Oxygen‐Vacancy‐Driven Persistent Luminescence in Mn²⁺‐Doped MgGeO₃: Insights from X‐Ray Absorption Fine Structure Spectroscopy

Abstract

Mn²⁺‐doped MgGeO₃ (MMGO) is a near‐infrared‐emitting persistent luminescence (PersL) phosphor. Its low luminescence decay has been linked to vacancies within the MgGeO₃ host, likely associated with Ge, although no direct correlations have been confirmed. In this study, the Ge K‐edge X‐ray absorption fine structure (XAFS) spectroscopy is employed to investigate a series of MMGO submicron‐sized particles with varying afterglow durations. These samples are synthesized using a recently developed hydrothermal method, where it is found that the PersL duration is related to the annealing temperature and the presence of codopants. It is first demonstrated that the XAFS spectra obtained using an in‐house XAFS spectrometer are comparable in quality to those acquired at synchrotron facilities. Analysis of the local Ge environment of the MMGO shows that the PersL duration is directly related to the Ge–O coordination number. Additionally, the introduction of Li⁺ extends the PersL duration of MMGO but does not alter the Ge–O coordination. The findings provide insights into the structural mechanisms driving luminescence and highlight the capability of in‐house XAFS systems to deliver high‐quality data for advanced materials research.

X‐ray absorption fine structure (XAFS) spectroscopy data obtained using an in‐house spectrometer (Lab‐XAFS) are shown to be comparable to synchrotron measurements. Lab‐XAFS is then used to investigate the role of oxygen vacancies surrounding Ge and the persistent luminescence of Mn²⁺‐doped MgGeO₃ under thermal annealing and codoping, providing insights into the structural mechanisms that govern luminescence behavior.

1. Introduction

Persistent luminescence (PersL) is an optical phenomenon characterized by a material's ability to emit light continuously for several seconds, minutes, or even hours after the excitation source is removed. Solid‐state inorganic PersL materials are crystalline hosts doped with transition metal or rare‐earth metal ions, which serve as activators and/or emitters. The wavelength, intensity, and duration of PersL depend on electron and energy transfer processes between the host material and dopants. Recently, near‐infrared (NIR) PersL nanoparticles have attracted significant attention for their wide range of applications in targeted drug delivery, biosensing, bioimaging, and anticounterfeiting.^{[ 1} , ^{2 ]} Transition metal ion dopants frequently employed as emitting centers to achieve NIR luminescence include Cr³⁺, Mn²⁺, and Mn⁴⁺. These dopants can be incorporated into various host matrices, such as silicates, germanates, aluminates, and gallates, to exhibit NIR PersL.^{[ 1 ]}

Mn²⁺‐doped MgGeO₃ (MMGO) was first discovered as an NIR emitter with a long decay lifetime by Iwasaki et al. in 2003. The authors initially referred to this PersL behavior as phosphorescence.^{[ 3 ]} They proposed that the long‐lasting luminescence was linked to the chemical basicity of the host lattice and the reactivity of GeO₂. Since then, several studies have demonstrated that MMGO is a highly effective PersL emitter.^{[ 4} , ^{5 ]} Recent advancements in synthesis techniques, particularly the hydrothermal method, have succeeded in synthesizing submicron‐sized MMGO with uniform morphology,^{[ 6 ]} making it a promising candidate for microscale optical devices. Beyond using Mn²⁺ as a single dopant, codoping MMGO with various ions, such as Yb³⁺, Bi³⁺, Eu²⁺, and Li⁺, has been found to be effective in modulating its photoluminescence (PL) and PersL properties.^{[ 7} , ⁸ , ⁹ , ¹⁰ , ^{11 ]} Notably, adding Yb³⁺ or Eu²⁺ introduces a second NIR emission band, and Bi³⁺ serves as a separate electron trap that extends the afterglow duration of MMGO.

