1. Introduction

Mechanoluminescence (ML) is light emission from an optically active material caused by mechanical stimulation [1,2]. The mechanical stimulus usually induces internal electric fields (for example, in a piezoelectric host material), thereby triggering electronic transitions between dopant-associated energy levels. This gives rise to light emission [3,4]. ML provides a straightforward way for visualizing (transient) mechanical strain; applications have been suggested in biophysical stress mapping and monitoring, self-reporting structures and optomechanical read-out of loading situations, information encryption and security labeling, or even photomechanical energy conversion [4–9]. Different ML materials are known, which emit light in different spectral regions from the ultraviolet (UV) to the visible (Vis) and near infrared (NIR). Mechanical stimuli range from uniaxial compression to impact loading, friction, scratching and ultrasonic excitation [10–12]. For quantitative load monitoring, the ML intensity or a characteristic band position are explored. For this broad versatility, considerable efforts are being devoted to the discovery of materials exhibiting ML, including polycrystalline ceramics [7], metal-organic frameworks (MOFs) [13] and functional polymers [14]. For now, however, the selection of materials with efficient ML emission is still limited [4]; only a few high-performing materials are currently available, for example, ZnS:Mn²⁺, SrAl₂O₄:Eu²⁺,Dy³⁺ and CaZnOS:Mn²⁺. Most of these materials are produced through conventional ceramic powder processing methodology, leading to polycrystalline (porous, microstructured) bulk or composite products.

As an alternative, glass processing would enable dense, homogeneous objects of very high geometrical versatility, e.g., from microbeads to fiber, sheet or hollow shapes. In addition, the glass route offers very high compositional flexibility, for example, in the concentration of dopant species [15], and unmatched surface quality such as required for adapted mechanical durability [16]. However, glasses do normally not exhibit ML activity. This is due to the high non-radiative transition probability and the enhanced bandwidth of defect (trap) levels caused by structural disorder [17]. Two ways to overcome this limitation are the fabrication of glass-ceramic composites (GCCs) or glass ceramics (GCs). ML GCCs have been demonstrated, for example, incorporating ML ceramic particles into a phosphate glass matrix with sufficiently low melting temperature [11], however, the compatibility and stability of both phases presents a major obstacle to such phosphor-in-glass materials. GCs are produced through controlled precipitation of crystals from a glass-forming matrix phase, leading to a material comprising one or more crystal phase(s) embedded in a residual glass matrix. In principle, GCs combine the advantageous optical properties of the crystalline phase with the processing ability, chemical flexibility and microstructural homogeneity glasses [18–24]. However, they require the improbable mix of high glass forming ability and rapid crystal nucleation, which needs to be well separated from the kinetics of crystal growth. The former is in order to be able to process the glass. The latter is in order to induce homogeneous crystal precipitation in a controllable way manner, especially when aiming for nanoscale crystallites in order to obtain visually transparent or translucent GCs.

To this end, we now consider broadband ML emission addressing the spectral range of about 650-800 nm within the transmission window of biological tissue, using Chromium (Cr³⁺)-doped ZnGa₂O₄ (ZGO) nanocrystals in a germanate glass ceramic.

