19206
J. Phys. Chem. C 2010, 114, 19206–19213
Structural Characterization of Nanocrystalline Sb-Doped SnO2 Xerogels by Multiedge X-ray Absorption Spectroscopy V. Geraldo,†,‡ V. Briois,§ L. V. A. Scalvi,*,† and C. V. Santilli| Departamento de Fı´sica FC, State UniVersity of Sa˜o Paulo (UNESP), C.P. 473, 17033-360 Bauru, SP, Brazil, Laborato´rio de Nanomateriais, Federal UniVersity of Minas Gerais (UFMG), Belo Horizonte, MG, Brazil, Synchrotron SOLEIL, L’Orme des Merisiers, BP48, 91192 Gif-sur-YVette, France, and Instituto de Quı´mica, State UniVersity of Sa˜o Paulo (UNESP), C.P. 355, 14801-907 Araraquara, SP, Brazil ReceiVed: June 29, 2010; ReVised Manuscript ReceiVed: September 22, 2010
Multiedge XAS data are presented for Sb-doped SnO2 xerogels dried at 200 °C and fired at 550 °C, aiming to determine the location of antimony doping in the host matrix and then to understand the role that the intentional impurity plays in the structure and properties of the tin oxide based materials. Xerogel processing at 200 °C leads to the original trivalent antimony, used for the xerogel preparation to the SbV oxidation state, mainly for low doping levels ([Sb] e 4 atom %). As the doping level increases, a significant amount of antimony remains in the trivalent oxidation state. Upon firing at 550 °C, the antimony is present mainly as SbV, independent of the doping level. The analysis of EXAFS data recorded at the Sn and Sb K edges leads to the conclusion that doping with Sb for a level of less than 4 atom % favors crystallite growth, concomitant with a strong dominance of the SbV oxidation state. Besides, the pentavalent antimony is located at internal sites of the SnO2 nanocrystallites. The EXAFS spectra of the higher ([Sb] > 4 atom %) doped samples can always be fitted by a linear combination of the spectra corresponding to the SbV site in solid solution and the spectra corresponding to the sample where SbIII is chemically grafted at the surface of the SnO2 crystallite. This SbIII segregation also agrees with the low electrical conductivity reported for sol-gel deposited films, because it generates a barrier at the grain boundary, inducing a very high electron scattering and, thus, a low electron mobility. 1. Introduction Tin dioxide is a wide band gap semiconductor (3.6-4.0 eV),1,2 with a tetragonal structure of the rutile type, with many applications in optical and electronic devices. A tin dioxide thin film presents high transparency in the visible range and high reflectivity in the infrared range.3 Its processing normally gives birth to defects such as oxygen vacancies and interstitial tin atoms, which act as donors in the SnO2 matrix, increasing the electron density in the conduction band, leading to fair n-type conduction, even in the undoped form.4 However, the electrical properties of films prepared by the sol-gel dip-coating process are worse compared to those of films deposited by other techniques. This inferior performance is generally attributed to a combination of effects mainly associated with nanosized crystallites of sol-gel materials.5-7 A considerable density of crystallites is present in the material, which makes electron scattering at grain boundaries the most relevant mechanism to film conductivity.6,7 Doping this nonstoichiometric compound with antimony strongly modifies its intrinsic conductivity. The resistivity of Sb-doped thin films decreases when the doping rate increases until about 8 atom % (about 0.4 × 10-4 Ω · cm compared to the typical 10-1 Ω · cm of undoped films). However, the resistivity increases for higher Sb doping,8 which suggests a limit for simple substitutional incorporation along with micro* To whom correspondence should be addressed. Phone: 55 14 31036084. E- mail:
[email protected]. † Departamento de Fı´sica FC, UNESP. ‡ UFMG. § Synchrotron SOLEIL. | Instituto de Quı´mica, UNESP.