The original hypothesis attributing the origin of PersL in MMGO to oxygen vacancies within the MGO lattice has been widely accepted and echoed by many researchers.^{[ 5} , ⁷ , ^{12 ]} As illustrated in Figure  1 , upon ultraviolet (UV) radiation (Step 1), electrons are promoted to the conduction band and then can be captured by oxygen vacancies within the MgGeO₃ (MGO) host (Step 2). These vacancies serve as traps and temporarily store the electrons. Over time, the trapped electrons are released into the excited states of Mn²⁺ (Step 3), followed by radiative decay to the ground state, producing long‐lasting luminescence, or PersL (Step 4). Therefore, the intensity and duration of PersL are directly influenced by the density and depth of electron trap states present in the host lattice. Several characterization techniques, such as thermally stimulated luminescence and electron paramagnetic resonance spectroscopy, have confirmed the presence of deep electron traps.^{[ 12 ]} Not only in MGO, oxygen vacancies in the host lattice have also been reported to play a role in modulating the optical properties of several other PersL phosphors.^{[ 13} , ¹⁴ , ^{15 ]} However, there has been no direct experimental evidence correlating oxygen vacancies with PersL duration, and whether these vacancies are specifically associated with a specific element (Ge in the case of MGO). Additionally, while the extended PersL duration in codoped MMGO has been attributed to additional electron traps brought by these codopants,^{[ 9} , ^{16 ]} there has been no discussion on whether changes in the host lattice vacancies might also contribute to the prolonged PersL duration.

Figure 1.

Schematic illustration of the PersL mechanism for MMGO based on previous studies. The numbers represent the electron transfer process: 1. Excitation, 2. electron trapping, 3. electron de‐trapping, and 4. radiative decay.

X‐ray absorption fine structure (XAFS) spectroscopy has become an increasingly popular tool for studying the local electronic structure of solid‐state and nanomaterials. Being specific to the element of interest, XAFS spectroscopy analyzes how a target element coordinates with its nearest neighbors, with no prerequisite for materials being crystalline. XAFS spectroscopy has been used to characterize metal ion‐doped luminescent materials, but previous studies have primarily focused on the chemical environment of the dopants rather than the host lattice.^{[ 17} , ¹⁸ , ^{19 ]} We previously conducted an XAFS study on MMGO codoped with Yb³⁺ and Li⁺, and analyzed the Mn²⁺ chemical environment in ternary Mg–Ge–O compounds with varying stoichiometries.^{[ 20 ]} In this work, we focused on analyzing the local environment of Ge in MMGO synthesized under different annealing temperatures and codoped with Li⁺.

The principle of XAFS spectroscopy relies on using X‐rays with tunable energies near the absorption threshold of a core electron of an element for spectral acquisition. Conventionally, the technique has been limited to synchrotron facilities due to the requirement for high‐intensity, tunable X‐rays. Another key contribution of this work is demonstrating that laboratory‐based XAFS (lab‐XAFS) can achieve spectra comparable in quality to those obtained at synchrotron facilities. We first compared the lab‐XAFS data with synchrotron‐based XAFS, followed by a comprehensive analysis of MMGO using lab‐XAFS instrumentation.

2. Results and Discussion

2.1. Comparison of Synchrotron‐ and lab‐XAFS Spectra

Figure  2 presented the Ge K‐edge XAFS spectra of commercial GeO₂ powder, acquired using both the lab‐XAFS system and a synchrotron beamline for comparison. Both measurements were performed using the transmission mode, with samples pressed and sealed between X‐ray transparent tapes. The synchrotron XAFS spectrum was collected with a single scan that took ≈20 min, while the lab‐XAFS data were obtained from an average of five scans, totaling 195 min of acquisition time. Despite the longer collection time required for lab‐XAFS, the X‐ray absorption near‐edge structure (XANES) region was successfully reproduced. The characteristic peaks, denoted with three arrows in Figure 2a, appearing after the main absorption peak, correspond well to GeO₂ in the quartz structure.^{[ 21} , ^{22 ]} Further comparisons between the spectra obtained from the lab and synchrotron‐based XAFS were shown in Figures 2b and 2c, after the spectra were converted into the k‐space and R‐space, respectively. The lab‐XAFS exhibited a lower signal‐to‐noise ratio at k‐values above 10 Å⁻¹. This is primarily because of the lower photon flux from the X‐ray tube source compared to synchrotron radiation, particularly at high k‐values where the signal is inherently weaker. Despite this limitation, the Fourier‐transformed XAFS (FT‐XAFS) spectra remained highly comparable within the radial distance range of 0 Å to 4 Å, indicating that the local electronic structural information is well preserved. The results aligned closely with existing literature,^{[ 23 ]} reinforcing the reliability of lab‐XAFS measurements. In the following discussion, all XAFS data for MMGO were acquired using the lab‐XAFS spectrometer.