2. Experimental

2.1 Materials

Glass samples with nominal composition of 65GeO₂-5K₂O-15Ga₂O₃-15ZnO-0.1Cr₂O₃ (in mol %) were prepared by conventional melt-quenching in air. This glass composition is adapted from previous studies of ZGO GC fabrication in a germanate matrix [24–26]; it corresponds to a binary potassium germanate near the local density maximum [27], to which about 35 wt.% of stoichiometric ZnGa₂O₄ are added. As an assumption, the solubility of heavy metal species with large ionic radii (such as Ga³⁺) is reduced in such high-density germanates, what facilitates phase separation and crystal precipitation. Compared to silicate matrices for which ZnGa₂O₄ precipitation is intensely studied [28], the germanate offers very high nucleation rates, leading to prominent nanocrystallization with very low crystal size but, at the same time, high crystal volume fraction [29]. We used analytical-grade GeO₂, Ga₂O₃, K₂CO₃, ZnO, CrCl₃(H₂O)₆ as starting materials. Batches were melted at 1550 °C for 30 min in Pt crucibles covered by a lid. The melt was poured onto a stainless-steel plate and rapidly quenched by pressing with a metal stamp to form a disk-shape slab of precursor glass (labeled PG). Subsequently, PGs were heat-treated at 640 °C for 6, 20 and 48 h to form glass ceramics (GC), GC6, GC20 and GC48, respectively. Individual samples were cut or broken from the slabs and polished to optical grade using CeO₂ suspensions for further characterization. The sample size is about 6*6*1 mm³

2.2 Characterization

Room temperature ex situ X-ray diffraction (XRD) patterns were collected with Cu K_α radiation (MiniFlex 600, Rigaku, λ = 1.54059 Å) over the angular range 10°≤ 2θ ≤80° at a step size of 0.01°. In situ high temperature XRD characterization was performed on powdered samples using a multi-purpose X-Ray diffractometer (Rigaku Smartlab, 3 kW) in Bragg Brentano configuration, equipped with a high temperature chamber (Anton Paar HTK1200) and a hybrid pixel detector (Rigaku HyPix 3000) operated in 1D integration mode. Diffraction patterns were collected with 0.04° step size over a 2θ range of 20° - 65° at a scanning speed of 50 °/min. For the scanning experiments, a heating rate of 5 K/min was employed. Optical transmittance spectra of PG and GC samples were evaluated on a double-beam spectrophotometer (Cary 5000 UV-Vis) over the spectral range of 200 to 800 nm. Raman spectra were collected with a dispersive confocal Raman microscope (Renishaw inVia) for wavenumbers ranging from 80 - 1000 cm⁻¹ at step widths of 1.5 cm^-1, using the 514 nm excitation line of an Ar-laser. Microstructural observations were conducted by transmission electron microscopy (TEM) using a 200 kV FEI Tecnai G2 FEG equipped with an Oxford 80 mm² energy-dispersive silicon drift X-ray detector, and a Gatan UltraScan 2 k CCD camera. The photoemission spectra were recorded by a computer controlled spectrofluorometer (Fluorolog, Horiba Jobin-Yvon) upon excitation with a Xe lamp. Impact ML was examined on GC powder in a custom ball-drop experiment, using a compact fiber spectrometer (Shamrock 163, Andor) equipped with a CCD camera (iDus 420, Andor). Thereby, the impact energy was varied through the drop height of a 16.705 g stainless steel ball.

3. Results and discussions

A differential scanning calorimetry (DSC) curve of PG is presented in Fig. 1(c). From these data, glass transition temperature (T_g), onset of crystallization (T_x) and peak crystallization temperature (T_p) were obtained at 550, 650 and 679 °C, respectively, for a DSC heating rate of 10 K/min. In order to obtain a high crystallite density (which means that a high nucleation rate and limited crystal growth are required), we initially conducted isothermal heat treatment at a temperature slightly below T_x, i.e., 0.99T_x. Further optimization of the PG-to-GC transformation treatment was not in the scope of this study.

 

Fig. 1. (a) Sample photographs under daylight and under UV LED illumination (365 nm, as labelled). (b) Optical transmittance spectra for a specimen thickness of 1.0 mm. (c) DSC upscan of a 65GeO₂-5K₂CO₃-15Ga₂O₃-15ZnO-0.1Cr₂O₃ glass obtained at a heating rate of 5 K/min. (d) Bright-field TEM image of GC20, mounted on a carbon film covering a Cu grid. (e) HRTEM image of GC20, showing lattice fringes corresponding to the spinel structure. The crystal in the lower left is viewed along the [110] zone axis, depicting (111) lattice fringes, while the crystal in the upper right displays (400) lattice repeats. (f) Selected area electron diffraction pattern of crystalline matter and the glassy matrix shown in Fig. 1(d).

