Vis and NIR Diffuse Reflectance Study in Disordered Bismuth Manganate—Lead Titanate Ceramics

Abstract

This work shows a correlation between light reflectance, absorption, and morphologies of series of bismuth manganate–lead titanate, (1 − x) BM–x PT, (x = 0.00, 0.02, 0.04, 0.08, 0.12, 0.16, 0.24, 1.00) ceramics composite. Low reflectance in the Vis-NIR range corresponds to ‘black mirror’ features. The modified Kubelka-Munk function applied to measured visible-near infrared (Vis-NIR) diffuse reflectance enabled the estimation of the energy gaps magnitude of the order of 1.0–1.2 eV for BM-PT. Histograms of grains, obtained using a scanning electron microscope, enabled finding the correlation between grains size, reflectance magnitude, and PT content. The magnitude of energy gaps was attributed to electronic structure bands modified by crystal lattice disorder and oxygen vacancies.

1. Introduction

Perovskite ABO₃ oxide materials attract attention for solar cells research and development and photovoltaic studies. The energy gap, E_gap, which separates the energies of the valence band, VB, maximum, and conduction band minimum, determines the material possibilities for such applications [1]. Electronic structure tuning is necessary for many materials to obtain suitable optical E_gap demanded for optical applications. The power conversion high efficiency also depends on the long lifetimes of excited charge carriers, which allows travel across long distances prior to recombination. Moreover, in accord to the Shockley–Queisser limit, the highest efficiency is reached at a band gap energy of 1.34 eV [2,3].

The ABO₃ VB is formed by oxygen O 2p states hybridized with states of metal 3d, 4d, and 5d placed in the B position. The 4d and 5d metal ions placed in B sublattice usually form a wide gap in the electronic structure. For instance, several perovskite titanates and niobates show insulator E_gap of the order of 3 eV and ultraviolet, UV, light range only is available for energy harvesting, due to limited light absorption in Vis range. It is worth noting that the light energies higher than the Vis-UV limit are insufficient for photovoltaic applications.

The wide gap ferroelectric perovskites provide low electrical conductivity and negligible dielectric losses. Moreover, it should be noticed that ferroelectric ordering enhances absorption efficiency due to empty d⁰ shells crucial for the formation of the ferroelectric state, which exerts a weak force on the moving charge carriers. Simultaneously, when the electron-hole pairs are formed after photon absorption, the electrons and holes can be separated and pushed towards opposite directions by the electric fields of the internal polarization [4,5,6].

Contrary, the narrow band perovskites can be obtained from the oxides of the metal ions from the 3d group, e.g., Fe, Mn, Cr, and Co. They offer better ability, including absorption in Vis and NIR, range. However, narrow optical E_gap results in low activation energy in electric conductivity, which induces marked electrical cross-current flow and losses [7,8,9,10,11,12,13,14,15,16,17,18,19].

There are several routes to modify the optical E_gap of oxide ABO₃ materials. Demanded properties can be obtained when materials exhibiting different E_gap magnitude are combined. Consequently, composites formed from oxide materials, which show large and narrow E_gap, respectively, allow modifying the optical properties because the narrower gap may enhance light absorbance efficiency in the Vis-NIR range. The substitution of the metal ions can lead either to solid solutions or the formation of precipitates. Minor phases may appear when the solubility limit is overstepped, and/or a transformation of the phases occurs [15,20]. The electronic structure can also be modified by an additional treatment that can change the concentration of the oxygen vacancies, V_O defects [21,22,23]. The V_O forms shallow and deep energy levels within the E_gap, increasing both absorbance at lower energies and electrical losses [24]. For instance, the active role of the Mn²⁺/Mn³⁺ in capturing photons in the NIR spectrum in the case of, e.g., multilayer Bi(Fe_1−xMn_x)O₃ photonic crystals and Mn-doped TiO₂ was analyzed [20,25,26,27].

From such points of view, narrow gap manganates and wide gap ferroelectric perovskites attract attention for photovoltaic applications [6,17,20,28]. In our study, we have chosen BM and PT to obtain ceramics composite. We presumed that ferroelectric PT addition to semiconductor BM would affect the electronic structure and influence light reflectance and absorption efficiency.

