structure. In this paper, detection of a pH and enzyme-free H2O2 sensing mechanism has been investigated by using a GdO x membrane in a simple EIS structure for the first time. Polycrystalline grain, Gd content, and oxidation states (Gd2+/Gd3+) have been confirmed by transmission electron microscope (TEM), X-ray photo-electron spectroscopy (XPS), and electron energy loss spectroscopy (EELS) on grain and boundary regions. The 15-nm-thick GdO x membrane detects H2O2 whereas both 3-nm-thick GdO x and bare SiO2 membranes do not sense H2O2. Due to the strong catalytic activity of Gd, a low detection limit of 1 μM is obtained. Both time- and concentration-dependent H2O2 sensing and its mechanism have been investigated.

Methods

p-type 4-in. Si (100) wafer was cleaned by the Radio Corporation of America (RCA) process. Prior to thermal growth of SiO2, HF dip was used to remove native oxide from the surface. After the cleaning process, a 40-nm-thick SiO2 layer was grown as an insulating layer by dry oxidation process at 950 °C. Then, the back-side-grown SiO2 layer was removed by using a buffer oxide etching (BOE) solution. To fabricate the EIS chip, a 300-nm-thick Al film was deposited on the back side of the Si wafer. The sensing membrane area was defined by standard photolithography process using a negative photoresist-SU8. Then, EIS devices were attached on a printed circuit board having copper lines. An epoxy layer was used to encapsulate the EIS structure and the copper line. Therefore, a sensor (S1) using SiO2 membrane was fabricated. Our fabrication process of EIS structure can be found elsewhere [28]. This SiO2 sensing membrane was modified by deposition of 3-nm- (S2) and 15-nm-thick (S3) GdO x films. The GdO x film was deposited by electron beam evaporation. The Gd2O3 granules were used during deposition, and the deposition rate was 6 nm/min. A schematic view of the Gd2O3- (or GdO x (x < 1.5)) modified SiO2 sensor is shown in Fig. 1. To probe the thickness and microstructure of GdO x films, low-voltage spherical aberration corrected field emission TEM (Cs-corrected FE-TEM) was performed. The model number is JEOL JEM-ARM200F with accelerating voltages of 60, 120, and 200 kV. In addition, a Cs-corrected FE-TEM Oxford energy spectrometer (energy-dispersive spectroscopy, EDS) and electron loss EDS (EELS, Model 965 QuantumERTM) were used to observe the elemental composition on polycrystalline grain and boundaries. The ambient temperature of our laboratory was 21 ± 3 °C and relative humidity was 50 ± 10 %. The elemental composition was investigated by XPS analyzing chamber. The vacuum of the XPS chamber was 1 × 10−9 Torr. The spectra were recorded by using an Al K∝ monochrome X-ray at an energy of 1486.6 eV. The scanning energy range from 0 to 1350 eV was used. All spectra were calibrated by C1s spectrum at a centered peak energy of 284.6 eV. After depositing the GdO x films on the SiO2/Si substrates, the samples were transferred immediately to the XPS chamber. The capacitance-voltage (C-V) measurements were performed by using Agilent 4284A LCR meter and an Ag/AgCl reference electrode was used. The measurement frequency was 100 Hz. The sweep voltage was applied on the Ag/AgCl electrode. The reference voltage (Vr) was measured at 50 % of accumulation capacitance.

Fig. 1
figure 1

Schematic view of our pH and H2O2 sensor using Gd2O3 (or GdO x (x < 1.5)) membrane and demonstration of H2O2 sensing mechanism