Figure 2.

Comparison of Ge K‐edge XAFS spectra of GeO₂ powder acquired using lab‐based and synchrotron‐based XAFS. a) XANES, b) k‐space XAFS, and c) FT‐XAFS.

2.2. The Effect of Annealing Temperature on the Structure and Luminescence of MMGO

The synthesis of MMGO was based on a hydrothermal protocol, followed by annealing at 825, 950, and 1050 °C, respectively. The crystal structures of the resulting products were characterized by X‐ray powder diffraction (XRD), as shown in Figure  3a. All MMGO samples had an orthorhombic structure corresponding to the Pbca phase of MgGeO₃ (JCPDS 04‐008‐8425), indicating that the overall crystal structure remained unchanged across the annealing temperatures and that no secondary phases were present. The diffraction peaks were broad in MMGO_825, indicating lower crystallinity, but became sharper as the annealing temperature increased above 950 °C. With improved crystallinity, MMGO emitted red luminescence under 254 nm excitation, appearing as a single broad peak centered at 680 nm (Figure 3b). The origin of this luminescence is from the ⁴T₁(⁴ G) → ⁶A₁(⁶S) 3 d−3 d transition of Mn²⁺ in an octahedral environment. Because of the weak luminescence from MMGO_825, PersL was only detected from MMGO_950 and MMGO_1050, with the latter exhibiting a slower decay (Figure 3c).

Figure 3.

a) XRD patterns of MMGO annealed at 825 °C (MMGO_825), 950 °C (MMGO_950), and 1050 °C (MMGO_1050), with the standard MgGeO₃ pattern (JCPDS 04‐008‐8425) shown for comparison. b) PL spectra of MMGO_825, MMGO_950, and MMGO_1050 under 254 nm UV light. c) Decay curves of PersL intensity (log scale) as a function of time over 100 s.

The Ge K‐edge XANES spectra of MMGO were presented in Figure  4a. For comparison, the spectrum of unannealed MMGO was also included. Since the K‐edge transition originated from the excitation of Ge 1s electrons to unoccupied p states, the white‐line intensity is related to the p state occupancy. When Ge is surrounded by more oxygen atoms, the local electron density around Ge decreases, leading to an increase in white‐line intensity. For example, the four‐coordinated Ge–O in GeO₂ of a quartz structure has a lower white‐line intensity than the six‐coordinated Ge–O in GeO₂ of a rutile structure.^{[ 22} , ²³ , ²⁴ , ^{25 ]} The evolution of the white‐line intensity revealed changes in the Ge local environment during the hydrothermal synthesis of MMGO.

Figure 4.

XAFS spectra of unannealed MMGO, MMGO_825, MMGO_950, and MMGO_1050 at the Ge K‐edge. a) XANES, b) FT‐EXAFS, and c) fitted first‐shell EXAFS plotted in R‐space. The dotted lines represent the experimental data and the solid lines represent the fitted spectra.

Conventionally, MMGO was synthesized by the solid‐state method: a mixture of GeO₂ and MgO undergoes high temperature heating, facilitating the formation of a new solid compound through thermally induced atomic rearrangement and diffusion.^{[ 3} , ^{26 ]} However, in solution‐based synthesis, such rearrangement occurs among solvated ions. During the synthesis of MMGO, GeO₂ was first dissolved in NH₄OH before the introduction of Mg²⁺ salts. Previous studies suggest that in ammonia‐based basic solutions, GeO₂ forms soluble complexes, such as Ge₅O₁₁ ²⁻. In these complexes, Ge has a six‐coordinated structure.^{[ 27} , ²⁸ , ^{29 ]} Upon mixing with Mg²⁺, a precipitate formed, but some of the Ge remained in an octahedral coordination state.

