Metal Oxysulfides: From Bulk Compounds to Nanomaterials

This review summarizes the syntheses and applications of metal oxysulfides. Bulk compounds of rare earth and transition metals are discussed in the section Introduction. After a presentation of their main properties and applications, their structures are presented and their syntheses are discussed. The section Bulk Materials and Their Main Applications is dedicated to the growing field of nanoscaled metal oxysulfides. Synthesis and applications of lanthanide-based nanoparticles are more mature and are discussed first. Then, works on transition-metal based nanoparticles are presented and discussed. Altogether, this review highlights the opportunities offered by metal oxysulfides for application in a range of technological fields, in relation with the most advanced synthetic routes and characterization techniques.


INTRODUCTION Definition
A "metal oxysulfide" is a compound composed of at least a metal, oxygen and sulfur, with negative oxidation states (e.g., -II) for both oxygen and sulfur. The generic formula for ternary oxysulfide is M x O y S z . Due to its negative oxidation state, sulfur forms no bounds with oxygen in oxysulfides, in contrast with more common metal sulfates M x (S VI O 4 ) y where the sulfur is +IV.
In 1951, Eastman et al. recommended the following distinction (Eastman et al., 1951): M x O y S z compounds should be designed by the general term "oxide-sulfide" and named after the similarities of their crystalline structure with the corresponding oxide or sulfide. If the oxide-sulfide has the same crystalline structure than the oxide, it should be named "thio-oxide"; if its crystalline structure is the same than the sulfide, it should be called "oxy-sulfide" and if its structure is none of the two, it should be called "sulfoxide." However, in an article of 1958 published in French, Flahaut et al. questioned this nomenclature (Flahaut et al., 1958). They argued that all the Ln 2 O 2 S (Ln, lanthanide) compounds crystallize in the same structure and showed similar chemical properties. With Eastman's nomenclature, because Ln 2 O 3 oxides crystallize in the two different structures Ce 2 O 3 and Tl 2 O 3 , the Ln 2 O 2 S compounds would have been named "thioxyde" (French word for thio-oxide) from lanthanum to praseodymium and "sulfoxyde" (sulfoxide) for the others.
Although the terms "thio-oxide" and "oxide-sulfide" are still present in the literature, "oxysulfide" is now employed in a large majority of the works to name any combination on one or several metals to oxygen and sulfur anions.
Altogether, metal oxysulfides represent a class of compounds that is independent from both metal oxides and metal sulfides, though common properties may be punctually identified depending on the metal, the crystal structure, and the anion substitution scheme.

Discovery and First Phases
Oxysulfides are scarce in nature and are the most often synthetic. One reason for this is the competitive formation of sulfates, which are found in numerous minerals and are more stable toward oxidation. This competition between sulfate and sulfide is also at stake when designing a synthetic route.
To the best of our knowledge, the first occurrence of an oxysulfide compound was reported in 1827 by Mosander, who was working on the sulfidation of Ce 2 O 3 into Ce 2 S 3 using H 2 S (Figure 1). He noticed the presence of oxygen and sulfur combined with the metal in a single product, along with the formation of cerium sulfate. Later, Sterba (1904) and Biltz (1908) also reported this observation. Biltz even proposed the formula Ce 2 S 2.5 O·S as he identified remaining sulfur as polysulfide on its final product. In 1907, Hauser prepared using H 2 S on oxides two oxysulfides of tetravalent metal, namely ZrOS and ThOS based on their composition (Hauser, 1907). Hauser indicated that the zirconium and thorium oxysulfides were pyrophoric. Without knowing it, Klemm et al. were probably the first to obtain a pure phase of Ln 2 O 2 S by heating Er(SO 4 ) 3 in H 2 S and consequently getting Er 2 O 2 S, that they only described as pale pink and resistant to other heating treatments in H 2 S (Klemm et al., 1930).
The first crystalline oxysulfide structures were elucidated by Pitha et al. (1947) (La 2 O 2 S) and Zachariasen (1949a) (La 2 O 2 S, Ce 2 O 2 S, and Pu 2 O 2 S). The samples often contained impurities and were prepared either by reducing the corresponding sulfate Ln 2 (SO 4 ) 3 using H 2 or by gently heating in air sesquisulfide compounds (Ln 2 S 3 ). The two authors noticed that the metal was coordinated to seven atoms: four atoms of oxygen and three atoms of sulfur. The Ln 2 O 2 S structure derives from the hexagonal oxide Ln 2 O 3 and crystallizes in the P-3m1 space group. This lamellar structure can be described as alternating sheets of [Ln 2 O 2 ] 2+ and S 2− (Figure 2). Since this discovery, the entire series of lanthanide oxysulfide Ln 2 O 2 S (except promethium) was prepared (Flahaut et al., 1958). In 1948, McCullough et al. established the structure of cubic ZrOS (McCullough et al., 1948) prepared similarly to Hauser and in 1962, Jellinek described a new tetragonal form with the same composition (Jellinek, 1962). In the 1960's and the 1970's, more M 2 O 2 S compounds were also reported. The work on radioactive elements gave the actinide oxysulfides Np 2 O 2 S (Marcon, 1967a), Am 2 O 2 S (Haire and Fahey, 1977), Cm 2 O 2 S (Haire and Fahey, 1977), Bk 2 O 2 S (Haire and Fahey, 1977), and Cf 2 O 2 S (Baybarz et al., 1974). Similarly to Ln 2 O 2 S compounds, An 2 O 2 S (An, actinide) materials crystallize in the P-3m1 space group. On the contrary, Sc 2 O 2 S crystallizes in the hexagonal P6 3 /mmc space group. Its structure remains very close to Ln 2 O 2 S with a coordinence of seven for scandium atoms and a structure based on alternative layers of [Sc 2 O 2 ] 2+ and S 2− (Julien-Pouzol et al., 1978).
In 1949, Zachariasen described the structure of tetravalent actinide oxysulfides ThOS, UOS, and NpOS as presenting a PbFCl structure type with tetragonal symmetry in P4/nmm space group. The tetragonal form of ZrOS described by Jellinek is isostructural of these compounds (Zachariasen, 1949b). On the contrary, cubic HfOS (isostructural to cubic ZrOS) was first identified by Stocks et al. (1980) and prepared as a pure phase by Eisman and Steinfink (1982). The latter were also able to prepare solid solutions of zirconium and hafnium oxysulfides Zr 1−x Hf x OS (0 ≤ x ≤ 1) such as Zr 0.25 Hf 0.75 OS and Zr 0.75 Hf 0.25 OS.
A few years later, Khodadad et al. and Ballestracci demonstrated the existence of several Ln 2 O 2 S 2 compounds (Ln = La, Pr, and Nd), where the disulfide [S 2 ] 2− anion is present (Khodadad et al., 1965;Ballestracci, 1967;Wichelhaus, 1978a). They have to be distinguished from AnOS (An, actinide) compounds in which the actinide is at the +IV oxidation state while the lanthanide in Ln 2 O 2 S 2 remains at the +III oxidation state.
A sulfur-rich phase was also discovered by trying to solve the crystalline structure of what was thought to be β-Ln 2 S 3 . It happened to be Ln 10 OS 14 (Ln = La, Ce, Pr, Nd, Sm) that formed because of traces of water or oxygen during the reaction Besançon, 1973). Besançon et al. showed that the oxygen content of Ln 10 S 14 O x S 1−x can be lowered down to a value close to 0.1 mol% for La, Ce, and Pr . Later, Schleid et al. also reported the gadolinium compound Gd 10 OS 14 (Schleid and Weber, 1998).
The work of Marcon with actinides led to the first description of more complex compounds, namely M III 2 M IV 2 O 4 S 3 (Pu 4 O 4 S 3 , U 2 Pu 2 O 4 S 3 , U 2 Gd 2 O 4 S 3 , and Ce 4 O 4 S 3 ), based on composition analysis (Marcon, 1967b). He also completed the work of Zachariasen by obtaining PuOS (Marcon, 1967b). In the same time, based on the work of Marcon, the compositions of cerium oxysulfides Ce 4 O 4 S 3 Wichelhaus, 1978b) and Ce 6 O 6 S 4 (Dugué et al., 1979) were confirmed and their structures were elucidated by X-Ray diffraction on monocrystals by Dugué et al. and Wichelhaus. Ce 4 O 4 S 3 and Ce 6 O 6 S 4 monocrystals were obtained by heating Ce 2 O 2 S and sulfur or CeO 2 , Ce 2 S 3 , and sulfur together. In the lanthanide series, only cerium allows both oxidation states +III and +IV. In Ce III 2 O 2 S, partial oxidation of cerium led to Ce III 2 Ce IV 2 O 4 S 3 and Ce III 4 Ce IV 2 O 6 S 4 .
A decade after the discovery of Bi 2 O 2 Se (Boller, 1973), Koyama et al. published in 1984 a study about the combination of bismuth with chalcogens. They obtained the ternary oxysulfide Bi 2 O 2 S from Bi 2 O 3 and Bi 2 S 3 via a hydrothermal synthesis (Koyama et al., 1984). The Bi 2 O 2 S structure differs from Ln 2 O 2 S (Ln, lanthanide), as it crystallizes in the Pnnm space group.
The coordination number of the bismuth is eight: bismuth is bound to four atoms of oxygen and four atoms of sulfur (Figure 3). In comparison with Ln 2 O 2 S in which Ln forms four Ln-O and three Ln-S bonds, bismuth-oxygen bonds are in the same length range (between 2.2 and 2.5 Å) but bismuth-sulfur  bonds are significantly longer (3.4 Å for Bi 2 O 2 S, <3 Å for Ln 2 O 2 S). Further works showed that bismuth can form several oxysulfides, leading to superconductive Bi 4 O 4 S 3  (containing both sulfide and sulfate ions) and to Bi 9 O 7.5 S 6 (Meng et al., 2015).

BULK MATERIALS AND THEIR MAIN APPLICATIONS Toward More Complex Structures: Quaternary Oxysulfides and Selective Bonding
We already cited the work of Marcon who isolated actinide oxysulfides U 2 Pu 2 O 4 S 3 and U 2 Gd 2 O 4 S 3 (Marcon, 1967b). These structures contain U IV and Ln III . This mixed valence allowed the formation of the An IV 2 Ln III 2 O 4 S 3 and An IV 2 Ln III 4 O 6 S 4 (of general formula An IV 2 Ln III 2n O 2+2n S 2+n ) structures by a shearing mechanism of the Ln 2 O 2 S structure when similar mixed-valent uranium-lanthanide oxysulfides were obtained (Tien et al., 1988). With Okabe et al. (1988), they also exhibited a series of U 2 La 2n−2 O 2n S n+1 compounds.
Besides, in the 1980's, a considerable amount of quaternary oxysulfides containing other metals than lanthanides or actinides were synthesized. Firstly, the idea was to insert another metal in the lamellar structure of a lanthanide oxysulfide Ln 2 O 2 S. The easiest way to get a quaternary oxysulfide was to put the other metal in the layer of sulfur anions, and consequently obtain a structure composed by sheets of lanthanide oxide and metal sulfide. This compound, in which oxygen is bound only to the lanthanide and sulfur only to the additional metal, exhibits a particular order that one can call selective bonding. As the quaternary oxysulfides can be formed with a large variety of precursors (mainly oxides and sulfides, but also elemental sulfur, H 2 S, metals, . . . ) and not only using lamellar preformed structures such as Ln 2 O 2 S, this selective bonding can be extended to any resulting oxysulfide in which one of the anions is preferentially bound to one of the metals and conversely. It generally led to layered compounds. On the contrary, when no such order is present in the structure (at least one metal site in the structure is bound to the two anions), the compound exhibits unselective bonding.
To illustrate this difference, we chose to study a family of quaternary oxysulfides Ln 2 Ti 2 S 2 O 5 (with Ti IV ) reported in the late 1990's ( Figures 4A-D). These structures turned out to be defective Ruddlesden-Popper phases which alternate [Ln 2 S 2 ] 2+ and [Ti 2 O 5 ] 2− layers ( Figure 4A). However, it is also possible to get compounds where both metals are equally bound to both oxygen and sulfur without particular arrangement (unselective bonding). It can be illustrated by the previously reported quaternary titanium oxysulfides La 4 Ti 3 O 8 S 4 and La 6 Ti 2 S 8 O 5 that do not show any selective bonding (Figures 4B,D; Cody and Ibers, 1995). In the 1980's, the study of the La w Ga x O y S z compounds already started the reflexion on the selectivity of the bonds in quaternary oxysulfides (selective bonding for LaGaOS 2 -α, La 4 Ga 1.33 O 4 S 4 , and La 3 GaOS 5 ; unselective bonding for LaGaOS 2 -β and La 3.33 Ga 6 O 2 S 12 , Table 1; Guittard et al., 1985).