The PT shows a tetragonal P4mm phase at room temperature. Ferroelectric-paraelectric phase transition, which occurs in 763 K, points to the high stability of this compound ferroelectric phase. PT crystal is the ferroelectric insulator showing measured optical E_gap = 3.4 eV [29,30,31]. The tuning of the energy gap was achieved by defects and doping control. The Mn-doped PT crystals exhibited optical E_gap lowered to 2.0–2.7 eV, attributed to the occurrence of V_O and dipole Mn^k+-V_O defects [32]. The defected crystal, that is, N-V_O co-doped PT, showed the experimental E_gap = 2.93 eV [29]. The PT powder, which was used for the sintering of the BM-PT ceramic, was obtained from hydrothermal growth [33].

BM structure depends on growth conditions [34,35,36]. We sintered BM ceramics in standard high-temperature ambient air conditions, and a thermodynamically stable two-phase self-composite compound was obtained. Occurrence of two phases in equilibrium: 38% of the cubic I23 with the Bi₁₂MnO₂₀ composition and 62% of the orthorhombic Pbam with the BiMn₂O₅ composition were determined [37]. The self-composite BM was an electrically non-polar compound, which exhibited antiferromagnetic order in a low-temperature range with T_Néel ≈ 39 K [38]. In accordance with the literature, the BiMn₂O₅ phase exhibited E_gap = 1.24 eV for thin films [39], and E_gap =1.61 eV, for nano-grain material [40]. The E_gap = 0.78 eV [41] and E_gap = 1.23 eV [42] was calculated dependably on the modeling method. The sillenite Bi₁₂MnO₂₀ exhibits intrinsic E_gap = 1.86 eV, intermediate band related to the gap of ~1.07 eV, and VB placed 1.4 eV below the Fermi level [43]. It should be noted that the E_gap = 1.1 eV was estimated for the thin film perovskite BiMnO₃ variance [27,44]. Moreover, differences in energy states related to a material’s particular phases would be determined since their contribution can manifest as a superposition in the spectral response [44]. Therefore, diffuse reflectance of the (1 − x) BM–x PT composite ceramics should be determined in the Vis and NIR ranges of the photon spectrum.

A series of (1 − x) BM–x PT, (x = 0.00, 0.02, 0.04, 0.08, 0.12, 0.16, 0.24, 1.00) ceramics was sintered previously and their phases were determined using XRD, SEM, and dielectric impedance tests [45,46,47]. In this work, we studied diffuse reflectance in the Vis-NIR range to determine the PT content influence on the BM-PT composite absorption coefficient magnitude. It should be noted that the reflectance magnitude also depends on the properties of the surface exposed to the incident light. Hence, we determined ceramics’ morphology using scanning electron microscopy. This manuscript discusses the correspondence between light reflectance, estimated absorption coefficient energy gap, ceramics’, and morphology.

2. Materials and Methods

2.1. Preparation of Composite Ceramics

A series of (1 − x) BM–x PT (x = 0.00, 0.02, 0.04, 0.08, 0.12, 0.16, 0.24, 1.00) ceramics was produced via standard high temperature sintering. A two-stage procedure was performed. BM ceramics [48] and PT powder [33] were prepared separately at the first step. BM ceramics were produced using high-temperature sintering in the air (T_S = 1170 K for 2 h), according to the nominal BiMnO₃ formula [49]. They were graphite-like in color. The stoichiometric PT powder, which showed grains of sizes on submicron level (<500 nm), was obtained from the hydrothermal synthesis. It exhibited tetragonal P4mm symmetry [33]. The reference PT pellets were produced, and they were light yellow in color. Then the BM and PT powders were mixed in stoichiometric ratio, milled for 1 h, and pressed to pellets at 20 MPa. The ceramics were sintered at T_S = 1170 K in ambient air for 1 h [46,50]. The sintered BM-PT ceramics were hard, porous, and graphite-like in color. The structure of phases was determined using X-ray diffraction patterns in previous works [46,48,50].

