Incorporation of Cd-Doping in SnO₂

Abstract

Tuning the electrical properties of materials by controlling their doping content has been utilized for decades in semiconducting oxides. Here, an atomistic view is successfully employed to obtain local information on the charge distribution and point defects in Cd-doped SnO₂. We present a study that uses the time-differential perturbed gamma–gamma angular correlations (TDPAC) method in samples prepared by using a sol–gel approach. The hyperfine field parameters are presented as functions of the annealing temperature in pellet samples to show the evolution of incorporating Cd dopants into the crystal lattice. Additionally, the system was characterized with X-ray fluorescence, electron dispersive spectroscopy, and scanning electron microscopy after the probe nuclei ¹¹¹In(¹¹¹Cd) decayed. The TDPAC results reveal that the probe ions were incorporated into two different local environments of the SnO₂ lattice at temperatures up to 973 K for cation substitutional sites.

1. Introduction

Stannic oxides crystalize in a rutile structure with a tetragonal lattice of space group P4₂/mnm. As these are semiconductor oxides, their bandgap is approximately E_g ~ 3.6eV Compared with other oxides, these have a higher transparency over the visible range and a higher electrical conductivity. Furthermore, stannic oxides present a good chemical stability with high melting and boiling points with low production costs. These can be applied as gas sensors, humidity sensors, and transparent electrodes for solar cells, varistors, optoelectronic devices, capacitors, flat-panel displays, high capacity anodes in lithium-ion batteries, and many other electronic device applications [1,2,3].

In this system, defects and small concentrations of impurity dopants affect the SnO₂ conductivity over several orders of magnitude, which is particularly sensitive to the presence of oxygen and interstitial Sn vacancies [1]. For example, the addition of iridium to SnO₂ improve the electrochemical properties of this material result in good electrodes for electrocatalysis [4,5,6].

The replacement of Sn⁴⁺ by a metal with a smaller valence state can induce the formation of oxygen vacancies, thus improving the sensor efficiency of SnO₂. In this sense, Cd²⁺ doping in stannic oxide decreases its resistivity and enhances its molecular sensing performance, particularly for ethanol and H₂ [7]. Sensing H₂S is enhanced by more than 20 times compared with pure SnO₂ [8]. It is clear that metal ion doping’s effectiveness and associated manufacturing processes in SnO₂ are very complex and necessitate an atomistic view. Therefore, we have applied the time-differential perturbed angular correlations (TDPAC) method with gamma rays to better understand the influence of Cd as a dopant in the proposed system.

The influence of the implantation process for Co-doped SnO₂ thin films was studied by using the TDPAC technique by presenting indirect evidence that In and Co diffuse to different depths during the annealing process [9]. The possibility of achieving electrical doping via Cd/In ion implantation in SnO₂ thin films was investigated by using TDPAC in non-pre-doped samples and ab initio simulations [10].

Experiments that were carried out with TDPAC for SnO₂ began in the late eighties. Desimoni et al. (1987) studied the incorporation of In from an ¹¹¹InCl₃ solution while forming SnO₂ from the oxidation of Sn metal during annealing treatments [11]. They reported two different local environments. The first was at the substitutional site for ¹¹¹In in the sample that was subject to an intense defect distribution centered at ω_Q = 35.6 (1.8) Mrad/s with η = 0.40. The second was assigned to a minor component characterized by ω_Q =17.2 (0.6) Mrad/s with η = 0.77 and was interpreted as small In₂O₃ precipitates. Renteria et al. (1991) performed TDPAC experiments on SnO₂ thin films by using an ¹¹¹In/Cd probe and identified the two Sn 2⁺ and 4⁺ oxidation states for different SnO and SnO₂ phases [12]. The nuclear quadrupole interactions with v_Q1 = 117(1) MHz and η = 0.18(2) were assigned to In/Cd at defect-free lattice sites of SnO₂, which agreed with previous results obtained by Wolf et al. (1986) in powder samples [13]. Later, Ramos et al. (2010) [14] and (2013) [15] reported two local environments for ¹¹¹In probe atoms in thin films and nano-structured powder SnO₂ samples, as characterized by the hyperfine parameters ν_Q1 = 115 MHz and η ∼ 0.1 that were assigned to ¹¹¹In probe atoms at substitutional Sn sites and ν_Q2 = 146 MHz and 0, 4 ≤ η ≤ 1, as assigned to defects and a disordered SnO₂ phase. Munoz et al. (2008) performed TDPAC measurements with SnO pellets after ¹¹¹In diffusion and observed a single major fraction for Cd in a neutral charge state [16]. On the other hand, the other fraction was also assigned to the cation-site, but this was done so in a charge state of q = +1. More recently, a study was published that combined the dynamic and static quadrupole interactions, and this was called an on–off model [17]. The measurements were complemented with ab-initio calculations.