Quaternary Oxysulfides: A Large Catalog of Compositions
Using high temperatures and long reaction times, monovalent (Cu I , Ag I ), trivalent (Cr III , Ga III , As III , Sb III , Bi III ), tetravalent (Sn IV ) and pentavalent elements (Nb V ) were shown to be able to crystallize along with a lanthanide in various types of oxysulfide compounds ( Table 1). In some cases, the second metal can also present mixed oxidation states (Ti III,IV , V III,IV ).
The study of the quasi-binary system La 2 O 2 S-AgGaS 2 (La 2 O 2 S -0.75 Ga 2 S 3 -0.75 Ag 2 S) by Carcaly et al. (1981) led to the formation of La 4 Ag 1.5 Ga 1.5 O 4 S 5 in which silver and gallium are randomly distributed in the same sites. Along with La 3 MO 5 S 2 (M = Nb, Ta;  (Eisman and Steinfink, 1982 (Tranchitella et al., 1996). La 5 Ti 2 MS 5 O 7 (M = Cu, Ag), an alkaline-free structure with perovskite layers was also evidenced by Meignen et al. (2004b) and studied for its photocatalytic properties for water reduction and oxidation (Suzuki et al., 2012). Meignen et al. (2005) (Rutt et al., 2003). As a perspective, in 2015, Yee et al. designed by DFT modeling a new high-temperature superconductor Ca 2 HgCuO 2 S 2 whose superconducting transition temperature should be close to mercury cuprates' ones (Yee et al., 2015).

Transition Metal Oxysulfides
For a long time, ternary oxysulfides M x O y S z were limited to lanthanides, actinides, and bismuth. Despite the presence of numerous metals in quaternary oxysulfides, the transition metals did not give any crystalline ternary oxysulfide (except ZrOS and HfOS) until the synthesis of ZnO 1−x S x in the 1990's. This phase is the most often found as crystalline thin films. It is also the case for titanium, tungsten and molybdenum oxysulfides except that they are amorphous.
The first-raw transition metals ternary oxysulfides represent a challenge, because the coordination of the metal commonly does not exceed six, and consequently cannot bear the M 2 O 2 S structure of Ln 2 O 2 S (Ln, lanthanide) where the lanthanide coordination is seven or the Bi 2 O 2 S structure where the coordination of bismuth is eight. Alternative crystal structures may be obtained in the case of first-raw transition metals.  (Meyer et al., 2013). These were obtained using radio-frequency magnetron sputtering (RFS) with a copper target and a flow of O 2 and H 2 S with various gas ratios. The authors showed that for x > 0.39, the compounds did not crystallize in the cubic structure of Cu 2 O and became amorphous. The lattice constant of cubic Cu 2 O 1−x S x evolved with the composition toward bigger values because of sulfur insertion. The variation was linear only up to x = 0.13 and did not follow the Vegard's law. Unfortunately, direct information about the sulfur oxidation state is missing: the oxysulfide nature of the compound remains unsubstantiated.

Zinc oxysulfide ZnO 1−x S x
Despite their electronegativity and size differences, sulfur atoms are able to replace the oxygen atoms of the wurtzite structure which progressively turns into the ZnS blende structure. It evidences another challenge of metal oxysulfide identification: they could be isostructural of metal sulfides or metal oxides.
Zinc oxysulfide was first reported as thin films grown by atomic layer deposition (ALD) in 1992 by Sanders et al. The oxygen and water traces in the gases were responsible for the oxygen in the resulting film. Since 2010, extensive characterization of ZnO 1−x S x thin films were reported, not only involving ALD (Bakke et al., 2012) but also pulsed-laser deposition (Deulkar et al., 2010), chemical spray pyrolysis (Polat et al., 2011a(Polat et al., ,b, 2012Thankalekshmi and Rastogi, 2012) or thioacetate-capped ZnO nanocrystals (Lee and Jeong, 2014). Because of the active research on bandgap engineering, zinc oxysulfide was envisaged as buffer layer in solar cells (Platzer-Björkman et al., 2006;Sinsermsuksakul et al., 2013). X-Ray photoemission spectroscopy (XPS) showed that the sulfur in these films is reduced and thus in agreement with the announced oxysulfide nature (Thankalekshmi and Rastogi, 2012;Lee and Jeong, 2014).

Molybdenum oxysulfides Mo x O y S z
In 1986, Inoue et al. crystallized two Mo x O y S z compounds while studying the MoS 2 :MoS 3 system (Inoue et al., 1986). The deep-bluish crystal of MoO 2.74 S 0.12 (otherwise written as Mo 4 O 10.96 S 0.48 ) was isostructural to γ-Mo 4 O 11 and exhibited charge density wave instabilities similar to these of quasi-2D materials. The similar properties of MoO 2.74 S 0.12 and γ-Mo 4 O 11 supported the hypothesis of a true oxysulfide compound. Also, reddish crystals of MoO 1.88 S 0.15 were obtained and presented structural and electronic similarities with monoclinic MoO 2 .
The decomposition of molybdenum oxodithiocarbamate as a single source precursor also enabled the formation of crystalline thin films (Olofinjana et al., 2010). Rutherford backscattering spectroscopy (RBS) indicated a pure phase. Unfortunately, the final product shared the XRD patterns of Mo 8 O 23 , Mo 9 O 26 , and Mo 2 S 3 but the structure was not fully solved.

Amorphous Titanium, Tungsten, and Molybdenum Oxysulfides
In this section are referenced the oxysulfides of three elements: titanium, tungsten, and molybdenum. In the 1990's, thin films of these oxysulfides were obtained and studied for their electrochemical properties.
In 1993, Tchangbedji et al. announced the formation of a hydrated amorphous phase of vanadium oxysulfide by reacting Na 2 S·9H 2 O and VOCl 2 (Tchangbedji et al., 1993). The first described formula for this compound was V 2 O 4 S·2H 2 O, but was adjusted to V 2 O 3 S·3H 2 O in latter studies (electron paramagnetic resonance and XANES at V K-edge demonstrated the presence of V IV species; Tchangbédji et al., 1994;Ouvrard et al., 1995). Water was believed to stabilize the compounds, as its evaporation was accompanied by the loss of the sulfur in the structure. Unfortunately, the authors did not provide enough convincing arguments to justify the oxysulfide nature and the purity of their phase without ambiguity. In particular, the absence of the IR and XANES at S K-edge spectra, which are discussed in the articles, is detrimental. Because of this lack of information, we did not focus on this phase.

Titanium
Titanium oxysulfides were obtained under the form of thin films to serve as positive electrode material for solid state batteries. Reported for the first time in 1989 by Meunier et al. (1989Meunier et al. ( , 1991 they were extensively characterized in the same group by X-ray photoemission spectroscopy (XPS) that was shown well-adapted for thin films characterization .
Titanium oxysulfides (TiO y S z ) of various compositions were obtained using RFS of hydrolyzed TiS 2 targets. The composition can be adjusted via the partial pressure of oxygen during the sputtering process. XPS showed that titanium oxysulfides thin films contain three titanium species (Ti IV as in TiO 2 , Ti IV as in TiS 2 and Ti in mixed environment) and three sulfur species (S −II of S 2− anions, S −I in disulfide S 2 2− ions and undefined S n 2− ions). For high oxygen contents (TiOS for instance), S VI species of sulfate ions attributed to surface species were also observed, although in a lesser extent due to mechanical erosion (Gonbeau et al., 1991;Dupin et al., 2001;Martinez et al., 2004;Lindic et al., 2005a) Besides, the presence of ordered domains, observed by TEM and XRD, revealed the existence of TiS 2 nanocrystals in the amorphous materials (Lindic et al., 2005b). Lithiated titanium oxysulfides thin films were recently obtained with RFS using LiTiS 2 targets (Dubois et al., 2017). Their characterization show similar properties than TiO y S z . Their capacities of around 85 µAh.cm −2 .µm −1 made them usable in a Li-ion cell. Aside these thin films, "sulfur-doped TiO 2 " can be obtained by reacting TiO 2 with thiourea or hexamethyldisilathiane, for instance. However, in this case, the products should not be named "oxysulfides, " because they only contain oxidized sulfur under the form of S IV and S VI species (Yang et al., 2012;Ramacharyulu et al., 2014;Smith et al., 2016).

Tungsten
Similarly to titanium oxysulfides, amorphous tungsten oxysulfides thin films with adjustable composition were obtained by RFS on WS 2 targets and mainly characterized by XPS (Martin et al., 1999). Along with the three species of sulfur described in the titanium section, three different species of tungsten (W VI as in WO 3 , W IV as in WS 2 , and W V in a mixed environment of O 2− , S 2− , and S 2 2− ) were observed Martinez et al., 2004). TEM and XRD showed the presence of nano-crystallites of WS 2 , but the polymorphs 3R-WS 2 and 2H-WS 2 could not be distinguished (Martin-Litas et al., 2002). The incorporation of lithium in these thin films and their electrochemical properties were studied (Martin et al., 1999;Martin-Litas et al., 2001). It revealed that 1.1 lithium atoms per formula can be incorporated, providing a capacity of 75 µA.cm −2 . XPS also demonstrated that the tungsten ions are reduced to W (0) and that sulfide ions participated to the redox process with irreversible behaviors (Martin-Litas et al., 2003).

Molybdenum
Abraham, Pasquariello et al. synthesized various MoO y S z amorphous compounds from the thermal decomposition of ammonium dithiomolybdate (NH 4 ) 2 MoO 2 S 2 (Abraham et al., 1989;Pasquariello et al., 1990). This precursor was obtained by bubbling H 2 S on ammonium paramolybdate [(NH 4 ) 6 Mo 7 O 24 ·4H 2 O] in an ammonia solution. Depending on the thermal treatment (temperature, number of steps), significant amounts of hydrogen and/or nitrogen could be found in the solids. Reacting a mixture of [(NH 4 ) 6 Mo 7 O 24 ·4H 2 O] and (NH 4 ) 2 MoS 4 also led to a solid precursor whose thermal decomposition yielded MoO y S z . Based on the electrochemical properties of these amorphous compounds, the authors suggested different structures for them, with different O/S ratios and involving both S 2 2− and S 2− anions (Abraham and Pasquariello, 1993). Infrared spectroscopy and XPS supported the presence of Mo-O and Mo-S bonds in the solid, but Mo-Mo bonding could not be evidenced.
The solution obtained by reflux of (NH 4 ) 2 Mo 2 S 12 in acetone dispersed in different aqueous electrolyte solutions led to original morphologies of amorphous molybdenum oxysulfides (water/acetone = 1/10 v/v; Afanasiev and Bezverkhy, 2003). For instance, tubular morphologies (with the following electrolyte: 10% KCl, 10% NH 4 SCN), hollow spheres [with 10% (NH 2 OH)H 2 SO 4 ] and fractal sponge-like solids (with 20% NH 4 SCN) were obtained. A solid was collected by evaporation of the solvent. EXAFS at Mo K-edge spectra showed one oxygen atom and four sulfur atoms in the first coordination shell of molybdenum atoms. XPS supported the hypothesis of mainly reduced sulfur, even if a broad peak in the region 167-171 eV indicated oxidized species. In the same group, Genuit et al. (2005) performed the condensation in acidic medium of MoO 2 S 2 2− to amorphous MoOS 2 . The addition of HCl in a (NH 4 ) 2 MoO 2 S 2 aqueous solution led to MoOS 2 .
Similarly to titanium and tungsten, RFS gave amorphous thin films of molybdenum oxysulfides (Schmidt et al., 1994(Schmidt et al., , 1995a. The target was a pellet of MoS 2 . Pure oxygen was flowed into the chamber to get oxygen-rich oxysulfides (MoO ∼1.3 S ∼1.9 ), but the traces of oxygen in the glovebox were originally sufficient to get MoO y S z thin films. For oxysulfides with a low content of oxygen (MoO ∼0.5 S ∼2.0 ), TEM showed ordered domains that are isostrucural of MoS 2 , based on electronic diffraction. XRD evidenced both MoS 2 and MoO 2 phases when the films were annealed under inert atmosphere. As shown by Buck for contaminated MoS 2 films, substitution of sulfur by oxygen atoms is likely to explain the changes in lattice parameters observed in the MoS 2 -like phase (Buck, 1991). Later, XPS analysis provided clues about the oxidation states of molybdenum and sulfur in MoO y S z films which strongly varied with the film composition (Levasseur et al., 1995;Schmidt et al., 1995b;Dupin et al., 2001). For y < 0.6 and z > 2 (oxygen-poor oxysulfides), Mo IV cations and S −II (as in MoS 2 ) were dominant. For y > 3 and z < 1 (oxygen-rich oxysulfides), only Mo VI in octahedral sites (as in MoO 3 ) was observed. For 0.6 < y < 3 and 1 < z < 2, Mo V was observed in addition to Mo IV and Mo VI and was likely surrounded by O −II (O 2− ) and S −I (S 2 2− ) species. Moreover, in MoO 0.6 S 1.9 thin films, extended X-ray absorption fine structure (EXAFS) at molybdenum K-edge also showed the presence of oxygen atoms in the coordination sphere of molybdenum atoms (Schmidt et al., 1995b).
During the same decade, useful XPS and IR references for molybdenum oxysulfides were established by Muijsers et al. (1995) and Weber et al. (1996) in the study of MoO 3 films sulfidation. The formation of oxysulfide intermediate species with their corresponding probable structures was detailed.