The XRD pattern of unannealed MMGO was shown in Figure  5 . The compound exhibited broad diffraction patterns that do not correspond to MgGeO₃. Instead, the patterns closely resembled the pattern of magnesium germanium hydrate (Mg_4.25Ge₄O_18.25H₁₂, JCPDS 00‐012‐0063).^{[ 5} , ^{30 ]} One peak at 2θ = 28.3° matches the (110) diffraction plane of GeO₂ of a rutile structure (where (110) corresponds to the Miller Indices notation (hkl)), in which Ge has an octahedral coordination. This further suggested that six‐coordinated Ge was involved during the formation of MMGO. As the annealing temperature increased, the white‐line intensity decreased, suggesting a reduction in the Ge–O coordination number. This indicated that octahedrally coordinated Ge were progressively converted into GeO₄ tetrahedral coordination, consistent with the structure of MgGeO₃.

Figure 5.

XRD pattern of unannealed MMGO in comparison with standard XRD patterns of Mg_4.25Ge₄O_18.25H₁₂ (JCPDS 00‐012‐0063), rutile GeO₂, and MgGeO₃.

To further examine changes in the Ge–O coordination number, EXAFS spectra were analyzed. As shown in Figure 4b, the Fourier‐transformed EXAFS (FT‐EXAFS) spectra of all MMGO samples exhibit one distinct peak which corresponds to Ge–O bonding. Based on a qualitative comparison, the peak amplitude, which correlates with the coordination number of Ge, was highest in unannealed MMGO. As the annealing temperature increased, the peak intensity lowered, indicating a reduction in the Ge–O coordination number. To quantitatively assess the Ge–O bonding environment, a first‐shell fit was conducted. The fitted spectra are presented in Figure 4c, and the full fitting parameters are summarized in Table  1 . The unannealed MMGO had a Ge–O coordination number of 4.6, confirming the coexistence of GeO₄ tetrahedra and GeO₆ octahedra. Upon thermal annealing, the coordination number decreased, reaching 4.0 for MMGO_950. Interestingly, further annealing at higher temperatures resulted in the formation of undercoordinated Ge, indicating the presence of oxygen vacancies. Both MMGO_950 and MMGO_1050 exhibited sharp diffraction peaks characteristic of orthorhombic MgGeO₃, yet MMGO_1050 showed stronger PL and a longer PersL duration compared to MMGO_950. PL and PersL are governed by different factors: PL intensity is primarily influenced by crystallinity and the presence of nonradiative quenching centers, while PersL is controlled by the nature and concentration of deep trap states. High‐temperature annealing enhances both crystallinity (eliminating nonradiative quenching centers) and the formation of oxygen vacancies, the latter act as effective deep traps. Therefore, the improved PersL in MMGO_1050 is attributed to oxygen vacancies near Ge atoms that facilitate long‐lived charge‐trapping and delayed recombination. These findings provide direct evidence supporting the role of oxygen‐vacancy‐related defects in governing the PersL behavior of MMGO.

Table 1.

First shell EXAFS fitting parameters of unannealed MMGO, MMGO_825, MMGO_950, and MMGO_1050 at the Ge K‐Edge.

	Scattering path	C.N.	R [Å]	ΔE [eV]	$\sigma$ 2 [Å2]	R factor
Unannealed MMGO	Ge—O	4.6 (0.1)	1.747 (0.026)	2.739 (4.024)	0.005	0.048
MMGO_825	Ge—O	4.3 (0.2)	1.692 (0.044)	−4.070 (8.083)	0.005	0.018
MMGO_950	Ge—O	4.0 (0.1)	1.736 (0.028)	2.107 (4.569)	0.005	0.021
MMGO_1050	Ge—O	3.2 (0.2)	1.743 (0.015)	1.944 (2.361)	0.005 (0.001)	0.016

NOTE: C.N.: coordination number; R: radial distance; ΔE: energy shift; σ ²: Debye–Waller factor.