This work uses the γ–γ TDPAC method to measure the hyperfine parameters in SnO₂ samples that are doped with 1.4% Cd by using a sol–gel approach, which complements previous systematic investigations of structural defects from an atomic perspective [18]. It is acknowledged that these TDPAC measurements have not been previously conducted.

2. TDPAC Methodology

This section briefly describes the TDPAC methodology; further details can be found in references [19,20]. The TDPAC technique is based on the correlation between the nuclear spin of the excited level of the probe nuclei and the emission direction of gamma radiation that is modulated by the hyperfine interactions with extra nuclear magnetic field and/or electric field gradient (EFG) from the electronic neighborhood. The electric hyperfine interaction can be written in terms of the major component of the electric field gradient (V_zz) and the asymmetry parameter η = (V_xx − V_yy)/V_zz, which describes the asymmetry of the electric field gradient (EFG) tensor. From the TDPAC measurements, it is possible to obtain the perturbation function $\mathrm{R}\left(\mathrm{t}\right)\approx {\mathrm{A}}_{22}{{\displaystyle \sum }}_i{f}_i{G}_{22}^i$ that takes into account the fractional site population f_i of probe nuclei and its respective perturbation factor $G_{22}^i\left(\mathrm{t}\right),$ which describes the hyperfine interaction. A₂₂ is the anisotropy of the γ–γ cascade of the probe nuclei that was used in the measurements. The electric quadrupole interaction perturbation factor, for polycrystalline samples while considering the nuclear spin 5/2 and $\eta \ne 0$, is given as shown in Equation (1).

$${\mathrm{G}}_{22}\left(\mathrm{t}\right)={\mathrm{s}}_0+{\displaystyle \sum }_{n=1}^3{\mathrm{s}}_{\mathrm{n}}\left(\eta \right)\cos \left[{\omega }_n\left(\eta \right)t\right]$$

(1)

Furthermore, the major component of the EFG tensor V_zz can be obtained by the observable spin dependent quadrupole frequency ${\omega }_Q$, expressed as:

$${\omega }_Q=\frac{eQ{V}_{zz}}{4I(2I-1)\hslash },$$

(2)

with Q being the nuclear quadrupole moment. The EFG is primarily determined by the electrons involved in the bonding with the nearest neighbors. For 5/2 nuclear spins, the transition frequencies are given by ${\omega }_n={\omega }_Q{C}_n\left(\eta \right)$. The coefficients $C_n$ and $s_n$ (including S₀) can be numerical calculated for a known $\eta$. The broadness of frequency distribution is parametrized with ${\delta }_{\omega }$, and, for the quadrupole electric interaction, Lorentz broadening is assumed.

3. Experimental Methods

The Cd-doped SnO₂ powder samples were prepared by using a sol–gel process. The detailed preparation procedure and the TDPAC results of the nanostructured SnO₂ powder without doping can be found in our previous work [14]. We emphasize that the addition of ¹¹¹In and Cd was performed before the gel formation with a 1.4% Cd content. The thermal treatment under a nitrogen atmosphere was performed for 10 h prior to the TDPAC measurements. The system was further characterized at the Laboratório de Caracterização Tecnológica (LCT), University of Sao Paulo, Brazil, after the probe nuclei ¹¹¹In(¹¹¹Cd) decayed by using X-ray diffraction (Philips Xpert and Malvern Panalytical Ltd., Kassel, Germany), electron dispersive spectroscopy (EDS) (FEI (Quanta 650FEG), Hillsboro, OR, USA), and scanning electron microscopy (SEM) (FEI (Quanta 650FEG), Hillsboro, OR, USA), for the secondary electrons (SE) mode.

4. Experimental Results

Figure 1 shows the ¹¹¹In(¹¹¹Cd) TDPAC R(t) spectra for the 1.4% Cd-doped SnO₂ powder samples; these measurements were carried out at room temperature (RT). The measurements were performed as a function of annealing temperature from 295 to 973 K in vacuum. The R(t) spectra attenuated much more than observed in the spectra that were obtained from the thin film and the undoped powder sample [14,15]. Figure 2 shows the temperature dependence of the corresponding measured hyperfine parameters.