Applications of Bulk and Thin Film Metal Oxysulfides
In the 1980's, potential applications for doped Ln 2 O 2 S materials were identified and led to their use as lamps, lasers, scintillators, screens, etc. For example, they can be found in X-ray detectors used for tomography or medical imaging. They have also been used for oxygen storage. Often, Y 2 O 2 S, La 2 O 2 S, or Gd 2 O 2 S are used as the lattice and doped with one or several lanthanide ions to obtain the desirable luminescence features. The oxysulfide was compared with the corresponding oxide Sm 2 Ti 2 O 7 that has a higher bandgap. Also, electrochemical properties of transition metal oxysulfides were investigated for their use in lithium-ion batteries. A brief summary is given below on these applications.

Screens
Doped oxysulfides were primary employed in cathode ray tubes (CRTs) of television screens and later in computer monitors. In 1968, Royce patented a "family of new cathodoluminescent phosphors" by describing the potential use of doped Y 2 O 2 S and Gd 2 O 2 S (Royce, 1968). Lutetium and lanthanum were also envisaged as efficient matrixes for the doping ions (mainly Sm III or Eu III ).
Three classes of phosphors, respectively, associated to red, blue, and green, are necessary for a proper screen to emit the colors of the visible spectrum. Thanks to their good luminescence properties, lanthanide doping ions equip the main phosphors used for industrial applications . Red color is provided by Eu III : Y 2 O 3 :Eu, Y 2 O 2 S:Eu, YVO 4 :Eu, Y 2 (WO 4 ) 3 :Eu; blue emission is enabled by Eu II in compounds such as: Sr 5 (PO 4 ) 3 Cl:Eu, BaMgAl 11 O 7 :Eu, Sr 2 Al 6 O 11 :Eu; and green is emitted thanks to Tb III : CeMgAl 11 O 19 :Tb, (Ce,Gd)MgB 5 O 10 :Tb, (La,Ce)PO 4 :Tb, Y 2 SiO 5 :Tb, Y 3 Al 5 O 12 :Tb . However, for CRTs, blue and green are preferentially obtained with ZnS:Ag and ZnS:(Cu,Au), respectively.
In current computer monitors, the amount of europiumdoped yttrium oxysulfide Y 2 O 2 S:Eu (0.73% w/w for Eu, 13.4% w/w for Y) used for red emission has become large enough to implement and develop the rare-earths recovery (Resende and Morais, 2015).

Laser: Emission and Absorption Stimulated emission in lanthanum oxysulfide
The first study on metal oxysulfides as laser-emitting material was reported in the earliest years of the design of laser devices. "Laser" stands for light amplification by stimulated emission of radiation and is a general term for a device that emits light through a process of optical amplification based on the stimulated emission of electromagnetic radiation. It is characterized and differs from other light sources by the spatial and temporal coherences of the resulting light. Thus, there are countless applications of the laser devices. Two kinds of applications can be distinguished: information transfer (fiber-optic communication, length measurements, fingerprint detection, barcode scanner, thermometers, laser pointers, weapon guidance. . . ) and power transfer (cutting, welding, superficial fusion, marking materials. . . ).
In 1971, while the most famous laser crystal, namely YAG:Nd (neodymium-doped yttrium aluminum garnet), had already been extensively studied, Alves et al. (1971) carried out the first experiments dealing with an oxysulfide-based laser. They grew and studied millimetric La 2 O 2 S:Nd crystals with many defects, but also estimated the properties of crystals with less imperfections. Similarly to YAG:Nd, the stimulated emission takes place between the 4 F 3/2 and 4 I 11/2 energy levels of the Nd III ion in La 2 O 2 S:Nd with an emission wavelength of 1,075 nm (9,300 cm −1 ), while YAG:Nd emits at 1,064 nm (9,400 cm −1 ).
In 1990, Markushev et al. (1990) presented preliminary results on the stimulation emission kinetics at the temperature of liquid nitrogen for 1 mol% of neodymium. In 2012, the stimulated emission properties of La 2 O 2 S:Nd were studied with oxysulfide powders (Iparraguirre et al., 2012). In particular, Iparraguirre et al. (2012) estimated and experimentally investigated the influence of the doping ion concentration and pumping wavelengths on the different laser properties.

Laser absorption of samarium oxysulfide
The counterpart of laser emission is laser absorption. Because of the coherence of the emitted light, laser devices can be harmful for human skin or eyes. Protecting glasses or clothes are then required for safety issues. Absorption materials must display a low reflectivity and a good thermal stability because of local heating induced by the laser beam.
The research on absorption devices deals with materials that can absorb the 1,064 nm radiation of the widespread YAG:Nd 3+ laser. In particular, samarium-based compounds were found to be efficient absorption materials because of electronic transitions between the ground state 6 H 5/2 to the 6 F 9/2 excited state.
Undoped Sm 2 O 2 S was found to absorb a large proportion of the 1,064 nm laser radiation with a reflectivity of around 0.74% and it is stable up to 2,000 • C (Zhu et al., 2016). In comparison, SmBO 3 presents a reflectivity of 0.6% but endures a phase transition at 1,200 • C (He et al., 2009). Doping with erbium or thulium may also be an efficient way to slightly enhance the absorption properties of Sm 2 O 2 S (Sun et al., 2017).

Scintillators
A scintillator is a material that emits light when it is excited by an ionizing radiation (X-rays or gamma rays for example). Scintillators are mainly used in the field of medical imaging. Their role is to lower the dose of X-rays endured by a patient during an analysis. To enable a good absorption of the Xray beam, the requirements for a good scintillator phosphor is the presence of heavy atoms (cadmium, bismuth, lanthanides, tungsten for instance), a high material density (≥ 4 g.cm −3 ) and a high stability regarding the radiations. The photon must be converted into photons in the visible range (500-800 nm) with a good efficiency, a fast decay and a short afterglow. Moreover, mechanical strength, absence of toxicity, and chemical stability are desired features (Rossner and Grabmaier, 1991).
Thus, a scintillator is generally composed by a dense ceramic and converts X-rays in visible light. It is connected to photodiodes that convert the visible photons in electrons that form an image on a layer of amorphous silica. Considering their efficient absorption of X-rays, Y 2 O 2 S:Tb, La 2 O 2 S:Tb, Gd 2 O 2 S:Tb were considered to replace CaWO 4 , which was commonly used as scintillator (Brixner, 1987). Gd 2 O 2 S:Tb was finally chosen for its higher density and better absorption properties in comparison to the other lanthanides (Brixner, 1987). Later, Gd 2 O 2 S:Pr was shown to be an efficient scintillator by Rossner et al. They demonstrated that the main differences between the Pr III and the Tb III doping lie in the incident beam conversion efficiency (for a 40-80 keV X-ray beam, 8.5% for Gd 2 O 2 S:Pr, Ce, F, and 15% for Gd 2 O 2 S:Tb) and the luminescence lifetime of the doping ion (∼3 µs for Gd 2 O 2 S:Pr, Ce, F; 600 µs for Gd 2 O 2 S:Tb; Rossner and Grabmaier, 1991). Pr III shows a very rapid decay, cerium decreases the trap states and fluorine causes an important decrease of the afterglow. Gd 2 O 2 S:Eu was also studied. Its absorption and luminescence properties were competitive enough and it enables the emission of red photons (instead of green photons for Pr and Tb) which can be useful for compatibility issues with digital imaging systems (Michail et al., 2010).

Lithium-ion Batteries
Lithium intercalation and electrochemical properties of bulk metal oxysulfides were discussed because of possible oxidoreduction reactions with transition metals such as titanium or molybdenum. We already mentioned that titanium (Meunier et al., 1989;Lindic et al., 2005a;Dubois et al., 2017) or tungsten (Martin et al., 1999;Martin-Litas et al., 2001 thin films were studied as cathodes in solid-state lithium-ion batteries. As cathodes, molybdenum oxysulfide thin films were also developed (Abraham et al., 1989;Pasquariello et al., 1990;Gonbeau et al., 1991;Abraham and Pasquariello, 1993;Levasseur et al., 1995;Schmidt et al., 1995a;Yufit et al., 2003;Golodnitsky et al., 2006a,b) More recently, a TiO 2 @MoO y S z composite was investigated as anode material (Qiao et al., 2013). The external layer of molybdenum oxysulfide was supposed to enhance the conductivity of the hybrid material.

Conclusion
At the end of this section, we wanted to underline several crucial points: (i) Oxysulfide materials are mainly reached by chemical synthesis and much fewer compositions were obtained compared to monochalcogenide compounds. (ii) The lack of oxysulfide compositions is mainly based on the strong differences between oxygen and sulfur. Metals tend to preferentially bind to one compared to the other. (iii) Transition metals oxysulfides are particularly rare and their crystalline phases even more.
In this context, a periodic table showing the reported oxysulfide compounds is presented in Figure 5.

NANOSCALED TERNARY LANTHANIDE OXYSULFIDES Ln 2 O 2 S Introduction
Following the hype for nanotechnology in the twenty-first century, researchers have recently worked on producing metal oxysulfide materials as nanoparticles (Figure 6). This trend is justified by the applications that could emerge from nanomaterials, especially in the domain of biology and medicine.
In particular, nano-scale objects can cross biologic barriers and be metabolized by living beings. Also, because of their wide range of morphologies, compositions, and grafting, the nanoparticles can reach targeted zones using specific interactions to provide local information, deliver drugs at precise places or stimulate organs and tissues with an internal or external stimulus. Lanthanide oxysulfide nanomaterials present many advantages for imaging in biological medium. They have a good chemical and thermal stability. Their size and shape is highly tunable from very small crystals around 5 nm to micrometer spheres, rods, belts, tubes and so on. Moreover, the Ln 2 O 2 S crystalline phase bears many lanthanide/transition metal or lanthanide/lanthanide substitutions, which guarantees a generous variety of luminescent properties.

Examples of Oxysulfide Nanoparticles' Applications Upconversion
In the fields of therapy and in vivo imaging, using direct light composed of high energy photons, typically X-rays or gamma rays, leads to potential harmful effects for the patient. Organic dyes, radioisotopes and quantum dots are currently used in order to perform bioimaging. However, toxicity of radioactive isotopes, and quantum dots is problematic. Also, organic fluorophores and quantum dots (QD) are sometimes excited through ultraviolet (UV) irradiation that can lead to autofluorescence (excitation of natural targets, such as elastin, collagen...), photobleaching (destruction of the dye), and luminescence blinking.
Another indirect but efficient way to excite phosphors at low energy for bioimaging is infrared (IR) irradiation, taking advantage of the biological transparency windows: 750-950 nm  Frontiers in Chemistry | www.frontiersin.org (BW−1), 1,000-1,450 nm (BW-II), and 1,500-1,700 nm (BW-III). The main advantage is the high signal-to-noise ratio, because biological tissues (containing melanin, hemoglobin and water) absorb less light in these spectral ranges (Shi et al., 2016). Consequently, IR bioimaging does not result in parasitic fluorescence. Moreover, it causes low tissue damage and enables local irradiation along with high penetration depth.
Lanthanide-based upconverting phosphors are based (in the simplest case) on the combination of two absorbed low-energy photons in one of a higher energy, resulting for instance in the absorption of IR wavelengths and emission of visible light (Auzel, 2004). This way, many advantages are conferred to the imaging system (Ajithkumar et al., 2013): the chemical stability and low toxicity of rare-earth compounds, the absence of photobleaching, the low and easy available required energy.
Oxysulfide nanomaterials based on the upconverting properties of lanthanide dopants have been studied as potential upconverting phosphors for biomedical imaging. Ytterbium and erbium co-doped materials are being investigated in detail, but other dopants, such as holmium and thulium have also been reported for upconverting materials.

