Effect of the dopant on the structural and hyperfine parameters of Sn0.95M0.05O2 nanoparticles (M: V, Mn, Fe, Co)

In this work, we present a comparative study of the structural and hyperfine parameters on tin dioxide nanoparticles doped with different transition metal elements. The nanopowders with the stoichiometry Sn0.95M0.05O2 (M: V, Mn, Fe, Co) were synthesized by simple co-precipitation method. The characterization was carried out by conventional x-ray diffraction technique, transmission 119Sn Mössbauer spectroscopy and x-ray photoelectron spectroscopy (XPS). The effect of dopant element on structural parameters of tin dioxide and, particularly, on hyperfine parameters of Sn was analysed. It was found that dopants, except for Fe, were in two valence states. For the Mn-doped SnO2 sample, it was found the strongest influence of M cation on Sn hyperfine parameters, whereas Co-doped sample resulted in the least altered one as compared to doped and un-doped SnO2. We propose an explanation of the changes of the hyperfine parameters observed over the modification of the structural changes, based on the size of the dopant elements, whose oxidation states were identified by XPS. Additionally, it should not be discarded the relevant role of oxygen vacancies whose presence on the surface was indirectly witnessed by XPS.


Introduction
Tin dioxide (SnO 2 ) is a wide band-gap (about of 3.6 eV for the bulk material at room temperature) n-type semiconductor and a transparent conducting oxide, which has many technological applications related to the combination of its characteristic properties. Therefore, this system has been of great interest in research in recent years. At ambient pressure, SnO 2 has a tetragonal rutile-type structure (low pressure phase, space group P4 2 /mnm). This structure is composed of chains of edge-sharing SnO 6 octahedral units lying along the c-axis with corner-sharing cross linkages along the perpendicular directions.
The incorporation of dopants in the SnO 2 system produces important changes in its properties. For instance, the use of transition metal (TM) elements as dopant has attracted a great attention because of interesting magnetic properties of the doped semiconductor, which makes it a promising candidate as a diluted magnetic semiconductor (DMS) [1,2].
From other approach, regarding its application as a gas sensor, the incorporation of TM elements changes its sensitivity, selectivity and time response [3]. Besides TM elements, trivalent rare earth (RE) elements are used as dopants because they provide advantageous features without altering the main properties that define the system as a transparent-conducting oxide [4][5][6]. Additionally, incorporation of a RE element can induce highly attractive optoelectronic properties positioning these materials as possible candidates for the development of different devices such as LEDs and lasers [7].
It is well known that materials in the nanoscale range can exhibit unique properties compared to their bulk counterparts mostly because of the significant ratio surface/volume. In this connection, differences between surface properties and the ones corresponding to the core, are usually found [8]. For instance, conductivity of SnO 2 is very sensitive to the surface state, as well as absorption properties and reactivity [9].
Therefore, a combination of both surface-and bulk-sensitive techniques can result very fruitful giving a more complete information, especially structural, about a nanoparticle (NP) system under research. In this work, we have unified the potentiality of two spectroscopic techniques, 119 Sn Mössbauer spectroscopy in transmission geometry and x-ray photoelectron spectroscopy, in addition to the conventional x-ray diffraction technique. We have identified the chemical state of the 3d dopants and their impact on SnO 2 NP structural parameters and also on the hyperfine parameters of tin in the host matrix.

Experimental
Sn 0.95 M 0.05 O 2 nanoparticles (M: V, Mn, Fe, Co) were prepared by the wet chemical co-precipitation method following the procedure described elsewhere in [10]. Here, we only recall the chlorides of the transition metals used for the syntheses: VCl 3 , MnCl 2 , FeCl 3 and CoCl 2 .
Energy dispersive spectroscopy with Oxford Instruments X-MaxN 50 SDD was carried out in order to get insight of the elemental analysis. Conventional x-ray diffraction was carried out with a Rigaku D/max diffractometer equipped with a vertical goniometer, using a Bragg-Brentano geometry (θ-2θ coupled arms) and Cu-Kα radiation in the 20°2θ80°range, measuring with a 0.05°step and sweeping with a 0.4°/min velocity.
The samples were analyzed by Mössbauer spectroscopy, at room temperature, using the 23.875 keV γradiation from a Ca 119m SnO 3 source with transmission geometry. In all the samples, the isomer shift is reported relative to CaSnO 3 at 300 K. The calibration was carried out employing the 57 Fe 14.4 keV γ-radiation from a 57 Co(Rh) source with sodium nitroprusside as absorber.
FlexPS equip (Specs) was used during the XPS measurements employing a monochromatic Al K α (hν=1486.6 eV) source, 100 W power and 10 kV voltage. The scans were acquired with a pass-energy of 20 eV, 20 scans per sample. The instrument work function was calibrated using C1s photoelectron. Spectra were analyzed using Casa XPS software. Gaussian (70%)-Lorentzian (30%) profiles were used for fitting core levels. A standard Shirley background was used for all reference sample spectra.

