Exploration of Dopant Species for Lanthanum Polyphosphate

Doped lanthanum polyphosphate (LaP 3 O 9 ) exhibits relatively high proton conductivity. For the practical applications such as the electrolyte of fuel cells, however, its conductivity must be improved by 2 orders of magnitude. Protons are introduced into matrix by lower-valent cation doping, and proton conductivity depends on dopant species. To date, LaP 3 O 9 has been doped with only Ca, Sr and Ba. In this work, we tried to dope LaP 3 O 9 with Na + , K + , Mg 2 + and Pb 2 + , as the new dopant species, due to their close ionic radii to La 3 + . Among them, only Pb could substitute for La at a comparable concentration to those of alkaline earth metals and its highest doping level was 6.4 mol% (Doping level is deﬁned as the concentration ratio of dopant (M) to host cation (La) site in matrix ( ≡ M/(La + M) × 100 (mol%))). Though Pb can exist as either divalent or tetravalent state, Pb in LaP 3 O 9 was identiﬁed to be divalent state by XPS analysis. Proton conduction was demonstrated by H/D isotope effect. The electrical conductivity of Pb-doped LaP 3 O 9 increased with Pb-doping level, owing to the increase in proton concentration. The conductivity of 4.5 mol% Pb-doped LaP 3 O 9 was about one order of magnitude lower than that of 7.9 mol% Sr-doped LaP 3 O 9 . © The Author(s) 2015. Published by ECS. This an open access article distributed the terms of the Creative Commons Attribution 4.0 License (CC BY, http://creativecommons.org/licenses/by/4.0/), which permits

One of the problems preventing the conductivity enhancement is limited choices of dopant species. Proton conductivity is proportional to concentration and mobility of protons in matrix. Positive defects of interstitial protons are introduced into matrix to maintain electroneutrality, when negative defects are formed by lower valence cation doping. 15,16 Thus, proton concentration depends on dopant species and increases with dopant concentrations, though dopant concentrations have upper limits which are different by dopant species. 4 Proton mobility is also dependent on dopant species and dopant concentration. In fact, it has been previously reported that the conductivity of 1 mol% Sr-doped LaP 3 O 9 is 1 order of magnitude higher than that of 1 mol% Ca-doped LaP 3 O 9 . 7 This conductivity difference of LaP 3 O 9 indicates that proton concentration and/or proton mobility can be different by orders of magnitude, depending on dopant species. To date, however, LaP 3 O 9 has been doped with only Ca, Sr and Ba. 7 In this study, new appropriate dopant species which can increase concentration and/or mobility of proton has been explored for further conductivity enhancement.
The new dopant species must be monovalent or divalent cations, because they must form negative defects by substituting for La 3+ . In addition, the doping levels comparable to those of alkaline earth metals should be attained (The highest value ever reported is 7.9 mol% by Sr-doping 10 ). For achieving higher doping levels, the ionic radius of dopant is generally required to be similar to that of host cation. In light of this empirical rule, Na, K, Mg and Pb were selected as the candidates of the new dopant species in this work (La 3+ :1.16 Å, Na + :1.18 Å, K + :1.51 Å, Mg 2+ :0.89 Å, Pb 2+ :1.29 Å in eight-fold coordination 17 ).
For the synthesis of doped LaP 3 O 9 , precipitation method was employed, because it has several advantages over solid state reaction method (SSR) for enhancing the conductivity. For example, higher doping levels can be achieved. 10,18 Additionally, larger and c-axisoriented grains can be obtained, 19 while grains are randomly oriented in the case of SSR. Larger grains reduce the density of grain boundary, which generally has an unignorable resistance against proton conducz E-mail: adachi.yoshinobu.36z@st.kyoto-u.ac.jp tion, and c-axis orientation enhances the bulk conductivity, because c-axis has one of the highest conductivities in LaP 3 O 9 crystal. 10

Charge Carrier Identification
Monovalent or divalent cation-doping can introduce not only interstitial proton (H i • ), but also oxygen ion vacancy (Vo •• ) and electron hole (h • ). 12 Since Vo •• and h • can also function as charge carriers, electrical conductivity (σ) is given by the total of the three partial conductivities.