Persistent luminescence
The phenomenon of persistent luminescence is the emission of light by a material after excitation has stopped. It must be distinguished from fluorescence and phosphorescence. Its mechanism is complex and still debated (Jain et al., 2016). In persistent luminescence, the origin of the extended emission in an insulator or semi-conductor is the entrapment of electrons or holes that are progressively released (Leverenz, 1949). Either an electron is trapped in an energy level near the conduction band or a hole is trapped in an energy level near the valence band.
The traps can be point defects with intrinsic defects of the lattice such as vacancies, interstitial defects, antisite defects, or extrinsic defects when doping ions substitute lattice atoms or occupy interstitial sites. Extended defects (dislocations, surface, or grain boundaries) of the lattice can also play the role of traps.
Oxysulfide materials containing titanium and europium have been developed for persistent luminescence. Here, the doping ions substitute the rare-earth of the matrix and correspond to extrinsic defects. Y 2 O 2 S:Ti in 2005 was the first example (Zhang et al., 2005b), but numerous articles focused on the promising properties of Ln 2 O 2 S:Eu 3+ , Mg 2+ , Ti 4+ (Ln = Gd, Y) which will be named Ln 2 O 2 S:Eu, Mg, Ti for simplification (Mao et al., 2008;Li et al., 2010;Cui et al., 2013aCui et al., , 2014aLiu et al., 2014a).

Magnetic probes
Because of their remaining 4f electrons, most of the lanthanide ions present magnetic properties. Lanthanide oxysulfides were found to be paramagnetic in a large range of temperatures, and their magnetic properties at low temperatures were extensively studied (Ballestracci et al., 1968;Quezel et al., 1970;Biondo et al., 2014).
Lanthanides can exhibit high magnetic susceptibility, which is major interest for chemicals that can be injected in a living organism. For instance, Gd III complexes are used as positive contrast agents in magnetic resonance imaging (MRI) due to the 4f 7 electronic configuration of the ion (µ = 7.94 µ B ). The role of a contrast agent is to enhance the MRI signal by locally perturbing the magnetic field. The spin relaxation time of Gd III is long enough to optimize the dipole-dipole interactions of electron and protons (biological tissues, water) in the neighborhood of the contrast agent. The MRI signal is then enhanced by the acceleration of the spin relaxation of the protons caused by these interactions. Gadolinium ions in molecular complexes are toxic because of polarizing effects and competition with calcium. Special hydrosoluble complexes were then developed to prevent the toxicity of Gd III (Tóth et al., 2002).
An alternative to lanthanide complexes is lanthanide nanoparticles. A better detection occurs as the consequence of the concentration of several thousand atoms in a little volume. Iron oxide nanoparticles have been widely studied and used as negative contrast agents, but many artifacts were observed on the resulting images (Bulte and Kraitchman, 2004). Gd 2 O 3 nanoparticles were found to have a similar or better relaxivity than gadolinium complexes, without the drawbacks of iron oxides. They were then chosen for the precise visualization of locally injected cells (Engström et al., 2006;Petoral et al., 2009).
With doping ions, gadolinium oxide nanoparticles were then applied for bimodal imaging (MRI and luminescence) (Kryza et al., 2011). Because of their very good luminescence properties, similar results are expected for oxysulfide Gd 2 O 2 S nanoparticles. Bimodal agents are especially useful to get various information of the environment of the nanoparticles from the luminescence properties (wavelength, lifetime, and so on) in short times coupled with long term data and precise localization with magnetic resonance imaging (Cherry, 2006). Ajithkumar et al. (2013) demonstrated the possibility of performing multimodal bioimaging using oxysulfide material choosing the Gd 2 O 2 S:Yb, Er phosphor. Besides, Gd 2 O 2 S:Eu micronic particles were used as a colloidal solution for X-ray Luminescence Computed Tomograghy (XLCT), a technique that could be applied in vivo (Pratx et al., 2010a,b). Drug delivery can also be tracked in vivo. Gd 2 O 2 S:Tb nanoparticles coated with SiO 2 were employed as radioluminescent markers to evaluate the release of doxorubicin as a function of pH, using X-ray Excited Optical Luminescence (XEOL) (Chen et al., 2013).

Catalysis
Recently, sub-micronic powder of Sm 2 Ti 2 O 5 S 2 was used as a stable photocatalyst for water oxidation and reduction under visible-light irradiation, and this was later further extended to other Ln 2 Ti 2 O 5 S 2 (Ln = Pr, Nd, Gd, Tb, Dy, Ho, and Er) phases (Ishikawa et al., 2002(Ishikawa et al., , 2004. Moreover, because the majority of lanthanides are often restricted to the +III oxidation state, catalysis based on oxido-reduction reactions is not the preferential application of oxysulfide materials. Nevertheless, cerium (Ce III and Ce IV ) and europium (Eu II and Eu III ) are notable exceptions. In particular, Ce 2 O 2 S nanoparticles on carbon was tested for oxygen reduction reaction (ORR) . Also, Eu 2 O 2 S nanoparticles showed catalytic activity for the water-gas shift reaction (reaction of CO with water that yields CO 2 and H 2 ) (Tan et al., 2016). They can also act as a peroxidase mimic for the catalytic oxidation of 3,3 ′ ,5,5 ′tetramethylbenzidine (TMB) (Ghosh et al., 2016).

Synthetic Strategies for Lanthanide Oxysulfide Nanoparticles
General pathways toward Ln 2 O 2 S nanoparticles Several strategies can be employed to yield oxysulfides nanoparticles. Historically, bulk oxysulfides were formed by partial sulfidation of oxides, oxidation of sulfides or reduction of sulfates (Figure 7). However, solid-gas or solid-solid reactions at high temperatures inevitably lead to sintering and large particles. This should be avoided to control the growth of nanoparticles. Moreover, avoiding sulfates is challenging: their formation is thermodynamically favored.
Four major strategies are employed to yield Ln 2 O 2 S (bulk and nanoparticles). The two first methods are the sulfidation of an oxygenated phase such as an oxide or a hydroxide (Figure 7, pathway A) and the oxidation of sulfides (Figure 7, pathway B). In the latter case, the term "oxidation" names a substitution between sulfur and oxygen and does not imply oxido-reduction processes. This process is challenging: the partial oxygenation of sulfides is hard to control because sulfates are easily formed. To the best of our knowledge, only bulk materials were synthesized this way.
The reduction of sulfates and oxysulfates is also possible (Figure 7, pathway C). It is generally excluded for the formation of nanoparticles as it demands high temperatures (≥800 • C). Finally, another way to achieve the synthesis of metal oxysulfides is the co-insertion of oxygen and sulfur. Decompositions of organic precursors containing oxygen or sulfur are especially helpful for this method (Figure 7, pathway D). For syntheses in which oxygen rate has to be finely controlled, inert atmosphere assured by N 2 or argon is mandatory.
Since 20 years, a broad spectrum of techniques has been developed to yield Ln 2 O 2 S nanoparticles, which remains by far the center of the oxysulfide research. Here, we chose to classify them in three groups mainly depending on the reaction medium: water, organic solvent, and others. As we focused our study on the synthesis of nanomaterials, we excluded the works dealing with particles which were systematically sub-micronic or micronic (>700-800 nm).

Typical oxygen and sulfur sources in Ln 2 O 2 S nanoparticles syntheses
Oxygen source The oxygen source for the formation of Ln 2 O 2 S nanoparticles highly depends on the synthetic route (Figure 7).
Commonly, in the water-based syntheses, oxygen is brought by hydroxide ions with the precipitation of an intermediate oxygenated phase in basic medium. Oxygen insertion in sulfides SCHEME 1 | Sulfur sources typically used for the sulfidation processes leading to oxysulfides (mp, melting point; bp, boiling point; Sol., solubility). The first neighbor of sulfur in the molecule is indicated.
Ln 2 S 3 has never been performed for nanoparticles, to the best of our knowledge. Molecular precursors such as lanthanide formate or lanthanide acetylacetonate contain enough oxygen for the targeted composition. In organic medium, the use of ketones as ligands enables the formation of in situ water when an amine is present. The thermal decomposition of single-source precursors with sulfide ligands can be performed in air or pure dioxygen to give Ln 2 O 2 S nanoparticles. In the case of reduction of sulfates and oxysulfates, no additional source of oxygen is required.
Sulfur sources (Scheme 1). In water, sulfidation is mainly carried out by solid-gas reaction with H 2 S or in situ formed CS 2 using elemental sulfur heated in graphite or in presence of carbon. Nevertheless, a significant amount of syntheses also use sulfur sources soluble in water, such as thiourea or thioacetamide that initiate the sulfidation process. Elemental sulfur can also be used in organic medium especially dissolved in amines. Recently, substitution of oxygen by sulfur was carried out by ammonium sulfide and hexamethyldisilathiane (HMDTS).

Exotic Syntheses
Classical nanoparticles syntheses consist in heating hydrophobic or water-soluble inorganic precursors in aqueous or organic media, possibly sealed and/or pressurized and often followed by a thermal treatment which helps sulfidation and/or crystallization. In marge of these techniques, unconventional synthetic methods can be found. They involve unusual solvents, like molten salts, or are performed in uncommon conditions (electrospinning, combustion, and so on). This section describes such syntheses.

Boron-Sulfur Method
In 2008, Huang et al. adapted the boron-sulfur method, originally destined to the synthesis of sulfides, to the synthesis of La 2 O 2 S and Nd 2 O 2 S (Huang et al., 2008). In this synthesis, nanowires of the lanthanide hydroxide Ln(OH) 3 (formed by reaction between Ln(NO 3 ) 3 and KOH) are directly heated in presence of boron and elemental sulfur S 8 placed in a neighboring crucible. The driving force of the reaction is the strong affinity of boron with oxygen, which leads to the formation of B 2 O 3 as a by-product.
When the reaction is maintained for 24 h at 400 • C, LnS 2 nanowires are obtained. Using shorter reactions times (500 • C, 10 min), sulfidation of the wire is partial and Ln 2 O 2 S can be obtained ( Figure 8A).
This solid-state reaction preserves the shape of the precursor. Also, it is one of the rare techniques that enable the formation of Ln 2 O 2 S 2 nanomaterials using in some conditions an excess amount of sulfur compared with the targeted stoichiometry to ensure complete reactions. Nevertheless, only a small quantity of reactants were loaded in the crucible, leading to <15 mg of product per reaction. Also, the remaining species (B 2 O 3 , sulfur in excess) were washed with toxic CS 2 .

Combustion Synthesis
In order to get a swift synthesis, Hakmeh et al. (2015) developed a combustion synthesis by mixing lanthanide nitrates [La(NO 3 ) 3 , Er(NO 3 ) 3 and Yb(NO 3 ) 3 ] with thioacetamide in ethanol. The precursors were rapidly inserted in a furnace at 500 • C. Two successive flames evidenced first the ignition of ethanol, then the exothermic decomposition of the organic compounds, leading to an increase of the temperature and eventually to the formation of particles. A post-treatment at high temperature was also necessary (H 2 S in N 2 , 2 h, 1,000 • C) and resulted in large particles with a typical size around 300-500 nm ( Figure 8B).

N,S Dual Doped Carbon Supported Ce 2 O 2 S
Recently, an original catalyst for oxygen reduction reaction (ORR) was obtained by using the thermal decomposition of a vegetal, which provides the carbon support for the inorganic catalyst . Cerium nitrate [Ce(NO 3 ) 3 ] was dissolved in water along with thiourea and then pomelo skins were added to the solution in order to form a gel. After drying, the gel was annealed at 900-950 • C for 2 h to get Ce 2 O 2 S supported on carbon doped by nitrogen and sulfur. When the reaction temperature was set to 850 or 1,000 • C, the reaction led to the formation of CeO 2 . The TEM observation of the catalyst SCHEME 2 | Examples of surfactants employed by Wang et al. to synthesize Y 2 O 2 S:Eu in molten NaCl .
shows 50-100 nm crystals of Ce 2 O 2 S disseminated on the surface of the samples (Figure 8C). The porous structure, inherited from the pomelo precursor and the oxygen vacancies evidenced by the authors make this material suitable for the ORR.

Emulsion Liquid Membrane System (ELM)
Emulsion Liquid Membrane System (ELM) employs a waterin-oil-in-water (W/O/W) double emulsion. Originally, ELM was applied to separate metals. Here, the double emulsion is used for the formation of doped yttrium and gadolinium oxalates. These intermediates are converted to oxysulfides, Y 2 O 2 S:Yb, Er, and Gd 2 O 2 S:Eu, by a solid-state reaction with sulfur vapor (Hirai et al., 2002;Hirai and Orikoshi, 2004). Typically, a first emulsion is obtained by mechanical agitation of an organic phase containing kerosene with bis(1,1,3,3tetramethylbutyl)phosphinic acid (DTMBPA) (or 2-methyl-2-ethylheptanoïc acid, VA-10) as extractant and sorbitan sesquioleate as surfactant and an aqueous phase containing oxalic acid. This emulsion is then added to the external water phase which contains the metal ions (chloride or nitrates) and the double emulsion is produced by mechanical stirring. The oxalate compounds are thus produced at ambient temperature, and the system is demulsified using ethylene glycol. Oxysulfides nanoparticles of 50-100 nm are then obtained by annealing the powders at 600-1,000 • C in sulfur vapor generated at 200 • C by elemental sulfur and carried by a N 2 flow ( Figure 8D).