structure phenomena. There is some agreement in the literature on the interpretation of these resistivity features. The resistivity decrease for lower Sb incorporation is related to the free carrier concentration increase, coming from the replacement of SnIV by SbV ions, which act as donorlike centers in the SnO2 lattice.9 For high doping levels, the increase of resistivity is related to the presence of an increasing amount of SbIII, giving rise to acceptor levels trapping the electrons created by the substitution of SnIV by SbV 9,10 or by SbIII segregation at the particle surface.11 The electrical properties of films are explained taking into account structural (X-ray diffraction, Mo¨ssbauer, X-ray absorption spectroscopy) results from doped powders1 for concentrations close to the high doping rate (>8 atom % Sb), instead of low doping samples. Besides, X-ray photoelectron spectroscopy has been carried out to confirm the competition between the two oxidation states of Sb concerning the film electrical properties.12,13 However, the localization of antimony atoms with these two oxidation states in the SnO2 matrix remains a controversial issue, and the investigation of low doping samples is a very important matter in this direction. Recently, we concluded from XAS data taken for Sb-doped thin films (with 3, 9, and 16 atom % Sb) that the antimony atoms are mostly in the +5 oxidation state. Besides, the ab initio simulation of experimental EXAFS at the Sb K edge evidenced the presence of antimony atoms in a substitutional site located near the grain boundary surface region for a 9 atom % Sb-doped thin film.14 The increase of Sb concentration also increases the oxygen vacancy concentration. Although the material presents a high grain boundary scattering, the electron mobility can be slightly increased due to Sb doping, which is related to lowering of the grain boundary potential barrier, due to the reduced trapped
10.1021/jp106001x 2010 American Chemical Society Published on Web 10/22/2010
Characterization of Sb-Doped SnO2 Xerogels
J. Phys. Chem. C, Vol. 114, No. 45, 2010 19207
charge by interaction with oxygen vacancies. The general effect is to increase the film conductivity with Sb doping. In this paper, XAS results for Sb-doped SnO2 xerogels treated at two distinct temperatures (200 and 550 °C) are presented, aiming to elucidate the location of the Sb ion inside the SnO2 casiterite structure in relation to its oxidation state and the dependency on the doping level. 2. Experimental Section The preparation of the Sb-doped tin dioxide colloidal suspension was done from the reagents SbF3 and SnCl4 · 5H2O, which have been mixed in aqueous solution under magnetic stirring. NH4OH was added until the pH reached 11,7,14 followed by hydrolysis and condensation reactions. The remaining precipitate was submitted to dialysis against distilled water for approximately 10 days to eliminate Cl- and NH4+ ions. This procedure led to a stable suspension (sol) of SnO2/Sb. Powdered xerogels were obtained through conventional drying of the solutions, which were concentrated by the use of a rotary evaporator device. After drying at 100 °C, the obtained powders were treated at 200 °C for 30 min or 550 °C for 1 h. This latter temperature was chosen to be the same as that used for the treatment of the films and coincides with the elimination of oxygen species from the boundary layer and surface.15 The former temperature (200 °C) is just an intermediate temperature, used for firing of films between each layer. Additionally, a sample with 7.6 atom % SbIII ions chemically grafted onto an undoped SnO2 xerogel was prepared and dried at 110 °C. XANES and EXAFS measurements were carried out at room temperature, at HASYLAB (Hamburg Synchrotron Radiation Laboratory) (4.5 GeV, 140 mA), Germany. Xerogels were ground and mixed with boron nitride before being pressed into pellets. XANES measurements at the Sb L1 edge (4698 eV) were carried out on the E4 station, using as the monochromator a Si(111) double crystal. The XANES data for xerogels were recorded in total electron yield detection mode using the detector developed by Tourillon et al.16 XANES spectra were normalized far from the edge (4750 eV) and analyzed in comparison to Sb2O3 and FeSbO4 references to determine the Sb oxidation in the Sb-doped SnO2 xerogels. EXAFS data were recorded at the Sb K edge (30491 eV) and Sn K edge (29200 eV) on the ROMO II station at HASYLAB, equipped with a Si(311) double-crystal monochromator which was set to 60% of the maximum of its rocking curve to eliminate higher harmonics. Sn and Sb K edge data were both recorded in transmission mode using the same pellets. The quantity of powdered xerogels was chosen to obtain a edge jumps around 1 in transmission mode at the Sn K edge. At the Sb K edge, the obtained edge jumps were equal to 0.022 and 0.35, depending on the Sb loading. Sn and Sb K edge EXAFS spectra were extracted according to the procedure fully described elsewhere17 using the EXAFS98 software developed by Michalowicz.18 Structural parameters were determined from the fitting procedure of Sn and Sb K edge EXAFS spectra using the Artemis software.19 Scaling parameters, S02, which takes into account the multielectronic effects, and E0, used for the matching of the experimental k scale with the theoretical scales, were first checked on crystalline SnO2 and Sb2O3 reference samples. The Sb-O coordination number (NO), interatomic distance (RO), and Debye-Waller factor (σ2) were fitted for the first coordination shell for all the samples. For some xerogels, the contributions at larger distances observed for the Sb K edge Fourier transform (FT) EXAFS spectra were analyzed considering a model of solid solution between Sb and Sn into an idealized cassiterite SnO2 crystallite. In this case, the different substitution sites described
Figure 1. XANES spectra recorded at the Sb L1 edge (a) for references with oxidation states III (Sb2O3) and V (FeSbO4) and (b) for xerogels treated at 200 and 550 °C for several Sb doping concentrations. Inset: example of linear combination for SnO2 doped with 9 atom % Sb using 45% SbV and 55% SbIII.
in ref 20 for Sb into SnO2 were considered in the ab initio theoretical FeFF621 model used in the Artemis interface.19 Single scattering paths until the path length was equal to 5.84 Å were considered in the calculations for modeling an EXAFS χ(k) data range (∆k) and Fourier transform range (∆R) of approximately 2.1-13.5 Å-1 and 1-6 Å, respectively. NSn was constrained to the degeneracy values based on the crystal structure of SnO222 considering the limited size of the idealized crystallite, whereas the distance and Debye-Waller factor for each considered path were determined in the fit. A total of 15-21 parameters were considered in the model for 36 independent data points. 3. Results and Discussion 3.1. Antimony Oxidation State. Sb L1 edge XANES spectra for the SnO2 xerogels doped with Sb at different levels and treated at 200 and 550 °C are compared in Figure 1 with Sb2O3 and FeSbO4 reference spectra of SbIII and SbV, respectively. The L1 edge of antimony (4698 eV) corresponds to the electronic transitions from the Sb 2s level to the unoccupied p-like density of states derived from atomic 5p states. The comparison of the Sb2O3 and FeSbO4 reference spectra clearly evidences the well-
19208
J. Phys. Chem. C, Vol. 114, No. 45, 2010
Geraldo et al.