Here, σ H+ , σ O 2 − and σ h indicate the partial conductivities of proton, oxide ion and hole.
For the utilization as the electrolytes, the transport number of hole conduction must be ∼0. In this work, these three partial conductions (especially hole conduction) have been discussed, by examining the partial pressure dependence and H/D isotope effect of electrical conductivity.  20 Especially when the concentration of oxygen ion vacancy (C(Vo •• )) can be regarded as constant (in other words, hydration level is low.), those of interstitial proton (C(H i • )) and electron hole (C(h • )) depend only on p H 2 O or p O 2 , and can be expressed as below.
If the mobility of each carrier is independent on p H 2 O or p O 2 , their partial conductivities are expressed as follows.
H/D isotope effect.-Proton conduction can be verified also by examining H/D isotope effect on electrical conductivity. The conductivities of proton conductors in hydrogen atmosphere can be larger

Experimental
Synthesis.-LaP 3 O 9 and PbP 2 O 6 were precipitated in homogeneous condensed phosphoric acid solutions, containing lanthanum oxide and hydrogen carbonate of monovalent dopant (Na, K) or oxide of divalent dopant (Mg, Pb). The starting materials used for the synthesis are showed in Table I. The detailed procedure is available elsewhere. 19 First, starting materials were mixed at a proper ratio in perfluoroalkoxy (PFA) beakers or glassy carbon (GC) crucibles, and the mixtures were kept at 190 • C for several days in air, so as to dissolve solid reagents into phosphoric acid. Then the temperature was raised up to 230-270 • C. By the solutions held at those temperatures in air for several days, LaP 3 O 9 and/or PbP 2 O 6 precipitated at the bottom of the crucibles. By washing the precipitates with the hot deionized water (∼90 • C), the remaining phosphoric acid solutions were removed from the surfaces of the precipitates.
In the case of the direct precipitation of the dense polycrystals of LaP 3 O 9 for conductivity measurements, the following procedure was applied using mirror-polished GC crucibles. During the precipitation, the water vapor partial pressure (p H 2 O ) was kept constant at 3.2 kPa by flowing the air, which was bubbled through liquid water kept at room temperature. Then, the temperature was gradually raised from 190 to 250 • C at the rate of 1 • C h −1 , and then the solutions were held at 250 • C for four and a half day. After this procedure, plate-like dense polycrystal of LaP 3 O 9 formed at the bottom of the crucibles.
Characterization.-Phase identification was carried out at room temperature via X-ray diffraction (XRD) analysis on PANalytical X'Pert-Pro MPD using Cu-Kα radiation. The lattice volume was determined by the Rietveld method using X'Pert HighScore Plus software (Version 2.2c). The morphology of precipitates was observed using KEYENCE VE-7800 scanning electron microscopy (SEM).
The compositions of the precipitates were analyzed by energy dispersive X-ray Spectroscopy (EDX) on EDAX Genesis XM2, Wavelength dispersive X-ray Spectroscopy (WDX) on Microspec WDX-3PC or inductively coupled plasma atomic emission spectrometry (ICP-AES) on Seiko Instruments SPS3500. An accelerating voltage of 20 kV was used during EDX and WDX analyses. As the standard materials of WDX analysis, NaPO 3 (95%, Wako Pure Chemical Industries) and undoped LaP 3 O 9 were used. Undoped LaP 3 O 9 was precipitated at 250 • C from a phosphoric acid solution with a composition of La:P = 0.8:15. It has been reported that the doping level of Sr-doped LaPO 4 synthesized by precipitation method depends on precipitation period. 18 Thus, doping levels in LaP 3 O 9 were derived from the average compositions of whole precipitates analyzed by ICP-AES.