Synthesis in Molten Sodium Chloride
The synthesis in molten salts is an emerging technique which consists in the use of one or several salts as solvents for an inorganic reaction. An eutectic mixture can even be used to benefit from a lower melting point. Molten salts are typically suitable for reaction temperatures between 300 and 1,000 • C, which enable the formation of nanoparticles while avoiding their sintering (Portehault et al., 2011;Gouget et al., 2017). After cooling, the particles are obtained in a matrix composed by the salts that are washed with water or alcohols.
Molten sodium chloride (melting point: 801 • C) was chosen for the one-pot synthesis of Y 2 O 2 S:Eu. Y(NO 3 ) 3 , Eu(NO 3 ) 3 and NaOH were mixed and stirred before the addition of NaCl, S 8 and a surfactant . After grinding, the mixture was heated to 850 • C in a CO atmosphere for 4 h, and then cooled and washed.
Depending on the surfactant, the particles were either sub-micrometric or nanoscaled, but the morphology was quite irregular and the size polydisperse in all cases. For instance, polyoxyethylene (9) nonylphenyl ether (Scheme 2) gave 150-250 nm particles while sodium dodecylbenzenesulfonate (Scheme 2) gave 0.5-1.5 µm particles.

Composite-Hydroxide-Mediated Method
The composite-hydroxide-mediated method is also a synthesis in molten salts, but with hydroxides. Thirumalai et al. (2011a) adapted this method to the synthesis of Eu-doped yttrium oxysulfide by heating concentrated yttrium acetate Y(CH 3 COO) 3 in an eutectic mixture of NaOH and KOH in an autoclave. As the eutectic of the mixture is 165 • C, the autoclave was heated at 200 • C to yield Y(OH) 3 nanobelts (48 h) and nanorods (24 h) (Figure 9). Europium and S 8 were then mixed with the Y(OH) 3 nanomaterial at 70-80 • C and underwent an undescribed sulfidation process. In any case, the product was then annealed at 600 • C for 2 h in an (Ar or N 2 )/sulfur atmosphere to form Y 2 O 2 S:Eu. Interestingly, the final product retained the morphology of the Y(OH) 3 precursor. On the other hand, the step where the product was sulfidated was particularly unclear here, as three sulfidation processes are mentioned.

Electrospinning
Electrospinning is based on the application of a high potential difference between a polymer solution or a polymer melt and a collector. The electrical field creates charged threads that can be assembled depending on the experimental parameters such as tension, temperature, relative humidity (RH), concentration of the precursors, viscosity, distance between capillary screen and collection screen, etc.
Lanthanide nitrates Y(NO 3 ) 3 , Yb(NO 3 ) 3 , and Er(NO 3 ) 3 and polyvinyl pyrrolidone (PVP) were dissolved in DMF and stirred 8 h (Han et al., 2015a). Fibers were produced by electrospinning. They were annealed twice: (i) at 700 • C for 8 h under air to get Y 2 O 3 :Tb, Er fibers and (ii) at 800 • C for 4 h in a CS 2 atmosphere (obtained by heating S 8 in presence of carbon) to yield Y 2 O 2 S:Yb, Er hollow nanofibers ( Figure 8E). The same strategy was used to yield Y 2 O 2 S:Er hollow nanofibers (Han et al., 2015b). With slightly different electrospinning parameters, full nanofibers of Y 2 O 2 S:Yb, Er with a diameter comprised between 80 and 140 nm were obtained and studied by Lu et al. (2015).

Anodic Aluminum Oxide Template
In 2013, Cui et al. (2013a,b) elaborated a synthesis for doped oxysulfide nanoarrays using an anodic aluminum oxide template SCHEME 3 | Urea decomposition in water. Isocyanic acid is slowly obtained by urea thermolysis and ammoniac release. A second ammoniac molecule is released by hydrolysis which gives aqueous carbon dioxide "H 2 CO 3 ." (AAO). A nitrate solution obtained by dissolution of Y 2 O 3 , Eu 2 O 3 , and Mg(OH) 2 ·4MgCO 3 ·2H 2 O in hot HNO 3 (65%) was diluted by ethanol. Titanium doping was then obtained by adding the reaction product of Ti(OBu) 4 with acetylacetone. The pH was adjusted to 1 with HNO 3 . The sol was eventually obtained by evaporation at 80-90 • C. The AAO template was dipped in the sol, dried, calcined at 600 • C for 2 h and etched by NaOH (2.0 M) to give Y 2 O 3 :Eu, Mg, Ti nanoarrays. The whole process involved numerous steps and the resulting nanoarrays had to be sulfurated to Y 2 O 2 S:Eu, Mg, Ti using S 8 in graphite at 850 • C sharp (Cui et al., 2013a). Lower and higher temperatures were indeed not adequate: they resulted, respectively, in uncomplete sulfidation or oxide formation. Besides, an optimal concentration of europium dopant for the luminescence properties was determined (6.5 mol% Eu vs. Y) (Cui et al., 2013b).

Water-Based Syntheses
In the following syntheses, the reaction medium is water. It is an available, green, and ideal solvent for the dissolution of numerous metallic precursors, especially nitrates and chlorides.
Water also brings two main advantages: first, the availability of lanthanide precursors, and especially nitrates (that can be prepared from oxides in HNO 3 ) and water-soluble sulfur sources (thioacetamide, thiourea, ammonium sulfide, sodium sulfide, and so on; see Scheme 1); second, the substantial knowledge on inorganic polymerization in water. So far, in more than 90% of the articles dealing with Ln 2 O 2 S nanoparticles, the desired feature of the material was luminescence. Luminescence is due to a controlled doping of the oxysulfide phase (Ln 1 2 O 2 S:Ln 2 , M 3 , M 4 ) that is achieved by co-precipitation of the main cation (Ln 1 ) with the cations that trigger the luminescence and influence its properties (Ln 2 , and possibly M 3 , M 4 ,. . . ).
Water is however limiting metal oxysulfide synthesis by its relatively low boiling point. Even hydrothermal syntheses with autoclaves do not provide enough energy to obtain crystalline oxysulfide nanoparticles. In general, syntheses lead to an intermediate nanoscaled phase (which sometimes already contains sulfur) that is subsequently fully converted in oxysulfide nanoparticles with a solid-gas sulfidation (Figure 7). This last step remains an important drawback. It requires relatively high temperatures for nanoparticles synthesis (typically between 600 and 1,100 • C) and a large excess of inert gas and sulfur which is often present under the active but toxic gaseous forms of H 2 S or CS 2 . Also, it can affect the morphology of the solid by sintering or degradation of the desired phase.
The high-temperature sulfidation step remains the most challenging process here, but can be useful for other features. For luminescence purposes, the energy provided during the thermal treatment gives better-crystallized nanoparticles that present better photoluminescence properties. Moreover, doping ions can be inserted during this step.

Gelatin-Templated Synthesis
Reported in 2008 by Liu et al., this synthesis stands out through the original use of gelatin and the way the oxysulfide phase is obtained .
First, the appropriate amounts of lanthanum, terbium, and europium nitrates obtained from dissolution of La 2 O 3 , Tb 4 O 7 , and Eu 2 O 3 in nitric acid are mixed and heated with gelatin at 80 • C in H 2 O. The obtained translucent gelatin sol turns into a gel at 0 • C. Small pieces of the gel are soaked into NH 3 ·H 2 O and La(OH) 3 :Eu, Tb precipitates inside the gel. Violent stirring can then turn the gel into sol again, and (NH 4 ) 2 (SO 4 ) is added in stoichiometric amount. After drying and annealing at 500 • C for 2 h in air, a powder of oxysulfate La 2 O 2 SO 4 :Eu, Tb nanoparticles is formed. The oxysulfate nanoparticles are then converted to oxysulfide nanoparticles by solid-gas reaction using H 2 as reducing gas (700-800 • C, 2 h).
The pathway of oxysulfate reduction is quite rare in the oxysulfide nanoparticles literature, as it often requires high temperatures and long reaction times. Here, the nanoparticles however keep a reasonable 50 nm diameter. On the other hand, this synthesis comprises a myriad of steps, generates two intermediary phases and requires two heat treatments above 500 • C.

Sol-gel Polymer Thermolysis
This strategy is based on the elaboration of an organic network in which the inorganic nanoparticles nucleate and grow in a controlled way. The network is then burnt to free the nanoparticles. It is analogous to the Pechini method used for oxide synthesis for which a tridimensional polyester network is elaborated by reaction of trisodium citrate and ethylene glycol for instance (Pechini, 1967). Dhanaraj et al. published in 2003 a first version of a sol-gel polymer thermolysis strategy to yield Y 2 O 2 S:Eu nanoparticles (Dhanaraj et al., 2003). Y(NO 3 ) 3 and Eu(NO 3 ) 3 were obtained from the corresponding oxides. Urea, formaldehyde and elemental sulfur were then added and the network was formed at 60 • C. By condensation of urea and formaldehyde along with water evaporation, a gel was obtained. After thermolysis at 500 • C in sulfidating atmosphere, Y 2 O 2 S:Eu nanoparticles were formed. Based on the XRD pattern, the product was not pure (small peaks of impurities). Despite the treatment at 500 • C, the nanoparticles were quite small (around 30-50 nm) but presented an unclear morphology and aggregation. The work of Dai et al. in 2008 on La 2 O 2 S:Eu which deals with the effects of Eu 3+ concentration on the photoluminescence is based on the same synthetic route (Dai et al., 2008).
One year later, Dhanaraj et al. published a second version of the protocol that led to hexagonal nanoplates with a size between 7 and 15 nm, tunable via the reactants concentrations (Dhanaraj et al., 2004). The thermolysis process was divided in two steps: first, the sol/network solid was heated at 500 • C for 2 h to get Y 2 O 3 :Eu nanoparticles, and was subsequently digested by a thiosulfate solution. After water evaporation, a second thermal treatment at 500 • C (1 h) burnt the mixture to yield Y 2 O 2 S:Eu nanoparticles. The authors did not obtain a pure product yet, based on XRD analysis, but this time they identified sodium polysulfides as side-products. Later, Thirumalai and Nakkiran reused this strategy, succeeded in washing the byproducts  and deeply investigated the nanoparticles: optical (Thirumalai et al., , 2008a and electronic properties (Thirumalai et al., 2008a) were discussed as well as the photo-assisted relaxation of surface states (Nakkiran et al., 2007).

Syntheses in Water at Atmospheric Pressure
Because of the attractiveness of luminescent water-dispersible nanoparticles, the pursuit of doped oxysulfide nanoparticles led to the publication and the refinement of synthetic strategies in water. However, the reported syntheses illustrate the complexity of obtaining oxysulfides at low temperatures in water: most often, the authors choose to precipitate an unsulfurated intermediary doped phase [Ln(OH) 3 , Ln(OH)(CO 3 ) for instance] that can be amorphous or not. Thus, the syntheses presented in this section are worthwhile for oxide-, hydroxide-, or hydroxycarbonatebased nanomaterials. The intermediate nanoparticles are then sulfidated, most often with a solid-gas or alternatively with a solid-solid reaction.
Interestingly, the conditions for lanthanide oxysulfide nanoparticles syntheses in water are majorly optimized on Gd 2 O 2 S and Y 2 O 2 S because of their well-known luminescent properties and also maybe for the relatively low price of the related precursors ( Table 2).