TABLE 1: Relative Proportions of SbIII and SbV in the Sb-Doped Xerogels Deduced by Linear Combination of Sb L1 Edge Data SbIII/SbV ratio (%)
nominal amt of SbIII (atom %)
treatment at 200 °C
treatment at 550 °C
3 4 9 16
10/90 15/85 55/45 70/30
∼0/100 7/93 13/87 15/85
known separation of about 4 eV in the position of the maximum of the white line between the trivalent and pentavalent Sb oxidation states.1 As the position of the Sb L1 absorption edge presents a low influence of the closest neighbor around the antimony atom, it is possible to determine the SbIII/SbV ratio in the Sb-doped SnO2 xerogels by fitting the XANES data with a linear combination of reference spectra representative of each oxidation state. This linear combination procedure was also done by reconstruction of the first derivatives of the xerogel spectra presented in Figure 1b, with an error bar of (5%. The SbIII/ SbV ratios so determined are shown in Table 1. The comparison of the results summarized in Table 1 evidence that the initial SbIII salt used as the starting reactant has a strong tendency to be oxidized during the synthesis, forming SbV species. The second and third columns of Table 1 evidence that the SbIII/SbIV ratio continuously increases with the nominal amount of the initial SbIII. As a consequence of the oxidation effect induced by firing at 550 °C, pentavalent antimony is mainly present with a proportion varying from 100% to 85% as the doping level increases. This behavior is different from the constant SbIII/SbV ratio of about 1/4 reported by Rockenberger et al.1 for samples initially Sb-doped with 9.1 and 16.7 atom %. This difference in SbIII/SbV ratios after firing at 500-550 °C could result from a slightly different preparation route used by Rockenberger et al.1 and can be a consequence of the changes of the proportion of Sb dissolved in the SnO2 crystalline lattice and of Sb segregated at the surface of nanoparticles. It is important to point out that in spite of the high Sb doping level any crystalline antimony phase was detected by X-ray powder diffraction as evidenced in Figure S1 of the Supporting Information. 3.2. Sn and Sb Local Order. FTs of the EXAFS signal at the Sn K edge for xerogels doped with different Sb amounts are presented in Figure 2. These Fourier transforms have the characteristic FT shape of the cassiterite structure with two main contributions located in the ranges 1 Å < R < 2.2 Å and 2.4 Å < R < 4 Å. The first peak corresponds to the contribution of the first oxygen coordination shell around tin and the second broad one mainly to tin-tin contributions. It is noteworthy that the intensity reduction of the second contribution of the FTs of xerogels compared to a crystalline SnO2 reference is strongly indicative of the presence of nanometer-sized crystallites as reported in our previous works.17,20 As the intensity of the contribution mainly related to the Sn second neighbors is nearly constant whatever the doping level, it is possible to infer that the crystallite dimensions are little affected by the doping level for xerogels treated at 200 °C. On the other hand, for a higher firing temperature (550 °C), the highest intensity of the Sn second-neighbor contribution observed for samples doped with 1.5, 3.0, and 4.0 atom % Sb suggests the presence of a crystallite with size larger than those found in undoped or heavily doped samples. These features are fully confirmed by the least-squares fitting procedure done for the simulation of the Sn K edge EXAFS data. Relevant structural parameters determined by the
Figure 2. Fourier transforms of the Sn K edge EXAFS data recorded for Sb-doped SnO2 xerogels treated (a) for 30 min at 200 °C and (b) for 1 h at 550 °C.