In addition to the nearest neighbors, the second shell feature (signals between 2 Å and 3.5 Å) is also worth examining qualitatively. However, signals within this region have complex contributions. Referring to the structure of MgGeO₃,^{[ 31 ]} there are two unique Ge sites: the Ge(1) site has two Ge neighbors at a distance of 3.112 Å, while the Ge(2) site has one Mg neighbor at 2.882 Å, and two Ge neighbors at 3.173 Å. In contrast, for magnesium germanium hydrate (the sample before annealing), the second shell is primarily composed of Mg at distances ranging from 3.0 Å to 3.5 Å. To better visualize changes in the second shell of MMGO at different annealing temperatures, EXAFS k‐space oscillations were analyzed using wavelet transformation (WT‐XAFS), as shown in Figure  6 . The unannealed MMGO exhibited a stronger signal at high radial distances, which corresponds to Ge surrounded by Mg in the hydrate structure. In the annealed sample, the distance of the second shell moved closer to Ge, and the intensity decreased, indicating a reduction in the number of Ge neighbors in the final MgGeO₃ structure.

Figure 6.

WT‐XAFS of MMGO samples. a) unannealed MMGO, b) MMGO_825, c) MMGO_950, and d) MMGO_1050.

2.3. The Effect of Codoping on the Structure and Luminescence of MMGO

Increasing the annealing temperature effectively extended PersL duration by creating more oxygen vacancies near Ge, as confirmed by the XAFS results shown in the previous section. Next, we examined the effect of introducing Li⁺ codopants on the local structure of Ge and the luminescence properties of MMGO. Li⁺ and Mn²⁺ codoped MMGO was annealed at 1050 °C (denoted MMGO‐Li_1050) and compared with MMGO_1050. XRD analysis (Figure  7a) confirmed that introducing Li⁺ did not alter the crystal structure of MgGeO₃, as the diffraction patterns of both samples were nearly identical. Although the PL intensity of MMGO‐Li_1050 was slightly weaker in intensity, its decay was significantly slower compared to MMGO_1050 (Figure 7b), indicating an extended PersL duration.

Figure 7.

(Revised). Crystal structure, optical properties, and electronic structure comparisons between MMGO_1050 and MMGO‐Li_1050. a) XRD, b) PL spectra under 254 nm excitation, with an inset showing the PersL decay curve (intensity was plotted in log scale), c) the Ge K‐edge XANES, d) FT‐EXAFS with fitted spectra displayed in the R‐space, e) WT‐EXAFS of MMGO_1050, and f) WT‐EXAFS of MMGO‐Li_1050.

Figure 7c compared the Ge K‐edge XANES spectra of MMGO_1050 and MMGO‐Li_1050. The nearly overlapping spectra suggest that the introduction of Li⁺ has minimal impact on the oxygen vacancies surrounding Ge. This was further confirmed by the first‐shell fitting analysis, summarized in Table  2 , where the coordination numbers (CN) of Ge–O were found to be the same within the fitting error for both samples. This evidence suggested that the prolonged PersL duration in MMGO‐Li_1050 originates from a different mechanism rather than oxygen vacancies around Ge. The ionic radius of Li⁺ in an octahedral environment is 76 pm, which is comparable to that of Mg²⁺ (72 pm) in the same coordination environment.^{[ 32 ]} When incorporated into the MMGO lattice, Li⁺ could occupy the Mg²⁺ sites. However, the resulting charge imbalance acts as an electron trap, contributing to the extended PersL observed in MMGO‐Li_1050.

Table 2.

First shell EXAFS fitting parameters of MMGO_1050 and MMGO‐Li_1050 at the Ge K‐Edge.