Valence determination of Pb was carried out at room temperature via X-ray photoelectron spectroscopy (XPS) analysis on JEOL JPS-9010TRX using Mg Kα radiation. The spectrometer was calibrated using the photoemission line C 1s (binding energy 285.0 eV). As a standard material, PbP 2 O 6 (phosphate of divalent Pb) was precipitated from phosphoric acid solution, by mixing the reagents at the ratio of Pb:P = 1.5:15 and setting the precipitating temperature at 270 • C. The obtained precipitates were ground into powder, and pressed into pellets at 420 MPa. Though phosphate of tetravalent Pb was also tried to be prepared for a standard material, it could not be obtained.
Electrical conductivity measurement.-The electrical conductivity was measured by 2-probe AC impedance spectroscopy in humidified hydrogen atmosphere. For this measurement, ∼200 μm thick of plate-like dense polycrystals of LaP 3 O 9 were directly precipitated in phosphoric acid solutions. The obtained precipitates were held at 500 • C in air for 100 hour to evaporate the remaining phosphoric acid, because it has been recently reported that the remaining phosphoric acid solutions can effect on the conductivity measurement. 11 After that, Pt was deposited on the both sides of the plates as electrodes by sputtering using Eiko ion coater IB3. During the electrical conductivity measurements, humidified gas mixture of H 2 and Ar was flowed over the samples. The equilibrium oxygen partial pressure (p O 2 ) was controlled by changing p H2 / p H 2 O ratio. p H 2 O was varied from 3.2 to 31 kPa, by bubbling the gas through liquid water kept at appropriate temperatures. Especially for the isotope effect examination, p H 2 O and p D 2 O were kept at 5.0 kPa, by bubbling the gas through liquid water or deuterium water, kept at 33 or 35 • C, respectively. 22

Results and Discussion
The following abbreviations are used in this text. C init. is the initial concentration ratios of dopant to La in phosphoric acid solutions The powder X-ray diffraction patterns of the obtained precipitates are shown in Fig. 1. Single-phase LaP 3 O 9 precipitated from every solution in the case of C init = 10, while the secondary phases of NaLa(PO 3 ) 4 or MgP 2 O 6 precipitated with LaP 3 O 9 from Na or Mg-containing solutions, in the case of C init . = 20. Therefore, the upper limits of C init for obtaining single phase LaP 3 O 9 lie between 10 and 20 for Na-and Mg-containing solutions. In the case of K-containing solutions, the upper limit is larger than 20, but it might be of the same order as that for Na-containing solutions because Na and K are both alkali metals and the existence of KLa(PO 3 ) 4 has also been reported. The upper limit for Pb-containing solutions will be discussed later.
The concentrations of Mg, K and Pb on the surface of singlephase LaP 3 O 9 precipitates were analyzed by EDX, and that of Na was analyzed by WDX. Pb was detected from the precipitate and its local doping level was about 2.6 mol%, while Mg and K were not detected (Fig. 2). Na concentrations were analyzed at 7 points, and the average doping level was 0.7±0.9 mol%. (The error was defined as twice the standard deviation). Na doping level is considered to be ∼1 mol% at most. Thus, only Pb can substitute for La in LaP 3 O 9 at a considerable doping level, among Na, K, Mg and Pb.