simulation of the first and second main contributions of the FT presented in Figure 2 are displayed in Table 2, together with the crystallite diameter estimated from the tin coordination numbers at 3.20 and 3.73 Å considering a model of an isotropic crystallite.23 From the reported results we may conclude that the doping with Sb for a level of e4 atom % favors the crystallite growth, concomitant with a strong dominance of the SbV oxidation state. Information about the location of Sb atoms in the SnO2 nanocrystalline matrix is obtained through the EXAFS analysis of local order around this cation. The Fourier transforms of Sb K edge EXAFS spectra recorded for samples with different Sb doping concentrations treated at 200 °C (Figure 3a) and 550 °C (Figure 3b) are compared to the FTs of the EXAFS spectra of an Sb2O3 reference and an SbIII-grafted SnO2 sample. The shape of the spectra depends clearly on the treatment temperature and on the Sb concentration. FTs of EXAFS data recorded for xerogels treated at 200 °C (Figure 3a) present in the 2-4 Å R range a peak with two maxima, suggesting two distances for the Sb cation contribution as is encountered in the cassiteritelike structure. However, the intensities of the first (1-2 Å) and second (2-4 Å) peaks continuously decrease as the doping concentration increases with the result that the overall FTs
Characterization of Sb-Doped SnO2 Xerogels
J. Phys. Chem. C, Vol. 114, No. 45, 2010 19209
TABLE 2: Structural Parameters Determined by the Least-Squares Fitting Procedure of the Sn K Edge Data Presented in Figure 2a O
Sn
Sn
R factor (%) D (nm)
Undoped, 6.1(1) 1.2(1) 2.06(1) 3.21(1) 4.25(18) 3.83(41)
200 °C 2.8(2) 3.73(1) 4.65(33)
0.41
1.1(1)
1.5% Sb, N 6.2(1) 1.2(1) r (Å) 2.06(1) 3.21(1) σ2 (10-3 Å2) ) 4.38(18) 3.58(33)
200 °C 3.1(2) 3.73(1) 4.72(28)
0.42
1.3(1)
N r (Å) σ2 (10-3 Å2)
3% Sb, 200 °C 6.1(1) 1.2(1) 3.1(3) 2.06(1) 3.21(1) 3.73(1) 4.40(23) 3.57(49) 4.74(38)
0.55
1.3(1)
N r (Å) σ2 (10-3 Å2)
4% Sb, 200 °C 6.2(1) 1.4(1) 3.5(2) 2.06(1) 3.21(1) 3.74(1) 4.29(21) 3.46(38) 4.67(33)
0.48
1.4(1)
N r (Å) σ2 (10-3 Å2)
9% Sb, 200 °C 6.3(1) 1.2(1) 3.3(2) 2.06(1) 3.21(1) 3.73(1) 4.42(18) 3.42(34) 4.80(27)
0.44
1.35(10)
N r (Å) σ2 (10-3 Å2)
16% Sb, 200 °C 6.3(1) 1.3(1) 3.6(2) 2.06(1) 3.21(1) 3.73(1) 4.31(19) 3.49(35) 4.65(26)
0.44
1.5(1)
N r (Å) σ2 (10-3 Å2)
Undoped, 6.0(1) 1.2(1) 2.06(1) 3.20(1) 4.23(18) 3.31(31)
550 °C 3.4(2) 3.72(1) 4.37(22)
0.40
1.4(1)
N r (Å) σ2 (10-3 Å2)
1.5% Sb, 6.2(2) 1.6(1) 2.06(1) 3.20(1) 4.14(25) 3.17(35)
550 °C 4.9(3) 3.73(1) 4.27(23)
0.61
2.5(3)
N r (Å) σ2 (10-3 Å2)
3% Sb, 550 °C 6.1(2) 1.7(1) 5.0(3) 2.06(1) 3.20(1) 3.73(1) 4.18(26) 3.23(37) 4.36(24)
0.65
2.6(3)
N r (Å) σ2 (10-3 Å2)
4% Sb, 550 °C 6.2(2) 1.7(1) 5.2(3) 2.06(1) 3.20(1) 3.73(1) 4.19(27) 3.22(39) 4.50(25)
0.69
9% Sb, 550 °C 5.9(2) 1.5(1) 4.3(3) 2.06(1) 3.20(1) 3.73(1) Å2) ) 4.46(25) 3.44(38) 4.69(27)
0.60
2.0(2)
16% Sb, 550 °C 6.0(1) 1.3(1) 3.8(2) 2.06(1) 3.20(1) 3.73(1) 4.55(23) 3.48(40) 4.63(28)
0.53
1.6(2)
N r (Å) σ2 (10-3 Å2)
N r (Å) σ2 (10-3
N r (Å) σ2 (10-3 Å2)
2.8(4)
a For the simulations, the parameter S02 was set to 1.03 and the parameter E0 to 29211(2) eV.
progressively acquire the shape of the surface-grafted SbIII species. Thus, the observed intensity change could arise by the occupation of different substitution sites for Sb in SnO2 nanocrystallites from an inner site for the lowest doping level to a more superficial one for the highest doping level. However, as the Fourier transform of the EXAFS data recorded for the SbIII-grafted SnO2 xerogel does not present an intense Sb cation contribution, the presence of such an SbIII species chemically adsorbed at the surface of the crystallites in different proportions could also be responsible for the observed decrease in intensity of the Sb cation contribution when the doping level increases.
Figure 3. Fourier transforms of the Sb K edge EXAFS data recorded for Sb-doped SnO2 xerogels treated (a) for 30 min at 200 °C and (b) for 1 h at 550 °C. Data are compared to a standard reference of Sb2O3 and to a sample with a grafted surface of SbIII.
