Smart mechanoluminescent phosphors: A review of zinc sulfide-based materials for advanced mechano-optical applications

1.1 Research history

The historical development of ZnS as a luminescent material has a fascinating timeline indeed. Théodore Sidot's initial discovery in 1866 marked the beginning of its documented history,⁴⁰ his research findings being further developed by the renowned luminescence researcher A. E. Becquerel.⁴¹ The earliest research on ZnS ML dates back to 1904 when Wallace Goold Levison discovered that grayish chunks of sphalerite would emit colored light and composite luminescence when rubbed with fingertips, matches, fingernails, or steel brushes.^{42, 43} The ML intensity was substantial enough to be evident from several meters away. The cause of composite luminescence was attributed to various impurities in the mineral, marking the first report of natural sphalerite ZnS exhibiting ML properties.¹⁸

Curie and Porst pointed out that ML could originate from the cleavage of crystals when subjected to external stress. Later, in 1964, Alzetta et al. contradicted this interpretation.⁴⁴ They studied the ML properties of ZnS powder suspensions in various liquids under hydrostatic compression and discovered that ML was independent of the medium, whether insulating or conducting. Moreover, they proposed that the energy required to excite electrons to the excited state in ML materials originates from phonons produced when the vibrational equilibrium of the crystals shift to a higher energy level owing to pressure changes. Chudacek noted that ZnS ML was directly proportional to the time variation of applied pressure, suggesting that it might be a phenomenon of a destructive nature.^{45, 46} In summary, early ZnS ML research focused on the phenomena and discussion of ML fundamentals.⁴⁷ However, practical applications of ML materials were limited, primarily owing to challenges such as inadequate sample preparation techniques, lack of sensitive photon detection equipment, and the limited luminescence brightness of early synthesized ML materials. These factors collectively hindered progress in ZnS ML research during that time.^48-55

It was not until 1999 when Xu et al. proposed an effective sample preparation method and potential application of ZnS ML films for stress sensing as artificial skin that a new wave of ML research was sparked.²³ They found that, under the same preparation conditions, the ML intensity of nanoscale ZnS:Mn²⁺ artificial skin exceeded that of the bulk solid by an order of magnitude, thanks to the high orientation of ZnS:Mn²⁺ thin films and the nano-sized particles embedded within. In 2013, Jeong et al. reported a kind of elastomer composite—that is, ZnS:Cu@polydimethylsiloxane (PDMS)—with a brightness of up to 120 cd · m⁻² under moderate external forces (the pressure ranging from kilopascals to megapascals), and found that considerable levels of ML were capable of withstanding over 10⁵ repeatable mechanical stress cycles.²⁴ Moreover, they suggested that ZnS could produce tunable light emission with different luminescent dopants (such as Cu^+/2+ and Cu^+/2+/Mn²⁺), under specific experimental conditions.⁵⁶ ZnS:PDMS ML composite materials could also be fabricated into wind-driven ML devices, capable of emitting light with different color temperatures (including warm, neutral, and cool white), allowing for energy-efficient lighting and image displays (Figure 1).⁶³

Since 2015, researchers have focused on optimizing ZnS ML performance for fine applications in different fields. For example, Wang et al. fabricated hand-writable ZnS:Mn²⁺ films for electronic signatures²⁵; Qian et al. improved the deformability and ML intensity by incorporating SiO₂ into ZnS:Mn²⁺ and ZnS:Cu⁶⁶; Peng et al. increased the brightness to 2.2 times that of commercial ZnS:Mn²⁺ by heterogeneous compounding of ZnS:Mn²⁺ with CaZnOS:Mn²⁺, and explored the ZnS:Mn²⁺ bi-phase homojunction structure to enhance its ML capabilities³³; Hong et al. prepared nanoscale ZnS:Ag⁺/Co²⁺ ML particles for optogenetic therapy, showcasing the diverse applications of ML materials in biomedical fields⁶⁷; Hou et al. developed bite guards for human–machine interaction based on ZnS particles and introduced artificial neural networks (ANNs) to optimize data quality²¹; Mukhina et al. clarified the differences in the ML mechanisms of ZnS particles under low and high pressure conditions⁶⁸; and Zheng et al. investigated the spectral properties of ZnS/CaZnOS:Mn²⁺ under extreme conditions, suitable for temperature detection under extremely high pressures.³⁴ These studies collectively showcase the multifaceted applications and advances in ML research, ranging from material synthesis to biomedical applications and environmental sensing.

2.1 Crystal structure

ZnS is not only a typical material for ML applications but also a typical II–VI group semiconductor. Pure ZnS is an n-type semiconductor, although p-type semiconductors can also be obtained by doping ions such as Ag,⁶⁹ Tm,⁷⁰ and N.^{71, 72} At 300 K, the residual electron density in ZnS is approximately 10⁴ cm⁻³, considerably higher than the expected value produced by thermal ionization in wide-bandgap materials. This suggests that it is a compensating semiconductor characterized by a significantly higher concentration of donor (D) defects compared to acceptor (A) defects.⁷³ In the absence of an external field, the majority of these donors are occupied by electrons and thus remain neutral, while a smaller proportion of ionized donors ($${\mathrm{N}}_{\mathrm{d}}^{+}$$) offsets the charged acceptors ($${\mathrm{N}}_{\mathrm{a}}^{-}$$), thereby ensuring the charge neutrality of the material.⁶⁸

ZnS exhibits two common allotropes—that is, the hexagonal wurtzite structure (WZ, α phase, non-centrosymmetry) and the cubic zinc blende structure (ZB, β phase, centrosymmetry) (Figure 2). These two structures share a similar single cell. The phase transition temperature from β phase to α phase in undoped bulk ZnS is around 1020°C.⁷⁴ In the ZB-ZnS structure, tetrahedrally coordinated zinc and sulfur atoms stack in an ABCABC pattern, belonging to the cubic phase and face-centered cubic structure with central symmetry. The lattice parameters are a = b = c = 5.41 Å, Z = 4, and the space group is F-43m, making it an isotropic crystal with a bandgap width of 3.72 eV. In the WZ-ZnS structure, the same elements stack in an ABABAB pattern, belonging to the hexagonal phase and close-packed hexagonal structure. Its lattice parameters are a = b = 3.82 Å, c = 6.26 Å, Z = 2, and the space group is P63mc, making it an anisotropic crystal with a bandgap width of 3.77 eV.⁷⁵ Additionally, owing to its non-centrosymmetric structure, ZnS exhibits several piezoelectric properties with a piezoelectric constant of 8 pm · V⁻¹.⁷⁶

The size of ZnS particles can affect its bandgap and phase transition temperature. Compared to bulk materials, ZnS nanomaterials have a wider bandgap owing to the quantum size confinement effect of low-dimensional nanostructures.⁷⁷ Qadri et al. found that reducing the particle size of ZnS crystals to the nanoscale lowers the phase transition temperature.⁷⁴ In their experiments, the mixture of α and β phases in nano ZnS began to appear at temperatures as low as 350°C, much lower than the typical 1020°C. Additionally, with increasing annealing temperature, the particle size increased considerably, ranging from 2.8 to 243 Å.⁷⁴

Apart from particle size, the phase transition of ZnS crystals is also related to pressure,⁷⁸ temperature,⁷⁹ doping,⁷⁹ and preparation details.⁸⁰ For example, Wang et al. proposed that the irreversible phase transition from the WZ phase to the ZB phase occurred at pressures >14 GPa in ZnS:Mn²⁺/Eu³⁺.⁷⁸ Feng et al. mentioned that thorough ball milling of raw materials before sintering could reduce the required temperature for obtaining WZ-structured ZnS.⁸⁰ Wang et al. explored the influence of the proportion of Mn²⁺ in the ZnS host on the phase transition.⁷⁹ They synthesized nanorod-shaped ZnS systems via the hydrothermal method and explored the correlation between the β-phase proportion and the intensity of ML and PL. By adjusting the concentration of Mn²⁺ doping and the annealing temperature, it was possible to control the ratio of the two phases. Moreover, they indicated that the sample did not exhibit ML properties at calcination temperatures below the required levels.

2.2 Deformation properties related to microstructure

Oshima et al. discovered that ZnS single crystals could exhibit extraordinary plastic deformation effects at room temperature (297 K) in a completely dark environment, achieving a strain of ε_t = 45%, without undergoing fracture.⁸¹ However, when exposed to ordinary light or 365-nm ultraviolet light, the sample quickly fractured after exceeding its plastic deformation point (Figure 3a,b). This phenomenon could be attributed to the fact that the inorganic semiconductors were typically hard and brittle. During deformation, the optical bandgap of the crystal decreased from 3.52 to 2.92 eV (Figure 3c,d), and the crystal changed in color from colorless to gradually becoming orange. This suggested that the optical gap depended on the magnitude of deformation, owing to the introduction of dislocations in the deformed sample. Dislocation core regions in the crystal could have different energy band structures compared to the dislocation-free regions.⁸²

Further microstructural characterization elucidated the detailed processes involved. Based on the X-ray diffraction (XRD) results, the initially undeformed ZnS crystals exhibited approximately 50% twinning formation (Figure 4a–e).⁸³ As plastic deformation reached around 10% strain, accompanied by changes in the crystal domain volume—that is, the disappearance of twinning—further strain increases in darkness resulted in almost constant crystal domain volume. During the initial stages of plastic deformation, the movement of the crystal domain boundaries and the associated changes in the crystal domain volume were caused by the slip of inter-domain boundary dislocations. During the later stages of plastic deformation, the crystal domain volume remained constant, and an increase in the number of paired partial dislocations was evident as the plastic strain increased (Figure 4f–h). The transition in dislocation behavior could originate from an increase in the crystal domain volume—that is, the disappearance of twinning—leading to the formation of paired partial dislocations.

To validate the changes more directly in the domain volume of Domain 1 and Domain 2 (Figure 4b,c) measured by XRD during plastic deformation, transmission electron microscopy (TEM) observations of both undeformed and deformed specimens were conducted. In darkness, TEM images showed the volumes of either domain (Domain1 or Domain2) tended to increase as the plastic strain increased (Figure 4i–k). Moreover, the high-resolution transmission electron microscopy images and XRD results identified the two-type domains having a twin relationship (Figure 4m,n). Additionally, diffraction patterns were captured from the zone axis [110] for both undeformed and deformed sample (ε_t = 25%). The diffraction analysis of the undeformed sample revealed twin formations on {111}, while one of the two patterns from the zincblende structure, which had varying crystal orientations, nearly vanished following plastic deformation at ε_t = 25%. It is evident that the volume ratios of the observed black- and white-contrast areas in the TEM images (Vol.(b)/Vol.(w)) of the undeformed specimen are almost equal to 1, which deviate considerably (reaching approximately 1/6) in the specimen deformed up to ε_t = 25% (Figure 4o,p). By contrast, under light conditions, Figure 4l shows that a TEM image captured from an area displaying slip lines in the specimen, which was deformed up to ε_t = 3% under UV light illumination. It is worth mentioning that the Vol.(b)/Vol.(w) differs considerably from the others, approaching roughly 1/4 (Figure 4l).

In conclusion, it is evident that the crystal domain volume changed locally, indicating that local deformation was caused by the motion of boundary dislocations, similar to the initial deformation stages in darkness. The mobility of dislocations interacting with photoexcited charge carriers depends enormously on dislocation characteristics—that is, the atomic species in the dislocation core and the dislocation direction. The dependence of dislocation characteristics on their mobility can lead to non-uniform plastic deformation under light conditions.

When subjected to external forces, the WZ-crystal structure is prone to deformation, leading to the separation of positive and negative polarization charge centers and the generation of a polarization field. This field drives trapped electrons within crystal defects to move to the conduction band, thereby forming electron–hole pairs. The energy released from this process is transferred to the luminescent centers through non-radiative recombination, resulting in luminescence. Although the ZB-crystal structure is a centrosymmetric crystal and typically lacks piezoelectricity—often considered essential for solid materials to exhibit ML—several studies have indicated that piezoelectricity may not be indispensable. For instance, Ma et al. reported that Mn²⁺-doped ZB crystals also exhibited self-recoverable ML (RML) characteristics. They proposed that this phenomenon was attributed to the transfer of electronic energy to luminescent centers during the process of chemical bond reconstruction, leading to the emission of excited light.⁸⁴ This topic will be discussed in detail in Chapter 3.

2.3 ML and mechanical quenching (MQ)

The intense ML in ZnS typically requires doping to occur. Undoped ZnS crystals exhibit weak PL and EL and exhibit little ML owing to the absence of impurity energy levels. Commonly used doping ions that can act as luminescent centers include Mn²⁺,^{63, 66, 85-87} Cu⁺,²¹ Cu²⁺,^{56, 88-90} Cu/Cl,⁸⁶ Mn²⁺/Te,^{62, 91} Mn²⁺/Li⁺,⁹² Mn²⁺/Cu,⁵⁶ Ag⁺/Co²⁺,⁹³ Cu/Al³⁺,^{61, 86} and Cu/Mn²⁺/Al³⁺ (Figure 5, Table 1).⁶¹ The emissions in the visible light band are bright and clearly visible, especially the blue-light emission (which is unrivaled in brightness compared to other ML materials), thereby enabling the realization of white-light ML emission by combining three colors.^{63, 100} These advantages have also enabled ZnS to rapidly achieve widespread commercial use in the field of ML.