Achievable doping level of Pb.-In this experiment, the reagents were mixed at the ratio of (La + Pb):P = 1:15 (C init. = 0-60), and LaP 3 O 9 was precipitated at 230 • C in air. The powder X-ray diffraction patterns of the obtained precipitates are shown in Fig. 3. When C init. was small (C init. ≤50), single-phase LaP 3 O 9 precipitated, while the secondary phase of PbP 2 O 6 precipitated with LaP 3 O 9 when C init. was larger (C init. = 60). The compositions of single phase LaP 3 O 9 were analyzed by ICP-AES (Fig. 4). The Pb-doping level tends to increase with C init. , and reached to 6.4 mol% at C init. = 50. Though slightly higher Pb-doping levels may be achieved in the range of 50<C init. <60, Pb-doping level of 6.4 mol% is as high as the maximum Sr-doping level ever reported (7.9 mol% 10 ). 3 O 9 .-Comparing the XPS spectra of PbP 2 O 6 and Pb-doped LaP 3 O 9 (Fig. 5), the binding energies of Pb 4f 5/2 and 4f 7/2 were accurately matched (144 eV, 139 eV respectively). Based on this result, the valence of Pb in LaP 3 O 9 matrix should be the same as that in PbP 2 O 6 , i.e. +2. However, this result might not be conclusive, because the binding energies of tetravalent Pb in phosphates are not available.

Valence determination of Pb in LaP
In order to support the result of XPS analysis, the lattice volumes of undoped and Pb-doped LaP 3 O 9 are discussed (Fig. 6). The lattice volume increased with Pb-doping level. The ionic radius of Pb 4+ is smaller than that of La 3+ , while that of Pb 2+ is larger (Table II). If Pb existed as Pb 4+ in LaP 3 O 9 matrix, the lattice volume would decrease as Pb-doping level increased. Therefore, this result confirms that the valence of Pb is +2 in LaP 3 O 9 . Electrical conductivity.-Morphology of plate-like dense polycrystals.-For conductivity measurement, plate-like dense polycrystals of undoped and Pb-doped LaP 3 O 9 were directly precipitated from phosphoric acid solutions. The reagents were mixed at the ratios shown in Table III. All of the dense polycrystals were single-phase LaP 3 O 9 (Fig. 7a). Their Pb doping levels and relative densities are also shown in Table III, and their relative densities were over 97%.
It was previously reported that the plate-like polycrystals of Srdoped LaP 3 O 9 had two geometrically different surfaces, upper surface and lower surface. 19 Upper surface is the plane facing to the solutions during the precipitation, and consists of large columnar grains (∼100 μm). Since the columnar grains are unidirectionally crystallized along c-axis, (001) planes tends to be oriented parallel to upper surface. In contrast, lower surface (facing to the bottom of the crucibles) consists of relatively smaller grains (<10 μm), and (110) planes tended to be oriented parallel to this surface.
Pb-doped LaP 3 O 9 also followed the above characteristics. Orientation can be determined from Fig. 7b and 7c, and grain size can be measured using Fig. 8. Additionally, as Pb doping level increased,     (001) planes were more strongly-oriented parallel to upper surface, and were almost completely-oriented when Pb doping level is larger than 2.9 mol% (Fig. 7b). Impedance spectrum.-The impedance spectrum of LP(Pb4.5) is shown in Fig. 9 as a typical example. In the spectrum, there were one clear large arc and unclear arc at low frequency range. The diameter of the large arc was almost independent on hydrogen partial pressure (p H2 ), and its electric capacitance was calculated to be ∼10 −11 F. Based on p H2 -independence, this arc is not attributed to the reaction impedance at the interface between electrolyte and electrode. Then, it should be attributed to either bulk impedance or grain boundaryimpedance. In the case of typical ionic conductor of yttria-stabilized zirconia, the capacitances of the arcs attributed to bulk and grain boundary are reported to be ∼10 −12 F and ∼10 −9 F respectively (Section 4.1.3 in Reference 23). Thus, the large arcs in Fig. 9 should be attributed to bulk impedances. The arc attributed to the grain boundary impedance would be too small to be seen, due to the relatively low density of grain boundary (see Fig. 8). Thus, the bulk conductivity can be regarded as the total conductivity, in the case of this sample. Charge carrier identification.-In the cases of LaP 3 O 9 doped with alkaline earth metals, their electrical conductivities depend on neither p H 2 O nor p O 2 , but show H/D isotope effect. 