Download Full Size | PDF

Figure 2(a) shows ex situ XRD patterns of the Cr³⁺-doped PG and corresponding GCs. For PG, there are no distinct diffraction peaks visible in the diffraction pattern, indicating that the as-received material is X-ray amorphous. Characteristic diffraction peaks were observed in heat-treated samples, readily assigned to ZGO spinel (JCPDS No. 38-1240, space group Fd_3m, with tetrahedral Zn²⁺ and octahedral Ga³⁺). The octahedral site (with ionic radius of Ga³⁺, r_Ga = 620 pm) provides an almost ideal match for Cr³⁺ substitution (r_Cr = 615 pm); [30] Cr³⁺-doped gallate spinels are well-documented [31–33]. The average crystallite size D was estimated from the width of the (400) diffraction peak using the Scherer equation [34]. As shown in Fig. 2(c), we obtained D ∼ 10 nm, unaffected by prolonged annealing time. This observation is corroborated by measurements of mass density (Fig. 2(e), and also by analysis of the integrated diffraction patterns (Fig. 2(d)); both show that the glass-to-crystal fraction does not change with prolonged heat-treatment, hence, we may assume that the glass ceramics have reached a characteristic microstructural state. In the present case, crystallization of ZGO is highly incongruent; if ZGO completely precipitated from the precursor glass, the residual potassium germanate glass phase would have a K/Ge ratio of ∼ 13/87 (K₂O/GeO₂ ∼ 7/93), which is close to the density maximum of potassium germanate glasses at ∼ 3.8 g/cm³ [35], and also close to a local minimum in T_g (∼ 460 °C for the pure 7K₂O-93GeO₂) [27]. The density of the pure zinc gallate spinel is ∼ 6.2 g/cm³. [36] As an assumption, further crystallization and crystal growth are limited by chemical depletion and the kinetics of Zn/Ga diffusion. The fraction of crystalline ZGO could reach about 35 wt.%, according to nominal composition. Taking a linear mixture of the bulk densities of the pure potassium germanate glass and the ZGO spinel, this would result in a maximum density of the GC of ∼ 4.6 g/cm³. From XRD signal integration, we estimate a crystal volume fraction of ∼ 30% (corresponding to 30-40 wt.% for a residual glass density in between the ones of the pure germanate glass and the precursor glass). The measured mass density of the GCs is ∼ 4.3 g/cm³. Given the uncertainty of the density of the residual glass phase, the density of the nanoscale ZGO interface, and the applicability of a linear mixing relation, these observations indicate that the major part of the batched ZnO-Ga₂O₃ precipitated from the glass in crystalline form. This leaves a binary potassium germanate as the residual glass phase (potentially with minor fractions of dissolved Ga₂O₃ and ZnO). Closer inspection of the XRD pattern of GC48 reveals a number of weak reflexes appearing at around 2 θ ∼ 20 … 30°, in addition to the peaks assigned to ZGO (these reflexes are not visible on GC20 or GC6). Due to their low intensity and significant broadening, we refrain from specific assignment, however, we attribute crystalline potassium germinates as their generic origin. We assume that these germanates result from devitrification of the residual glass phase after prolonged heat treatment, i.e., > 20 h at 640 °C. As will be discussed in the following, their presence causes elastic light scattering, which affects the observed transmission and luminescence spectra.

 

Fig. 2. (a) Ex situ XRD patterns of PG and GCs. (b) Micro-Raman spectra of PG and GCs with varying heat-treatment time. (c) Estimated crystal size for GCs as a function of isothermal treatment time at 640 °C (0.99T_x). (d) Estimated crystal volume fraction for GCs. (e) Mass density of PG and GCs as a function of heat-treatment time. (f) Isothermal in situ XRD patterns for treatment at 800 °C. (g) In situ XRD patterns for heating from 25 to 800 °C at a rate of 5 K/min. (h) Intensity, (i) FWHM and (j) peak area of the (311) in situ XRD signal at 2 θ ∼ 35.6°.