Metal ions typically replace Zn atoms in the ZnS crystal lattice and serve as luminescent centers, whereas Cl⁻ ions fill defect levels within the ZnS crystal, thereby considerably altering the local piezoelectric constants around luminescent ions and making it easier to produce ML under pressure owing to piezoelectrically-induced detrapping of the charge carriers.^{86, 101, 102} Currently, the strongest luminescence intensity among ZnS materials is associated with the doping of Mn²⁺ ions, which has a luminescence peak at 585 nm (attributed to electron transitions from the ⁴T₁-⁶A₁ level of Mn²⁺).²³ Another common dopant element is Cu. Although it has exhibited blue, green, and orange ML emissions, as shown in Hou et al.'s research (ZnS:Cu⁺@ Al₂O₃ exhibits an ML peak at 475 nm, ZnS:Cu²⁺@Al₂O₃ at 525 nm, and ZnS:Cu²⁺/Mn²⁺@Al₂O₃ at 595 nm),²¹ its valence state and mechanism remain controversial.⁶³ First, Cu⁺ and/or Cu²⁺ could be involved in the ML process.¹⁰³ Second, the extraction process causing the blue emission remains unclear.²⁴ The PL spectra show that ZnS:Cu has two emission bands (blue and green).¹⁰⁴ The green emission originates from the transition from the shallow defect state to the t₂ state of Cu.¹⁰⁴ The blue emission has been attributed to various factors—such as the association with the e state,^{105, 106} the transition from the CB edge of ZnS to the acceptor-like t₂ state of Cu,¹⁰⁴ or native defect states,¹⁰⁷ as Peng et al. determined that undoped samples also exhibited blue emission near 450 nm.¹⁰⁷ The introduction of Al could also have a more complex effect on the ML of ZnS:Cu. For example, the overall ML color of ZnS:Cu/Al³⁺ changes from green to blue light under high-frequency forces.¹⁰⁸ Additionally, it has been reported that coating the surface of ZnS:Cu with Al₂O₃ is necessary to exhibit ML properties.¹⁰⁹

Beyond doping, ML can also be influenced by the phase of ZnS (Table 1). Generally, the WZ phase (which forms at high temperatures), displays a stronger ML intensity owing to its non-centrosymmetric crystal structure. However, for ZnS, the total energy difference between the WZ and ZB phases is minimal, making it a polymorphic material prone to stacking faults and mixed phase formation (WZ–ZB).¹¹⁰ This propensity can lead to variations in experimental results, as the precise proportion of the phases within samples is often insufficiently analyzed. Consequently, the interaction between the two phases in ML behavior does not merely reflect a straightforward sum of their individual contributions but involves a more intricate coupling mechanism.^{68, 110-112} This complexity suggests that the interplay between phases in ML requires additional detailed study to fully understand its dynamics and implications.

Here, we list several results of the relationship between the phase and ML from past research, as shown in Table 1. It is evident that ML has been reported in both WZ and ZB phases for ZnS:Mn²⁺,^{84, 86} ZnS:Cu,^{24, 89} ZnS:Cu/Mn²⁺,^{56, 98} and ZnS:Mn²⁺/Te.^{62, 91} Specifically, doping and particle size are also factors that can influence the existence of ML. For example, bulk crystals of WZ-ZnS:Cu/Cl exhibit plastico-ML (PML) instead of elastico-ML (EML); microcrystals of WZ-ZnS:Cu/Al³⁺ exhibit considerable EML. Finally, in nanocrystals, WZ-ZnS:Ag⁺/Co²⁺ exhibits UV-RML.⁹³

The ML spectra of ZnS can be influenced by the frequency and magnitude of the applied stress. This has been validated by multiple research groups, although the methods they used to vary the stress frequency and magnitude differ. Here, we list some of the conclusions observed. Through the high-pressure testing system (specifically mentioned in Section 5.3), Zheng et al. applied high pressure to ZnS:Mn²⁺/Eu³⁺ and found that under an applied stress frequency of 0.5 GPa · s⁻¹, increasing the stress from 2 to 3.6 GPa results in the ML peak shifting from 585 to 615 nm.⁷⁸ High-frequency mechanical (or electrical) signals typically cause the spectrum to shift toward blue, the extent of this influence differing according to the doping ions. For instance, with a stretching–releasing testing system, jeong et al. found that the ML peak of ZnS:Cu/Mn²⁺ remains virtually unchanged when the stretching–releasing circle per minute (S–R cpm) increases from 200 to 500 (Figure 6a). Moreover, its EL peaks do not shift when the electrical frequency increases from 80 Hz to 10 kHz (a sinusoidal voltage, 170 V).⁵⁶ However, under another doping condition, ZnS:Cu, it exhibits a 3-nm blue shift under the same frequency of S–R cpm (Figure 6b),²⁴ and its EL peak changes from 508 to 460 nm as the electrical frequency increases from 50 Hz to 20 kHz at a fixed applied voltage of 100 V.¹¹³ Subsequently, Zhou et al. were able to observe spectral changes under high-frequency mechanical signals (in the kHz range) by applying high-frequency electrical voltages to piezoelectric materials, revealing the influence of high-frequency forces on the spectrum.¹⁰⁸ For ZnS:Cu/Al³⁺, with rectangular wave voltage with fixed strength of 165 V, an initial increase in frequency led to enhanced ML in both the blue and green bands, with the blue band showing a more substantial increase, surpassing the green band in intensity at 3.3 kHz, with both bands beginning to quench at 48 kHz (Figure 6c,d). For ZnS:Mn²⁺, with rectangular wave voltage with fixed strength of 200 V, an increase in frequency initially led to enhanced ML in the orange band, peaking at 4 kHz and almost disappearing at 9 kHz; moreover, as the frequency continued to increase, near infrared (NIR) ML began to dominate, peaking around 30 kHz, and began to quench with further frequency increases (Figure 6e,f).

Upon analysis, it was concluded that the blue shift in the ZnS:Cu/Al³⁺ spectrum was due to increased electron–hole recombination rates for both V_s (S vacancies)-Cu_Zn (Cu replacing Zn) (blue light) and Al_Zn (Al replacing Zn)-Cu_Zn (green light) pairs; whereas for ZnS:Mn²⁺, the blue shift was caused by high-frequency signals promoting the formation of (Mn)_n clusters (n = 2–6) and the energy transfer from excited Mn²⁺ ions to (Mn)_n clusters.^{95, 108} Finally, the saturation and subsequent quenching of ML in both cases could be attributed to the carrier recombination rate failing to match the excitation frequency.

The ML of ZnS possesses an exceptionally low threshold. However, this threshold can be influenced by several factors, including the doping ions,⁸⁶ particle size,⁸⁶ surface morphology,¹¹⁴ crystal structure,^{68, 110} testing conditions (whether compounded with a matrix, the rate of load application), and the structure of the fabricated devices.^{115, 116} Consequently, variations in threshold values are evident across different studies. Here, data observed in several seminal studies are summarized. In the research conducted by Chandra et al.,⁸⁶ it was mentioned that for ZnS powder, smaller particles led to a lower threshold. For example, ZnS:Mn²⁺ bulk crystals (diameter: 1.2 mm, length: 3 mm) exhibited an EML threshold of 20 MPa, whereas its microcrystals had an EML threshold of 1 MPa. In the case of ZnS:Cu/Cl, the bulk crystals (2 × 2 × 4 mm³) only exhibited PML with a threshold of 30 MPa, whereas microcrystals exhibited EML with a threshold of 1.5 MPa. On the one hand, bulk crystals are less prone to deformation and have fewer internal defects near impurity centers, which can lead to a higher ML threshold and a lack of EML in ZnS:Cu/Cl. On the other hand, the introduction of Cl ions into ZnS:Cu results in the filling of negative ion vacancies by the Cl⁻ ions.¹¹⁷ Consequently, the number of local dipoles is reduced, leading to the disappearance of EML.¹¹⁷ Wei et al. fabricated thin-film structures of ZnS:Cu using polymethyl methacrylate (PMMA) and fluorinated ethylene propylene (FEP), measured the ML through friction and discovered that the ML threshold could be even lower than 10 kPa. They attributed this extremely sensitive ML to triboelectroluminescence (TIEL).¹¹⁵ In the study by Mukhina et al.,⁶⁸ it was found that some ZnS:Mn²⁺ microcrystals had a threshold below 1 MPa, with a few microcrystals exhibiting a threshold of just 233 ± 60 kPa. It is noteworthy that in their experiment, 44 samples of ZnS:Mn²⁺/PDMS were tested, with the ZnS:Mn²⁺ having a diameter of 4.6 μm, various surface structures, and a load application rate of 0.125 μm · ms⁻¹. They attributed the exceptionally low thresholds observed in certain particles to the enhancement of the local piezoelectric field resulting from the WZ–ZB phases. Subsequent research by Wang et al. proposed a similar viewpoint, suggesting that ZB-ZnS stacking faults in WZ-ZnS crystals caused band bending, which reduced the gap between shallow donors and the conduction band, thereby enabling the material to generate ML under minimal stress excitation.¹¹⁰ In this model, the variability in threshold values among different ZnS particles could be explained by the differing relative proportions of the WZ and ZB phases, leading to varying degrees of conduction band bending and, consequently, different ML thresholds.

ZnS ML also exhibits linear and quadratic characteristics. Linearity refers to the fact that the emitted ML intensity is proportional to the magnitude of (P-P_th)².¹¹⁸ Overall responses can be fitted with linear functions, with the slope and intercept of this function varying depending on factors such as the doping and testing conditions (Figure 7a).⁸⁶ Considering that ML responds to the strain rate—that is, dynamic forces are needed to trigger ML—the response curve can also be influenced by the strain rate (the amount of deformation per unit time). For instance, Jeong et al. observed considerable differences in ML intensity when applying stress at strain rates ranging from 200 to 500 cpm (Figure 7b).²⁴ However, controlling the strain rates, especially under rapid force impact, can be challenging. Hence, current research commonly employs the stress magnitude as the independent variable for the response function. Additionally, the quadratic characteristic imply that during the EML process, a second ML peak emerges after releasing the applied pressure (Figure 7c),⁴⁴ which is more intense than the light emitted during pressure application.²³ This could be due to higher strain rates upon the release of pressure.⁸⁶ However, the generation and appearance of this secondary ML peak are notably quicker, posing a challenge for its precise capture and characterization.

The rise and fall times of ZnS ML are typically in the microsecond range, but results can vary owing to different testing methods. In Chen et al.'s research, the material was mixed with epoxy resin, and a force signal was applied using the drop-ball method. They measured the ML lifetime of ZnS:Mn²⁺ to be approximately 0.97 ms, with the ML intensity curve undergoing a rapid increase to a peak at approximately 35 μs, followed by a slow decay over several milliseconds.¹²⁰ In Zhou et al.'s study, the material was compounded using a UV-curable adhesive matrix, and a high-frequency force signal was applied using a piezoelectric material.¹⁰⁸ They found that the rise and fall times were both related to the frequency of the applied force. When the excitation frequency was below 12 kHz, the rise and fall times were 9.7 and 35.1 μs, respectively. As the excitation frequency increased, the decay process was interrupted by the strain release, and the fall time was no longer calculated. The rise time began to gradually decrease only when the excitation frequency exceeded 30 kHz, reaching 7.2 and 5.8 μs at 40 and 50 kHz, respectively.¹⁰⁸

The repeatability of ML is also crucial for its application. Initially, Xu et al. conducted sliding-stress tests on crystal thin films of ZnS.²³ Later, Jeong,²⁴ Wang,²⁵ Wei,¹¹⁵ Peng,^{33, 35, 84, 121} and others performed repeatability tests under sliding/stretching/bending conditions on films compounded with ZnS powder (Figure 7d–f), confirming its excellent repeatability.²⁴

It is worth mentioning that ZnS exhibits an effect opposite to ML—known as MQ¹¹⁹—which refers to the phenomenon where the afterglow (AL) of the material diminishes after being subjected to stress (Figure 7g,h). The principle behind this phenomenon shares certain similarities with ML, the main difference lying in the stress-induced piezoelectric field driving electrons to recombine with holes at non-radiative centers. Observing MQ requires pre-charging the material with UV light to generate AL. Upon applying stress, a momentary ML occurs, followed by a weaker AL at the stressed sites compared to unstressed ones, indicative of MQ. Pre-charging also leads to stronger ML compared to when not pre-charged, although the enhancement diminishes with the continued application of stress. This characteristic is very similar to the ML of SrAl₂O₄:Eu²⁺,⁵ suggesting that ZnS's MQ can be seen as having dual modes of generation—that is, mechanical energy converted to light energy and mechanical energy inducing the release of stored energy in the form of light, with the latter being less pronounced.

It is important to note that MQ is more easily observed in undoped ZnS,¹¹⁹ but the introduction of Mn doping induces ML in the material, reducing the AL, which in turn makes the MQ effect more likely to be overlooked.

2.4 Multi-stimuli responsive luminescence

ZnS exhibits multimode luminescent characteristics beyond ML, including PL, EL, thermoluminescence (TL), X-ray-induced luminescence (XIL), and cathodoluminescence (CL) (Figure 8). These luminescence modes share close ties with ML, their spectra being consistent with that of ML, so playing an important role in understanding the underlying mechanism within ML.

PL refers to the phenomenon in which luminescent materials emit light after absorbing photons (electromagnetic radiation). If the emitted light has a shorter wavelength than the excitation light, it is called down-conversion or down-shifting luminescence; conversely, if the emitted light has a longer wavelength, it is known as up-conversion luminescence or anti-Stokes photoluminescence, which requires the simultaneous absorption of two or more photons. Down-shifting luminescence is more common in ZnS-based fluorescent materials and can be observed in various main group, transition metal, and lanthanide ion-doped ZnS materials, such as Mn²⁺,²³ Co²⁺,^{122, 123} Cu^+/2+,^{21, 24} Sb³⁺,^124-126 Pb²⁺,^127-130 Al³⁺,^{126, 131-133} Sn²⁺,^{126, 134, 135} Bi³⁺,¹³⁶ Ag²⁺,¹⁰⁰ Au⁺,^137-139 and Ln^2+/3+ = Eu^2+/3+, Sm³⁺, Tb³⁺, Er³⁺,^{140, 141} among others. Their spectra essentially cover the visible light range. Additionally, up-conversion luminescence can be observed in ZnS nanocrystals doped with various lanthanide ions, such as Er³⁺,¹⁴² Yb³⁺,¹⁴³ and Ho³⁺,¹⁴⁴ among others. The Zn²⁺ and S²⁻ ions in ZnS are also prone to defects during the synthesis process, introducing intrinsic defects and trap levels for capturing electrons and holes, characteristics that contribute to the versatile luminescent properties of ZnS materials.