	Scattering path	C.N.	R [Å]	ΔE [eV]	$\sigma$ 2 [Å2]	R factor
MMGO_1050	Ge—O	3.2 (0.1)	1.743 (0.015)	1.847 (2.331)	0.005	0.012
MMGO‐Li_1050	Ge—O	3.2 (0.1)	1.746 (0.044)	2.209 (3.867)	0.005	0.017

This also explained the decreased PL intensity of MMGO‐Li_1050 compared to MMGO_1050. If both Li⁺ and Mn²⁺ compete for Mg²⁺ sites, MMGO‐Li_1050 is expected to incorporate fewer Mn²⁺ dopants. To verify this, energy dispersive X‐Ray (EDX) spectroscopy was performed to obtain the elemental composition of MMGO_1050 and MMGO‐Li_1050. As shown in Table  3 , the Mn content (%) was slightly lower in MMGO‐Li_1050. Furthermore, WT‐EXAFS analysis (Figure 6f) revealed a weakened second shell contribution (at radial distance ≈2 Å) in MMGO‐Li_1050, further suggesting Li⁺ substitution at Mg²⁺ sites. Since Li⁺ is a lighter scatterer compared to Mg²⁺, the substitution resulted in a decreased amplitude in the EXAFS signal.

Table 3.

Elemental composition of MMGO_1050 and MMGO‐Li_1050 from EDX analysis.

	Atomic [%]				Ratio
	Mg	Ge	O	Mn	Mn/[Mg+Mn]	Mn/Ge
MMGO_1050	13.93 ± 1.07	14.50 ± 1.26	46.63 ± 1.39	0.48 ± 0.10	0.033 ± 0.006	0.033 ± 0.005
MMGO‐Li_1050	13.96 ± 0.65	14.32 ± 1.39	48.96 ± 2.14	0.29 ± 0.10	0.020 ± 0.006	0.020 ± 0.005

3. Conclusion

In summary, this study demonstrated the effectiveness of lab‐XAFS as a convenient and reliable technique for obtaining synchrotron‐quality data for materials analysis. Our findings revealed that during the hydrothermal synthesis of MMGO, Ge initially adopts a GeO₆ octahedral coordination. Unlike conventional hydrothermal syntheses, where postannealing primarily enhances crystallinity, in MMGO synthesis, annealing is required to convert all Ge to tetrahedral coordination. We confirmed that high‐temperature annealing produces oxygen vacancies around Ge, providing direct experimental evidence for the previously proposed PersL mechanism in MMGO. These vacancies serve as charge‐trapping sites, contributing to the slow decay of luminescence. Additionally, we demonstrated that introducing Li⁺ as a codopant significantly enhances the PersL properties of MMGO. While Li⁺ doping did not affect the existing oxygen vacancies around Ge, it substituted for Mg²⁺ in the lattice, introducing new trap sites that extended the PersL duration. The outcomes of this study provide fundamental insights into the structural evolution and defect chemistry underlying the luminescence properties of MMGO while also highlighting the effectiveness of lab‐XAFS for precise materials analysis.

4. Experimental Section

4.1.

4.1.1.

Materials

Magnesium nitrate hexahydrate (Mg(NO₃)₂ · 6H₂O, ACS reagent >99%), germanium oxide (GeO₂, hexagonal phase, ≥99.99% trace metal basis), lithium nitrate (LiNO₃, ≥99.99% trace metal basis), and ammonium hydroxide (NH₄OH, 35 wt%), were obtained from Sigma‐Aldrich. Manganese (II) chloride tetrahydrate (MnCl₂ · 4H₂O, ≥99.99% metal basis) was purchased from Alfa‐Aesar. Type‐1 water (resistivity: 18.2 MΩ.cm) obtained from Thermo Scientific was used as the solvent in the synthesis. All chemicals were used as received.

Synthesis Procedure

MMGO submicron particles were synthesized using a hydrothermal method developed by our group.^{[ 6 ]} The Ge⁴⁺ precursor solution was prepared by dissolving 1.308 g of GeO₂ in 35 mL of Type‐1 water and 0.6 mL of concentrated NH₄OH with the assistance of ultrasonication. The Mg²⁺ precursor solution was prepared by dissolving 3.205 g of Mg(NO₃)₂ · 6H₂O in 10 mL Type‐1 water. The two solutions were mixed, and 250 μL of 0.5 mol L⁻¹ MnCl₂ · 4H₂O was added to the solution. The pH of the solution was adjusted to 8 using NH₄OH, reaching a total volume of 50 mL. After 2 h of stirring at room temperature, the solution was transferred into a Teflon‐lined autoclave for hydrothermal treatment at 200 °C for 12 h. The solid products were collected by centrifugation at 8000 rpm for 3 min, washed three times with Type‐I water, and annealed at 825, 950, and 1050 °C, respectively. During high‐temperature annealing, the heating rate was maintained at 200 °C/h, with a 1 h holding time at the target temperature. After annealing, the crude products were suspended in 5 mM NaOH and ultrasonicated for 1 h. The final products were washed three times with Type‐I water and dried at 60 °C. The synthesized samples were denoted as MMGO_825, MMGO_950, and MMGO_1050.