For xerogels treated at 550 °C (Figure 3b), the similarity in shape of the FTs recorded at the Sb K edge with those recorded at the Sn K edge (Figure 2) strongly evidence that Sb and Sn cations are embedded in the same cassiterite-type local order. For a low doping amount ([Sb] e 4%), the intensities of the first and second FT contributions are nearly identical, suggesting that the local order around Sb is quite similar. However, the remarkable decrease in the intensity of the first and second FT contribution observed for 9 and 16 atom % doping is evidence of the change of the antimony local order. Deeper knowledge about the local order around Sb atoms was obtained by the least-squares fitting of the filtered EXAFS signal related to the oxygen first coordination shell. The structural parameters of the Sb-O coordination are gathered in Table 3 as a function of the doping level and treatment temperature. Irrespective of the doping level, a lower oxygen coordination number NO is observed for a given sample treated at the lowest temperature. In fact, for samples treated at 200 °C a pronounced decrease of NO was observed, which is stronger when the doping level is higher than 4 atom %. For the highest doping level (16 atom %) the NO value of 3.9 ( 0.2 along with the dominant amount of SbIII species (see Table 1) strongly
19210
J. Phys. Chem. C, Vol. 114, No. 45, 2010
Geraldo et al.
TABLE 3: First Sb-O Coordination Shell Structural Parameters Determined by the Least-Squares Fitting Procedure of the Sb K Edge Dataa sample
NO
σ2 (10-3 Å2)
RO (Å)
R factor (%)
Sb2O3 grafted 7% Sb 1.5% Sb, 200 °C 3% Sb, 200 °C 4% Sb, 200 °C 9% Sb, 200 °C 16% Sb, 200 °C 1.5% Sb, 550 °C 3% Sb, 550 °C 4% Sb, 550 °C 9% Sb, 550 °C 16% Sb, 550 °C
3 3.1(3) 5.4(5) 5.3(5) 5.0(3) 4.1(3) 3.9(2) 5.7(8) 5.6(6) 5.7(6) 5.4(4) 5.3(3)
2.68(80) 5.04(101) 4.67(99) 5.04(109) 4.64(81) 4.79(69) 5.57(66) 4.00(167) 4.18(116) 4.55(111) 5.36(90) 5.48(75)
1.97(1) 1.96(1) 1.98(1) 1.98(1) 1.98(1) 1.98(1) 1.98(1) 1.98(1) 1.99(1) 1.99(1) 1.99(1) 1.98(1)
3.00 1.77 1.80 2.28 1.22 0.94 0.71 4.51 1.99 1.75 1.11 0.85
a
S02 was set to 1.15 and E0 to 30500(2) eV.
suggests that a considerable proportion of Sb atoms are not incorporated into the SnO2 lattice. On the contrary, the values close to 6 found for the oxygen coordination number concomitant with the fact that the second broad contribution of the FTs looks like a Sn-Sn contribution observed at the Sn K edge for the FTs of the same samples (Figure 2b) strongly suggest that the antimony is incorporated into the SnO2 lattice after firing at 550 °C. In this case the increasing deviation of the coordination number NO from the 6-fold coordination observed by increasing the doping level can reflect the increase of (i) the oxygen vacancy concentration, (ii) the amount of Sb atoms located at the surface region of SnO2, and/or (iii) the SbIII/SbV proportion. These hypotheses will be considered in the following section. 3.3. SbIII and SbV Location in the SnO2 Nanoparticles. The reported results so far strongly evidence the modification of the antimony environment and oxidation state by increasing the firing temperature and the doping level. The former favors the incorporation of antimony into the SnO2 lattice, while the latter favors antimony demixing with a segregation of doping near to the surface region of SnO2 nanoparticles. Furthermore, we have also evidenced that the incorporation of antimony in substitution for Sn in the SnO2 lattice is concomitant with a change in the Sb oxidation state from III to V. Indeed, the pentavalent Sb state presents a smaller ionic radius for a 6-fold coordination (RSbV ) 0.60 Å) than the trivalent Sb state (RSbIII ) 0.76 Å), allowing an easier incorporation into the SnO2 lattice (RSnIV ) 0.69 Å).24 In this context, and according to the Sb L1 edge results (Table 1), we propose that xerogels Sb doped with 3 atom % are archetype samples in which Sb atoms are substituted for Sn in the cassiterite SnO2 crystallites. Although there is a lack of L1 Sb results for 1.5 atom % antimony-doped SnO2 xerogels, we assume that for this doping level antimony is as well in the substitutional solid solution in the cassiterite structure. These hypotheses were checked by simulating the Sb K edge EXAFS spectra of xerogels containing 1.5 and 3 atom % antimony assuming different sites for antimony atoms in a cluster of cassiterite structures idealized elsewhere.20 Figures 4 and 5 present the best simulation for the 1.5 and 3 atom % Sb-doped xerogel, respectively, treated at 200 and 550 °C. The substitutional a-g sites displayed in Figure 7 of ref 20 were used as a model for the different samples. Additionally, a “bulk” model in which Sb has the same tin and oxygen coordination shells until a distance of 5.84 ( 0.02 Å from the absorbing atom as that described for SnO2 was also computed. For all these models, the Sb-O distance of the first coordination shell is the only distance different from those encountered in
Figure 4. Comparison of the FT of the Sb K edge EXAFS spectrum (full line) recorded (a) for the xerogel doped with 1.5 atom % Sb and dried at 200 °C and (b) for the xerogel doped with 1.5 atom % Sb and fired at 550 °C with the FT of the simulated EXAFS spectrum (dotted line) corresponding to the site giving the best R factor.