Similar to PL, when a material emits light owing to X-ray radiation, it is referred to as XIL. X-rays can induce phenomena such as the photoelectric effect and Compton scattering, and pair production of electron–hole pairs within the atomic lattice of the luminescent material, generating a large number of hot electrons and deep holes. These low-energy charge carriers gradually migrate to luminescent centers, leading to luminescence emission. Materials exhibiting XIL properties are also called scintillators. As the excitation mechanisms for XIL and PL are similar, their spectra are also similar, such as the XIL spectrum of ZnS:Cu/Co nanoparticles,¹⁴⁵ which falls in the range of 400–650 nm with a peak at 510 nm, closely matching the PL spectrum.

Moreover, ZnS exhibits excellent EL properties, with ZnS:Cu demonstrating superior performance when subjected to an external alternating current (AC) electric field. This phenomenon was first discovered in 1936 when Destriau dispersed the ZnS:Cu phosphor in castor oil.^{146, 147} Usually, it is necessary to formulate a multi-layer structure to achieve ZnS alternating current electroluminescent (ACEL). The core phosphor layer is either the ZnS powder itself (using the deposition method) or a composite of ZnS, dielectric particles (e.g., BaTiO₃ ¹⁴⁸), and matrix (e.g., resin,¹⁴⁸ polyvinylidene fluoride [PVDF],¹⁴⁹ or PDMS¹⁵⁰). Structurally, ACEL devices can be classified according to their dielectric layers into two types—that is, symmetric and asymmetric structures (Figure 9).

In symmetric structures, the electron and hole generation layers can be inserted between the phosphor layer and dielectric to facilitate carrier generation. In simpler asymmetric structures, holes and electrons can be injected alternately from the same electrode in different electric field directions.¹⁵¹

The dielectric layer and Cu_xS needles play important roles in the process of ZnS ACEL. Under the influence of an AC electric field, defects at the interface or within the phosphor layer are activated as charge carriers and subsequently injected into the phosphor layer. As shown in Figure 10a, the dielectric directly contacts ZnS, creating substantial interfacial states with energy levels distributed within the forbidden band gap of the semiconductor. These activated hot electrons are injected into the conduction band through tunneling and/or field-enhanced thermal excitation of trapped electrons (Figure 10a). Conversely, the Cu_xS needles not only facilitate the formation of Schottky barriers between Cu_xS and the ZnS semiconductor but also intensify the local electric field by one to two orders of magnitude. This intense local electric field causes electrons and holes to be injected from both ends of the Cu_xS needles through tunneling (Figure 10b).

After carrier injection, theories that explain the subsequent light emission include the impact ionization and impact excitation theories. In the impact ionization theory, high electric fields accelerate thermal electrons to collide with the phosphor lattice, generating electron–hole pairs (Process 1, Figure 10a). These pairs are subsequently captured by donors and acceptors. The recombination of these electrons and holes then produces light emission, or electron–hole pairs transfer their energy to doping ions (such as Mn²⁺), causing electron transitions to excited states and subsequent light emission upon returning to the ground state. In the impact excitation theory, thermal electrons directly collide with local luminescent centers, exciting ground state electrons to higher energy levels (Process 2, Figure 10a), producing light emission upon their return to the ground state.

TL is a phenomenon where a material emits electromagnetic or ionizing radiation (previously absorbed and stored as photon energy within the lattice defects) when heated. This process is also known as thermally stimulated luminescence. The sources of radiation energy that can be absorbed include visible light, X-rays, β-particles, or α-particle radiation, among others. The TL intensity of ZnS material increases as the crystal size decreases because smaller particle sizes result in increased surface states or defect sites that can trap carriers.¹⁵² When heated, more charge carriers are released, combining with holes, and the energy is transferred to the luminescent centers, resulting in enhanced light emission. Additionally, reducing the particle size can increase the overlap between electron and hole wave functions, potentially accelerating the rate of charge carrier recombination, and further enhancing luminescence.¹⁵² It is worth noting that the ZnS TL peak does not vary substantially with decreasing particle size, which could be because the trap depth of surface states is insensitive to particle size.¹⁵³

CL is the phenomenon where luminescent materials emit light when excited by a high-energy electron beam, thus offering higher spatial resolution (up to several nanometers) and applicability to wide-bandgap semiconductors compared to other luminescent methods. CL imaging is a powerful tool for visualizing the spatial distribution of excitons, defects, and other phenomena within a material, thereby facilitating detailed analyses of the crystal lattice and band structure. For the same material, different morphologies, doping ions, and temperature variations can result in different CL spectra.¹⁵⁴ Generally, ZB-ZnS exhibits strong and stable 337-nm ultraviolet emission in CL spectra at room temperature.⁵⁹ Liu et al. studied the morphology, doping, and temperature dependence of one-dimensional ZnS nanostructures in CL.⁵⁸ Chen et al. successfully synthesized pyramid-structured ZB-ZnS using thermal evaporation, and its CL spectra exhibited strong and stable 337-nm ultraviolet emission, indicating high crystallinity.⁵⁹ This emission could be attributed to band emission from the ZB structure. Moreover, compared to other structures—such as nanobelts, nanowires, and nanorods—pyramid-structured ZB-ZnS exhibited a lower turn-on field, higher field-enhancement factor, and high stability with low fluctuation of field-emission current, suggesting its potential suitability for optoelectronic applications.⁵⁹

3.1 Piezoelectric effect model

The piezoelectric effect model is the most common model used to explain ZnS ML in different doping situations, such as Mn²⁺,¹¹⁷ (Mn)_n,¹⁰⁸ Cu,¹¹⁷ Cu/Mn²⁺,⁷⁷ Mn²⁺/Te,⁹¹ and UV-RML in ZnS:Ag⁺/Co²⁺.⁹³ This model suggests that ML in ZnS is intimately linked to the excitation–deexcitation process of the built-in electric field and luminescent centers during deformation, which is also supported by the consistency observed in the PL, EL, and ML spectra. Regarding the transfer of mechanical energy from ZnS crystals to luminescent centers during the EML process, Chandra et al. noted that when the crystal underwent elastic deformation, the piezoelectric constant near the impurity center increased considerably compared to other positions in the crystal.¹⁰² Consequently, with the non-centrosymmetric structure of ZnS, a strong piezoelectric field is generated in the crystal's vicinity, leading to a reduction in trap depth within the crystal. This reduction allows trapped electrons to escape the confines of the traps and move into the conduction band. Some of these escaped electrons recombine with holes, releasing energy that can be transferred to the luminescent center, which then emits energy in the form of light waves. Moreover, the self-recovery characteristic of EML can be achieved through the recapture of carriers drifting in the presence of the piezoelectric field.¹¹⁷ Chandra coined this ML excitation model the “piezoelectrically-induced trap-depth reduction model,” which was also suitable for ML of ZnS crystals of different sizes (bulk crystals, microcrystals, and nanocrystals).^{94, 158, 159} However, ML is only applicable when the force varies with time rather than being static.^{102, 160}

This model is typically used to explain the orange ML emission of ZnS:Mn²⁺.⁹⁴ Although there are partial distinctions in other situations, the main triggering factor is usually the piezoelectric field induced by pressure. For ZnS:(Mn)_n, the NIR ML is due to the energy transfer from excited Mn²⁺ ions to (Mn)_n clusters under high-frequency mechanical excitation.¹⁰⁸ For ZnS:Cu, the luminescent center is Cu rather than Mn²⁺.¹¹⁷ For ZnS:Cu/Mn²⁺, Cu and Mn²⁺ together act as luminescence centers.⁷⁷ For ZnS:Mn²⁺/Te, the EML process involves energy transfer from Mn²⁺ to Te, where the transition of Te from ³T₁ to ¹A₁ emits red light.¹⁶¹ For ZnS:Ag⁺/Co²⁺, UV light excites electrons, which are captured by Co²⁺ and remain in a metastable state. Pressure-induced piezoelectric fields release electrons from Co²⁺, transferring energy to Ag⁺, causing a transition that emits photons with a wavelength of 470 nm.⁹³ Based on the principles of light-emitting diodes, heterojunction structures can considerably enhance the EML intensity of crystals.

Peng et al. combined p-type CaZnOS and n-type ZnS to create a p-CaZnOS/n-ZnS:Mn²⁺ piezoelectrically-induced luminescent heterojunction crystal. When an external force is applied, the built-in piezoelectric field causes a band offset between the ZnS/CaZnOS heterojunction, increasing the efficiency of electrons escaping from traps into the conduction band and forming electron–hole pairs (Figure 12a,b).

This enhances carrier mobility, thereby substantially increasing the EML intensity. Consequently, the ZnS/CaZnOS heterojunction structure exhibits highly reproducible EML, with an intensity 2.2 times higher than that of commercial ZnS:Mn²⁺ and 3.5 times higher than that of the classical quaternary EML material CaZnOS:Mn²⁺. It can emit visible light under a 1 N fingertip force excitation (Figure 12c). Additionally, full-spectrum tunable luminescence has been achieved covering the visible and NIR spectrum by incorporating Mn and lanthanide ions Ln³⁺ (such as Ce, Pr, Nd, Sm, Eu, Gd, Tb, Dy, Ho, Er, Tm, and Yb) into the structure. Based on the heterojunction principle, bi-phase homojunction ZnS:Mn²⁺ crystals (WZ and ZB phases) have also been synthesized, which exhibited EML intensities 3–6 times higher than pure WZ-phase ZnS:Mn²⁺(Figure 12d,e). Furthermore, they reported ML enhancement in SrZnOS using the ZnS/CaZnOS heterojunction, resulting in an EML intensity approximately 60 times higher than SrZnOS:Mn²⁺ (Figure 12f). Additionally, the ZnO/ZnF₂ heterojunction they constructed also considerably enhanced the EML intensity of ZnO:Mn²⁺.^{33, 37, 162}

3.2 Models related to dislocation motion

Covalent bond breakage-reformation model

The covalent bond breakage-reformation model can be used to explain the ML in β-ZnS:Mn²⁺. The majority of centrosymmetric crystal structures do not exhibit ML characteristics, because crystals with a center of symmetry undergo uniform deformation when subjected to stress, and the center of positive and negative charge mass remains overlapping, resulting in no polarization and, therefore, no piezoelectric effect. Whereas some centrosymmetric crystals may occasionally display ML, their ML is either caused by the piezoelectric phase of defects or is unrelated to piezoelectric effects, such as luminescence caused by the charge impact on gas molecules within the crystal and defect motion.¹⁶³

Ma et al. proposed this improved plastic DML model to explain the self-RML in β-ZnS:Mn²⁺.⁸⁴ When an external force is applied, defects within the crystal move, causing partial covalent bonds between Zn and S to break. This results in the redistribution of s and p electrons, leading to a considerable number of Zn²⁺ ions acquiring extra electrons. These extra electrons can be considered to be intermediate states for rebuilding covalent bonds, with an abundance of Zn dangling bonds in the Zn 4s orbital (CB) with extra electrons and extra holes in the S 3p orbital (VB). These conditions are sufficient to excite ML.

They also employed a carrier capture model to explain how carrier energy is transferred to Mn²⁺. To be specific, Mn²⁺ is involved in the recombination of electron–hole pairs. When an electron–hole pair migrates near Mn²⁺, Mn²⁺ sequentially captures the holes from the valence band and the electrons from the conduction band. This completes the energy transfer from the exciton (electron–hole pair) to Mn²⁺, resulting in Mn²⁺ being in an excited state. This process differs from the common pathway where Mn²⁺ directly transitions from the ground state to the excited state. In this model, there is a brief process of Mn³⁺ generation, similar to Mn²⁺ acting as a recombination center for electron–hole pairs.¹⁶⁴ Thus, it exhibits high luminescence efficiency. After the ML process is completed, the broken covalent bonds are reformed, allowing ZnS to exhibit good ML repeatability. This explains why the samples with the highest PL and ML do not necessarily have the same Mn²⁺ doping concentrations (Figure 13).

The mechanism can be explained through the emission and excitation spectra of β-ZnS and β-ZnS: Mn²⁺ (Figure 14a). In further experiments, they observed that the β-ZnS sample exhibited a broad UV excitation band at 470 nm, composed of excitonic transitions (ZnS:VB→CB) (I) and a defect-related excitation band centered at 340 nm (II). When monitoring the emission at 590 nm, corresponding to Mn²⁺, the excitation band of β-ZnS:Mn²⁺ included excitonic transitions (I), charge transfer transitions (Mn²⁺ 3d→CB), and a series of d-d transitions centered at 355 nm (II) (Mn²⁺:⁶A₁(⁶S)→⁴E (⁴D), ⁴T₂ (⁴D), (⁴A₁,⁴E) (⁴G), ⁴T₂ (⁴G), or ⁴T₁ (⁴G)) (III).¹⁶⁵

Two notable facts about β-ZnS:Mn²⁺ are the disappearing blue emission band under 355-nm excitation and the absence of the defect-related excitation band when monitoring at 470 nm. These observations suggest the presence of an efficient energy transfer process from the defect centers to Mn²⁺. Considering that ML may not be as efficient as PL in mechanical-to-optical energy conversion, under the condition of a fixed excitation wavelength at 300 nm, by further reducing the excitation power, the intensity ratio of blue to yellow emission (I₄₇₀/I₅₉₀) decreases (Figure 14b). At low excitation intensity (approximately 2% of the initial excitation), the blue emission band eventually disappears, resulting in the PL spectrum being similar to the ML spectrum (Figure 14c). Although the blue band caused by defects in β-ZnS and the yellow band caused by Mn²⁺ in β-ZnS:Mn²⁺ can be separately excited by 300-nm ultraviolet light, at relatively low excitation power, the efficiency of ultraviolet-to-yellow energy conversion is substantially improved. Consequently, they suggest that the RML process in β-ZnS:Mn²⁺ is similar to the inter-band excitation PL process involving Mn²⁺ under weak excitation, rather than being primarily influenced by any point defects.

To describe the luminescence mechanism, they further studied the luminescence decay behavior in β-ZnS:Mn. By adjusting the bandwidth to change the laser intensity (0.25 and 3 nm), they compared the luminescence decay curves (Figure 14d). Ignoring the later slow decay, after magnifying the initial decay curves (see inset in Figure 14d), they found that under weak excitation of band I or band II, the decay curves completely overlapped. The longer decay lifetimes compared to band III can be attributed to additional trapping and detrapping processes between the defect levels and the CB. Under strong excitation of band I or band II, accelerated decay of the 590-nm emission was evident owing to increased non-radiative electron–hole recombination. In conclusion, the above spectral results validated the rationality of their proposed ML model.