Download Full Size | PDF

Corresponding in situ XRD data are provided in Fig. 2(f) and (g) for heating from room temperature to 800°C and for isothermal annealing at 800°C, respectively. In the heating experiment, the ZGO phase appeared at ∼645 °C, consistent with DSC to within ± 5 K. The diffraction peak intensity does not increase during further heating. Isothermal treatment at 800 °C similarly led to saturating diffraction intensity, peak width and peak area (for 2 θ ∼ 36°, Gaussian peak fit, Fig. 2(h-j)). These observations confirm our ex situ XRD data, showing that crystal precipitation is rapidly completed near 650 °C, and that the ZGO crystals do not grow further even for heating up to 800 °C. Aside the ZGO spinel, we did not find signs of any secondary crystal phase in the ex situ or in situ diffraction experiments for treatment times below 20 h. Only for GC48, minor traces of devitrification were observed (see above).

A direct view at the microstructure of GC samples was obtained by TEM. TEM micrographs of GC20 are provided in Fig. 1. The bright-field image (Fig. 1(d)) shows nanoscale grains (dark shades) embedded in a residual glass matrix (gray region). The crystalline nature of the gallate nanograins is obvious from the strong change in diffraction contrast during tilting, while the surrounding glass matrix remains uniformly gray. The size of these crystals is in the range of 5 to 20 nm, in accordance with XRD studies. Furthermore, selected area electron diffraction (SAED, inset of Fig. 1(f)) confirms that the ZGO nanocrystals are of spinel structure. This is also in accordance with the HRTEM image in Fig. 1(e), which reveals well-defined lattice planes whose interplanar distances are consistent with the spinel structure. The lattice constant a was refined on the basis of six interplanar spacings (111, 220, 311, 400, 511, 440), yielding a = 8.417(1) Å.

Vibrational spectroscopy on PG and the GCs revealed an overall sharpening and splitting of bands as a result of crystal precipitation (Fig. 2(b)). The Raman spectra resembles the spectrum of a pure potassium germanate glass with low K₂O [37], that is, before the maximum of the ^IVGe-to-^VIGe conversion. We may hence expect germanium primarily in fourfold coordination, reflected by the strong band dominating the spectral region of 400-600 cm^-1, which has been assigned to ^IVGe-O-^IVGe and, and by the high-frequency shoulder, related to ^IVGe-O-^VIGe [38]. The further shoulder at ∼ 590-600 cm^-1could be due to Ge-O-Ge in GeO₆ units [39]. The two weaker bands at ∼ 830 cm^-1 and 930 cm^-1 (well-separated only in the GC samples) were previously assigned to transverse and longitudinal optical components (TO and LO modes), respectively, of the asymmetric stretching of Ge-O-Ge [40]. In PG, these modes are not readily visible, most probably due to structural interactions with zinc and gallium species in the multicomponent precursor glass, which lead to a broad, merged band spanning the range of 750-950 cm^-1. Raman spectra of the GC samples contain additive contributions of the residual glass phase and of the gallate spinel crystal phase. Overall, the spectral characteristics of a pure, low-alkali potassium germanate glass are much more developed in the GC samples, which confirms our previous assessment that the residual glass phase is close to a binary germanate with K₂O/GeO₂ ∼ 7/93. Aside this, the other primary difference between the GC and the PG samples is the additional band at around 680 cm^-1, which we assign to A_1g or E_g modes in crystalline ZGO [41].