Elemental analysis of the doped SnO 2
The elemental analysis of the doped samples was provided by energy dispersive x-ray spectroscopy (EDX) in order to corroborate their chemical composition. Meanwhile the corresponding results of Fe-and V-doped SnO 2 are mentioned in [11,12], respectively, here we show EDX results of Co-and Mn-doped SnO 2 samples. The values of 3.8 5 at% for Co and 5.2 9 at% for Mn were estimated being in each case the average of eight inspected zones (for example, figure 1, corresponding to Co-doped SnO 2 ). Figure 2 shows XRD patterns of Mn-and Co-doped samples, whereas the corresponding ones of Fe and V-doped samples were shown elsewhere [11,12]. In all the cases, cassiterite SnO 2 , which is isomorphic to rutile (space group P4 2 /mnm) was identified by the diffraction peaks presented in the shown patterns. The absence of secondary segregated phases in the XRD spectra is a good evidence that the doping element was incorporated into the SnO 2 lattice. Rietveld refinements of the experimental diffraction patterns allowed extracting the valuable structural properties of the system under research. In table 1 we show the goodness of the Rietveld refinements.

X-ray diffraction technique and transmission 119 Sn Mössbauer spectroscopy
On base of the extracted lattice parameters a and c, the volume of the cassiterite lattice (with the individual dopants) was calculated and compared with un-doped SnO 2 one (figure 3). The crystallite size of the doped-SnO 2 samples was under 20 nm. As it can be seen, in all the cases, except for Co-doped SnO 2 , doping SnO 2 leads to contraction of the rutile lattice.
Mössbauer spectroscopy is a very powerful technique for studying the local environment of Sn atom in Snsolid samples because one of the Sn isotopes, 119 Sn, is one of the best-suited probes for Mössbauer experiments. The local environment of tin can be described by means of the hyperfine interactions that the atom probe reveals in a solid matrix: isomer shift (IS) and quadrupole splitting (QS), while no magnetic splitting was detected.
Isomer shift is manifestation of electric monopole interaction between nucleus and its environment being the product of the nuclear charge distribution and the electronic charge density at the nucleus. IS gives a valuable   information about chemical state of the atom probe, that is, oxidation state. QS is manifestation of electric quadrupole interaction between the nuclear quadrupole moment and the tensor of the electric field gradients at the nucleus. Generally, QS is an indicator of the local symmetry around the atom probe. Mössbauer transmission spectra of the doped-SnO 2 samples are very similar among themselves and were fitted to a doublet, characteristic of a quadrupole electric interaction at nucleus of Sn 4+ as expected for this type of material [11,13]. In figure 4, we present only a Mössbauer spectrum (MS) of Mn-doped SnO 2 powders, whereas the fitted Mössbauer parameters of all the samples are gathered in figure 3 (including un-doped SnO 2 taken as a reference).
As it can be noticed, doping SnO 2 with any of four elements influences the values of the hyperfine parameters of the atom probe. It means that Sn is highly sensitive to the presence of the dopant in the samples. This can be another proof of the dopant incorporation to the host SnO 2 matrix. What is interesting to note is that Mn-doped SnO 2 revealed the maximum deviations from the hyperfine parameters of pure SnO 2 : the highest values of isomer shift and quadrupole splitting. On the other hand, Co-doped SnO 2 resulted to have a less notorious influence on tin.

XPS studies Dopants (V, Mn, Co)
We have also investigated the oxidation state of the dopants incorporated in SnO 2 structure by x-ray photoelectron spectroscopy (XPS). Relative quantity of different oxidation states for one element can vary from a surface to a core, being generally the state with highest oxidation state more populated on the surface than in the core.
It was revealed (figure 5) that V, Mn and Co were in two distinct chemical states. In particular, vanadium was identified as V 4+ and V 5+ , manganese as Mn 3+ and Mn 4+ , cobalt as Co 2+ and Co 3+ , oxidation state. The results of the deconvolution of the corresponding spectra (binding energies, BE) are shown in table 2, in accordance with those reported in literature [14][15][16]. Interestingly, in Sb-doped SnO 2, antimony species were in two distinct chemical state: Sb 3+ and Sb 5+ [17].