Urea-based syntheses
Decomposition of urea in water. Generally, the precipitation of the lanthanide salts is performed via the basification of the reaction medium. Thus, a significant amount of research has focused on the cheap, safe, highly available, and watersoluble urea. Urea is indeed known to decompose in ammonia [pKa(NH 4 + ,NH 3 ) = 9.25] and aqueous carbon dioxide which can carbonate aqueous lanthanide species (Scheme 3).
The concomitant release of ammonia and aqueous carbon dioxide is used in particular for the precipitation of lanthanide hydroxycarbonates Ln(OH)CO 3 that turned out to be a suitable precursor of lanthanide oxysulfide nanoparticles. In the absence of sulfur source, further decomposition of this intermediate leads instead to lanthanide oxide. This was demonstrated in the pioneering work of Matijević and Hsu (1987) in the context of the fabrication of well-calibrated lanthanide colloids.
Syntheses with urea in water. The first aqueous synthesis of oxysulfide nanoparticles was reported by Kawahara et al. (2006 ;  Table 3). Using yttrium and europium nitrates Y(NO 3 ) 3 and Eu(NO 3 ) 3 along with urea, an europium-doped hydroxide precursor Y(OH) 3 :Eu was obtained by heating the mixture possibly in the presence of a glycol (ethylene glycol, propylene glycol, or hexamethylene glycol). The isolated powder of Y(OH) 3 :Eu was then heated between 800 and 1,200 • C with Na 2 CO 3 and sulfur to create a sulfidating vapor and yield Y 2 O 2 S:Eu nanoparticles. XRD showed that the crystalline phase was pure Y 2 O 2 S. The obtained nanoparticles were facetted crystals of 100-300 nm length. Above 1,100 • C, sintering made the particles sub-micrometric (≥600 nm). Xing et al. (2009) then developed an inspiring but complex protocol to synthesize Y 2 O 2 S:Yb, Ho upconversion nanoparticles . A solution of lanthanide nitrates Y(NO 3 ) 3 , Yb(NO 3 ) 3 , and Ho(NO 3 ) 3 and a solution of urea were separately prepared. The latter solution was added to the first that had been pre-heated at 60 • C and the mixture was then heated at 82 • C. After cooling and aging during 48 h, a white amorphous precipitate [likely Y(OH)CO 3 ]  was dried and converted to Y 2 O 3 :Yb, Ho via calcination (600 • C, 1 h, air). Then, the oxide was sulfidated at 800 • C for 1 h with a sulfur vapor created by S 8 at 400 • C and conveyed by an argon flow. It enabled the formation of size-monodisperse and non-aggregated nanoparticles with an average diameter of ca 80 nm. The diameter could also be tuned by adjusting the reaction time (aging step). Several works are based on Xing's synthesis with slight modifications. Luo et al. added a small amount of oleic acid in the urea mixture and performed the sulfidation at only 600 • C to form the same Y 2 O 2 S:Yb, Ho nanoparticles (Luo et al., 2009). In the same group, Pang et al. (2010) reported additional reactions that coated the nanoparticles with functionalized silica using a derived Stöber process with polyvinylpyrrolidone (PVP), aqueous ammonia, teraethylorthosilicate (TEOS), and aminopropyltriethoxysilane (APTES) in a second step (Figure 10). Sulfidation of hydrated Ln(OH)CO 3 :Eu 3+ nanoparticles (Ln = Gd, Dy, Ho) was alternatively performed under a flow of H 2 S at 750 • C for 90 min followed by an annealing under Ar at 850 • C for 4 h (Verelst et al., 2010;Osseni, 2012) This constitutes the sole reported route to Dy 2 O 2 S and Ho 2 O 2 S nanoparticles, to the best of our knowledge.
Also based on Xing's work, Bakhtiari et al. (2015) later studied the effect of europium concentration on Y 2 O 2 S:Eu nanoparticles size and luminescence. Very recently, Tian et al. succeeded in forming upconverting core-shell nanoparticles Y 2 O 2 S:Er@Y 2 O 2 S:Yb,Tm by applying Xing's method twice to form the oxide-oxide compound Y 2 O 3 :Er@Y 2 O 3 :Yb,Tm as an intermediate . Solid-gas reaction with sulfur vapor at 800 • C finally provided the oxysulfide nanoparticles. After the shell formation, Y 2 O 3 :Er@Y 2 O 3 :Yb,Tm nanoparticles were well-separated ( Figure 11A). After sulfidation, the nanoparticles were aggregated because of sintering ( Figure 11B). Nevertheless, the shell prevented the quenching of the Er III luminescence and multicolor fluorescence was achieved thanks to Er III /Tm III co-doping ( Figure 11C).
Y 2 O 2 S:Eu, Mg, Ti nanoparticles were also synthesized for persistent luminescence applications by Ai et al. (2010a). Y(OH)CO 3 :Eu was obtained by heating a mixture of Y(NO 3 ) 3 , Eu(NO 3 ) 3 , and urea at 90 • C for 2 h. The final product is obtained by a two-step thermal treatment developed by Li et al. (2009), It involves first S 8 in graphite at 800 • C for 4 h, which creates in situ reactive CS 2 , and then solid-solid reaction with doping    (Osseni et al., 2011). After dissolution, the reactants were heated to 85 • C to form a doped hydroxycarbonate precursor Gd(OH)CO 3 ·H 2 O:Eu. After isolation and drying, a heat treatment in two steps was performed. First, sulfidation was performed by Ar/H 2 S at 750 • C for 90 min and then the nanoparticles were maintained at 850 • C for 4 h under argon atmosphere only. The final nanoparticles were crystalline and spherical. Diameter was tunable by varying the H 2 O/EtOH ratio and reaction time. Interestingly, two techniques of deposition of silica on the nanoparticles were presented. The shell was either formed of mesoporous silica using TEOS and cetyltrimethylammonium bromide (CTAB) or functionalized by a silica/APTMS shell using TEOS and 3-aminopropyltrimethoxysilane (APTMS). In particular, mesoporous silica was found to enhance the luminescence properties of the nanoparticles. Multimodal imaging was recently applied using these Gd 2 O 2 S:Eu nanoparticles (Santelli et al., 2018).
A slightly different strategy, close to the work of Xing et al. on yttrium, was adopted in 2013 by Yan et al. (2013a) for the formation of terbium-doped oxysulfide nanoparticles of gadolinium and yttrium. Tb(NO 3 ) 3 and Gd(NO 3 ) 3 were dissolved in water around 100 • C, and then urea was added. After filtration and drying, Gd(OH)CO 3 ·H 2 O:Tb was obtained. The sulfidation process was quite complex: the precursor is mixed with Na 2 CO 3 and sulfur but is also covered by a second mixture composed of Gd 2 O 3 , Na 2 CO 3 , and S 8 . The bottom layer was washed in hot water and filtrated after being fired at 900 • C for 1 h. The crystalline phases Gd 2 O 2 S or alternatively Y 2 O 2 S were pure (based on XRD) and the polydispersity of the diameter was significant (average diameter around 100-120 nm). Yan et al. also studied the role of the doping ions in the luminescence mechanism of Y 2 O 2 S:Tb, Er nanoparticles (Yan et al., 2013b). In 2016, Bagheri et al. fabricated a scintillator screen composed of Gd 2 O 2 S:Pr nanoparticles synthesized via a similar nitrate/urea reaction (Bagheri et al., 2016). However, the sulfidation process is a solid-solid reaction with S 8 at 900 • C for 1 h.
In 2014 Recently, Cichos et al. studied three different syntheses of europium-doped Gd 2 O 2 S nanoparticles starting from nitrates and urea: (i) heating water at around 100 • C for 2 h using an oil bath, (ii) heating a Teflon bottle at 100 • C for 24 h, and (iii) heating an autoclave at 120 • C for 12 h (see the autoclave section; Cichos et al., 2016). After reaction, the isolated solids were heated with an excess of sulfur under argon at 950 • C for 1 h to yield Gd 2 O 2 S:Eu particles. In case (i), the intermediary solid was amorphous but the particles were spherical and quite monodisperse in diameter ( Figure 13A). After sulfidation, crystalline Gd 2 O 2 S:Eu nanoparticles with a diameter close to 135 nm were obtained. The surface was rougher than the amorphous precursor's one. The Teflon bottle method [case (ii)] gave micrometric hydroxycarbonate Gd(OH)CO 3 particles ( Figure 13B) that were converted to Gd 2 O 2 S:Eu micrometric crystals and was thus not suitable for nanoparticles synthesis.

Other precipitation routes from aqueous solutions
Closely related to urea's precipitating method, an aqueous ammonia/ammonium hydrogenocarbonate precipitation of nitrates was reported by Tian et al. (2015). A NH 4 HCO 3 /NH 3 ·H 2 O solution was added dropwise to a nitrate solution including Y(NO 3 ) 3 , Yb(NO 3 ) 3 , and Er(NO 3 ) 3 . A white precipitate of Y(OH) x (CO 3 ) y :Yb, Er was obtained and dried. The Y 2 O 2 S:Yb, Er nanoparticles were obtained using sulfur vapor (S 8 heated at 400 • C) carried by N 2 at 900 • C for 1 h. The small but aggregated crystalline nanoparticles (Ø ≈ 30 nm) were phase-pure, based on XRD. Here, the use of ammonium hydrogenocarbonate and aqueous ammonia enabled the authors to carry out the reaction without heating whereas urea needed thermolysis.
Regarding upconverting oxysulfide nanoparticles, Fu et al. (2010) chose Na 2 CO 3 to form intermediate solids which were then sulfidated. After dissolution of Y(NO 3 ) 3 , Yb(NO 3 ) 3 , and Ho(NO 3 ) 3 , the nitrate solution was added in a 0.1 M solution of Na 2 CO 3 containing PEG 4000 as surfactant. A solid precipitated, was isolated and dried. It was heated at 600 • C to yield Y 2 O 3 :Yb, Ho. Then, the oxide was converted to oxysulfide using Xing's thermal treatment described in the previous section. Interestingly, Na 2 CO 3 enables the authors to work at ambient temperature in the first step whereas urea required thermolysis. However, two thermal treatments were necessary to reach the oxysulfide product. Moreover, an irregular faceted morphology and a significant polydispersity in size were found in the final sample.

Aqueous Reactions Under Autogenic Pressure
This section is dedicated synthesis in aqueous solution under pressure, in autoclave. We already mentioned the low boiling point of water as a strong limitation if we consider the temperatures commonly required for crystalline nanoparticles synthesis. Synthesis under pressure might be a way to overcome this limitation. Unfortunately, like the precipitation reactions at atmospheric pressure, the reported syntheses in hydrothermal conditions mainly focus on producing an intermediate solid that requires sulfidation in a second step (Table 4). Nevertheless, these syntheses expanded the range of available morphologies for the final oxysulfide nanoparticles.
In the late 2000's, Thirumalai et al. (2008bThirumalai et al. ( , 2009a reported the hydrothermal synthesis of Gd 2 O 2 S:Eu (Table 4, entries 1, 2). Starting with an amorphous precipitate (obtained by adjusting the pH of an aqueous solution of Gd(NO 3 ) 3 with NaOH), they obtained Gd(OH) 3 nanoscaled materials (hexagonal nanocrystals, nanotubes, nanobelts, . . . ) after the hydrothermal treatment. The influence of the pH of precipitation and the temperature and duration of the hydrothermal define the morphology of the material. After impregnation of the solid with Eu 3+ ions in a aqueous solution, sulfidation was performed using a CS 2 atmosphere generated by reaction of sulfur and carbon. The morphology of Gd(OH) 3 was retained in the final Gd 2 O 2 S:Eu nanopowder, with only slight size decreases. The nanomaterials are well-crystallized and the morphology is finely adjustable varying the reaction conditions. Unfortunately, an undescribed sulfidation process is performed before the annealing step. It is probably similar to the one mentioned before for the composite hydroxide method conducted by the same group (Thirumalai et al., 2011a). Moroever, an original study on the photo-induced impedance is presented. Interestingly, the morphology is retained also with other lanthanides, as similar results were obtained by Thirumalai et al. (2009b) on Y 2 O 2 S:Eu (Table 4, entry 3).
The oxysulfides nanoparticles obtained by hydrothermal syntheses were also extensively studied by Li, Ai, Liu et al. who obtained Y 2 O 2 S:Eu,Mg,Ti nanoparticles (Table 4, entries 4, 5, and 6;Li et al., 2009Li et al., , 2010Ai et al., 2010c). This combination of doping ions is typical for persistent luminescence. Aqueous ammonia NH 3 ·H 2 O was used as a base for precipitation of hydroxides. The authors then inserted the dopants by solid-solid reaction in the annealing step with Eu 2 O 3 , Mg(OH) 2 ·4Mg(CO 3 )·6H 2 O, and TiO 2 . Moreover, they noticed that using CS 2 formed in situ, rather than solid S 8 , is crucial to keep the morphology. With S 8 , the Y(OH) 3 nanotubes turned into hexagonal nanoparticles after the annealing step.
The group of Cui and Liu also put great efforts on the characterization of such nanoparticles (Table 4, entries 7 and 8; Cui et al., 2014b;Huang et al., 2014;Liu et al., 2014a,c). Soluble sources [Eu(NO 3 ) 3 , Mg(NO 3 ) 2 and Ti(OBu) 4 ] were employed as reactants rather than solids for doping. Thus, a moderated sulfidation annealing temperature (≤800 • C) was employed to yield Y 2 O 2 S:Eu, Mg, Ti nanotubes. The synthesis , the influence of earth-alkaline or metal M II ion (Huang et al., 2014;Liu et al., 2014b), the effect of the relative concentration of Mg II and Ti IV , and the Eu III concentration were separately studied . Later, Yuan et al. (2016) also reported mild conditions to synthesize composite Y 2 O 3 :Eu/Y 2 O 2 S:Eu nanoparticles starting from soluble nitrate precursors (Table 4, entry 12; Yuan et al., 2016). The particles were crystalline but presented an irregular morphology. They were incorporated in dye-sensitized solar cells that were fabricated by the group. An enhancement of the cell efficiency was measured thanks to the light scattering properties of the nanocomposite.
A rare example of Lu 2 O 2 S:Eu nanocrystals was reported in 2015 by Wang et al. (Table 4, entry 9). PVP K30 followed by a solution of thiourea in ethanol were added to lutetium nitrate dissolved in a mixture of water and ethylene glycol. Perfectly regular nanorods were obtained after a thermal treatment with a sulfidizing atmosphere. Here again, the sulfidizing step mechanism was not studied in detail.
In 2016, Cichos et al. tested an hydrothermal synthesis ( Table 4, entry 10) to yield doped Gd 2 (CO 3 ) 3 :Eu particles in comparison with reactions at atmospheric pressure (Cichos et al., 2016). The authors noticed that this method is rather not adapted to the synthesis of nanoparticles: several populations are obtained, including micrometric irregular crystals ( Figure 13C).
The synthesis reported by Rosticher et al. (2016) is a promising exception ( Table 4, entry 11). The crucial difference lies in the sulfidation method. An excess of water-soluble thioacetamide was incorporated before the hydrothermal heating after precipitation of amorphous Gd(OH) 3 :Eu, Mg, Ti with NaOH. This allowed incorporation of sulfur before the annealing step, which could conveniently be performed under inert atmosphere. Its role was only to improve the cristallinity and the luminescence performances of the powder.