the cassiterite structure since this distance was found at 1.98 ( 0.02 Å, in agreement with the distances gathered in Table 3. Figure 6 displays the R factor values for the simulations performed for the different sites for the different samples, whereas structural parameters for the best agreement are gathered in Table S1 in the Supporting Information. The ab initio calculations presented in Table S1 are not a fit of the experimental data since the number of scattering atoms in each path was fixed to the idealized value considering a given cluster. We compare the agreement between constraining models for different sites and the experimental data, limiting the number of variable parameters to the Debye-Waller factors and distances of the considered single scattering paths. It is noteworthy that, for fitting experimental data, the theoretical model should include, in addition to the considered single scattering paths, the most significant multiple scattering paths, as demonstrated in ref 25 for bulk SnO2 material. Whatever the antimony loading, these simulations evidence that, in xerogels treated at 200 °C, the mainly pentavalent Sb species (Table 1) substitutes for SnIV located closer to the
Characterization of Sb-Doped SnO2 Xerogels
Figure 5. Comparison of the FT of the Sb K edge EXAFS spectrum (full line) recorded (a) for the xerogel doped with 3 atom % Sb and dried at 200 °C and (b) for the xerogel doped with 1.5 atom % Sb and fired at 550 °C with the FT of the simulated EXAFS spectrum (dotted line) corresponding to the site giving the best R factor.
Figure 6. Evolution of the R factor of the iFeFFIT EXAFS simulation of Sb substituting for Sn in SnO2 for different sustitutional sites in the cassiterite structure according to the idealized crystallite presented in ref 20.
crystallite surface in the SnO2 lattice. Indeed, sites d-f in the crystallite model presented elsewhere20 are the best ones (Figure 6) for simulating the 1.5 and 3 atom % Sb experimental data.
J. Phys. Chem. C, Vol. 114, No. 45, 2010 19211
Figure 7. Comparison of EXAFS spectra recorded at the Sb K edge for the xerogels doped with 4, 9, and 16 atom % Sb and treated (a) at 200 °C and (b) at 550 °C (full line) with the linear combination of the EXAFS spectra where Sb is mainly pentavalent (3 atom % Sb xerogel, 200 and 550 °C, respectively) and the EXAFS spectra of the sample where Sb is trivalent (chemically grafted surface xerogel) (dotted line).
For thermal treatment at 550 °C, the best agreement between experiment and theory is obtained taking into account the socalled bulk or a sites of the idealized cassiterite structure (Figure 6). Nevertheless, except for the more superficial site g, the R factors obtained for the other sites are not so far from the values obtained for the internal crystallite sites of the SnO2 lattice, making the dependency of the R factor curve on the site occupancy relatively flat. This finding suggests that the dominant pentavalent antimony (Table 1) in 1.5 and 3 atom % Sb-doped xerogel homogeneously substitutes for Sn in the SnO2 lattice. Parts a and b of Figure 7 show a comparison between EXAFS spectra recorded at the Sb K edge for the 4, 9, and 16 atom % Sb-doped xerogel, treated at 200 and 550 °C, respectively, with linear combinations of EXAFS spectra of the 3 atom % Sb xerogel samples where Sb is mainly (200 °C) or totally (550 °C) pentavalent and the EXAFS spectra where the Sb is purely trivalent (xerogel with an SbIII-grafted surface). The obtained proportions are gathered in Table 4. Independent of the firing temperature, the EXAFS spectra of the higher doped samples (4, 9, and 16 atom %) can be fitted by a linear combination of
19212
J. Phys. Chem. C, Vol. 114, No. 45, 2010
TABLE 4: Relative Proportions of Grafted SbIII and SbV in the Sb-Doped Xerogels Deduced by Linear Combination of the Sb K Edge kχ(k) EXAFS Data SbIII/SbV ratio (%)
nominal amt of SbIII (atom %)
treatment at 200 °C
treatment at 550 °C
4 9 16
13(1)/87(1) 47(1)/53(1) 67(1)/33(1)
8(2)/92(2) 23(2)/77(2) 34(2)/66(2)
the spectra corresponding to the SbV site in the bulk site solid solution and the spectra corresponding to the sample where SbIII is adsorbed at the surface of SnO2 (Figure 5). It is noteworthy for xerogels treated at 200 °C that the SbIII/SbV proportion obtained through the analysis by linear combination of the Sb K edge EXAFS principal components is in very good agreement with the proportion reported in Table 1 from the XANES analysis recorded at the Sb L1 edge (Figure 1). For xerogels treated at 550 °C, good agreement is as well achieved for the 4 atom % Sb-doped xerogel, whereas higher SbIII/SbV ratios are obtained for 9 and 16 atom % Sb-doped xerogels from the Sb K EXAFS linear combinations compared to the L1 Sb XANES linear combinations. This feature results from the fact that the SbIII component is overestimated by the linear combination of the Sb K edge EXAFS spectra to take into account the smaller crystallite size deduced from the intensity of the cation-cation contributions of the FTs at the Sb K edge for the high Sb-loaded xerogels (Figure 3). These results evidence that our linear combination approach at the Sb K edge is valid as long as no crystallite size effect must be considered. In this latter case, i.e., for xerogels doped with 9 and 16 atom % and treated at 550 °C, more accurate SbIII/SbV ratios are obtained from the analysis of the Sb L1 XANES data. 3.4. Doping Structure