The spinel ZGO has a wide bandgap of ∼ 4.4-5 eV [42]. The Vis-NIR optical transmission spectra of PG and GC samples are therefore determined by the Cr³⁺ dopant (Fig. 1(b)). Furthermore, there are no notable variations among PG and the GC samples in terms of elastic scattering; the slopes at the UV edge of all samples are pretty much identical, with even a slight blue-shift of the UV edge for the GC samples. This is due to the very low crystallite size, even though density variations among the ZGO spinel and the residual glass phase are significant.

Thermal treatment led to notable changes in the absorption behavior of the Cr³⁺ species. For the Cr³⁺-doped PG sample, absorption bands of ⁴A₂ to ⁴T₂ and ⁴T₁ are located at ∼635 and ∼ 420 nm, respectively. Both bands blue-shift upon crystal precipitation, i.e., to 560 nm and 400 nm, respectively; although higher than in the glass, the crystal field of the octahedral Ga-site still has a relatively low strength (with the Ga-O bond dissociation energy of ∼ 374 kJ/mol, for example, as compared to that of octahedral Al with Al-O ∼ 501 kJ/mol). The change in the absorption spectra manifests in a change of sample coloration from deep green (PG) to a lighter tint in the GCs (Fig. 1(a)). The intensity of the absorption band appears to increase for GC48, what could be indicating crystal rearrangements and variations in Cr³⁺ site occupation [43], or simply be a result of elastic scattering caused by partial devitrification of the residual glass phase (see above).

The change of the Cr³⁺ environment from the weak crystal field of the glassy matrix to a stronger field in crystalline ZGO induced great variations in the photoluminescence behavior. Photoluminescence emission (excited at λ_ex = 410 nm) and excitation (monitoring the emission at λ_em = 697 nm) spectra are displayed in Fig. 3(a) and (b), respectively. The characteristic NIR emission (which is related to the ²E → ⁴A₂ transition of octahedral Cr³⁺) is observed in all GC samples. The zero-phonon-line of Cr³⁺ in undistorted Ga³⁺ sites of ZGO is located at 690 nm (R-line) [44]. For Cr³⁺ ions distorted by neighboring cationic anti-site defects, it was reported at 697 nm (N2 line) [45]. Stokes (S) and anti-Stokes (AS) phonon side bands (PSB) occur at 709 and 715 nm, and at 662, 671 and 679 nm, respectively [32]. All these band can be clearly identified on the PL spectrum of the GC samples. Their intensity increased with prolonged annealing (sample GC48, Fig. 3(a)), whereby the emission peak positions remained unchanged (the ²E level in d³ does almost not depend on crystal field strength). The corresponding excitation spectra exhibit four bands located at 260, 320, 410 and 560 nm; the 260 nm band is due to Cr-O charge transfer. The latter three result from the ⁴A₂ to ⁴T_1(4P), ⁴T₁ and ⁴T₂ transition, respectively [46]. In accordance with the emission spectra, the excitation intensity increased for GC48, what again corroborates our interpretation of the XRD and UV-Vis transmission spectra in that prolonged heat-treatment results in elastic scattering on secondary crystals. The pristine PG sample presents similar but much weaker emission as compared to that of the GCs, indicating the presence of a minor, but non-zero fraction of PL active emission centers already in the as-received glass sample.

 

Fig. 3. (a) Photoluminescence emission spectra (λ_ex = 410 nm) of PG and GC samples, and (b) corresponding excitation spectra monitoring the emission wavelength λ_em = 697 nm. (c) Contour plot of photoemission intensity in sample GC48 (color code) as a function of excitation and emission wavelength. (d) Photoluminescence decay curves of the 697 nm emission for PG and GC samples (λ_ex = 410 nm).