2.2. Vis-NIR Reflectance

A USB-650 VIS-NIR Optic Spectrometer (OceanInsight, Inc., Orlando, FL, USA) equipped with a krypton light source lamp and OceanView 2.0 software was used. The Vis-NIR diffuse reflectance, R, spectra of the samples (~1 mm thick) were recorded in the wavelengths, λ, 400–1000 nm range at room temperature. The absorption coefficient was calculated from the reflectance spectra. The optical absorbance spectra were evaluated and presented using a Tauc plot of the modified Kubelka-Munk function to determine the optical band gap.

2.3. Surface Morphology

A scanning microscope JSM-5410 equipped with an energy dispersion X-ray spectrometer (EDS with Si (Li) X-ray detector) was used to characterize the morphology and elemental content. The measuring chamber vacuum was 10⁻⁴–10⁻⁵ Pa.

3. Results

3.1. SEM Morphology

The grains’ morphology of series x BM–(1 − x) PT ceramics was studied using SEM. The ceramics showed a variety of grain shapes: cubes, hexagon base polyhedrons, and tiny irregular forms. The voids and porosity were observed in ceramics surfaces.

Figure 1 shows secondary electron images obtained for the reference BM and PT ceramics. The BM map shows polyhedral-shaped hexagonal base grains with different sizes, varying predominantly in 0.2–2.0 μm in transverse dimension and 0.3–4.0 μm in longitudinal dimension, which is clearly visible in obtained polyhedrons grain size distribution (Figure 1a). There is a low number of grains with longitudinal size above 6.0 μm. The PT ceramics show homogenous morphology (Figure 1b). However, the grains’ size is markedly smaller, lower than 1.0 μm.

The 0.98 BM–0.02 PT ceramics (Figure 2a) shows variation in grain size similar to the pure BM ceramics. The obtained distribution shows that the grain transverse dimension is mostly between 0.5–2.5 μm. The longitudinal dimension is mostly between 0.2 and 3.5 μm. There is a low number of larger grains with sizes reaching 6.5 μm. On the contrary, the 0.96 BM–0.04 PT ceramics exhibits a higher number of small grains (Figure 2b). Moreover, obtained grains size distributions show an increased number of larger size grains compared to the previous composition. Namely, grain’ transverse dimension is mostly between 1.0–5.0 μm, and the longitudinal dimension is mostly between 0.5 and 4.0 μm. There are several large grains with longitudinal sizes above 10.0 μm.

Figure 3a,b show similar morphology of 0.92 BM–0.08 PT and 0.88 BM–0.12 PT ceramics, respectively. The obtained histograms show that the amount of the grain’ of transverse varies between 0.3–3.0 μm for both ceramics. The longitudinal dimension distribution is slightly different when compared for these two ceramics and varies between 1.0–5.5 μm for 0.92 BM–0.08 PT and 1.0–4.5 μm for the 12% of PT. There are no large grains with longitudinal size above 6.0 μm, which occur for the ceramics with a low amount of dopant PT (compare Figure 2 and Figure 3).

The obtained results for the high contents of PT are shown in Figure 4a,b. We can see that the 0.84 BM–0.16 PT structure contains both large and tiny grains. The transverse size of the grain is mostly between 0.3 and 2.0 μm. Moreover, there are grains, which dimensions reach 7.0 μm. Grains’ longitudinal dimensions vary from 0.2 μm to 12.0 μm (Figure 4a). In turn, the 0.76 BM–0.24 PT transverse size is mostly between 0.5 and 3.0 μm, while the grains’ longitudinal size is between 1.0 and 5.5 μm (Figure 4b).