Tin
From figure 6, it can be seen that Sn on the surface of nanoparticles has two binding energies, that is, there are two slightly different environments for Sn atom. The values of binding energies BE 1 are compatible with those reported usually for Sn 4+ in pure SnO 2 [18]. Whereas BE 2 values are assigned also to Sn 4+ atom with an oxygen

Discussion
For the sake of simplicity of understanding, we have gathered the values of the ionic radii in table 4, we mention in our work, from the Shannon radius table [21].

V-doped SnO 2
Both V 4+ with the ionic radius 0.58 Å and V 5+ (0.54 Å) can explain well the decreasing of the lattice volume of V-doped SnO 2 , compared with the un-doped one. This is the reason of the intermediated value of QS (0.62 mm s −1 ) originated by the presence of vanadium in the host of SnO 2 matrix.

Mn-doped SnO 2
In accordance with the XRD results (figure 3), the cassiterite structure in the Mn-doped SnO 2 sample has the smallest lattice volume among the samples here investigated. It can be understood by the fact that the ionic radius of Mn 4+ (0.53 Å) is considerably smaller than that of Sn 4+ (0.69 Å). In case of Mn 3+ , also of a smaller size than Sn 4+ , the ionic radius can be either of 0.58 Å (for low spin, LS) or 0.645 Å (for high spin, HS). Interestingly, the tin in Mn-doped sample revealed the higher quadrupole splitting (QS=0.77 mm s −1 ) value. In other words, the local environment of tin atom is mostly distorted when SnO 2 is doped with manganese.

Fe-doped SnO 2
In case of Fe-doped SnO 2 , we assume it is dealt with Fe in 3+ oxidation state (HS). We will argument that in the next section. Its ionic radius, of 0.645 Å, which, again, is smaller than Sn 4+ ionic radius, explains the volume contraction of Fe-doped SnO 2 sample in comparison with the un-doped one.

Co-doped SnO 2
In accordance with the XRD results, Co-doped SnO 2 powders have the biggest lattice volume among the doped samples, including the un-doped SnO 2 . This is an indication of the fact that a bigger ion substitutes for Sn 4+ . In case of Co 3+ , the radii are 0.61 Å (HS) and 0.545 Å (LS). From the other hand, Co 2+ has 0.745 Å (HS) and 0.65 Å (LS). Due to these considerations, only Co 2+ (HS) can be responsible for the expansion of the lattice of Co-doped SnO 2 . This is consistent with the fact that the salt used for doping was CoCl 2 where the cobalt clearly is in Co 2+ state. We believe that in the core of the nanoparticles cobalt remains mainly in the original oxidation state, while in the surface due to the exposition to oxygen, most of the Co get oxidized achieving Co 3+ oxidation state. From Mössbauer spectroscopy, in Co-doped SnO 2 sample, the tin is at a distorted environment (QS=0.56 mm s −1 ) with a quadrupole splitting value close to the one corresponding of the un-doped SnO 2 (0.52 mm s −1 ).