Results and Discussion

Figure 2 shows the cross-sectional TEM images of the S2 and S3 sensors. The thickness of SiO2 is 41.2 nm (Fig. 2a), and the thickness of the GdO x film is 3.3 nm (Fig. 2b). The TEM image of the S3 sensor shows that the thickness of SiO2 is 41.5 nm (Fig. 2c) and the thickness of the GdO x film is 14.8 nm (Fig. 2d). Therefore, the thickness of SiO2 is 40 ± 2 nm and the thickness of GdO x is 15 ± 0.5 nm. The thicker GdO x film shows clearly polycrystalline grains and its boundary [29, 30], which will help to detect H2O2. Elemental composition of the SiO2 and GdO x films is observed by XPS, which is shown in Fig. 3. The peak binding energy of Si2p spectra for the S1 sample is 103.35 eV (Fig. 3a), which is similar to the reported value of SiO2 at 103.58 eV [31]. The spectra are fitted by Shirley background subtraction and Gaussian/Lorentzian functions. The Si2p spectrum shows one characteristic peak after de-convolution. Similarly, one characteristic peak of O1s centered at 531.5 eV is also observed (Fig. 3d). Lower values of full-width half-maximum (FWHM) are found to be 1.84 and 1.64 eV for the Si2p and O1s spectra, respectively. The ratio of O:Si is 1.84, which signifies the stoichiometric SiO2. An XPS spectrum of GdO x shows Gd3d 3/2 and Gd3d 5/2 doublet with binding energy of 1220.5 and 1188.3 eV, respectively (not shown here). However, peak binding energies of Gd3d 3/2 and Gd3d 5/2 spin-orbits are reported as 1218 and 1186 eV, respectively [32]. XPS spectra of Gd3d 5/2 core-level electrons are 1189 eV for S2 (Fig. 3b) and 1188.7 eV for S3 (Fig. 3c) samples, which are identified to be Gd2O3 3d 5/2 or Gd2O3 films. Corresponding lower binding energy peaks at 1186.2 and 1185.8 eV indicate the metallic Gd3d 5/2 peaks for the S2 and S3 samples, respectively. The area ratios of Gd/Gd2O3 are found to be 0.64:1 and 0.69:1 for the S2 and S3 samples, respectively, which show higher percentage of Gd in the S3 samples owing to polycrystalline grains. However, the O1s core-level spectra show three distinct peaks for the S2 (Fig. 3e) and S3 (Fig. 3f) samples. The strong peaks at 531.5 eV correspond to the oxygen in the Gd2O3 film, whereas lower (O1s A) and higher (O1s B) binding energy peaks centered at 529 and 532.9 eV are attributed to the hydroxyl (OH) and carbonate groups in Gd2O3 films, respectively [33, 34]. Moreover, the lower binding energy peak corresponds to Gd-O bonding or GdO x [35]. The area ratios of O1s A and O1s B with respect to O1s are 0.04:1 and 0.48:1 for the S2 samples whereas those values are 0.08:1 and 0.1:1 for the S2 samples, respectively. Therefore, the S2 samples show higher percentage of O1s B owing to higher carbonate groups in the GdO x films, which is insensitive to H2O2 sensing. On the other hand, the S3 samples have higher percentage of O1sA owing to higher OH and higher Gd content in Gd2O3 film, i.e., GdO x film. So, oxygen can be bonded loosely with Gd on a polycrystalline grain boundary as well as a thicker GdO x film will help to sense H2O2, which will be explained below.

Fig. 2
figure 2

TEM images of a 3-nm-thick GdO x membrane on 40-nm-thick SiO2 layer (S2) and b zoom in view of a. TEM images of c 15-nm-thick GdO x membrane (S3) and d zoom in view of c. Thicker membrane shows clear polycrystalline grain

Fig. 3
figure 3

XPS characteristics of Si2p for a S1, b S2, and c S3 samples. Corresponding O1s spectra of d S1, e S2, and f S3 samples are shown. The S3 film shows higher Gd/Gd2O3 ratio or oxygen deficient and higher OH group which helps to sense H2O2