The ceramic composites studied herein exhibit many caverns in the fracture surfaces (Figure 5). Such cavities are typical for the reference bismuth manganate ceramics [49]. Their origin would be associated with the technological process and attributed to chemical reactions. The occurrence of these cavities, where well-formed grains are visible, is one common feature of the BM-PT ceramics. The other common feature is porosity. We can distinguish well-shaped hexagon base polyhedrons grains, which are loosely attached to each other. Any clear tendency in size grain distribution in respect to PT content contribution was not observed for the BM-PT composite series. The majority of grains exhibited transverse size, which did not exceed 2.0 μm. The main number of grains showed longitudinal size lower than 7.0 μm. The exception occurred for 0.96 BM–0.04 PT and 0.84 BM–0.16 PT composition, where longitudinal size reached dimensions up to 12.0 μm. We attributed the occurrence of the small number of grains of sizes below 1.0 μm to the presence of the PT compound in the ceramics composite that is clearly visible for ceramics with higher content of PT.

Survey spectra obtained for the reference BM and PT ceramics are shown in Figure 6. The BM sample showed the occurrence of manganate Mn, bismuth Bi, oxygen O, and gold Au. The occurrence of Au line relates to the gold-sputtered on the samples. The PT showed the occurrence of lead Pb, titanate Ti, O, and Au. The EDS analysis was performed for areas of 300 × 300 μm² (see

Table 1. Exemplar atomic concentration (at%) estimated for chosen ceramics surfaces.

Atom	BM	Atom	PT	Atom	0.96 BM –0.04 PT	Atom	0.88 BM –0.12PT	Atom	0.76 BM –0.24 PT
Bi	50.2	Pb	48.5	Bi	45.0	Bi	43.6	Bi	36.5
Mn	49.8	Ti	51.5	Mn	51.4	Mn	45.1	Mn	38.3
				Pb	1.5	Pb	6.2	Pb	13.8
				Ti	2.1	Ti	5.1	Ti	11.4

). For the metal elements, the nominal chemical composition ratio is: Bi:Mn = 50.0:50.0 for BM ceramics, and Pb:Ti = 50.0:50.0 for perovskite PT. The EDS analysis showed, for BM ceramics, a real composition close to the nominal one. In the case of PT we observed a higher amount of Ti. This effect could be related to the evaporation of Pb during the sintering because of the high volatility of lead oxides. For BM-PT ceramics, survey spectra are shown in Figure 7. The samples showed the occurrence of Mn, Bi, Ti, Pb, and Au. The EDS analysis for Pb and Bi content is vague because of similar energy for these two elements. Hence, quantitative estimation suffers from overlapping of Pb and Bi most intensive lines in the analyzed spectrum. However, the obtained element content is close to expected. There are no impurities detected in all studies samples.

3.2. Vis-NIR Optical Features

The room temperature diffuse reflectance R spectra were measured in the 400–1000 nm range of Vis-NIR spectrum. For the graphical estimation of the difference in energy states, E_gap, magnitude, the Vis-NIR spectra recorded in the R mode were transformed to Kubelka-Munk equation:

F(R) = (1 − R)²/2R

(1)

which is proportional to the extinction coefficient, α. Determination of E_gap was conducted using the Tauc method and relation of incident photon energy, E = hν, with the modified Kubelka-Munk function:

(α·hν)^1/n ≈ B(hν − E_gap)

(2)

We presumed the indirect allowed transitions in the BM-PT composite. Hence, (α·hν)^1/n vs. hν plots, where exponent n = 2, were applied for analysis. The E_gap was estimated by extrapolating the straight-line segment to the intersection with the hν axis, that is for (αhν)^1/2 = 0 [51,52].

We note that the PT, BiMn₂O₅, Bi₁₂MnO₂₀, and their derivative phases, which resulted from sintering, bring in individual electronic structure contributions. Moreover, the mutual migration ions and V_O defects may introduce intermediate band gap states and tails in the vicinity of the VB and conductivity band, CB, [24,43]. Therefore, any superposition spectrum, obtained for a doping-modified and two-phase semiconductor, is the ‘effective’ linear combination of the spectra of both components:

(α(hν))_eff = a·α₁(hν) + b·α₂(hν))

(3)

where a and b determine the components’ concentrations contributions, while α₁(hν) and α₂(hν) are the absorption coefficients of the assumed two phases contribution [6,44,52,53]. Therefore, an effective energy gap is determined using the Tauc relation:

{[(α(hν))_eff ] hν)}^1/2 = B(hν − E_gap)

(4)

Following the proposed approximation, we decided to conduct the spectra analysis using the Tauc method [44,51]. We performed extrapolation where straight-line slopes could be distinguished.