ML spatial models

In the piezoelectric effect model, the piezoelectric magnitude can be calculated using the formula⁶⁸:

$$${E}_{\text{piezo}}=\frac{{d}_{33}\sigma }{{\varepsilon }_{0}(1+\chi )},$$$ (1)

where d₃₃ denotes the piezoelectric coefficient, σ denotes the applied stress, ε₀ denotes the permittivity of a vacuum, and (1 + χ) denotes the relative dielectric constant of ZnS. However, this model cannot further explain the ML of ZnS:Mn²⁺ below 1 MPa,^{68, 166} as the built-in electric field induced by mechanical forces cannot generate sufficiently large band bending and a sufficient number of charge carriers. Additionally, no ML was evident in defect-free ZnS:Mn²⁺ nanocrystals.

To fill the research gaps, Mukhina et al. revealed the existence of three different ML mechanisms in single ZnS particles under low stress (<8.1 MPa).⁶⁸ They referred to these mechanisms as the ML induced by stacking faults exciton fields (Pz mechanism), ML induced by fully or quasi-reversible fault bending (μD mechanism), and ML induced by irreversible fault slip at higher pressures (D mechanism) (Figure 15). When the stress was below 8.1 MPa, ML generation was dominated by the Pz mechanism, and the μD mechanism also came into play. By contrast, at pressures above 8.1 MPa, charged fault motion became the primary source of excitation. This model could explain the EML in ZnS:Mn²⁺ at extremely low pressures (≈233 kPa). The mechanism relies on the local enhancement of the piezoelectric field produced by the stacking faults' interfaces of WZ and ZB phases along the c-axis direction of the ZnS crystal, where some interfaces lack electrons, enhancing the local piezoelectric field generated when an external stress is applied.⁶⁸

Assuming that the EML of ZnS requires a sufficiently large electric field for excitation, in cases where the electric field intensity from the piezoelectric effect is insufficient, alternative sources of enhanced electric fields need to be identified. Previously, it was believed that the EML of ZnS originated solely from the piezoelectric effect of the pure WZ phase.⁶⁸ However, Mukhina et al. observed the phenomena of WZ–ZB phase alternation in ZnS and proposed that such microstructural heterogeneity could generate a sufficiently strong intrinsic electric field. This, when stressed, could induce intense EML. In the faulted structure depicted in Figure 16, ZB is inserted into the WZ, forming a WZ–ZB–WZ structure.

In Figure 16a, owing to the absence of inversion symmetry along the c-axis, an electric field (E_WZsp) is generated within the WZ phase. Owing to the reverse piezoelectric effect, the insertion of the ZB phase produces an electric field (E_ZBpiezo) in the opposite direction. Consequently, an intrinsic sawtooth potential (Ψ_saw) is generated, causing free carriers (electrons) to drift toward the positive pole, leaving behind vacant D⁺ and A⁰ energy levels to compensate for the intrinsic field. The most likely candidates for the A⁰ center are the shallow hole states in S vacancies, which are involved in defect-related ZnS luminescence at room temperature.

Each interface between WZ–ZB encompasses an intrinsic barrier:

$$${{\Psi }}_{\text{bi}}={{\Psi }}_{\text{offset}}+{{\Psi }}_{\text{saw}}\,-\,{{\Psi }}_{\text{screen}},$$$ (2)

where Ψ_offset denotes the II-type band offset between the WZ and ZB phases,¹⁶⁷ Ψ_saw depends on the relative sizes of the WZ and ZB phases,¹⁶⁸ and Ψ_screen corresponds to a dual-depletion layer formed on either side of each interface owing to the compensation of Ψ_saw. Additional contributions to the potential difference could arise from piezoelectrically-induced interface strain, concurrent flexoelectric effects,^{169, 170} and the presence of extended defects. The WZ/ZB/WZ sequence depicted in Figure 16a–d represents the nanostructural unit of the ZnS model, where the lateral dimension along the c-axis defines the height of the intrinsic potential barrier and, consequently, the degree of compensation. The aforementioned value of Ψ_offset can be estimated for the hexagonal/cubic (2H/3C) interface. This combination yields the maximum potential obstacle since the 2H modification possesses the largest hexagon.¹⁷¹ However, the crystal could contain various modifications (such as 6H, 4H, and 2H), whereas the ZB phase only exists in the 3C modification.¹⁷² Combined with the size-adjustable Ψ_saw, considerable differences in Ψ_bi between different ZnS particles could be exhibited, corresponding to differences in the ML threshold between particles.

During compression (Figure 16b), a unipolar piezoelectric field pervades the entire particle. It aligns with the intrinsic field direction in the ZB phase and opposes the intrinsic field direction in the WZ phase. Consequently, the overall internal fields rearrange, resulting in a change in Ψ_bi. The most important alteration occurs at the WZ/ZB interface owing to intrinsic compensation, leading to electron depletion. Here, the potential profile (Ψ_piezo) induced by the piezoelectric effect cannot be screened by free electrons, and the bands of the WZ and ZB phases could contain relative potential jumps reminiscent of a piezoelectric p–n junction.^{173, 174} Hence, the increase in the local barrier owing to this phenomenon could exceed what is predicted by the intrinsic piezoelectric coefficients of WZ-ZnS. If the piezoelectric field induced by mechanical deformation along the c-axis surpasses the intrinsic field (E_piezo ≥ E_bi), free electrons are liberated from the potential wells at the WZ/ZB interface and can acquire sufficient kinetic energy to reach the conduction band floor of the WZ phase, subsequently recombining with holes at the A⁰ center (see Figure 16b). The energy released when accelerated electrons recombine with holes can potentially be transferred non-radiatively to nearby isoelectronic Mn²⁺,¹⁷⁵ subsequently exciting and radiating photons at 585 nm, resulting in instantaneous ML.

This mechanism can also be used to explain the second ML peak with the condition that E_piezo < E_bi, which is typically attributed to ML produced when ZnS recovers after deformation. At the moment of stress application, holes trapped in the A⁰ center (owing to intrinsic compensation) can be captured by the potential wells (generated by piezoelectric-induced potential jumps). Concurrently, certain donor centers in the ZB phase can be reoccupied by electrons at the outset of piezoelectric polarization. After the pressure is released, these potential jumps vanish, and holes are released into the valence band. Subsequently, holes can be captured by nearby Mn²⁺ ions with an attractive hole, and Mn²⁺ ions can be excited through various resonant energy transfer pathways— such as the energy of recombination between electrons (at the D⁰ center) and trapped holes (at nearby Mn²⁺ ions), or the energy of excitons X (formed when the electrons are released from the D⁰ center owing to recompensation) bound to the Mn²⁺. During the expansion process of ZnS:Mn²⁺ microparticles, as depicted in Figure 16c, ML excitation is similar to the compression process described above, except for the reversed roles of the ZB and WZ phases.

Throughout the entire mechanism, when the applied force is removed, the crystal structure can fully or partially recover, but irreversible changes occur at the electron level, including the generation and migration of charge carriers and their recombination within the crystal. However, the behavior of the actual systems is more complicated than that. The ML process depletes electrons engaged in compensating the intrinsic field, resulting in a reduction in Ψ_screen (Figure 16d). During the subsequent pressure cycle, the potential jumps generated become more substantial, facilitating the escape of electrons from the potential wells at the WZ/ZB interface. Moreover, the depletion of charge carriers leads to a reduction in charge at the WZ/ZB interface, which can be regarded as partial dislocations, and the decrease in dislocation charge further increases the mobility of charges.¹⁷⁶ Consequently, it can be concluded that each elastic ML excitation cycle increases the chance of elastic or microplastic events occurring at the WZ/ZB interface during the next pressure cycle, followed by the appearance of μD components. Over time, this process depletes the internal reservoir of charge carriers, ultimately resulting in the cessation of ML.

In the aforementioned ML process, Mn²⁺ ions can be excited by the energy transferred through non-radiative recombination of electron–hole pairs to some extent, which is consistent with the mechanism predicted by EML. Additionally, they can also directly participate in the process of hole–electron recombination, emitting light during the recombination–decomposition process. Therefore, the spatial models of ML can be mutually validated using the previously proposed mechanisms.

3.3 Triboelectroluminescence

This model posits that the ML is primarily EL induced by triboelectrification, and has been employed to clarify the ML of ZnS:Cu/PMMA/FEP under low pressure (<10 kPa), which also cannot be explained using the traditional piezoelectric effect model.¹¹⁵ When different materials come into frictional contact with ZnS, frictional charges can be generated on the surface. These charges are easily transferred into the interior of the ZnS crystal, leading to the excitation of ZnS phosphors and the generation of EL. Wei et al. conducted an in-depth analysis of the luminescent process of ZnS:Cu and found that it could be categorized into three stages.¹¹⁵ When the applied pressure was <3.8 MPa, the sample exhibited only TIEL. In detail, the ML of ZnS:Cu thin films could be divided into three phases (Figure 17a). During phase I, the applied pressure was low, and the TIEL increased rapidly, with a sensitivity of approximately 0.04 MPa. During phase II, the stress was further increased to 3.8 MPa, but the TIEL no longer increased owing to the saturation of the tribocharge density. During phase III, the pressure exceeded 3.8 MPa, and EML began to occur (dominated by the piezoelectric effect model), further enhancing the overall ML intensity.

The overall mechanism can be explained as follows: electrons excited by friction fall into shallow electron trap states and then transition to Cu impurity states, resulting in luminescence at a wavelength of 510 nm (Figure 17b). Figure 17c shows that with the onset of sliding, the density of frictional charges on the moving object's surface rapidly reaches a saturation value related to the material. During the early stages of sliding, owing to the accumulation of frictional charges, the surface potential increases considerably, indicating substantial charge transfer at the contacting surfaces. Subsequently, with continued sliding, the surface potential decreases sharply and eventually becomes indistinguishable from the background after sliding 2 cm, indicating that no more frictional charges are formed on the electrified layer. Figure 17d shows the potential along the thickness direction of the luminescent layer underneath the charged sliding object, represented along the z-axis in the inset of Figure 17d.

The ML of ZnS:Mn²⁺ could also be associated with triboelectric charges. Wang et al. fabricated a triboelectric nanogenerator (TENG) based on ZnS:Mn²⁺/PDMS.¹⁷⁹ They discovered that when the applied pressure was sufficient to induce ML, there was a substantial decrease in both the current and voltage of the TENG, accompanied by a reduction in the triboelectric charges on the PDMS surface. Conversely, when the pressure was too low to generate ML, the triboelectric charges would accumulate without dissipating. Using this TENG to attract foam microspheres, they observed that, compared to specific regions of PDMS composite films without ZnS:Mn²⁺ phosphors, the ZnS:Mn²⁺ luminescent layer lost its attraction for foam microspheres after ML, which perhaps suggests that electrical charges contribute to the ML process in ZnS.

Park et al. analyzed the TIEL during the stretching–releasing process of ZnS:Cu/PDMS.¹⁶⁶ When ZnS:Cu was combined with PDMS, air entered at the interface, forming natural pores. During the stretching–releasing process, the deformation of PDMS caused the pores to enlarge and then recover (Figure 18a), with friction occurring at the interface between ZnS:Cu and PDMS, causing TIEL. In these experiments, the force was applied indirectly via polytetrafluorethylene (PTFE). The specific ZnS:Cu/PDMS/PTFE structure could generate both internal and external electric fields during the stretching–releasing cycle (Figure 18b,c), enhancing the overall intensity of TIEL. Additionally, the primary reason for choosing PDMS and PTFE as substrates was their ability to generate a higher friction-induced output voltage.

It should be noted that bare exposed ZnS:Cu is less likely to exhibit TIEL, making it necessary to cover the surface of ZnS:Cu with a layer of AlO_x. Lee et al. discovered that ML in ZnS:Cu (D512C, bi-phase structure,⁹⁶ Shanghai Keyan Optoelectronic Technology Co., Ltd.)/PDMS could not be attributed to the frictional electrification between ZnS:Cu and PDMS. Instead, it arose from the frictional electric properties at the PDMS-AlO_x interface.¹⁰⁹ They compared the effects of coating the surface of ZnS:Cu with AlO_x and leaving it uncoated, and found that only the former exhibited an extremely pronounced ML (Figure 19a,b). Based on several common triboelectric materials (Figure 19c), they used wet chemical methods to grow different layers on the surface of bare ZnS:Cu (D512B, Shanghai Keyan Optoelectronic Technology Co., Ltd.), and combined them with two substrates—that is PDMS and polyurethane (PU)—which had similar mechanical properties (elastic constant ≈ 2 MPa) but different frictional electric characteristics. By analyzing the post-friction voltage, current, and charge density (Figure 19d–f), they found that PDMS-AlO_x and PDMS-MgO outperformed in all of these parameters, which corresponded to the strong ML evident in ZnS:Cu@AlO_x/PDMS and ZnS:Cu@MgO/PDMS (Figure 19g).

As mentioned earlier, the luminescence of ZnS:Cu in powder-based ACEL can be attributed to the presence of Cu_xS conductive needles in the ZnS:Cu phosphor. These needles significantly boost the AC electric field at their tips, allowing powder-based ACEL to operate at low voltages, as low as 10⁶ V · m⁻¹ (which is two orders of magnitude lower than the voltage required for thin-film ACEL), considerably reducing the threshold voltage for ZnS:Cu luminescence.^180-182 Consequently, the frictional charges generated at the AlO_x-PDMS interface during frictional motion could even induce EL in this context.