Figure 4a shows the C-V characteristics with pH values from 6 to 10 for the S2 and S3 sensors. The Vr values of the S2 sensors are −0.84, −0.75, and −0.63 V for pH 6, 8, and 10, while those values are 0.01, 0.1, and 0.23 V for the S3 sensors, respectively. The Vr values of the S3 sensor are shifted towards the positive direction and are lower than the Vr values of the S2 sensors. This is due to lower oxide charges for the thicker GdO x membrane (55 vs. 43 nm [36]) and polycrystalline grains with higher OH ions (Fig. 3f). The pH sensitivity values are found to be 51.2 and 54.2 mV/pH for the S2 and S3 sensors, respectively, which are higher than the pH sensitivity of approximately 35 mV/pH from pH 2 to 10 [28, 37] and 42 mV/pH from pH 6 to 10 for the S1 sensors. The pH sensitivity of a 30-nm-thick GdO x membrane is approximately 51.7 mV/pH (not shown C-V curves), which is slightly lower than the S3 sensors. The pH sensitivity value of our GdO x membrane is comparable with other reported values of 48.29 mV/pH by Wang et al. [27], 64.78 mV/pH by Chang et al. [38], and 55 mV/pH by Yang et al. [39]. However, the S3 sensors show the lowest drift rate as compared to the S1 and S2 sensors (2.12 mV/h vs. 3.12 mV/h and 2.16 mV/h), as shown in Fig. 4b. The drift characteristics were measured a long time up to 500 min at pH 7 buffer solution. Considering a low drift rate (2.12 mV/h), the pH detection limit of the S3 sensors is 0.039 pH, which is due to high pH sensitivity. It is interesting to note that the GdO x membrane will detect H2O2. Figure 4c shows the time-dependent response of H2O2 for the S3 sensors. A negligible Vr shift is observed for pH 7 buffer solution up to 10 min. By including H2O2 with a concentration of 1 μM, a good Vr shift of approximately 40 mV is observed because of Gd1+, Gd2+, and Gd3+ oxidation states (https://en.wikipedia.org/wiki/Work_function) [40]. On the other hand, both S1 and S2 sensors do not show H2O2 sensing. When in contact with H2O2, the Gd2+ changes to the Gd3+ oxidation state and provides electrons for the reduction of H2O2. H2O as a byproduct is observed (Fig. 1). However, the pH value is unchanged by adding H2O2 in the buffer solution. A short response time of <2 min is needed without enzyme. After washing out, the sensor does not show any Vr shift owing to the reduction from the Gd3+ to Gd2+ states. Therefore, this sensor can be used repeatedly for H2O2 sensing. Based on our knowledge, this is the first ever report of H2O2 detection with a polycrystalline GdO x membrane. Basically, the oxidation/reduction of the GdO x material in contact with H2O2 with buffer solutions is responsible for the Vr shifting, which is shown by chemical reactions below.

Fig. 4
figure 4

a C-V characteristics of the S1, S2, and S3 sensors are shown. b Drift rate characteristics of the S1, S2, and S3 sensors. c Time-dependent response of H2O2 and d reference voltage shift vs. H2O2 concentration for all sensors

$$ \mathrm{G}\mathrm{d}\leftrightarrow {\mathrm{Gd}}^{2+}+2{e}^{-}\leftrightarrow {\mathrm{Gd}}^{3+}+3{e}^{-} $$
(1)
$$ {\mathrm{H}}_2{\mathrm{O}}_2+{e}^{-}\leftrightarrow {\mathrm{O}\mathrm{H}}^{-}+{\mathrm{O}\mathrm{H}}^{\ast } $$
(2)
$$ {\mathrm{OH}}^{\ast }+{e}^{-}\leftrightarrow {\mathrm{OH}}^{-} $$
(3)
$$ 2{\mathrm{OH}}^{-}+2{\mathrm{H}}^{+}\leftrightarrow 2{\mathrm{H}}_2\mathrm{O} $$
(4)