The diffuse reflectance spectra of BM and PT reference ceramics are shown in Figure 8. The R(λ) increased when wavelength increased, more steeply for λ > 900 nm. A broad hump occurred in ~550–800 nm range, corresponding to red and dark red light. The most distinct difference occurred in the 400–550 nm range, where PT reflectance exhibited a steep change in R magnitude, contrary to the BM sample.

We discerned the straight-line segment in the Tauc plot obtained for the reference BM self-composite, and E_gap(NIR) = 1.20 eV was estimated for the NIR spectrum range (Figure 9a). E_gap(Vis) = 2.45 eV was estimated in the case of the reference PT compound (Figure 9b).

R(λ) spectra obtained for (1 − x) BM–x PT composites are shown in Figure 10. R magnitude decreased when the PT content increased, from ~50–80% to ~25–30% despite that the reference PT R(λ) varied in the 40–100 % range (compare Figure 6 and Figure 8). The R(λ) common feature was its magnitude increase when wavelength increased and a more steep change in R magnitude for λ > 900 nm. A more or less distinct broad hump occurred in ~550–800 nm range; it shifted toward NIR range and became negligible for high content of PT component.

Graphical representation of the modified Kubelka–Munk function allowed us to determine E_gap magnitude dependence on the (1 − x) BM–x PT ceramics composition (Figure 11 and Figure 12). The E_gap(NIR) of the order of ~1 eV was estimated for BM-PT ceramics, and it showed a tendency to fluctuate with PT content change (Figure 12).

4. Discussion

The reference for PT ceramics was the one-phase compound. The light-yellow color of this material suggested that the actual E_gap magnitude originated from a defected PT structure would be lower from E_gap = 3.4 eV reported for the ideal stoichiometric transparent crystal [29,30]. We note that the E_gap = 3.4 eV of such PT can only be detected in the UV range, which was beyond our study frames. Hence, the E_gap(Vis) = 2.45 eV (Figure 9b) should correspond to the effects of defects and disorder. One notes that V_O defects introduce shallow and deep energy levels in the perovskite ABO₃ structures [24]. Such deduction also is justified by the literature reports. It is consistent with the results of calculations provided for defected PT, reported in the literature, where the full potential linearized augmented plane-wave calculation exhibited that oxygen vacancies induce additional in-gap Ti 3d states -1.1 eV below the Fermi energy [54]. Indirect optical gap varied from 2.78 eV to 3.66 eV dependably on the PT surface termination [31,55]. The experimentally determined E_gap = 2.6 eV seems to be also related to the defects subsystem of PT [56]. Therefore, E_gap(Vis) = 2.45 eV should be attributed to the gap between the defect-induced sub-band and CB, which position and possible tails were determined by crystal lattice disorder. These literature E_gap data are shown in Figure 13.

The BM and BM-PT ceramics are multi-phase composites [37,48,49]. BM is the self-composite consisting of the major BiMn₂O₅ and the minor Bi₁₂MnO₂₀ phase. The BM-PT composite consists of phases, which structure changed because of mutual migration of ions when the ceramics were sintered from the reference BM and PT [46,50]. For instance, it would be noted that the Bi₂O₃ phase, iso-structural with Bi₂₀TiO₃₂, transformed from the Bi₁₂MnO₂₀ parent phase of the BM self-composite [50]. Moreover, local fluctuations in composition and the occurrence of oxygen deficiency cannot be excluded. The graphite-black color of these BM-PT compounds indicated marked light absorbance in the Vis range.

Therefore, the measured R(λ) spectra were considered as effective quantities, which included contributions from individual phases of the BM-PT system. It should be noted that the lower R magnitude occurred when the PT content in the BM-PT ceramics was higher. The R magnitude decrease from ~50–80% to ~25–30% indicated an active role of the PT component and influence porosity. Similar to a glassy black mirror, a standard flat and smooth surface usually offers reflectance on a 40–50% level. Lower R would be achieved in the case of a textured surface when light is scattered and absorbed more effectively [57]. It should be noticed that the SEM test exhibited porosity and concave voids in the BM-PT ceramics surfaces.