Li⁺‐doped MMGO (MMGO_Li_1050) was prepared following the same procedure, except after the addition of MnCl₂ · 4H₂O, 1.25 mL (0.1 mol L⁻¹) of LiNO₃ was mixed in, followed by adjusting the pH to 8 and the total volume to 50 mL. The annealing temperature for Li⁺‐doped MMGO was at 1050 °C.

Characterization

The crystal structure of the samples was characterized using XRD using an Inel XRG3000 generator and an Inel CPS 120 detector (Cu Kα tube source). PL spectra were measured using a fiber‐optic spectrometer (Avantes, AvaSpec‐ULS2048XL‐EVO) under 254 nm excitation. The PersL decay curves were obtained using a photodiode with an open filter (Hamamatsu, C10439–03), and the samples were pre‐excited via a 254 nm flashlight for 20 s prior to the measurement. Elemental composition was analyzed using EDX (Oxford Aztec X‐Max50 SDD X‐Ray analyzer) attached to a scanning electron microscope (SEM, Hitachi SU3500 Variable Pressure SEM) at Surface Science Western. Samples were placed on carbon tape and sputter‐coated with gold (Hummer VI Sputter Coater) before analysis. EDX measurements were performed at an accelerating voltage of 15 kV.

Synchrotron‐based XAFS spectra were collected at the National Synchrotron Radiation Research Center (NSRRC), Taiwan Light Source, Beamline 17C1. For lab‐based measurements, XAFS data were collected using an easyXAFS300+ spectrometer (easyXAFS LLC), which operates based on Rowland circle geometry using spherically bent crystal analyzers. The Si(1,3,9) crystal was used for the Ge K‐edge measurement. All spectra were collected in transmission mode using a silicon drift detector. For data normalization, the incident X‐ray was measured by running a scan on an open spot of the sample holder. Extended XAFS (EXAFS) analysis was performed using the Demeter software package.^{[ 33 ]} The R‐space EXAFS signal was obtained using a variable kⁿ‐weighted Fourier transform (n = 1, 2, 3) of the χ(k) signal. The Ge–O paths from all the samples were constructed from MgGeO₃ of orthorhombic Pbca phase. The analysis used a k‐range of 3 Å⁻¹ to 11 Å⁻¹, with the fit window spanning R = 1.0 Å to 2.0 Å. To reduce parameter correlation and ensure stable fitting, the Debye–Waller factor (σ ²) was fixed during the fitting process. The σ ² value was determined from an initial unconstrained fit of the MMGO_1050 sample, which served as a reference due to its high crystallinity and minimal disorder. The resulting σ ² value of 0.005 Ų was then applied consistently across the remaining samples to enable a reliable comparison of CN. Fits performed with σ ² left as a free parameter yielded similar trends in CN, but with larger uncertainties.

Conflict of Interest

The authors declare no conflict of interest.

Author Contributions

Sherry Cao: formal analysis (lead); investigation (equal); methodology (equal); writing—original draft (lead). Yihong Liu: formal analysis (supporting); investigation (equal); writing—original draft (supporting). Ruoxin Deng: investigation (supporting). Clement Lee: formal analysis (supporting); investigation (supporting). Lijia liu: conceptualization (lead); formal analysis (supporting); funding acquisition (lead); methodology (equal); writing—original draft (supporting); writing—review & editing (lead).

Cao Sherry, Liu Yihong, Deng Ruoxin, Lee Clement, Liu Lijia. ChemPhysChem. 2025; 26:e202500207. 10.1002/cphc.202500207

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.