6,7 p O 2 -independence indicates a negligible transport number of hole conduction, and the isotope effect indicates a considerable transport number of proton conduction, though the cause of p H 2 O -independence has never been clarified. In this manner, LaP 3 O 9 doped with alkaline earth metals are verified to be an ionic conductor with a considerable transport number of proton conduction. In isotope effect examination, however, one must consider a possibility that if a material has hole conductivity, its electrical conductivity might be different between hydrogen and deuterium atmospheres, depending on the equilibrium oxygen partial pressure. 3 Thus, with considering that possibility, the partial pressure dependence and isotope effect of electrical conductivity of Pb-doped LaP 3 O 9 were examined.  As shown in Fig. 10a and 10b, the electrical conductivity of LP(Pb4.5) depends on neither p H 2 O nor p O 2 . Additionally, as shown in Fig. 10c, the conductivity in hydrogen atmosphere was ∼1.08 times higher than that in deuterium atmosphere. Since the conductivity in hydrogen atmosphere was clearly higher even when p O 2 was similar (Fig. 10b), the conductivity difference in isotope effect examination should be attributed to the difference in charge carrier (H + /D + ).    Therefore, Pb-doped LaP 3 O 9 should be an ionic conductor with a considerable transport number of proton conduction. Dependence of conductivity on Pb doping level.-The bulk conductivities of undoped and Pb-doped LaP 3 O 9 are shown in Fig. 11. The conductivities proportionally increased with the Pb doping level. This conductivity enhancement could be associated with the increase in the proton concentration and/or in orientation strength of grains along c-axis (as mentioned in Morphology of plate-like dense polycrystals section). If it was mainly associated with the increase in orientation strength, the conductivities should have been almost constant when Pb doping level was larger than 2.9 mol%, since (001) planes were almost completely-oriented parallel to upper surfaces in that range of doping level. However, the conductivity clearly increased linearly even in that range. Therefore, the conductivity was enhanced mainly by the increase in the proton concentration, which is caused by the increase in Pb doping level. It should be noted that, in the plate-like polycrystals of Pb-doped LaP 3 O 9 , small gradients of Pb doping level was seen along the direction normal to the surfaces. In the case of LP(Pb3.5), for example, local Pb doping level varied from 1.4 mol% to 5.0 mol%. However, we estimated the effect of the doping level gradient on the apparent conductivity and found that the deviation of the apparent conductivity from that of the ideally homogeneous sample was only 10-20%. Temperature dependence of the conductivity.-The temperature dependence of the conductivity of LP(Pb4.5) was investigated in the range of 200-400 • C (Fig. 12). In that temperature range, the activation energy of LP(Pb4.5) was calculated to be 0.79 eV, and larger than that of LP(Sr7.9) prepared in the similar manner (0.67 eV). 10 The conductivity of LP(Pb4.5) was about one order of magnitude lower than that of LP(Sr7.9), even though the doping level was only twice lower. Since these two samples have the similar morphologies, the difference in conductivity is not attributed to that in morphology, but to those in proton concentration and/or proton mobility in LaP 3 O 9 matrix.

Summary
In this study, it was demonstrated that LaP 3 O 9 can be doped with Pb at a doping level of 6.4 mol%, comparable to those of alkaline earth metals. Pb substituting for La in LaP 3 O 9 matrix was identified to be divalent by XPS, and it was confirmed by the lattice volume change induced by Pb doping. It was also revealed that Pb-doped LaP 3 O 9 is an ionic conductor with a considerable transport number of proton conduction. Its electrical conductivity increased with Pb doping level. This trend can be attributed mainly to increase in the proton concentration, rather than to increase in the orientation strength of grains. As far as the authors know, this is the first time for proton conduction to be exhibited by Pb doping, in proton-conducting phosphates. Unfortunately, the result of electrical conductivity measurement implied that Pb doping reduce proton concentration and/or proton mobility in LaP 3 O 9 matrix, compared to Sr doping. However, it should still be worth trying to dope other rare earth containing materials with Pb for the conductivity enhancement.