General remarks
As it is well known, XPS technique is suitable for studying surfaces (nearly 3 nm deep), meanwhile XRD gives information about significantly deeper layers (microns). That means XRD technique 'sees' the whole crystallite (up to 20 nm-size), not its surface. In accordance with the model we propose, SnO 2 doped nanoparticle is a core-shell one, where the core contains the cation with lower oxidation state (for example, here, Co 2+ ), and the shell contains the cation with higher oxidation state (here, Co 3+ ). Because of that mentioned above, XPS spectra of the dopants (figure 5) do not give a whole quantitative panorama about the populations of the dopants in the doped nanoparticles. Instead, showing the surface state, it gives rather an idea about qualitative model that could take place from our point of view. It should be worth noting, the proposed model is compatible with XRD and MS results. Moreover, the value of cobalt content, extracted from EDX analysis, reinforces our suggestion about the core-shell model (where the doping elements in the core are of a lower oxidation state) for the system under investigation. Cobalt content in Co-doped SnO 2 nanopowders is somewhat lower that the nominal one, and even though that content (roughly 4 at%) was enough to have the SnO 2 lattice expanded (the only case from those investigated). That is possible only if a bigger ion (than Sn 4+ ) enters predominantly the host lattice. This is the argument in favour of the model proposed (here, Co 2+ (HS)).
Concerning over the valence state of Fe in SnO 2, we strongly believe that in 5 at% Fe-doped SnO 2 sample, there are no other irons for exception of Fe 3+ . On one hand, the synthesis of the sample implied FeCl 3 salt, where Fe is 3+. It would be improbable to expect Fe 2+ in such a situation because this would imply reduction of the oxidation state what is very unlikely in nature. So far, we do not find any chemical argument to a favour of the formation of Fe 2+ during the synthesis by co-precipitation method. On the other hand, in our experiments published previously [11], in 10 and 15 at% Fe-doped SnO 2 nanoparticles, only Fe 3+ species were detected. This last fact reinforces very strongly our initial guess on Fe 3+ oxidation state in our sample with 5 at% of Fe. If the sample with 5 at% had irons in 2+oxidation state, then the samples with more amount of the same species would naturally have more fraction of them in the same oxidation state. In those experiments, no Fe 2+ ions were detected by the Mössbauer analysis.
A higher isomer shift is related to an increase in the valence electron density (mainly, s electrons) at tin nuclei. Undoubtedly, one of the possible contributions to the value of IS is the change of the bonding length due to distinct dopant sizes substituting for Sn. Other possible contribution is well described by Aragon et al [8]. They found that the mean particle size reduction of un-doped SnO 2 leaded to an increase in the IS values, ascribing that effect to the generation of s-type electrons due to the increase of oxygen vacancies at the nanoparticle surface. Moreover, substituting Sn 4+ by divalent (for example, Co 2+ ) or trivalent (Fe 3+ ) ions leads to the generation of oxygen vacancies to maintain the charge balance. Interestingly, our un-doped SnO 2 and Co-doped SnO 2 powders have the largest size of the crystallite among all the samples (and the lowest values of the IS). So far, we cannot discard the above-mentioned factors as the ones that could influence the IS.
The increasing of the quadrupole splitting in ionic materials is usually attributed to a more distorted local environment around the probe atom. This is because QS results on the interaction between the nuclear quadrupole moment and the tensor of the electric field gradients at the nucleus. From figure 3, it can be appreciated how the dopant presence impacts the local environments of the Sn in SnO 2 matrix. The main reason is, undoubtedly, the difference in the size of the Sn and the dopant, but, also, the presence of oxygen vacancies could not be discarded as the other source of anisotropic charge density around Sn nucleus. The existence of the oxygen vacancies can be suggested by two binding energies for Sn atom as it was corroborated by the deconvolution of Sn 3d 5/2 photoelectron spectra.
Summarizing, we have shown that in sub-20 nm SnO 2 nanoparticle system when doped nominally at label of 5 at% of V, Mn, Fe or Co the dopant enters the host lattice as can be seen from XRD and MS techniques. XPS suggests the coexistence of V, Mn and Co ions in bi-valence state, where from the suggested model, the lower valence ions are inside of the nanoparticles. While the higher valence ions are more superficial ones. Besides that, XPS gives clear indication of Sn 4+ with two slightly distinct environments. One of them is in accordance with binding energy of Sn with six nearest oxygen, while the second one is ascribed to the tin atom with an oxygen vacancy nearby. The last could be the consequence of the substitution of Sn by the dopant element of the valence distinct from 4+.

Conclusions
Sn 0.95 M 0.05 O 2 nanoparticle systems (M: V, Mn, Fe, Co) were investigated by XRD, Mössbauer spectroscopy (MS) and XPS techniques in order to reveal the effect of the dopants on hyperfine parameters of tin in SnO 2 matrix. XRD showed that in all the samples studied, only the diffraction peaks of the rutile structure of SnO 2 without secondary phases were present, which is a sign that the dopants have been incorporated into the host lattice. From Rietveld analysis of the XRD patterns we took into account the modification of the cell parameters and, therefore, the unit-cell volume; this modification is a conclusive evidence that the dopant has been inserted in the lattice of SnO 2 . 119 Sn transmission MS revealed that the presence of the dopant indeed influences the hyperfine parameters of the tin in the host SnO 2 matrix: both IS and QS values. As it can be seen, there is a clear tendency for the change of the cell volume with IS and QS. With the information about the oxidation states given by XPS, we found that considering the individual dopant ionic radii (Shannon table), we could explain the modification of the lattice' size of SnO 2 and the hyperfine parameters of tin by the incorporation of the dopants.