Finally, although both AlO_x and MgO can serve as outer shells for ZnS:Cu, the mechanical performance of ZnS:Cu@AlO_x/PDMS proved to be superior in the experiments. Its ML intensity remained stable even after more than 5000 stretch–release cycles. By contrast, the ML intensity of ZnS:Cu@MgO/PDMS exhibited an 80% decline after 1200–1400 stretch–release cycles (Figure 19h). They attributed this to the incomplete growth of the MgO coating.¹⁰⁹

4.1 Micrometer-scale ZnS crystals

High-temperature solid-state method

This method offers a simple, high-yield process, but its drawbacks include incomplete reactions and high energy consumption.¹⁹² For the synthesis of ZnS:Mn²⁺ crystals, high-purity ZnS and MnCl₂ are commonly employed as starting materials. These materials are weighed, ground, thoroughly mixed in an agate mortar, and transferred to a clean alumina crucible. Subsequently, the mixture is then subjected to a N₂ flow reaction at high temperatures for 1 h. After the reaction, the product is cooled to room temperature, removed, and rinsed multiple times with deionized (DI) water or anhydrous alcohol to remove impurities.¹⁹³

For further improvement, the choice of the annealing atmosphere can also be crucial. Wang et al. proposed that introducing oxygen during the synthesis process could enhance luminous efficiency, as the presence of oxygen positively affects the formation of sulfur vacancies in the ZB structure of ZnS, facilitating the introduction of Mn²⁺ as luminescent centers and shallow impurity energy levels, thereby increasing the electron–hole recombination rate.¹⁹² Through experimentation, the optimal oxygen proportion and Mn²⁺ doping concentration were found to be 1%–20% and 1%–2%, respectively. Liu et al. reported that under a thermal carbon-reduction atmosphere, the ML peak of ZnS:Mn²⁺ shifted from 585 to 650 nm, resulting in an overall transition to red light emission, and for the samples doped with 3% Mn²⁺, the PL and ML intensities were 1.76 and 3.23 times higher, respectively, than those observed under a N₂ atmosphere.¹⁹⁴ Within the ZnS:Mn²⁺ framework, the distance between the two nearest Zn²⁺ sites occupied by Mn²⁺ ions is notably short (3.823 Å), facilitating the formation of exchange-coupled pairs. The introduction of carbon increases the probability of ferromagnetic coupling in Mn²⁺–Mn²⁺. In this scenario, the electron spins of the Mn²⁺–Mn²⁺ align in parallel, occupying the bonding and antibonding orbitals, respectively. This configuration decreases the energy from the lowest excited state to the ground state, causing the ML emission to red-shift. The optimal concentration of Mn ions in the experiment was 3%, at which the probability of Mn²⁺–Mn²⁺ forming in the ZnS matrix was 12%. Moreover, Gan et al. proposed that annealing in a reducing atmosphere could enhance the ML and PL by regulating the concentration of sulfur vacancies (which provided the trapped electrons).⁸⁹ Through their research, it was evident that for commercial WZ-ZnS:Cu (Osram Sylvania Inc. GG45), annealing at 250°C for 1 h in a pure hydrogen atmosphere (99.9%) could achieve the maximum enhancement.

Moreover, Feng et al. suggested that the pre-ball milling of raw materials could reduce the average particle size, increase the specific surface area, and enhance the reactivity of the raw materials. This method not only lowered the reaction temperature but also increased the content of WZ-ZnS in the product, thereby improving the luminance of both PL and ML (Figure 20a–c).⁸⁰ After synthesis, when it needs to be compounded with matrices (such as PDMS), nanoparticles can be added to improve the elastic modulus⁶⁶ or triboelectric properties,¹⁰⁹ enhancing the material's strength and sensitivity under low stress conditions. Available nanoparticles include SiO₂, Al₂O₃, ZrO₂, In₂O₃, MgO, and TiO₂, among which SiO₂, Al₂O₃, and ZrO₂ have shown notably effective results (Figure 20d–f).¹⁹⁵

Molten salt shielded method

The molten salt shielding method uses molten salts such as NaCl, KCl, and LiCl as a protective layer to synthesize ML materials without the need for costly inert gases.^{85, 111} In comparison to solid-state method, this method produces powder with a controlled morphology and a uniform particle size. Consequently, it interfaces more closely with polymer composites, enhancing the stability and durability of flexible ML devices.^{26, 84, 85, 111} Moreover, the inclusion of molten salt promotes better mixing of reactants and isolates the reaction from the surrounding air, which results in lower reaction temperatures and reduced experimental costs while maintaining the material's luminescent brightness. It is important to note that if gas is generated during the reaction, it can disrupt and compromise the sealing integrity of the molten salt layer, excluding gas-forming reactants from the reaction. Additionally, owing to the boiling point of salt, reaction temperatures should not exceed a certain threshold.

A specific procedure for the synthesis of homojunction ZnS:Mn²⁺ is shown in Figure 21a,¹¹¹ where the molten salt also serves as a dispersing medium for Mn²⁺, a reaction medium for ZnS crystal growth, and a diffusion medium for Mn²⁺ doping into ZnS crystals. Raw materials and salt were mixed together, and another layer of salt was applied to the surface (Figure 21b,c). Owing to the considerable difference in density between ZnS (4.102 g · cm⁻³) and NaCl (2.165 g · cm⁻³), and the lower melting point of ZnS (approximately 680°C) compared to NaCl (approximately 804°C) (Figure 21d), ZnS is deposited at the bottom during the reaction. The molten salt on the surface effectively isolates the air (Figure 21e–g). Lastly, the product was washed with deionized water to remove the salt.

4.2 Nano to micrometer-scale ZnS crystals

Wet-chemical methods

Wet-chemical methods encompass various techniques involving liquid-state reactions for material synthesis, such as the sol-gel, hydrothermal, solvothermal, and chemical co-precipitation¹⁹⁶ methods. These methods can all be employed for the synthesis of ZnS nanoparticles, which has application in areas such as QD luminescent materials,^197-200 catalysis,²⁰¹ antibacterial purposes,²⁰² and solar cells,^{203, 204} among others. To impart ML properties to the material, the final step often involves high-temperature sintering.

The sol-gel method is a technique that involves hydrolysis and polymerization reactions of raw materials in a liquid-state, resulting in the formation of a gel that can then be dried to obtain the desired product. This method allows for uniform and controlled doping, with relatively low reaction temperatures and has been employed to synthesize (ZnS)_1-x(MnTe)_x,^{205, 206} ZnS:Mn²⁺,²⁰⁷ and ZnS:Cu,²⁰⁸ among others. However, the drawbacks include higher raw material costs, as some materials are strong acids, strong bases, or organic compounds harmful to both human health and the environment. Experimentally, it also involves complex and relatively time-consuming steps.²⁰⁹ Differing from the atmospheric conditions of the sol-gel method, the hydrothermal method employs a high-pressure stainless steel reaction vessel as the reaction container. This method involves promoting the recrystallization or reaction of substances in a solvent under medium- to high-pressure conditions and at a particular temperature. A specific procedure is shown in Figure 22a.²¹⁰ When a material cannot be prepared in water, alternative solvents are commonly employed, based on the solvothermal synthesis method.¹⁹¹ While sharing similarities with the hydrothermal method, the solvothermal synthesis method offers the advantage of enabling control over the phase type of ZnS by varying the solvent type.¹⁹¹

These methods are suitable for controlling the size and morphology of ZnS nanoparticles. For instance, Tran et al. achieved particle-size variation by adjusting the molar ratio of Zn²⁺ to S^2- (Figure 22b–d).²¹⁰ Zhao et al. transformed the particle morphology between microspheres, urchin-like structures, and hierarchical flower-like superstructures by controlling the ratio of ethylenediamine (En)—a chelating agent capable of forming complexes with Zn²⁺/Mn²⁺—to distilled water and reaction time (Figure 22e,f).⁶⁰ Jiang et al. induced changes in the size (200, 350, up to 450 nm) (Figure 22g–j) and morphology (spheres, rods, microflowers, and biphasic structures) (Figure 22k–o) of ZnS by altering the solvents (ethylene glycol, ethylenediamine, and hexanol), modifying the reaction temperature, and adding HNO₃, among other conditions.¹⁹¹

The heating process can also take the form of microwave-assisted methods.²¹¹ Compared to infrared waves, microwaves have longer wavelengths, resulting in greater penetration ability. This allows for instantaneous and uniform heating of the entire reaction volume, ensuring rapid heating. Moreover, microwave heating eliminates the problem of the actual temperature lagging behind the set temperature, providing more precise temperature control. The microwave-assisted method can be advantageous for synthesizing well-dispersed nanomaterials as microwaves induce molecular collisions and friction among materials, preventing agglomeration.^{212, 213} This method has been employed to synthesize ZB-ZnS nanorods,²¹¹ ZnS nanoparticles doped with ions such as Ag⁺, Cu²⁺, Ce³⁺, and Sn⁴⁺,^{214, 215} as well as ZnS QDs.²¹³

The one-pot method is conceptually similar to the sol-gel method, and can be used to synthesize ZB-phase ZnS QDs or nanocrystals.^{183, 216} A typical procedure for this method is shown in Figure 23a,b, where gelatin plays an essential role as a capping agent, effectively controlling the growth of nanoparticles to maintain colloidal stability and prevent uncontrolled aggregation, ultimately leading to the formation of smaller nanoparticle nuclei.¹⁸³ The addition of gelatin also enables the composite to transition between a solid-state and liquid-state in response to external temperature changes (Figure 23c,d). The liquid ZnS QDs/gelatin nanocomposite can be used for fluorescence labeling and bioimaging (Figure 23c). Other substances can also serve as capping agents. In the study conducted by Zhang et al.,²¹⁷ 1-dodecanethiol was employed, which forms strong bonds with the surface atoms of semiconductor nanocrystals. Consequently, thiol ligands can render Mn²⁺ much less reactive, thereby aiding in the formation of stable, small-sized ZnS core nanoclusters.²¹⁷ Even with an experimental heating time of up to 1 h, the particle size of the product remained <2 nm, with a quantum yield (QY) of 65% (Figure 23f,g). In the research of Zhang et al.,²¹⁸ Cu:Zn-In-S/ZnS QDs synthesized using the one-pot method exhibited a high PL QY of up to 85%. Additionally, they could adjust the emission color by varying the In/Zn precursor ratio and Cu ion concentration (Figure 23h,i).²¹⁸

Biomineral-inspired suppressed dissolution approach

The biomineral-inspired suppressed dissolution approach is a unique method that draws inspiration from the demineralization processes of bone and enamel (Figure 24a,b).^{65, 188} Through analysis of the Gibbs free energy, it is known that minerals and ceramic materials exhibit a critical size in solution, below which they do not undergo dissolution. For most ceramic materials, this critical size ranges from 10 to 100 nm. ML materials can be considered a type of ceramic material, and most ML particles prepared through high-temperature solid-state methods are in the micrometer range, far exceeding the critical size of 10–100 nm. Consequently, spontaneous dissolution occurs in the solution, releasing nanocrystals. However, owing to high surface tension, these particles tend to spontaneously precipitate. This problem can be addressed using an undersaturated solution of citrate, which (with its lower pH), accelerates the dissolution process. The negative charge provided by citrate decomposition can stabilize ML nanocrystals, forming a colloidal solution. This approach initially requires the synthesis of micrometer-sized ML particles through conventional methods such as solid-state reactions. These particles can then be subjected to ball milling, before partially dissolving them in a solution. Unlike ball milling throughout the entire process, this method avoids problems related to luminescence quenching resulting from excessive stress and dislocations.¹⁸⁸

The synthesis of ZnS:Mn²⁺ nanoparticles is shown in Figure 24a,b. It efficiently yields water-soluble ML nanocrystals with diameters as small as 20 nm (Figure 24c), while retaining the strong ML emission synthesized using high-temperature solid-state methods. Additionally, these ML liquids can be recharged with UV light and excited with focused ultrasound (FUS) (Figure 24d). These characteristics contribute to deep tissue imaging and optogenetics achieved by intravenous injection of ML nanocrystals using FUS. Apart from ZnS, this method can also be applied to waterproof materials such as Sr₂MgSi₂O₇:Eu²⁺/Dy³⁺, and CaTiO₃:Pr³⁺. However, for some ML materials that react with water (such as SrAl₂O₄), process modifications may be necessary.¹⁸⁸

Combustion method

As a classical chemical synthesis method, the combustion method has been widely employed to synthesize hundreds of compounds, including ZnS nanocrystals and QDs, owing to its advantages such as energy efficiency, simplicity of operation, low cost, and high product reactivity.^{219, 220} The energy required for combustion reactions is derived from the reactions themselves, with heat propagating in the form of a combustion wave from the reaction zone to other regions, inducing reactions between different compounds. However, for reactions that cannot sustain themselves, chemical furnaces or preheating methods can be applied to prevent interruptions during the reaction. A specific procedure is shown in Figure 25a, where the particle size of the product can be adjusted by varying the oxidizer and fuel ratio (O:F) (Figure 25b).²¹⁹ However, it is worth noting that the combustion method may produce toxic gases during the synthesis process, which can pose an environmental burden.