Evolution. To take full consideration of the above results, we describe the dependence of the doping structure of powders on the Sb loading level and firing temperature/particle size considering the following scenario. Prior to thermal treatment at 550 °C, the dried powder consists of very small tin(IV) oxide particles with a nanocrystallite of similar size, smaller than 2 nm, irrespective of the nominal doping level and of the proportion of antimony oxidation state, SbIII or SbV. The SbIII atoms are grafted to the surface of the SnO2 nanocrystallite, while the SbV atoms lie at the surface (sites d-f20) rather than at the bulk, adopting the same position as the SnIV atoms. However, the fact that the mean oxygen coordination number NO is systematically less than 6 indicates the SbV ions occupy the in-plane 5-fold sites rather than subsurface octahedral sites. This hypothesis is corroborated by the good agreement between all NO values determined by EXAFS (Table 2) and the values calculated considering the SbV/ SbIII proportions (Table 1) and assuming 5-fold and 3-fold coordination for oxygen in the first shell of pentavalent and trivalent antimony atoms. This configuration is in agreement with the structure of SbV segregated at the (001) face of the cassiterite crystal found by atomistic computer modeling.26 Interesting aspects emerging from this bimodal segregation configuration are (i) the charge compensation effect caused by the presence of an electron donor state created by the replacement of SnIV by SbV and by the presence of an SbIII electron trap state explains the almost invariance of electrical conductivity with the doping level generally observed for samples fired below 400 °C10-16 and (ii) the relatively high surface energy (which is indicative of low thermodynamic stability) and high attachment energy (which is indicative of kinetic stability) of the (001) face, which rapidly disappears upon firing at 550 °C,26
Geraldo et al. favors the bulk incorporation of SbV segregated at the (001) face. The coherent attachment of a nanocrystal by the doped (001) face explains the improved crystallite growth observed upon 550 °C firing of powder doped with 1.5, 3.0, and 4.0 atom % Sb (Figure 1). This mechanism is also consistent with the development of a textured morphology during isothermal treatment of Sb-doped SnO2 films.24 For Sb doping higher than 4 atom % the increasing amount of SbIII segregated at the external surface evidenced at 200 °C hinders the contribution of the nanocrystal attachment process on the crystallite growth upon firing at 550 °C. Thus, in this case the bulk diffusion process is only involved in the formation of a homogeneous distribution of SbV atoms in the substitutional solid solution. It is important to note that this behavior is opposite the behavior reported for copper-doped xerogels for which a phase separation was observed by increasing the temperature since the doping atoms moved from the substitutional inner sites to the surface of the crystallite.20 This feature could be responsible for the best sintering properties of CuIIdoped ceramics20 compared to those of undoped and Sb-doped SnO2 in which grain coarsening is the main phenomenon occurring during firing.24,27 4. Conclusion The knowledge of the doping atom location in a matrix host is of paramount importance for understanding the role played by the doping in the change of the properties of the materials. Using a multiedge X-ray absorption approach, we have investigated the location of antimony atoms in doped SnO2 xerogels obtained by a sol-gel route starting from an SbIII precursor salt, which is shown in this paper. First, we have unambiguously evidenced that the change of the antimony oxidation state from SbIII to SbV upon firing is a prerequisite for the formation of a substitutional solid solution involving the replacement of SnIV by SbV in the SnO2 lattice. The proportion of SbV substituting for SnIV in SnO2 nanocrystals depends both on the doping level and on the firing temperature. For low doping levels (e4 atom %), the drying of Sb-doped SnO2 xerogels at 200 °C leads to more than 85 atom % of the initial SbIII changing to the SbV oxidation state, whereas as the doping level increases, more than 50 atom % antimony remains in the precursor salt oxidation state, SbIII. At 550 °C, independent of the doping level, the antimony is mainly present in the SbV state, with a varying proportion between 100 and 85 atom %, decreasing for increasing doping level. It has been evidenced that after treatment at 200 °C SbV cations mostly occupy superficial substitution sites whereas the increase of firing temperature at 550 °C favors the homogeneous incorporation of SbV into the SnO2 lattice. The remaining SbIII cations are grafted at the surface of the crystallite. The varying proportion of SbIII and SbV depending on the doping level and firing temperature is responsible for the observed electrical properties of Sb-doped SnO2 nanomaterials. Furthermore, the SbIII/SbV ratio as well as the location of the trivalent and pentavalent antimony cations in the SnO2 crystallite allows us to understand the driving forces giving rise to crystallite growth. Acknowledgment. We acknowledge the following Brazilian financial sources: CAPES, CNPq, FAPESP, and PROPeUNESP/Santander. DESY and the European Community are acknowledged for financial support for access to HASYLAB (Contract RII3-CT-2004-50600, IA-SFS).