The fits followed a two fractions model with probes that interacted given the two main EFG₁ and EFG₂ distributions. The primary distribution, EFG₁, was extremely damped and accounted for f₁ > 80% of the ¹¹¹In/Cd probes. The EFG₁ and EFG₂ increased with annealing temperature. The η₁ was higher than η_2. The decrease in the EFG distributions indicates that In progressively experienced a more regular local environment at higher annealing temperatures. Variations in the EFG parameters showed an irreversible annealing temperature dependence that suggests this process was due to a cumulative annealing effect. The temperature dependence of the parameters further confirms that the observed R(t) damping was primarily due to the random distribution of the 1.4% Cd atoms in the SnO₂ provided by each ¹¹¹In/Cd atom sitting at Sn substitutional sites. These experienced slightly different charge distribution environments, even if stable Cd atoms were several unit cells away from the radioactive species.

The EFG₂ was observed to be nearly undamped after annealing at 873 and 973 K, which presented the minority fraction and was consistently assigned to lattice sites of an undoped SnO₂ lattice. We believe that the low fraction of EFG₂ occurred due to inhomogeneities of the non-radioactive Cd distribution in the sample.

A further indication of a continued diffusion of stable Cd was the decreases in the η, δ₁ and δ₂ parameters with the corresponding increase of f₁. This reveals that stable Cd was more uniformly distributed and into zones where it was not previously seen, such as a more regular local environment.

Figure 3 (top; left and right) shows micrographs for the secondary electrons (SE) mode of the powder samples under a high vacuum at two magnifications (scale bars in the figure) taken at a 10 kV accelerating voltage. The bottom figure is an EDS spectrum with no indication that Cd exists in the analyzed region. A Pt cover was used to increase the conductivity for the SEM imaging. The 1.4% Cd doping was only detected by using the X-ray fluorescence technique. A spectrometer (Axios Advanced from PANalytical) was used, as it can identify chemical elements from beryllium to uranium with a 10 ppm detection limit.

Figure 4 shows the XRD patterns that suggest the SnO₂ followed a characteristic rutile crystalline structure without other detectable phases. The nanoparticle sizes were obtained from the Debye–Scherrer equation as 26 nm for the Cd-doped samples.

Table 1. Lattice parameters obtained from the XRD technique for pure SnO₂ samples [21] or doped with Cd by using the sol–gel method.

Sample	a = b (Å)	c (Å)
1.4% Cd	4.7424	3.1873
SnO2 Ref. [21]	4.7374	3.1864

shows the obtained lattice parameters, which are close to those obtained as reported by Bolzan et al. (1997) [21]. The slight difference in the lattice parameters was an effect of the Cd doping that shifted the distance between the crystallographic planes. The expansion observed in the Cd-doped SnO2 can be ascribed to the replacement of Sn atoms by Cd due to the larger ionic radius of Cd²⁺ than that of Sn⁴⁺, as reported by Ahmad et al. [22].

The TDPAC technique with ¹¹¹In(¹¹¹Cd) or other atomic probes provides increasingly useful and important information of local environments in condensed matter systems [23]. Additional experiments with ^111mCd can be performed to better understand the influence of electron capture nuclear transmutation for the ¹¹¹In probe of the examined samples. The measurements could be subjected to the electron conversion process, which is part of the ^111mCd probe decay. This probe can be implanted at ISOLDE-CERN [19] within their solid-state physics program [24]. Moreover, it is possible to perform simultaneous ¹¹¹In and ^111mCd implantations [25].

5. Conclusions

It is clear from the TDPAC experiments that the atomistic behavior of Cd dopants strongly depends on the sample doping process. The Cd content, crystal lattice structure, and sample morphology were confirmed by using conventional characterization techniques. Based on the thermal treatments, one local environment was assigned to the Cd atoms at substitutional Sn sites, which were nearly free of point defects in their local neighborhoods. Segregation was not observed for the Cd-doped powder samples with an apparent increase in the stability of the incorporated Cd as a function of annealing temperature. The selected thermal treatment could play the role of decreasing the oxygen vacancies and reducing the mobility of Cd atoms. This enables the possibility of achieving electrical doping via the sol–gel method.