4.3 Thin films and nanostructures

Thermal evaporation deposition

Thermal evaporation deposition is a process in which a material is heated to the point of evaporation, and the evaporated atoms or molecules condense onto a substrate to form a thin film (Figure 26a). Thin films of ZnS are more susceptible to stress and can be easily sprayed or adhered to target objects, enabling efficient mechanical-to-optical energy conversion.^{66, 221, 222} Moreover, this method can also be used to efficiently prepare ZnS of various morphologies, such as nanowires, nanobelts, nanorods,⁵⁸ thin films, and even pyramid shapes (Figure 26b–e).⁵⁹ The raw material used is ZnS, and the heating source can be a tube furnace,^{59, 190} or laser,²²³ etc. The heating atmosphere can be selected as NH₃ or Ar, and the substrate can be made of quartz,⁵⁹ Si,¹⁹⁰ GaAs,⁵⁸ optical glass, steel, aluminum oxide, silicon nitride ceramics,²²⁴ or piezoelectric materials,⁹⁵ among others. During the preparation process, catalysts such as Au, Sn, and Bi can be added to the substrate or raw materials to assist in nucleation.²²⁵ Currently, the application of thermal-deposition synthesized ZnS thin films and nanowires is focused primarily on areas such as nanofriction generators and electronic sensors.^226-228

Using the process of synthesizing array-type one-dimensional ZnO-ZnS nanowires by Lu et al. as an example,¹⁹⁰ first, 1 g ZnS powder was placed in an alumina boat and positioned at the center of a tube furnace. Then, a silicon wafer coated with a 15-nm Au film (as nucleation sites to facilitate the formation of continuous films) was used as the substrate. After evacuating to a 100 mTorr vacuum, high-purity Ar gas was fed into the system at a flow rate of 50 sccm, maintaining the pressure within the tube at 5 Torr during the synthesis. Subsequently, the tube furnace was heated to 800°C, increasing the temperature at a rate of 50°C · min⁻¹, held for 10 min, before being further heated to 1100°C at a rate of 20°C · min⁻¹, followed by a 60-min holding period, before finally being allowed to cool naturally. Owing to residual oxygen during the synthesis process, ZnO continues to grow at the end of the ZnS nanowires, forming ZnO-ZnS heterostructures, while pure ZnS nanowires can be obtained through selective etching.¹⁹⁰

Similarly, materials can also be evaporated to form thin films using an electron beam.²²⁴ Xu et al. deposited ZnS films on different substrates using this method and achieved a twofold enhancement in ML intensity by adjusting the growth process parameters.²²⁴ The orientation and surface morphology of ZnS films can be further controlled through vacuum annealing processes (500–700°C), enhancing the adhesion between the film and the substrate, thereby improving the piezoelectric field energy conversion efficiency.^{23, 229-231} To overcome the problem of ZnS film detachment during ML testing, a ZnO buffer layer can be added between the ZnS film and the substrate.²³¹

Ultrasonic spray pyrolysis

Ultrasonic spray pyrolysis is a technique for synthesizing nano-sized crystals or producing thin-film materials. It has been employed in the preparation of metal sulfide nanoparticles,²³² carbon nanocapsules,²³³ metal oxide ML particles,²³⁴ ZnS films^{235, 236} and nanoparticles (ZnS:Ni²⁺,⁵⁷ and ZnS:Eu^{3+ 237, 238}), among others. In this method, precursor solutions are atomized using an ultrasonic frequency nebulizer, resulting in a fine vaporized mist that facilitates the formation of extremely-thin-film layers. The size of the droplets in the spray jet decreases as the nebulizer's frequency increases, enabling adjustment of droplet size through frequency modification.

Using the synthesis setup shown in Figure 27 as an example,⁵⁷ the process begins with the use of a high-frequency ultrasonic spray system to atomize a precursor solution (prepared by dissolving ZnNO₃, CH₄N₂S (thiourea), NiNO₃, and colloidal silica aqueous solution in a stoichiometric ratio). Subsequently, an inert gas (Ar) is employed to transport the atomized droplets into a high-temperature furnace for a reaction at 700°C. As the water rapidly evaporates, due to the limited solubility of ZnS and its precursors, the droplets decompose and form an outer shell rich in ZnS, with the inner core comprising primarily silica particles. The silica particles can be dissolved in hydrofluoric acid (HF) ethanol solution, resulting in the formation of nanostructured WZ-ZnS:Ni²⁺ hollow microspheres (as shown in Figure 27b,c). It is important to note that when the furnace temperature increases from 700 to 1000°C, the final product (after HF treatment), comprises aggregated nano-sized particles with the microspheres losing their structural integrity, as shown in Figure 27d,e.⁵⁷ In a control experiment, when the same precursor solution devoid of colloidal silica is processed in a tube furnace at 1000°C with an Ar gas flow, only micrometer-sized ZnS:Ni²⁺ solid microspheres are produced, as depicted in Figure 27f,g.⁵⁷

Although this method has advantages (such as the low consumption of raw materials and low reaction temperatures), controlling the thermal treatment process of atomized droplets within high-temperature furnaces can be challenging, and the quantity synthesized in a single run is relatively small. Consequently, it is still necessary to improve the overall process.

5.1 Impact testing method

The impact testing method has been widely used to test ML signals. It involves the direct impact of a steel ball or other objects on ML materials, the collection and quantification of the ML signal and its intensity. It has the advantages of simplicity, high repeatability, and low cost. However, it can pose challenges when testing liquid samples. Additionally, signal collection typically occurs only once per impact, resulting in a relatively short contact time between the sample and the steel ball and a weaker ML signal.

In a setup used by Fontenot et al., a pin acted as a switch to control the free fall of a steel ball, which impacted a plexiglass plate evenly coated with ML powder.²⁴¹ The emitted light from this impact was collected using a PD positioned below the glass plate (Figure 29a). Similarly, the scheme developed by Lin et al. exhibited greater precision and compactness (Figure 29b).²⁹ They used an electromagnet as a switch to control the free fall of the steel ball. The ML powder was mixed with PMMA and encapsulated in polyvinyl chloride (PVC) as a film to prevent measurement errors caused by the movement of the powder during the impact. The emitted light was then collected using a CMOS sensor located below the glass plate.

For scenarios involving high speeds and large impact forces, additional devices are needed to impart kinetic energy to the impacting ball. Haider et al. used an airsoft electric gun to accelerate the ball to speeds of several tens of m · s⁻¹ (Figure 29c).²⁴⁵ Other components of the system included a speedometer, target cylinder, condenser followed by an optical fiber, PMT, and digital oscilloscope. The airsoft electric gun imparted kinetic energy to the ball, shooting it toward the target material film from a distance of 20 cm. The PMT captured photons through optical concentrators and fiber optics, with the oscilloscope recording the light intensity. The optimized system reduced the lateral area and allowed for easy adjustment of the projectile's mass to regulate the mechanical energy. More powerful acceleration devices could achieve higher speeds, as demonstrated by Bergeron et al., who used a two-stage ultra-high-speed light gas gun to accelerate the ball to 5–6 km · s⁻¹,²⁴⁶ and Fontenot et al., who achieved a speed of 410 ± 9 m · s⁻¹ with their setup.²⁴⁷

5.2 Friction, deformation and ultrasonic testing method

Owing to the fact that a multitude of ML materials exhibit TML rather than DML, frictional methods are often used to obtain ML signals. Typically, prior to testing, samples require a degree of preprocessing, with samples prepared as films either directly encapsulated or mixed with a matrix to prevent powder movement during loading. Subsequently, a mechanical device applies a certain frequency of friction to the sample, and the ML signal is collected and analyzed through the PMT or CMOS devices.

Using the setup by Peng et al. as an example, the material was reciprocally rubbed in the horizontal direction to generate ML, as shown in Figure 30a.³³ In this testing system, the ML powder, mixed with ultraviolet-curable adhesive, was encapsulated as a film and affixed to a transparent quartz plate to secure the sample. A metal probe with adjustable pressure served as the force applicator. Friction was controlled using a programmable linear motor to regulate both the distance and frequency of friction. The resulting ML signal was captured using a fiber optic probe through the quartz plate and transmitted to the spectrometer for subsequent data collection.

Deformation methods include tensile and compression deformation tests, differing primarily in the direction of the applied force on the sample. Specifically, sample preprocessing is also required to prepare samples as thin films or coat them onto external deformable substrates. The samples are then subjected to tensile or compression forces using a linear displacement platform or a loading machine. In the testing conducted by Shohag et al. (Figure 30b),²⁴⁰ an 858 MTS testing machine was employed, with a mechanical force dynamically applied in the vertical direction. The ML signals were amplified using a PMT before entering a data acquisition device, with a computer program converting the output voltage into digital signals.

In cases where the ML material is liquid, forces cannot be directly applied; consequently, an indirect method of application through ultrasound can be used. In the apparatus used by Terasaki et al.,²⁴² the ML particles combined with an adhesion matrix were fixed at the bottom of a sample tube, and ML was excited in an ultrasonic cleaning bath, with light being collected using a PMT and photon counter (Figure 30c). Moreover, the testing apparatus by Jiang et al. exhibited a higher degree of automation.²⁴³ They constructed a circulatory system with a peristaltic pump serving as the power source (Figure 30d). At one end of the system, a UV light charged the ML colloidal solution, and at the other end, a FUS transducer excited the ML, with the light being transmitted to a spectrometer for detection. The ML colloidal solution was then recirculated back to the charging end by the peristaltic pump, with the entire process being controlled by software (such as LabVIEW and MATLAB), facilitating cyclical testing.

5.3 Extreme condition testing method

In the case of extremely small or exceedingly large stresses, special equipment is required for testing. An AFM probe can be used for the observation of ML at the particle level, but its complexity and high cost are drawbacks.²⁴⁴ Nonetheless, it enables the observation of ML processes at the microscopic level and allows for the visualization of particle fracture after force application.

Sakai et al. used this method to investigate ML emissions from SrAl₂O₄ particles under conditions of elastic deformation, friction, and destructive deformation.²⁴⁴ They employed a cantilever with 10-nm tip radius as the force applicator, varying the applied force by adjusting the magnitude of the deflection voltage (Figure 31a). Additionally, owing to molecular forces, when the cantilever approached the ML particles, it would suddenly draw closer, leading to a sudden change in the actual applied pressure at that moment. Consequently, the experimental setup also incorporated laser deflection methods to measure the deformation of the AFM cantilever, providing data on the force during these abrupt changes. The stress applied using this method could be as low as the μN level. However, owing to the minimal number of photons collected (approximately a few hundred), the PMT was necessary to amplify the signal, although this method exhibited a relatively low signal-to-noise ratio, requiring multiple measurements.

High-pressure testing can be achieved by applying extreme pressure using exceptionally hard materials, such as diamond.^{34, 78} In the apparatus developed by Zheng et al.,³⁴ a 532-nm laser served as the excitation source, and a diamond anvil cell (DAC) was employed as the pressure application end for ML testing, with a maximum pressure reaching 19.20 GPa. The samples were mixed with micrometer-sized rubies and loaded into the DAC chamber, while the pressure during PL testing was transmitted through a pressure transmitting medium (a mixture of methanol/ethanol/water). The generated ML signal was collected using a CCD after having passed through a long-pass filter and lens, as shown in Figure 31b,c.

6.1 E-signature and anti-counterfeiting applications

Signatures are considered an effective anti-counterfeiting technology, with the main principle being authentication based on individual signing habits and characteristics.²⁶³ Consequently, ZnS-based electronic signature and anti-counterfeiting systems represent a category of anti-counterfeiting solutions. Compared to traditional electronic solutions, they offer the advantages of being passive and having simple structures, resulting in longer lifespans. One of their key advantages is the real-time recording of the signer's signing habits, which can render the signature more difficult to forge. Even in extreme environments (such as prolonged exposure to humidity), these sensors can maintain high response speeds and sensitivity.^{97, 264}

The spatial resolution of an ML anti-counterfeiting system depends on the micro-patterning technology used during manufacturing. Wang et al. used a standard photolithography process to fabricate large-scale ZnS ML photoresist arrays on a polyethylene terephthalate substrate, achieving a high spatial resolution of handwritten fonts up to 100 μm (254 dpi) (Figure 32a,b).²⁵ Additionally, it is possible to enhance the resolution by creating molds with unit cells through photolithography and filling them with ML powder. This fabrication method offers the advantages of avoiding crosstalk between adjacent units and mitigating any adverse effects that can result from the direct mixing of ML powder and photoresist (Figure 32c–e).^{115, 260} Moreover, they developed an ML signature anti-counterfeiting system capable of visualizing handwritten signatures (Figure 32f–h). This system enabled single-point dynamic pressure recording and 2D planar pressure mapping within the range of 0.6–50 MPa, achieving rapid response times of <10 ms.²⁵

ACEL devices based on ZnS can also be used for signature and anti-counterfeiting systems. Moreover, such devices can effectively prevent triplet-triplet or triplet-charge annihilation, which occurs in traditional direct current injection-driven devices at high charge densities.²⁶⁵ This is because the continuous flipping of the applied electric field can prevent electron accumulation. Ji et al. demonstrated a new three-phase electric-power-driven EL (TPEL) device structure, that could be easily integrated into 110/220 V and 50/60 Hz AC power systems, considerably improving system integration.²⁶⁵ These TPEL devices can be thought of as three parallel capacitors, each phase being 120° out of phase with the others. When a polar electrode bridge (such as a water layer) is attached to the top of the luminescent powder layer, these capacitors become a series of connected capacitors, with the luminescent powder layer sandwiched between the electrodes and polar electrode bridge. During AC excitation, the direction of the dipole moment above the polar electrode bridge continuously flips, inducing an electric field perpendicular to the luminescent powder layer and resulting in EL. Researchers have demonstrated a lollipop-shaped electrode pattern (Figure 33a), which can be written on with water, recording the writer's handwriting. It does not require complex back circuits and pressure sensors, and the marks can be restored to their initial state through drying (Figure 33b,c).²⁶⁵ Zhu et al. developed a flexible self-powered patterned EL device based on TENGs to overcome the problem of traditional QR codes being invisible in the dark. The device comprised a ZnS:Cu/PVDF-HFP composite material (light-emitting layer), flexible upper indium tin oxide electrodes, and patterned lower laser-induced copper electrodes. It generated electricity through friction from the TENGs, enabling custom pattern display and visual information interaction (Figure 33d).²⁶⁶

Usually, ML signature anti-counterfeiting systems rely on capturing ML using an image acquisition unit and analyzing it using custom software, lacking the capability to store information. To address this challenge, Zuo et al. proposed a solution by applying force to specific areas of an ML film in advance and then exposing the film to other mechanical stimuli, which triggered visible light emission to reproduce the information (Figure 33e–k).²⁶⁷ This was because the microstrain in the prestressed area was invisible when the external force was removed, becoming visible only when subjected to external mechanical forces (such as stretching, wind, and ultrasound)—that is, this device could accurately remember the locations where it experienced stress and invisibly store these changes, which could then be reproduced as visible light through opening/closing external stimuli.²⁶⁷ Water mist could also reproduce information stored in the film, without the need for additional equipment or power units, making the entire system compact. This was due to the static charge generated from friction between the water droplets and substrate material, stimulating ZnS material to emit friction-induced luminescence.^269-271

Aside from electronic signatures, ML ink and paper also have enormous potential in anti-counterfeiting applications. Ma et al. prepared ML paper by mixing ZnS:Mn²⁺ ML powder with pulp, which could have patterns directly printed on it (such as floral patterns) using a commercial laser printer (Figure 33l–n).¹¹¹ Under a fluorescence microscope, these ML particles emitted yellow light. Owing to the partial penetration of ML from internal particles, they exhibited an enhanced luminescence response to external stress. Self-adhesive label stickers made from the ML paper were white but turned yellow under a UV lamp (Figure 33o,p), emitting bright yellow light under the friction of a finger (Figure 33q), demonstrating sensitivity to external forces. The ML powder could also serve as a functional additive in the printing ink. After printing, only a faint pattern remained, primarily because the ink penetrated the paper and the printed pattern contained only a small quantity of ML powder. If printed on plastic or metal, the pattern tended to be transparent.¹¹¹

Combining ML materials with fingerprint technologies could also be a feasible solution. Yin et al. proposed a fingerprint recognition touchpad based on the ML effect, using ZnS:Cu/SiO₂ as the ML material to convert finger pressure into optical signals.²⁶⁸ These signals could be received by visible light sensors and further converted into electrical signals. By integrating this with an established unlocking interface and processing, analyzing, and matching signals from different channels, it could be determined whether the electronic password lock had been opened (Figure 33r).