Characterization of Sb-Doped SnO2 Xerogels Supporting Information Available: Figure S1 giving the X-ray diffraction data for SnO2 xerogels doped with several concentrations of Sb and Table S1 giving the structural parameters used for the simulations presented in Figures 4 and 5. This information is available free of charge via the Internet at http://pubs.acs.org. References and Notes (1) Rockenberger, J.; zum Felde, U.; Tischer, M.; Tro¨ger, L.; Haase, M.; Weller, H. J. Chem. Phys. 2000, 112, 4296. (2) Giraldi, T. R.; Lanfredi, A. J. C.; Escote, M. T.; Longo, E.; Varela, J. A.; Ribeiro, C.; Chiquito, A. J. J. Appl. Phys. 2007, 102, 034312. (3) Senguttuvan, T. D.; Malhotra, L. K. Thin Solid Films 1996, 289, 22. (4) Rai, T.; Senguttuvan, T. D.; Lakshmikumar, S. T. Comput. Mater. Sci. 2006, 37, 15. (5) Aegerter, M. A.; Reich, A.; Ganz, D.; Gasparro, G.; Piitz, J.; Krajewski, J. T. J. Non-Cryst. Solids 1997, 218, 123. (6) Messias, F. R.; Vega, B A. V.; Scalvi, L. V. A.; Siu Li, M.; Santilli, C. V.; Pulcinelli, S. H. J. Non-Cryst. Solids 1999, 247, 171. (7) de Souza, A. E.; Monteiro, S. H.; Santilli, C. V.; Pulcinelli, S. H. J. Mater. Sci.: Mater. Electron. 1997, 8, 265. (8) Terrier, C.; Chatelon, J. P.; Roger, J. A. Thin Solid Films 1997, 95, 295. (9) Ferry, F. J.; Laundy, B. J. J. Chem. Soc., Dalton Trans. 1981, 1442. (10) Shokr, E. Kh.; Wakkad, M. M.; Abd El-Ghanny, H. A.; Ali, H. H. J. Phys. Chem. Solids 2000, 61, 75. (11) Messad, A.; Bruneaux, J.; Cachet, H.; Froment, M. J. Mater. Sci. 1994, 29, 5095. (12) Terrier, C.; Chatelon, J. P.; Roger, J. A.; Berjoan, R.; Dubois, C. J. Sol-Gel Sci. Technol. 1997, 10, 75.
J. Phys. Chem. C, Vol. 114, No. 45, 2010 19213 (13) McGinley, C.; Borchert, H.; Pflughoefft, M.; Al Moussalami, S.; de Castro, A.R. B.; Haase, M.; Weller, H.; Mo¨ller, T. Phys. ReV. B 2001, 64, 245312. (14) Geraldo, V.; Briois, V.; Scalvi, L. V. A.; Santilli, C. V. J. Eur. Ceram. Soc. 2007, 27, 4265. (15) Yamazone, N.; Fuchigami, J.; Kishikawa, M.; Seiyama, T. Surf. Sci. 1979, 86, 355. (16) Tourillon, G.; Dartyge, E.; Fontaine, A.; Lemonnier, M.; Bartol, F. Phys. Lett. A 1987, 121, 251. (17) Briois, V.; Santilli, C. V.; Pulcinelli, S. H.; Brito, G. E. S. J. NonCryst. Solids 1995, 191, 17. (18) Michalowicz, A. Exafs pour le Mac Logiciels pour la Chimie; Socie´te´ Francaise de Chimie Paris: Paris, 1991; p 102. (19) Newville, M. J. Synchrotron Radiat. 2001, 8, 322. (20) Santilli, C. V.; Pulcinelli, S. H.; Brito, G. E. S.; Briois, V. J. Phys. Chem. B 1999, 103, 2660. (21) Ankudinov, A. L.; Ravel, B.; Rehr, J. J.; Conradson, S. D. Phys ReV. B 1998, 58, 7565. (22) Baur, W. H.; Khan, A. A. Acta Crystallogr. 1971, B27, 2133. (23) Serrini, P.; Briois, V.; Horillo, M. C.; Traverse, A.; Manes., L. Thin Solid Films 1997, 304, 113. (24) Santilli, C. V.; Rizzato, A. P.; Pulcinelli, S. H.; Craievich, A. F. Phys. ReV. B 2007, 75, 205335. (25) Kelly, S. D.; Hesterberg, D.; Ravel, B. Methods of Soil Analysis. Part 5. Mineralogical Methods; Ulery, A. L., Drees, L. R., Eds.; SSSA Book Series, No. 5; Soil Science Society of America: Madison, WI, 2008; Chapter 14. (26) Slater, B.; Catlow, C. R. A.; Gay, D. H.; Williams, D. E.; Dusastre, V. J. Phys. Chem. B 1999, 103, 10644. (27) Santilli, C. V.; Pulcinelli, S. H.; Craievich, A. F. Phys. ReV. B 1995, 51, 8801.
JP106001X