By following the above Eqs. (1), (2), (3), and (4), the oxidation state of Gd changes from Gd2+ to Gd3+. The H+ ions are supplied by buffer solutions. The Vr shift increases with increasing H2O2 concentration from 1 to 200 μM because the generation of Gd3+ ions increases (Fig. 4d). A moderate sensitivity of 0.13 mV/μM is obtained from a linear range of 1 to 200 μM whereas it is 82 mV/μM from a linear range of 0.5 to 1 μM. Our detection limit of 1 μM is inferior than the published results [912, 15, 16, 4143], comparable with the published results [4447], and superior than the published results [13, 17, 18, 20, 4852] in literature by using different sensing methods, as shown in Table 1. Further study is needed to improve the detection limit in the future. However, our sensing method’s surface potential is changed when in contact with H2O2 because of the catalytic activity of Gd. It is known that Gd2O3 material is n-type and the energy difference in between the Fermi level and the conduction band (Ec) is 2.71 eV [53]. The electron affinity of Gd2O3 is 1.45 eV by considering the conduction band offset of 2.6 eV with Si [54]. The work function of Gd increases from 2.9 eV (https://en.wikipedia.org/wiki/Work_function) to 4.16–4.76 eV [5355] after oxidation. This suggests that the work function of GdO x is modulated by oxidation/reduction or Gd3+ concentration as well as the energy band bending of Si is changed. In consequence, the Vr is needed to bring Si energy bands to be flat. On the other hand, the S1 and S2 sensors do not show H2O2 detection because they do not have redox properties. The thinner GdO x film (S2) has a smaller crystalline grain with less Gd content (Fig. 3), while the S3 sensor has larger crystalline grain (Fig. 5a) with higher Gd content. Figure 5b shows electron energy loss spectroscopy of Gd measured at polycrystalline grain (P1) and amorphous region or grain boundary (P1). The regions of P1 and P2 are marked on Fig. 5a. The edges of the Gd M-4 and M-5 peaks at the P1 region are located at 1216.8 and 1187.5 eV, while those values at the P2 region are 1216.5 and 1187 eV, respectively. Du et al. [56] have reported the M-4 and M-5 peak values of 1217 and 1185 eV for the Gd(OH)3 nanorods. The edges of the O-K peak at both P1 and P2 regions are located at 538.5 eV, as shown in Fig. 5c, which is close to the reported value of 536.5 eV [56]. It is interesting to note that another peak of crystalline grain (P1) is located at 532.9 eV, which is shifted downwards to 3.9 eV. Egerton has reported the reduced energy gap of SiO x at the SiO2/Si interface with energy shift downwards to 3 eV [57]. In our case, this reduced energy gap is observed in the polycrystalline grain region. Therefore, the crystalline grain is GdO x (or Gd2+) and the amorphous region or grain boundary is Gd2O3 (or Gd3+). When in contact with H2O2, the oxidation state of the S3 sensor changes from Gd2+ to Gd3+ and the crystalline grain takes a major role, which is confirmed by EELS spectra. So, the thicker crystalline GdO x membrane can sense H2O2 repeatedly which will be useful to detect human disease in the near future.

Table 1 Comparison of linear range and detection limit of H2O2 published in literature [913, 1518, 20, 4152]
Fig. 5
figure 5

a TEM image for EELS spectra of the S3 membranes. The edges of b Gd and c O-K are plotted for the P 1 and P 2 regions marked on a. The polycrystalline grain corresponds to Gd2+ and the grain boundary corresponds to Gd3+ oxidation states

Conclusions

Higher pH sensitivity (54.2 m/pH) and the enzyme-free H2O2 sensing characteristics have been investigated by using 15-nm-thick GdO x membranes for the first time. The polycrystalline grain and thickness of the GdO x /SiO2 film have been observed by TEM image. XPS characteristics of the S3 membrane show higher Gd/Gd2O3 ratio than the S2 membrane (0.69/1 vs. 0.64/1). The S3 membrane shows GdO x and higher OH content in the crystalline grain, which help to sense H2O2 whereas both S1 and S2 sensors do not show H2O2 detection. Therefore, a larger polycrystalline GdO x grain has oxidation/reduction properties when in contact with H2O2, which is confirmed by EELS. During oxidation, the Gd2+ changes to the Gd3+ state and the amount of Gd3+ ions increases with increasing H2O2 concentration from 1 to 200 μM. A low defection limit of 1 μM is obtained owing to the catalytic effect of Gd. The time-dependent response and the sensing mechanism of H2O2 have been explored. Due to the short time detection of H2O2 in the EIS structure, this novel GdO x sensing membrane paves a way to diagnose other diseases of the human body in the near future.