6.2 Wearable stress sensor

With the advantages of excellent luminescent repeatability, self-recovery capability, and high environmental tolerance, ZnS can be used for the fabrication of superior-performing flexible wearable devices in fields such as human–machine interaction,²¹ smart textiles,^{166, 272-274} and electronic skin,^{275, 276} among others.²⁷⁷ Wearable ZnS-based devices can generate light through electrical or mechanical energy, leveraging their ML or EL properties. A typical fabrication approach involves compounding ZnS with PDMS, weaving it into textiles, or preparing it as optical fibers, combined with image processing systems for applications.²⁷⁸ To further enhance the ML intensity and overall mechanical performance, the efficiency of the stress transfer can be improved by altering the composite structure of the substrate or fibers.^{279, 280} Additionally, combining deformation luminescence with friction-induced luminescence can enhance the overall luminous intensity.²⁸¹

Jeong et al. fabricated flexible and stretchable ML fibers using ZnS:Cu/PDMS, as shown in Figure 34a.²⁸⁰ To enhance adhesion, a surface-treated cross-shaped rubber fiber framework was employed, offering a larger contact area compared to circular fibers. Moreover, a primer based on organosilane coupling was used to increase adhesion between the fibers and ZnS/PDMS. After curing the ZnS/PDMS, a transparent silicone adhesive layer was applied to cover the fiber surface, serving to fill any cracks or defects in the ZnS/PDMS. Figure 34b shows an enlarged image of a warp yarn filled with the ZnS/PDMS mixture, and Figure 34c shows the ML effect during the stretching process.²⁸⁰

Yang et al. used a novel thermoplastic polystyrene material instead of PDMS to prepare ML fibers.²⁸¹ This material could be dissolved in a mixed solvent of cyclohexane and paraffin oil, allowing for the recycling of ML phosphors and repeated textile production, meeting both durability and environmental requirements. Figure 34d shows the process of ML fiber fabrication, where hollow core yarns are introduced into the extrusion head, and the core and ML composite material are extruded and thermally stretched on a hollow spinning plate. By adjusting the stress of the ML composite material, a helical structure can be introduced into the straight core yarn owing to friction between the sheath and core materials. This structural design allows DML and TML to occur simultaneously during yarn stretching–retraction, synchronously producing light and electrical signals. Consequently, the overall ML intensity improved by approximately 30% compared to DML alone, and by optimizing the spatial and surface charge polarization coupling effects through altering the ML powder ratios, the overall triboelectric performance increased by 100%. Figure 34e,f shows woven luminescent fabrics with alternating ML fibers and spandex as warp threads and cotton yarn as weft threads, offering high tensile properties and comfort in the warp direction. Figure 34g shows the excellent water-resistant and luminescent properties of the ML fiber.²⁸¹

Biting can be used to generate signals in special situations, providing precise control and a new form of interaction for individuals with limited mobility. Hou et al. designed an interactive orthodontic device embedded with ZnS:Cu²⁺/Mn²⁺ (595 nm), Cu²⁺ (525 nm), and Cu⁺ (475 nm) ML phosphors within optical fibers, strategically positioned at predetermined locations (Figure 34h).²¹ These phosphors could modulate orange, green, and blue ML when distributed pressure was applied externally to the optical fibers. By detecting the fluorescence luminance ratio of unique bite points on the orthodontic device, various mechanical deformations on the lateral position could be distinguished. When combined with machine learning algorithms, this device could convert complex bite patterns into precise data input, enabling the interactive orthodontic device to operate computers, smartphones, and wheelchairs.

Figure 34i shows the pressure distribution and light propagation characteristics using a 4 × 4 single-layer array under different force modes. Figure 34j shows a three-dimensional (3D) printed soft orthodontic device incorporating a 2 × 3 single-layer sensor array with six pressure points. Two highly sensitive color sensor chips are positioned at the sensor's end and linked to a flexible processing circuit, allowing effective bending to accommodate the orthodontic device's interior, achieving high-precision sensing. Similarly, Liang et al. and Zhou et al. integrated ML materials with PDMS coated on the surface of optical fibers to create high-precision strain sensors²⁸² and distributed optical fiber strain sensors,¹²¹ demonstrating stress durability exceeding 10,000 and 8000 cycles, respectively.

The EL characteristics of ZnS are also frequently used in the fabrication of smart textiles with large-area display capabilities.²⁸³ When activated by AC electric fields on a polymer matrix, these textiles can serve as real-time communication tools for individuals with speech or language difficulties. The improved EL fibers are compatible with existing textile processes,^{64, 283} demonstrating good abrasion resistance and resistance to washing. Shi et al. presented a comprehensive textile system, including a display screen, keyboard, and power source (Figure 35a,b).⁶⁴ They used an industrial loom to weave conductive warp yarns and luminescent weft yarns together with cotton yarn. Each interwoven warp and weft yarn formed an EL unit, and when they came into contact, the circuit was completed, resulting in light emission. These textiles featured evenly spaced EL units, with less than an 8% relative deviation in brightness among 600 EL units. Even after repeated folding in different directions, most EL units exhibited intensity variations of <15%, and EL units along the fold line maintained stability after 10,000 folds, which indicated higher durability compared to traditional thin-film displays, with average brightness comparable to commercial flat-panel displays. As these fibers were woven, the density of EL units could be readily modified by adjusting the weaving parameters, which alters the spacing between the warp and weft intersections. They could also be sewn using nylon and cotton threads to accommodate various applications. Moreover, apart from creating EL fibers and then weaving or incorporating them into existing textiles, more lightweight and highly extensible smart textiles could be obtained by altering the physical structure of the textile. For example, Wu et al. demonstrated an open framework structure for ultra-thin textiles based on ZnS:Cu, reaching the strain limit of the textiles (200%). They designed electrode patterns on the textile and combined them with a soft contact layer to create luminescent textiles (Figure 35c–e).²⁸⁴

Owing to the excellent mechanical properties of ZnS itself, ZnS-based ACEL devices typically exhibit strong deformation resistance, as long as the composite materials are appropriately chosen. For example, PVDF is a polymer with a high dielectric constant that can effectively concentrate the electric-field-on fluorescent powder particles, thereby increasing the charge on the fluorescent powder particles and enhancing the light intensity.²⁷⁹ Zhou et al. introduced ZnS into a PVDF-HFP elastomer to create stretchable dielectric nanocomposites and prepared them as stretchable ACEL devices.²⁷⁹ These devices could operate at low voltages ranging from 10 to 35 V. They were mounted on the back of the hand using a silicon-based adhesive, ensuring conformal and intimate interaction with the skin. When this device displayed numbers (such as a stopwatch), it exhibited excellent deformation resistance, allowing it to continue functioning as the wearer made various hand gestures (Figure 35f). Go et al.²⁸⁵ combined Ecoflex with ZnS to create electronic skin and ACEL devices. These devices could still emit light when driven by AC even under various deformations such as bending and twisting (Figure 35g–j).

As shown in Figure 35k, during the initial stages of stretching, as the strain increases, the brightness increases to 395 cd · m⁻² owing to reduced emission layer thickness and subsequent electric field enhancement. For stretches exceeding 800%, the brightness decreases with rapidly increasing stretch, attributable to the increased electrode resistance. However, at 1400% stretch, the measured brightness remains at 234 cd · m⁻², comparable to the initial state. Importantly, the emitted light maintains its color throughout the stretching process.²⁸⁵ Moreover, if the ACEL devices are combined with self-healing polymer materials such as polyacrylic acid and PU, shape-deformable and self-healing EL display can be achieved (Figure 35l–n).^{149, 287, 288} Finally, as was the case for the ACEL devices' electrode materials, the materials here included Ag nanowires,^{283, 289, 290} carbon nanotubes,^{275, 291, 292} conductive polymers,^{149, 293} graphene,²⁹⁴ and their hybrids.^{295, 296} Moreover, the electrode shape could be designed in a wave-like form to enhance their overall stretchability and durability.²⁹⁰

TENGs are also an important means of obtaining AC electric fields, enabling self-powering capabilities for devices.²⁹⁷ Zhao et al. created an electronic skin comprising a ZnS:Cu/Al³⁺ ML layer, aluminum electrodes, insulating layer (Kapton), shielding layer (aluminum), and PDMS substrate.²⁹⁸ This electronic skin could sense pressures as low as 20 kPa and, when integrated with a microcontroller, could be used as a programmable touch interface supporting over 156 interactions. In some studies, the ML effect has been used to enhance the sensitivity of electronic skin. Jang et al. fabricated pressure-sensitive MoS₂ field-effect transistors incorporating ZnS ML particles and an air dielectric layer as a tactile pressure sensor array. Pressure applied to the ZnS elastic material generated ML, considerably enhancing its sensitivity to pressure and producing additional photocurrents in the MoS₂ channel, extending the detection range from 200 Pa to 5 MPa (for a wide range of pressure distribution).¹¹⁶

6.3 Battery-free display

ZnS light sources exhibit multimode conversion properties, allowing for visualization of not only mechanical forces but also responses to airflows,^{63, 195} magnetic fields,²⁹⁹ AC electric fields,¹⁵⁰ and even moisture,³⁰⁰ boasting potential applications in different scenarios and extreme environments.^{279, 301} Their implementation is primarily achieved by fabricating devices through the composite of ML materials with substrates such as PDMS. Tuning the ML color can be accomplished by blending different ML materials or by combining them with fluorescent dyes.^{56, 263, 302}

In outdoor environments, wind energy is one of the most common forms of mechanical energy that can be used to drive ML. Jeong et al. demonstrated an ML display driven by airflows, suitable for large-scale passive outdoor displays.⁶³ The pixel units of this display comprised ML rods, which were formed by mixing ZnS with PDMS and curing them in molds. Under the impact of N₂ airflow, the rods vibrated and generated ML along the direction of the airflow (Figure 36a). It is worth noting that rods close to the airgun did not vibrate and thus did not produce ML. As the distance increased, the irregularity of the wind direction increased owing to interactions between the rapidly diffusing gas and the aligned rods. Typically, ML emission was evident only twice—that is, during the gas injection and gas cessation. However, natural variations in the wind direction can be complex, so further improvements in the display structure are needed to adapt to different wind directions.⁶³

In this research, the mechanical strength of the ML rods cured directly from ZnS/PDMS was limited. If the size of ML rods was reduced to improve the resolution, there was a risk of detachment. Consequently, it was determined that an appropriate microcolumn diameter should be 1 mm (approximately 12.7 ppi). This problem could be partially addressed by preparing the material in a thin-film structure, which can withstand greater friction and is easier to integrate with other devices.³⁰³ Furthermore, aside from direct display applications, ML can also be used to charge solar cells at night through the photoelectric effect,³⁰⁴ thereby enhancing the energy storage efficiency of solar cells.³⁰⁵

ZnS-based devices can achieve mutual conversion among ultrasound, optical energy, and electrical energy through clever structural design. Li et al. reported highly porous composite materials, prepared from PVDF-HFP and ZnS, capable of converting ultrasound into optical signals.²⁵⁴ Their efficient ML performance resulted not only from the porous structure of the films but also from the strong interaction between the PVDF-HFP and ZnS. The porous structure facilitated the conversion of mechanical forces into electrical energy and induced ML in the ZnS. Moreover, the particle–film interaction induced a high proportion of the β-phase (a notably effective piezoelectric phase) in the PVDF-HFP film. Consequently, this composite film could be excited to produce ML in a laboratory ultrasonic cleaning tank (Figure 36b). Figure 36c shows an array based on this composite material, forming the letter Z under ultrasound vibration.²⁵⁴

Zhang et al. grew ZnS:Mn²⁺ thin films on a piezoelectric Pb(Mg_1/3Nb_2/3)O_3-xPbTiO₃ (PMN-PT) substrate, enabling the simultaneous conversion of electrical signals into optical signals and ultrasound (Figure 36d,e). This was achieved through the inverse piezoelectric effect in PMN-PT.³⁰³ This novel dual-mode light source for ML and ultrasound emission, integrated with a detection system on a single chip, holds considerable promise for the development of hybrid systems for tissue diagnostics.^{108, 306, 307}

Similarly, dielectric elastomer actuators (DEAs) can generate mechanical motion when subjected to an electric field, displaying outstanding mechanical actuation performance with area strains exceeding 200%.^{308, 309} Wang et al. integrated DEAs with stretchable EL devices, demonstrating a self-shaping ACEL device.¹⁵⁰ Compared to traditional ACEL devices, this special device exhibited dynamic shape display capabilities and could achieve approximately 55% area strain under an excitation voltage of 5 kV. Cho et al. created a stretchable sound display device capable of producing synchronized audio and visual information (Figure 36f).¹¹³ The device incorporated a stretchable EL loudspeaker based on ZnS:Cu emitting layers and a matrix of DEAs positioned between stretchable electrodes (Figure 36g). When an AC bias voltage was applied to this speaker, the emitting layer underwent EL, generating audible sound waves through the vibration of the DEA matrix. Consequently, the intensity of both light emission and sound could be adjusted by varying the bias voltage parameters.

