Figure 4a compares the intensity of QNS peaks recorded at a few characteristic temperatures for highly dense (99% of theoretical density) Ln-modified SrZrO_3−δ ceramics using a Time-of-Flight spectrometer. The elastic peak intensity variations are directly proportional to the hydrogen content, according to the following formula (Equation 2):

where σ_incoh is the incoherent neutron cross sections; and I the elastic peak intensity for protonated and deprotonated (or non-protonated) compounds, respectively.

As it can be seen in Figure 4a, the highest intensity is detected at room temperature in agreement with the presence of free water adsorbed on the ceramic surface as well as the surface (traces of hydrocarbonates, hydroxides, etc.) and bulk protonic species. Heating above 300 °C, especially under dynamically high vacuum conditions, allowed the ceramic surface to be cleaned from the surface protonic species [24,25,27]. A comparison of the intensity values recorded at RT and 400 °C clearly shows that in the case of very high density ceramics, the contribution of surface protonic species is low, much lower than the content of bulk protonic species. The decrease of the intensity detected above 600 °C reveals the deprotonation process, i.e., the departure of protons. The comparison of intensity values measured at 400 °C and 660 °C allows the calculation of the content of bulk hydrogen species. Note that the peak observed at 660 °C is an elastic contribution of the host perovskite lattice. The bulk proton content is low, i.e., ~0.005 mole/mole. Hempelmann et al. [33,34] determined that the proton content in cerate ceramic is 10 times higher, i.e., ~0.02. The difference can be explained first of all by the different nature of the sample, i.e., strontium cerate and its lower density (~95%). However, one may ask a question: Did Hempelmann et al. [33,34] only measure the bulk protonic species or both the bulk and surface protonic moieties?

The bulk proton content determined with the neutron technique is consistent with the results of TGA measurements. Note that the TG analysis is performed in a He atmosphere using a Pt crucible in order to enhance the accuracy of detecting very low mass losses in the temperature range where heat transfer is controlled by gas convection. Figure 4b compares the TGA curve characteristics of the high density (99%) and the lower density (94%) Ln-modified SrZrO_3−δ ceramics. As it can be seen in Table 1 both ceramics were protonated in different conditions in order to determine the influence of the protonation method on the bulk and surface protonic species contents. Note that both samples were preliminarily dried at 300 °C in order to eliminate the contribution of free surface water. Two or three characteristic mass losses can be detected as a function of the sample. The first one, detected near 400 °C can be attributed to the departure of the surface protonic species (higher departure temperature, higher chemically bonded (acidic) water); the second one—above 600 °C—is related to the elimination of the bulk conducting species whereas the third one observed above 800 °C, characteristic of ceramics containing the carbonates only, corresponds to carbonates decomposition [28]. As we can see, the 94% density sample possesses higher values of surface protonic species in relation with its higher active surface area. The very high density sample shows traces of surface protonic species only in good agreement with neutron measurements (Figure 4a). The concentrations of bulk and surface protonic species for both samples are given in Table 1.

Another important aspect is the proton distribution in a ceramic. The homogeneity of the proton distribution can be followed using the so-called Raman profilometry method [23]. This method is based on a careful analysis of an intense background, centered near 2,500 cm⁻¹, assigned to electronic defects associated with protons inserted at temperatures below ~300 °C, which can be detected in the Raman spectrum [23,24]. The comparison of the Raman study of zirconate and titanate perovskites revealed that the broad Raman background intensity is proportional to the content of the protonic species as well as roughly to the conductivity of the material. Figure 5a,b show the Raman spectra with characteristic broad backgrounds recorded on the section—from surface to its center—of BaZrO₃ and SrZrO₃ ceramics (see Table 1) protonated in different conditions. The background intensity as the function of the position along the sample section (Figure 5c) reflects the profile expected from Fick’s diffusion laws [23]. As it can be clearly seen, the curves exhibit different forms. The first one, characteristic of Ln-modified BaZrO₃ ceramic (Figure 5a), reveals that the protonic species are present at/near the ceramic surface only. In the lack of easy diffusion of the O²⁻/V_O species requiring high temperature, it is highly probable that the filling of vacancies is limited to the near-surface layers only. On the contrary, the curve presented in Figure 5b, characteristic of Ln-modified SrZrO₃, clearly shows a very homogeneous distribution of the protonic species in the sample, which confirms a successful protonation.

Since the neutronography method is a suitable tool for the detection of hydrogen elements in a dense material matrix, we performed the first tests on high dense perovskite ceramics containing the protonic species. The results of neutronography tests performed at Léon Brillouin Laboratory (Saclay, France) using radiographic films as detector on two Ln-modified SrZrO_3−δ ceramics are presented in Figure 6. The presence of protonic species is detected as white contrast. As it can be clearly seen the ceramic presented in Figure 6a contains protonic species mainly on the surface whereas in the case of Figure 6b the protons seem to be distributed throughout the ceramic.