Conclusion
The formation of oxysulfide nanoparticles in water encounters several limits. Because of the aqueous solvent, excess oxygen favors the formation of intermediary phases such as hydroxides, hydroxycarbonates, or oxides. Only an adequate sulfidation annealing step at high temperatures enables the formation of the oxysulfide nanoparticles. Nevertheless, it can affect the morphology of the nanoparticles with aggregation and sintering.
Moreover, the synthesis of the intermediary phases is also challenging. Precise reaction parameters have to be employed, with long optimization processes. In Figures 12, 13, we reminded for instance the works of Hernández-Adame et al. and Cichos et al. on the synthesis of doped gadolinium oxysulfide nanoparticles with urea. Not only the reaction temperature and time had a great effect on the final morphology of the intermediates: concentrations of the reactants, pre-heating temperatures, heating techniques are also crucial to obtain the desired product.
Working in organic medium then seems to be a suitable solution to overcome the excess available oxygen.  Thomson et al. (2011) Water traces can also react with the thioamide to form the corresponding amide and H 2 S similarly to reaction (b).

Syntheses in Organic Medium
The following section is dedicated to the reactions mainly performed in organic medium. Thanks to the availability of high boiling-point solvents, the temperature of the reaction medium can reach 200-300 • C much easier than in water. Moreover, the control of the nanoparticles size and morphology in organic solvents is easily attainable using surfactants.
In the case of lanthanide oxysulfide nanoparticles, we also note several benefits of organic medium for the stoichiometry: (i) the control of the oxygen concentration that is assured by the absence of excess reactive oxygen brought by water and reactions under inert atmosphere, (ii) the use of molecular sulfur sources with for instance the possibility of decomposing hydrophobic single-source precursors (typically, lanthanide complexes with sulfurcontaining ligands) or the dissolution and activation of elemental sulfur in primary amines, as they react to form reactive alkylammonium polysulfides which release in situ H 2 S (Scheme 4; Thomson et al., 2011).
In both cases, the amount of reactive anions can be set to the desired value by playing on the concentration and the nature of the reactants. In water-based reactions, an excess of water in the precipitation step was followed by an excess of sulfur during the annealing step. Thus, organic medium brings the possibility to finely control the stoichiometry of the anions and one could expect that it leads to different oxysulfide compositions apart from thermodynamics considerations.  (Zhao et al., 2006a). In a mixture of organic solvents and surfactants typical for colloidal synthesis (1octadecene, oleic acid and oleylamine), [Eu(phen)(ddtc) 3 ] (phen = 1,10-phenanthroline, ddtc = diethyldithiocarbamate; Scheme 5) was decomposed under air at 290 • C in 45 min, forming anistropic nanocrystals (15 × 1.7 nm (Flahaut et al., 1958; Figure 14). For the first time, the observation of self-assembled oxysulfide nanoplates to nanowires is made (Figures 14A-C). The nanoplates are piled one above each other, because of the hydrophobic interaction between the surface surfactant chains of oleic acid (oleylamine-metal bonds are weaker than oleic acid-metal bonds; Cheon et al., 2004). Interestingly, EuS (Eu II ) nanocrystals were obtained with the same synthesis but under inert atmosphere with oleylamine alone (which played the role of reducing agent; Zhao et al., 2006b). A more detailed study on the pyrolysis of the [Ln(phen)(ddtc) 3 ] precursor and the nanoparticles properties was also reported. A noticeable work using the same strategy was conducted by Tan et al. (2016). In comparison with europium, the decomposition of [La(phen)(ddtc) 3 ] and [Pr(phen)(ddtc) 3 ] only yielded LaS and PrS. From oxidation of the sulfides, oxysulfates nanoparticles of La 2 O 2 SO 4 and Pr 2 O 2 SO 4 were obtained. The nanoparticles of Eu 2 O 2 S, La 2 O 2 SO 4 , and Pr 2 O 2 SO 4 were then tested for the water-gas-shift reaction. Lin et al. (2016) obtained europium-and terbium-doped Gd 2 O 2 S and europium-doped Tb 2 O 2 S by decomposition of the same precursor. Although the morphology of the nanoparticles was not perfectly regular, and the crystallinity not optimal, an extensive luminescence study was performed and biologic tests (in vivo imaging, cell viability) were conducted. The latter required a coating with 3-aminopropyltriethoxysilane ( Figure 10) and grafting of methoxy-polyethyleneglycol and Alexa Fluor 660 (photostable red dye which emits photons in the wavelength range of 630-650 nm).

Decomposition of Sulfur-Containing Single-Source Precursors
In 2012, He et al. (2009) described a similar decomposition of a precursor formed in situ. The reaction yielded europium oxysulfide nanorods. In this synthesis, europium oleate, oleylamine, 1,10-phenanthroline, and dodecanethiol were heated at 320 • C under inert atmosphere before hot injection of diethylammonium diethyldithiocarbamate (dea-ddtc, Scheme 5) dissolved in oleylamine. Nanorods were isolated after 1 h of reaction. The oxygen source was not explicitly discussed, but it was likely the oleate ions in the europium-oleate complex. Even if dea-ddtc is the most probable sulfur source, the introduction of dodecanethiol was not discussed. Nevertheless, this report showed that forming the single-source precursor in situ was a viable strategy. The non-stoichiometric character of the Eu 2+x O 2 S nanoparticles was evidenced by the Eu/S ratio measured by EDS. Non-stoichiometry is attributed to Eu II in the solid and was already observed for the bulk phase in the 1960's by Ballestracci and Quezel who estimated that 1% of the europium atoms were divalent thanks to neutron diffraction and magnetic measurements (Ballestracci et al., 1968;Quezel et al., 1970). He et al. (2009) also described magnetic properties of europium oxysulfide nanoparticles and confirm the Eu II /Eu III ratio, even if the average composition Eu 2.11 O 2 S corresponds to about 15% of Eu II . Moreover, an electrophoretic deposition of the nanorods was proposed. Ghosh et al. reported another precursor to obtain Eu 2 O 2 S nanoparticles (Ghosh et al., 2016). According to the authors, La 2 O 2 S and Nd 2 O 2 S can also be prepared with a similar procedure. Synthesized from europium nitrate, triethylamine (Et 3 N) and 2-aminocyclopentene-1-dithiocarboxylic acid (Hacda), [(Et 3 NH) + (Eu(acda) 4 ) − ] was decomposed via three different methods (Scheme 6).
By heating the precursor in an OM/OA/ODE mixture, ultrathin nanoplates of Eu 2 O 2 S were obtained. However, the 0.3 nm reported thickness is quite surprising, as it would represent a single monolayer of the solid, and there is no correlated peak extinction in the corresponding XRD pattern. Using similar conditions with OM only led to rodlike nanoparticles (7 × 3 nm 2 ). Finally, hot injection of [(Et 3 NH) + (Eu(acda) 4 ) − ] and trioctylphosphine (TOP) led to polydisperse nanospheres with an average diameter of 13 nm. The catalytic activity of Eu 2 O 2 S, and especially its activity as a peroxidase mimic, was deeply investigated. Because Eu 2 O 2 S catalyzed the oxidation of 3,3 ′ ,5,5 ′ -tetramethylbenzidine (TMB) in presence of H 2 O 2 and neither La 2 O 2 S nor Nd 2 O 2 S succeeded in it, the authors concluded to a mechanism involving the Eu III /Eu II redox couple.

Syntheses With High Boiling-Point Organic Solvents at Atmospheric Pressure
Also colloidal synthesis in organic solvents have been used for years in the synthesis of metal and metal oxide nanoparticles, the first report for metal oxysulfides was published by Ding et al. (2011). Lanthanide acetylacetonate Ln(acac) 3 (1 equiv.), elemental sulfur (1 equiv.), and sodium acetylacetonate (1 equiv.) were added in an OM/OA/ODE mixture and heated for 45 min. at 310 • C under inert atmosphere after degassing under vacuum at 120 • C (Figure 15). Size-monodisperse hexagonal nanoplates of Ln 2 O 2 S were obtained. They were thin (a few monolayers) and 5-40 nm wide depending on the lanthanide. The composition of the powder showed a lack of sulfur (Na 0.4 La 1.6 O 2 S 0.6 ), which was attributed to terminal [Ln 2 O 2 ] 2+ layers. The crucial advantage of this method is its high versatility: La 2 O 2 S, Pr 2 O 2 S, Nd 2 O 2 S, Sm 2 O 2 S, Eu 2 O 2 S, Gd 2 O 2 S, Tb 2 O 2 S were prepared. The sodium ions, added in stoichiometric amounts, were proposed to help the crystallization and favor the oxysulfide formation. The hypothesis of the authors is that the close ionic radii of sodium [r(Na I (VII)) = 1.26 Å] and larger lanthanide ions [r(La III (VII)) = 1.24 Å to r(Tb III (VII)) = 1.12 Å] enables cation exchanges in the solid and favors the oxysulfide crystallization. Lithium ions were tested and were efficient for Y 2 O 2 S synthesis. In 2013, a more complete study (experimental study and calculations based on density functional theory) on the alkaline additives on the formation and morphology of the obtained nanocrystals also showed the possible use of potassium to synthesize oxysulfide nanoparticles (La 2 O 2 S, Eu 2 O 2 S, Gd 2 O 2 S, and Yb 2 O 2 S; Zhang et al., 2013). The hypothesis of alkali insertion in the crystal structure was recently disputed: the alkali would serve as a stabilizing species for the formation of a lamellar alkali-oleate phase (as observed with sodium) rather than as a doping ion in the Ln 2 O 2 S structure (Larquet et al., 2020). In 2017, Lei et al. investigated the roles of yttrium and sodium in the formation and growth of Gd 2 O 2 S, by using them separately or combined. They also demonstrated that a large excess of sulfur allows forming gadolinium oxysulfide nanoplates without adding sodium ions (Lei et al., 2017).
In 2017, Tan and Li announced the formation iron/sodium co-doped lanthanum oxysulfide nanoparticles (Na, La) 2 O 2 S:Fe (Tan and Li, 2017). Such doping with light transition metal is very rare due to ionic radii mismatch. Thus, according to the authors, only a limited amount of iron would have been able to substitute lanthanum, and surprisingly, no contraction of the lattice was observed despite the ionic radii difference [r(La III (VII)) = 1.24 Å; r(Fe III (VII)) ≈ 0.85 Å]. Even though such iron doping would be very interesting for catalytic features, it is quite unclear that iron was well-inserted in the La 2 O 2 S phase. In 2015, Jiang et al. employed Ding's synthesis and demonstrated the possible use of La 2 O 2 S:Eu nanoparticles as optical temperature sensors ("nanothermometer") (Jiang et al., 2015). Our group recently investigated the reactivity of Ln 2 O 2 S hexagonal nanoplates formed with a stoichiometric amount of sulfur and demonstrated different oxidation processes in bimetallic Gd 2(1−y) Ce 2y O 2 S nanoparticles. It enabled us to prove that highly unstable Ce 2 O 2 S nanoparticles can also be formed with the help of sodium ions, as long as it is isolated and stored under strict inert conditions (Larquet et al., 2017). The thermal stability of the nanoparticles was investigated under inert and oxidizing atmosphere, highlighting the possibility to remove the oleate surface ligands by a mild thermal treatment (Larquet et al., 2019a). Moreover, both magnetic properties (Larquet et al., 2019b) and optical ones (bandgap) (Larquet et al., 2019c) could be tuned as a function of the Gd:Ce ratio. Gu et al. managed to obtain yttrium, gadolinium, erbium, and ytterbium oxysulfide nanoplates using oleylamine as only solvent and H 2 S as sulfurating agent. Ln(acac) 3 and oleylamine were degassed at 120 • C and then heated at 280 • C for 1 h under a H 2 S/N 2 flow (20/80 v/v, 60 mL/min) to yield Ln 2 O 2 S nanoplates . Again, sodium ions were shown to help the crystallization of the nanoplates but were not necessary in this case. Y 2 O 2 S, Eu 2 O 2 S, Gd 2 O 2 S, Er 2 O 2 S, and Yb 2 O 2 S were prepared by this route.
The amine/thiol ratio influenced the morphology. When the TETA/DT ratio was 3:1 or 1:1, micronic structures were obtained. Interestingly, other precursors [La(NO 3 ) 3 , LaCl 3 , La 2 O 3 , and La(OH) 3 ] were not selective enough or did not completely react [impurities of La(OH) 3 or La 2 O 3 based on XRD]. Also, shorter reaction times with TETA/DT = 3:1 for the exhibited the rare La 10 OS 14 intermediary phase with LaOOH and La 2 (SO 4 ) 3 . Although the crystals are large and the selectivity can be improved, it is to the best of our knowledge the only occurrence of a promising protocol for nanoscaled Ln 10 OS 14 .