The detection of magnetic fields is crucial for environmental monitoring, mineral exploration, and safety surveillance applications, and ML materials can be applied in this context.³¹⁰ Compared to traditional magnetic sensors, they offer advantages of non-contact, non-destructive, and non-invasive detection.^{61, 311} Wong et al. embedded soft magnetic particles into PDMS to create a composite layer, which could produce magnetostrictive strain in a magnetic field, triggering adjacent ZnS/PDMS layers to emit ML, as shown in Figure 37a. When the magnetic field frequency was modulated from 50 to 470 Hz, the emission color shifted from green to blue, and the intensity changed too (Figure 37b).³¹⁰ Listyawan et al. developed a magnetic field-driven lighting device without electricity sources. This device comprised a Kirigami-shaped ZnS/PDMS composite and a magneto-mechano-vibration cantilever beam, capable of generating ML under magnetic field drive (Figure 37c). This structure reduced rigidity and increased stretchability, promoting ML generation (Figure 37d).²⁹⁹

The aforementioned magnetic-mechanical-optical energy transformation exhibited low conversion efficiency and produced weak light emission. In another approach, Yang et al. leveraged the EL properties of ZnS:Cu to achieve battery-free, high-brightness light emission, applying electromagnetic-electric-optical energy transformation, which could be more suitable for harnessing electromagnetic energy.^{148, 312} They developed a sophisticatedly structured interactive fiber to realize this function. First, this kind of fiber could serve as an antenna, and the electromagnetic field in the environment (simulated by applying a 30-dBm/∼20-kHz transmitter) could induce an electromotive force within the fiber, charging a capacitor. Subsequently, when the human body contacted the fiber, a closed-loop circuit was formed between the fiber, the human body, and the earth. The electrical energy stored in the capacitor was then rapidly discharged (Figure 38a). Owing to the high dielectric constant of the human body, when the finger was in close proximity to the interactive fiber, the momentary high voltage caused air breakdown, resulting in local plasma discharge (Figure 38b,c), leading to the simultaneous emission of electromagnetic waves and light, with the light emission facilitated through the EL properties of ZnS:Cu. Under 21.1°C and 46.8% relative humidity conditions in office, the brightness could reach up to 270 ± 11 cd · m⁻². This kind of intensity makes it possible to use the technology in various battery-free display scenarios, such as interpersonal information exchange and friction-driven luminous carpets (Figure 38d–h), holding great promise for the integration of ML properties to enhance the overall luminosity and energy conversion efficiency.

Finally, as is evident from the applications discussed in the previous three chapters, printing technology has been used extensively, making it one of the most common methods for fabricating ML materials into fibers, arrays, patterns, and other application forms.^{34, 305, 313-315} For ZnS, 2D printing is often done using screen printing to form thin-film structures on paper or ceramic pieces.^{111, 162} 3D printing techniques commonly used for ZnS include direct-write printing and extrusion-based 3D printing. For example, Qian et al. prepared printable inks based on ZnS/SiO₂/PDMS, where SiO₂ nanoparticles added to the ink fully infiltrated and adhered tightly to the PDMS polymer chains (forming Si-O bonds),⁶⁶ allowing for better transmission of pressure to the ML crystals. Zhao et al. used extrusion-based 3D printing and ZnS:Cu/SiO₂/PDMS ML ink to create periodic honeycomb structures.³¹⁶ When subjected to uniaxial stretching, these two structures alternately produced anisotropic and isotropic light emission.

Moreover, Li et al. developed a universal strategy for laser direct printing of inorganic nanomaterials, which eliminated dependence on photocurable resins during the printing process, enhanced the content of ML materials, and improved the overall ML performance of composites.³¹⁷ Moreover, this method exhibited extremely high compatibility with materials and offered considerable printing freedom. In their research, they printed over 10 types of semiconductors, metal oxides, metals, and their mixtures, including CdSe/ZnS core-shell QDs. Colloidal nanocrystals served as building blocks and could be chemically bonded through their native ligands. Without the need for resins, this bonding process could produce arbitrary 3D structures with over 90% inorganic mass fraction and high mechanical strength. This technique opens up new possibilities for designing and fabricating advanced materials and devices with tailored properties and structural complexity.

6.4 Biomedical imaging

In the field of biomedical research, ZnS has been applied primarily in the areas of bio-inspired skin, vivo imaging, and therapy.^318-320 As biological tissues and organs are opaque, it can be practically impossible to directly observe the physiological conditions of internal organs with the naked eye. Consequently, biomedical sensors are essential for real-time imaging and observation of internal organs. ML sensors have the advantage of being able to directly convert mechanical forces within the biological body into light emission. Compared to traditional biomedical sensors, they offer simplicity in structure, high complexity, good self-recovery properties, and excellent biocompatibility.

Kim et al. developed an artificial cochlea (Figure 39a,b) using a ZnS:Cu/PDMS composite film as the human basilar membrane (BM), which could simulate the phenomenon of upward wave propagation on the BM in the human cochlea.³²¹ When the ML BM was subjected to acoustic waves, it produced its maximum wave amplitude at different positions, inversely proportional to the frequency of the sound waves. By measuring the position of ML emission on the BM using a non-contact image sensor, the frequency information of the sound waves could be obtained. This method had a theoretical resolution that was infinite (continuous), as it was not limited by traditional artificial cochlea BM arms or hair cell numbers, and could achieve real-time imaging and wireless sensing of mechanical stress within the biological body. As shown in Figure 39c, with increasing excitation frequency, the peak position moves from the apex to the base of the ML BM. The wave travels from the base to the top, and the position of the maximum wave amplitude shifts considerably from the top to the base as the stimulus frequency increases. At high frequencies (150 Hz), the shape of the envelope of the wave tilted slightly near the base, whereas at low frequencies (25 Hz), it tilted slightly near the apex. However, the shape of the wave envelope did not change significantly at low frequencies. This observation confirmed the velocity input response of the ML BM to the oval window at different frequencies. Simulation observations from the Finite element analysis (FEA) clearly showed that the prototype artificial cochlea using the ML BM exhibited similar frequency selectivity compared to the human cochlea.³²¹

In terms of in vivo light sources, Hong et al. used extracorporeal ultrasound-focused excitation to trigger luminescence from ZnS:Ag⁺/Co²⁺@ZnS ML particles in the blood, enabling non-invasive detection from inside to outside (Figure 39d,e).^{67, 93} Initially, the ZnS:Ag⁺/Co²⁺@ZnS core-shell nanoparticles were intravenously injected into mice as light sources. As they flowed with the bloodstream into brain vessels, the continuous blood flow ensured a constant supply of luminescent light sources to the brain vasculature. Through excitation using FUS applied externally to the mice, via the intact skull and skin, the ZnS:Ag⁺/Co²⁺@ZnS nanoparticles in the mouse's blood emitted blue visible light, which could be charged using 400-nm UV light through surface-level veins. Subsequently, under the FUS excitation, 470-nm light emission could be generated, enabling non-invasive diagnosis of the optogenetic status of the mouse's brain (Figure 39e). Imaging sensors were used externally to visualize the luminescence within the body, equivalent to an effective power density of approximately 1.2 mW · mm⁻² (Figure 39f).⁹³ Moreover, the composite of ZnS ML nanomaterials with graphene could be used to create pulse sensors capable of sensing forces from 0 to 20 N.²⁰⁷

Photodynamic tissue bonding (PTB) is a new technology for trauma closure. This method employs light and photosensitizers to enhance collagen cross-linking, minimize inflammation, and lessen scarring. One frequently used photosensitizer in PTB is Rose Bengal dye. This dye captures energy from green light and reacts with collagen, producing radicals that facilitate the formation of covalent bonds among collagen molecules. Kim et al. used the long AL of ZnS:Ag⁺/Co²⁺ to stimulate the collagen matrix inside cut tissue for PTB (Figure 39g).³²² In the depth direction of skin closure incisions, the emission intensity of ZnS:Ag⁺/Co²⁺ particles at 540 nm remained robust, while the intensity of the green laser at the same wavelength decreases exponentially with skin depth. Within 30 s, the AL intensity of ZnS:Ag⁺/Co²⁺ particles significantly exceeded that of the green laser. This indicates that the green light from ZnS:Ag⁺/Co²⁺ particles can be efficiently transmitted to deeper tissue layers for sustained tissue bonding. Figure 39h shows the consistent increase in collagen fiber formation over a stable time point of 6 min. The results confirmed that using intermittent UV irradiation with a hyaluronate-RB/ZnS:Ag⁺/Co²⁺ mixture could accelerate collagen cross-linking more effectively than traditional PTB using green light.³²²

6.5 Deep learning and optical information processing

To date, machine learning and deep learning methods have been used primarily to extract information from ML signals or as auxiliary tools to enhance the accuracy of the information retrieval process. In this chapter, various models are introduced for processing ML information related to handwritten digits, crack propagation, and limb movement. It should be noted that the models currently applied to ML data processing are relatively simple, partly owing to the ongoing research in this field and partly limited by the accuracy of the CMOS and ML sensors themselves.

Zhou et al. used the random forest machine learning algorithm to recognize handwritten digits generated by ZnS:Cu sensors (Figure 40a–c).³²³ In their experiments, the network was pre-trained using the MNIST handwritten database, and handwritten-digit images were generated by extracting and overlaying consecutive frames from videos of handwritten digits. The green component was isolated to eliminate the influence of red and blue light, and after binarization, it was input into the network for digit recognition. The experiment compared the accuracy of three machine learning methods—that is, the random forest, K-nearest neighbors, and decision tree models—with the random forest model achieving the highest accuracy. After training with 60,000 handwritten digits, it achieved approximately 96.87% recognition accuracy on another dataset of 10,000 handwritten digits.

Compared to traditional methods, such as digital image correlation or FEA methods combined with ML for crack inspection—machine learning, and deep learning algorithms can achieve higher accuracy and real-time quantitative analysis. Zhang et al. proposed a machine-learning-based fatigue crack propagation detection method, where seven different models were used to classify cracks. The decision tree model exhibited the highest accuracy, with a crack length measurement precision of 0.6 mm.³²⁵

Ahn et al. employed the generative adversarial network (GAN) and convolutional neural network (CNN) to process ML and FEA images. The GAN was trained to convert real ML images into virtual FEA images (or vice versa), whereas the CNN was trained to derive a K_p value from the FEA or ML data as an alternative to traditional numerical solutions. This dual AI approach provided more accurate $${K}_{p}^{\text{AI}.\text{FEM}}$$ results compared to pure ML solutions. Moreover, it could also provide far-field stress distribution, achieving similar accuracy to real computed FEA while substantially reducing time and costs.³²⁶

In the design of a wearable bite device by Hou et al., two ML optical fibers were used, each containing three ML blocks that emitted green, yellow, and orange light.²¹ The spectral information of these three lights was combined into three stimulation values using a color sensor, resulting in six data points (X₁, Y₁, Z₁; X₂, Y₂, Z₂). These data were transmitted via Bluetooth to a computer for analysis (after conversion to CIE XYZ color space data [x₁, x₂; y₁, y₂]) (Figure 40d–f). There were a total of 21 bite modes, of which 14 were used for testing and training, with over 300 data points obtained for each mode. If a simple scheme of threshold classification based on CIE values was adopted, the overall recognition accuracy was only 89%. Not only was it difficult to set the threshold, but the set threshold also struggled to meet the goal of precise classification. Consequently, they employed a two-layer feedforward ANN for data analysis, with the ANN comprising 4 input nodes, 14 output nodes using a SoftMax function, and 10 hidden nodes with a sigmoid activation function. After training, this approach increased the accuracy to 98%.²¹

Deep learning is not only used for processing ML data but can also handle AL variation data, which is important for materials like SrAl₂O₄. Timilsina et al. used two CNN-based neural networks—that is, the Visual Geometric Group network and residual neural network (ResNet-34), to learn from fluorescent images.³²⁴ They achieved recognition accuracies of 98.5% and 98.75% for 10 pieces of information, including seven words, two letters, and a smiling mouth (Figure 40g,h). This work involved attaching eight long AL patches around a person's mouth emoji. When the person expressed different meanings, such as “around” or “ground,” the motion patterns of the eight fluorescent patches varied. Ten frames were selected and overlaid for each scenario, converted to grayscale, and then input into the neural network to extract information. To avoid overfitting, each stacked image was augmented with five copies generated by random transformations including rotations, scaling, flipping, and translations during training.

Combining target recognition neural networks with the variation of ML images may be used for a new generation of motion force sensors and systems (Figure 41a). ML images from objects have the feature that their luminescence intensity is proportional to the magnitude of the motion force, and they exhibit sustainable repeatability and self-recovery (Figure 41b,c). With the development of deep learning networks, precise capture and recognition of motion image information has already been achieved (Figure 41d).^{328, 329} Therefore, soft sensors based on ML materials can be worn on different parts of the body. When a part of the body moves and emits light, the neural network can recognize the location and amplitude of the motion force, achieving precise motion force sensing (Figure 41e,f). Chun et al. have proposed an interactive wearable display by integrating multicolor perovskite and ZnS materials, powered by an external energy source (Figure 41g,h).²⁷⁶ If self-powered ML materials are used instead of external energy, it is expected to create an interactive motion sensing system that meets the requirements of low energy consumption and high precision. Hou et al.'s research has preliminarily verified the feasibility of this concept.²¹ They achieved motion force pattern analysis of tooth occlusion in a small area of the oral cavity to output control signals (Figure 41i). Looking ahead, expanding this approach to large-scale motion force monitoring could result in a more comprehensive and interactive intelligent sensing system.

7.1 Conclusions

This paper provides a comprehensive review of the history, materials, performance, synthesis, and application of ZnS ML, highlighting its considerable advantages in passive visual sensing, self-powered displays, and innovative light sources. ZnS demonstrates enormous potential in various fields, including flexible optoelectronics, materials science, engineering, and energy, owing to its luminescence mechanism that differs from traditional methods such as EL and PL. As a classic ML system, ZnS serves as an inspiration for the development of new intelligent ML systems, LED-like heterojunction high-brightness ML, and bio-inspired mechanical excitation luminescence models.

7.2 Future prospects