Download Full Size | PDF

Figure 3(c) shows a contour plot of the photoemission intensity in sample GC48 as a function of excitation and emission wavelength. Photoemission in the Cr³⁺-doped GC covers the first transmission window of biological tissue at ∼ 650-950 nm. Excitation is possible through sunlight, but also through high-power UV LEDs (e.g., at 415 nm). Decay curves of the 697 nm emission (²E → ⁴A₂) (λ_ex = 410 nm are shown in Fig. 3(d). We obtained effective lifetimes τ of the ²E level of 9.46, 55.13, 58.96 and 62.26 µs for PG, GC6, GC20 and GC48, respectively. The increase in excited state lifetime from PG to GC is another indicator for the incorporation of Cr³⁺ into the ZGO crystal phase (with low phonon energy). Furthermore, the lifetime continues to increase slightly with prolonged heat-treatment time.

The ML performance of Cr³⁺-doped ZGO GCs was demonstrated through ball-drop tests at variable impact energy (depending on drop height). As shown in Fig. 4(a), no ML response was detected from the Cr³⁺-doped PG sample. Yet, ML was observed from GC6, GC20 and GC48 samples, respectively, under mechanical stimulation. With increasing thermal treatment time, the ML intensity of GC48 was enhanced in accordance with PL and UV-Vis data. The spectral ML characteristics resemble those of PL shown in Fig. 3(a). However, the underlying energy transfer process is fundamentally different for ML as compared to direct violet excitation of PL. While 410 nm light is directly absorbed through ⁴A₂→ ⁴T₁ in Cr³⁺ ions, impact ML involves host lattice transfer reactions; the observed ML intensity is comparable to the one from Cr³⁺-activated ZGO polycrystals prepared by solid state reaction (Fig. S1 and S2).

 

Fig. 4. (a) Impact ML spectra of PG and GC6, GC20 and GC48 taken at a shutter (integration) time of 3s for an impact energy of 0.12 J. (b) ML intensity as a function drop-height in ball-drop impact tests at a ball weight of 16.705 g (diameter 16 mm), taken on GC48. (c) Time-resolved impact ML response: single drop (upper curve) with an integration time of 0.01 s, and five consecutive drop events on the same sample (G48) with an integration time of 0.01 s (d) Impact ML reproducibility in in ten repetitions of identical ball-drop tests on the same sample.

Download Full Size | PDF

Figure 4(b) presents the load-dependent ML for GC48 as a function of impact energy. With an increase in mechanical stimulation, also ML intensity increases gradually. Within the observed range of impact energy, the emission intensity scales linearly with impact energy (R = 0.952). From time-dependent response studies (Fig. 4(c)), we estimated a lifetime of ∼ 60 ms of the ML signal following the impact event. This is much higher than the PL lifetime of the Cr³⁺ emission (∼60 µs), and confirms the role of host-lattice transitions. The achievable ML intensity decreases gradually during five consecutive drop events conducted on the same sample (without intermediate re-charging), reflecting continuous depopulation of the charged defect states responsible for ML emission. When re-charged by UV illumination between each repetition, the impact ML response is repeatable for consecutive impact events on the same sample (Fig. 4(d)). This demonstrates efficient ML from highly-transparent germanate ZGO:Cr³⁺ glass ceramics with high crystal volume fraction.

We demonstrated mechanoluminescence from the ZGO: Cr³⁺ glass ceramics with a close-to-binary residual glass phase of potassium germanate. Broadband NIR ML emission covered the first transmission window of biological tissue. The GC material was highly transparent, with a crystallite size of ∼ 10 nm and a number density of ∼ 10²³ m^-3 of phase-pure ZGO. Crystallization occurred in an extraordinarily narrow temperature window, leading to a stable microstructure even at > 150 °C above the crystallization onset. This may enable glass-like processing of the material into various shapes and geometries. PL and ML observations indicate that the optically active Cr³⁺ ions are embedded on octahedral sites of the ZGO nanocrystals, with direct excitation in PL and host-lattice energy transfer excitation in ML. ML active glass ceramics offer new playground of glass ceramic design, with potentially significant implications for stress field imaging, but also light delivery, for example, to within biological tissue.