We note that the reference PT powder grain was much smaller than the BM-PT compounds grain (compare Figure 1b to Figure 1a, Figure 2, Figure 3, Figure 4 and Figure 5). Hence, we deduce that tiny grains of the PT powder introduced to the sintered BM-PT ceramics also would be responsible for this effect. We conclude that both the small grains of the PT component and the porous texture of BM-PT ceramics surface may facilitate scattering, decrease R to ~25%, and enhance light absorption. We noticed a correlation between R and ceramics morphology. The BM, 0.96 BM–0.04 PT, and PT ceramics showed R higher magnitude and more homogenous morphology than the other doped BM-PT ceramics. Moreover, the BM and PT samples were characterized by the smaller grains and the highest reflectance. Therefore, despite any uncontrolled and random factors that cannot be excluded a priori, we deduce that the low reflectance of the BM-PT ceramics can relate to the rough surface features.

We applied the Tauc procedure within the NIR following the assumed effective reflectance spectrum superposition obtained for the BM and BM-PT multi-phase composites [43,44,58]. The E_gap(NIR) = 1.20 eV, estimated for the BM self-composite, correlates to the literature data. The studied herein BM consists of the major BiMn₂O₅ phase and minor Bi₁₂MnO₂₀ phase. One notices that the E_gap = 1.24 eV was reported BiMn₂O₅ compound [44]. The case of Bi₁₂MnO₂₀ compound offers more possible charge transfers determined by Wu et al. [43]: the charge excitation between VB and CB of intrinsic E_gap = 1.86 eV, other from the VB to the intermediate band, IB, of ~0.79 eV, and also excitation from the intermediate band to the CB of ~1.07 eV (see Figure 13). Therefore, we do not assign the effective E_gap(NIR) = 1.20 eV to specific phase. Such the choice would be vague due to the similarity of the energy gap size; either E_gap = 1.24 eV of BiMn₂O₅, or E_gap = 1.07 eV of Bi₁₂MnO₂₀ can be chosen. We note that the coincidence of estimated herein E_gap(NIR) = 1.20 eV to the literature reports on the perovskite BiMnO₃ compound (E_gap = 1.1 eV) is random due to this unstable perovskite phase was not detected in case of the studied herein BM. In the case of the (1 − x) BM–x PT compounds, the obtained magnitude of E_gap fluctuated on E_gap(NIR) = 1.00 ± 0.15 eV level (see Figure 12). We deduce an effective role of local disorder, which would influence the electronic structure of the studied material. The V_O occurrence would induce an additional sub-bands within the energy gap [24]. In case of the Bi₁₂MnO₂₀ phase, the V_O creates the IB band, mentioned above, which can be responsible for the charge transfer. However, in case of the BiMn₂O₅, additional IB energy levels, leading to lower than 1 eV excitation energy, were not observed, using the applied NIR range. We propose that defects, most probably caused by mutual diffusion of ions between the phases, resulting from the final sintering procedure [46,50], create tails of states in the VB and CB that lead to the detected narrowing of the energy gap (see diagram in Figure 14).

In summary, we would emphasize the most important properties of studied herein composite ceramics.

• In the case of BM and PT reference materials, it was possible to attribute determined E_gap magnitudes to electric charge excitation from the VB to CB (BiMn₂O₅), to transition from IB to the CB (Bi₁₂MnO₂₀), and from VB to IB (PT);
• The (1 − x) BM–x PT composite exhibited E_gap(NIR) = 1.00 ± 0.15 eV lower than E_gap magnitudes of the reference compounds. This gap narrowing effect originated from a disorder;
• Determined properties of (1 − x) BM–x PT, especially the narrowing of the E_gap magnitudes, of the order of 1 eV, which corresponds to marked light absorption in Vis and NIR, suggest potential optoelectrical applications.