The ignorance of the differentiation between bulk and surface protonic species can have serious consequences. First of all, the important presence of protonic species adsorbed on a material’s surface may significantly distort experimental data. Conductivity measurements can be cited as a good example. The surface secondary phases, such as hydroxides, especially in liquid-hydrolyzed form, are very good proton conductors. Jalarvo et al. [39] showed that in the case of Sr₄(Sr₂Nb₂)O₁₁·nH₂O and Sr₄(Sr₂Ta₂)O₁₁·nH₂O proton conducting perovskite, the secondary phases such as Sr(OH)₂ contribute to and even dominate the proton conduction. In consequence, the Sr content in the pristine perovskite material is strongly decreased, which leads to a deterioration of the sample. IR analysis of similar compounds show strong ca. 1,450 cm⁻¹ band characteristic of carbonated surface species [40,41].

Our study revealed that the conductivity of a very dense ceramic, measured directly in a pressurized water steam electrolyzer, achieves 10⁻² S/cm below 600 °C and its value is stable during hundreds of hours [22,25]. On the contrary, low dense ceramics with a high level of surface protonic species have a very short lifetime under high H₂O pressure. Namely, the formation of OH⁻ species leads to the loss of mechanical properties or even to ceramic crumbling and structural decomposition in the case of highly substituted compounds [26].

The analysis/interpretation of X-ray and/or neutron diffraction experiments also suffers from the presence of surface protonic moieties. Despite intensive studies [42,43,44,45,46,47], the exact proton localization in the host perovskite structure has not been definitively determined yet. Indeed, an extremely high incoherent cross section together with an extremely low coherent cross section makes the proton site determination a big challenge even at low temperatures far from conduction conditions. The analysis of some attempts, however, suggests the source of these difficulties. First of all, the sample composition is estimated. Namely, the proton/deuteron content is fixed due to the estimation based on the substituting atom and expected oxygen vacancy contents. Consequently, the obtained reliability factors question the results of Rietveld refinement [48]. Moreover, most of the authors performed diffraction studies on powders or poorly densified ceramics at room or at very low temperatures [42,43,44,45,46,47], which significantly enhances the probability of the presence of surface protonic species. The presence of some secondary phases, mostly hydroxides, was evidenced even by diffraction [44,45]; note that if the second phases are limited to traces only, their detection is impossible in diffraction experiments [26,28].

The lack of differentiation between the bulk and surface protonic moieties also makes the true proton nature very ambiguous and hinders a clear analysis of QNS data. Recently, most of literature data has “accepted” the previous statement that protons in perovskite penetrate the oxygen covalence sphere to create OH groups. This thesis is based on the presence of important IR signatures detected in the OH-stretching region, i.e., 1,500–4,000 cm⁻¹ [49,50,51,52,53]. Note that such a rich IR signature is also considered to be undisputable proof of a successful sample protonation. The proton conduction is described according to the Grotthuss mechanism [49]. However, our results obtained on highly dense polished ceramics, which according to neutron and TGA analyses contain mostly bulk protons, allowed us to question the hydroxyl nature of protons. As presented in Figure 7a, the IR spectrum characteristics of highly dense SrZrO₃ does not show any signature in the O-H-stretching region, similarly to samples in non-protonated and deprotonated states. IR spectra recorded on optical clear glass pieces with similar thickness show characteristic hydroxyl ion and water signatures [54]. For clarity, Table 2 lists vibrational signatures of Sr-based carbonates, hydrates and hydroxides. The lack of specific signature on protonated polished ceramic confirms the absence of hydroxyl species. This reveals that the true bulk proton does not give any IR signature, which can be explained by the presence of another type of hydrogen species. The most reasonable hypothesis is interstitial protons, free of any specific strong hydrogen bonding. This type of proton was detected for the first time in 1997 by inelastic neutron scattering in some complex oxides and is called the ionic proton [21,55]. Another type of non-covalent bonded protons that do not show IR signature is a proton gas as observed in polyaniline by neutron scattering [21,56].

In order to go further into the comprehension of this discrepancy we simultaneously performed the IR study on low dense ceramics protonated in conditions favoring the adsorption of surface moieties. The low density sample—see Figure 7b—exhibits a significant IR signature in the 1,400–4,000 cm⁻¹ region (Table 2): A very broad structured massif similar to that recorded for well characterized materials hosting Mⁿ⁺(H₂O)_n species [3,28] and a very thin peak located near ~3,600 cm⁻¹ assigned to the OH⁻ vibrator [28]. Such a IR spectrum is very similar to that characteristic of Sr(OH)_x(CO₃)_y nH₂O powder.

It should be stressed that most of the IR spectra of different proton conductors presented in the literature [49,50,51,52,53] are very similar to that typical of low density ceramic and powder. This indicates that the authors analyzed mainly, or even exclusively, the IR signatures of second complex phases instead of those of conducting protons. The description of their samples (powder or poor sintered ceramics) confirms the presence of a very important active surface.