Solvothermal Syntheses in Autoclave
In 2000, Li et al. tested a direct and simple solvothermal sulfidation process for numerous lanthanide oxides. Ln x O y powders (Ln = Y, Sc, La, Pr, Nd, Eu, Sm, Gd, Ho, Er, Yb, or Lu) and S 8 were suspended in ethylenediamine and heated in autoclave at 150 • C for 8 h (Li et al., 2000). Aggregated and irregularly-shaped crystalline spherical nanoparticles of Pr 2 O 2 S, Eu 2 O 2 S, and Gd 2 O 2 S were obtained with this method (<50 nm). The authors suggested an anion-exchange mechanism between the O −II in the Ln x O y crystal and the S −II available in polyanions when S 8 reacts with TETA. They supposed next that oxysulfide nuclei could leave the surface of the oxide to grow apart.
For Ln = La, Nd, Sm, Ho, and Er, the conversion was incomplete (Ln 2 O 2 S and Ln x O y on the XRD pattern) and no change was observed with longer reaction times. Only the starting oxide was observed for Ln = Y, Sc, Yb, and Lu. In 2012, Gd 2 O 2 S:Eu and Gd 2 O 2 S:Er, Yb nanoplates were also obtained by Liu et al. in ethylenediamine using gadolinium nitrate and elemental sulfur . PVP (K29-32) or OM was added to a solution of the lanthanide nitrate in ethanol. The resulting solution was added dropwise into ethylenediamine and sulfur. The autoclave was heated at 220 • C for at least 4 h to form crystalline hexagonal nanoplates. With OM, subsequent aggregation in flower-like structures was observed. Separated nanocrystals of irregular shape were obtained with PVP. Y 2 O 2 S:Eu and Y 2 O 2 S:Er, Yb were obtained from yttrium acetate by the same group with PVP and a thermal treatment at 250 • C for 24 h . Various self-assemblies of the nanoparticles were formed, depending on the presence of PVP, sulfur concentration, and so on. Song et al. (2010) obtained Gd 2 O 2 S:Eu and Gd 2 O 2 S:Tb nanospheres from the solvothermal treatment of lanthanide nitrates in a mixture of ethanol and ethylene glycol, containing polyvinylpyrrolidone (PVP K30, M = 40,000 g/mol) and thiourea. The autoclave was heated at 200 • C for 24 h and the SCHEME 6 | Triethylammonium of tetra(2-aminocyclopentenedithiocarbamate) europium ([(Et 3 NH) + (Eu(acda) 4 ) − ]) decomposition in organic solvents by (Ghosh et al., 2016) isolated solid was then sulfidized in a N 2 /S atmosphere at 600-800 • C to form doped gadolinium oxysulfide nanoparticles. PVP is believed to be responsible for the spherical morphology, and polymer residues were evidenced on the rough surface of the nanoparticles. They presented a good crystallinity and a good monodispersity in diameter. Their size was tunable between 150 nm and 1.25 µm by varying the PVP content and the ethanol/ethylene glycol ratio. A similar strategy was used by Deng et al. to yield Y 2 O 2 S:Sm hollow nanospheres (Ø = 140-200 nm; Deng et al., 2012a). The thermal treatment was based on Li's work (Y 2 O 2 S nanoparticles hydrothermal synthesis, Table 4, entry 4; Li et al., 2009). The authors first proposed a mechanism involving H 2 S/CO 2 bubbles to explain the holes, but finally declared in a second paper on Y 2 O 2 S:Eu, Mg, Ti nanoparticles that NH 3 /CO 2 bubbles were more likely the templating agents (Deng et al., 2012b). Similarly to Song's spheres, the surface was rough and the nanoparticles seemed to be constituted with smaller units. Thirumalai et al. (2011b) have mainly focused their work on water-based syntheses, but also prepared various morphologies of Gd 2 O 2 S:Eu nanoparticles in oleylamine. GdCl 3 ·6H 2 O and EuCl 3 ·6H 2 O were introduced in hot oleylamine and various amounts of thioacetamide were added. The resulting solution was heated in an autoclave at 120-240 • C for 12-24 h. Flowerlike nanocrystals (≈10 nm), nanospheres (Ø = 5-10 nm) and nanorods of various lengths (Ø = 6 nm) were obtained depending on the reaction conditions and the thioacetamide amount. An excess of sulfur was proposed to be mandatory to ensure a high chemical potential, which promoted the formation of nanorods. Despite the good morphology control, the XRD patterns of the nanoparticles showed a poor crystallinity of the materials: only broad peaks were observed. It is intriguing because nanorods presented big crystal domains, and HRTEM images showed large and regular lattices.

TRANSITION METAL OXYSULFIDES NANOPARTICLES
Transition metal bulk oxysulfides are quite rare. Zinc, titanium, molybdenum, and tungsten oxysulfide were nevertheless obtained. Most of the time, they were obtained under the form of amorphous thin films or particles.
Because of the electronegativity and atomic number differences between the two anions, transition metals will preferentially bind to one of them (in the hard and soft acids and bases theory, oxygen is a hard base and sulfur a soft base). Also, keeping reduced sulfur id est avoiding sulfates or other oxidized sulfur species is highly difficult because of their good thermodynamic stability.

Challenging Synthesis, Tricky Characterization
In the previous section, we detailed numerous syntheses of Ln 2 O 2 S nanoparticles. It is an exception in the oxysulfide family, as it remains to the best of our knowledge the only structure for which monophasic crystalline nanoparticles could be formed. Today, a vast and promising land of metal oxysulfide nanoparticles, especially involving transition metals, must be explored. With such nanoparticles with transitions metals and chalcogens, new applications could emerge, such as heterogeneous catalysis, photocatalysis, battery materials, superconduction, and so on.
Several advantages are intrinsically brought by soft reaction conditions (compared with typical synthesis of bulk crystals) and the nanoscale. Mild temperatures and small grain size can unlock metastable structures. Also, diffusion processes are much faster over nanometric distances and lead to efficient substitution reactions with nanoscaled materials. It opens new synthetic strategies to transform preformed oxide, sulfide, or metal nanoparticles in oxysulfide nanoparticles.
However, synthesizing transition metal ternary oxysulfides is particularly challenging. The ionic radius difference between O −II (1.26 Å) and S −II (1.70 Å) associated with the variable affinities with the metal make the substitution reactions highly difficult. Energy input by heating is especially not recommended for nanoparticles synthesis because of excessive growth and sintering.
Despite these difficulties, transition metal oxysulfide nanoparticles were already prepared. A few examples will be detailed in the next section. The main issue consists of identifying FIGURE 15 | Ding's alkali-based synthesis of lanthanide oxysulfide and its derivatives. The number of equivalent ("eq" in the table) is the molar ratio between the reactant and metal. The term "hex." stands for "hexagonal" and "irreg." for "irregular." and evidencing the oxysulfide nature of the compound. Because excessive heating tends to stabilize sulfate or separate oxides and sulfides rather than crystallize an oxysulfide structure, the reported structures are mainly amorphous.
(EDS) are suitable techniques to evidence the presence of sulfur, but the determination of the nanoparticles' precise oxygen content remains a challenge. High resolution transmission electron microscopy and energy filtered transmission electron microscopy (EFTEM) constitute an elegant solution, but requires well-dispersed nanoparticles and will not provide accurate quantitative data. Moreover, it is hard to conclude about the precise localization of the atoms: are they in the whole particle or only at the surface (because of ligands for instance)?
The identification of the nature of the chemical bonds and oxidation states inside the material is a supplementary issue, yet this is required to differentiate oxysulfides from sulfates. It becomes highly problematic when the composition of a solid is unclear. Infrared and Raman spectroscopies are particularly appropriate for amorphous oxysulfide identification because M-O and M-S bonds generally present very distinguishable signatures. However, only qualitative analysis is possible. X-Ray photoemission spectroscopy (XPS) brings some clues but investigates only the very surface. X-ray absorption spectroscopy, such as XANES or EXAFS (at O K-edge, S K-edge, M K, L, or M-edge) is able to characterize the whole sample and gives precious information on the oxidation states and chemical bonds. However, surface and core cannot be distinguished and only average information is obtained, so that one should be very careful about hypothesis and interpretations. Furthermore, it must be noticed that the required energies for the different edges involves the use of different X-ray ranges (soft for oxygen, tender for sulfur, hard for the metal K-edge) and consequently the use of distinct beamlines. The analysis of the pair distribution function of the diffuse background of X-ray diffraction patterns (PDF) is expected to bring solutions as it can be applied to the analysis of amorphous compounds. Still, it remains a poorly studied technique in the field of nanoparticles analysis. Finally, we emphasized the fact that oxysulfide nanoparticles can be metastable or unstable phases. It reinforces the difficulty to store, transfer, manipulate, and characterize them (for instance in air-filled room atmosphere and devices or when heating upon irradiation by electron or X-ray beams).

Amorphous and Crystalline Cobalt Oxysulfide
In 2016, Nelson et al. reported the formation of cobalt oxysulfide CoO x S y hollow nanoparticles (Nelson et al., 2016). The strategy consisted in the substitution of oxygen anions by sulfur anions in cobalt oxide hollow nanoparticles, using ammonium sulfide dissolved in oleylamine at 100 • C ( Figure 17A). The sulfur content was adjustable via the nominal (NH 4 ) 2 S amount, with a saturation of the sulfur content at y ≈ 1.3. With low sulfur contents (y < 0.2), the particles keep the crystalline structure of CoO but with higher sulfur contents, the nanoparticles became amorphous ( Figure 17F). The hollow nanosphere morphology was preserved during the whole experiment (Figures 17B-E).
No direct proof of the oxidation state of sulfur is brought by the authors. Nevertheless, annealing the sulfur-rich nanoparticles led to the formation of cobalt sulfides (possibly in a mixture with CoO). It supported the presence of reduced sulfur in the nanoparticles.

Crystalline ZnO 1-x S x Nanoparticles
Crystalline zinc oxysulfide was obtained at the nanoscale. In 2009, Park et al. carried out the substitution of oxygen atoms in ZnO by sulfur using hexamethyldisilathiane (Scheme 1) and obtained ZnS crystalline hollow nanoparticles (Figure 18; Park et al., 2009). The driving force of the reaction with ZnO is the formation of very stable Si-O bonds.
During the substitution process, the authors were able to isolate ZnO@ZnS core-shell crystalline nanoparticles, which are composed by a core of ZnO and a shell of isostructural ZnS wurtzite structure.
The process was accompanied by the so-called "nanoscale Kirkendall effect, " which refers to the hollowing of the nanoparticles as a consequence of unbalanced diffusion rates (Wang et al., 2013). Because Zn II diffuse outwards faster than S −II inwards, the reaction finally led to a hollow ZnS structure. HRTEM and EFTEM also showed that the final ZnS nanoparticles were obtained through the formation of heteroepitaxial ZnO@ZnS intermediates that release the high interface energy by the diffusion of the core into the shell (Figure 19). The composition analyses of core-shell intermediates indeed showed that oxygen is not only localized in the core of the nanoparticle, but also in the shell. It suggested that the substitution process with hexamethyldisilathiane took place in the shell region where oxygen had migrated.
The reaction resulted in the formation of crystalline ZnO 1−x S x located in the shell. Furthermore, the authors were able to obtain pure hollow ZnO 1−x S x nanoparticles by thermally annealed the core-shell intermediates. Interestingly, the diffusion processes spontaneously occurred without sulfurating reagents and led to hollow ZnO 1−x S x alloys.
The research on zinc oxysulfide nanoparticles has grown in the last years. Pandey et al. (2013Pandey et al. ( , 2014 managed to obtain the whole composition range (0 ≤ x ≤ 1) of nanoparticles. They obtained ZnO 1−x S x crystalline nanoparticles by a solution combustion method. Zn(acetate) 2 and thiourea were incorporated in a mixture of ethanol and ethylene glycol (4:1) and then placed in a hot furnace (350 • C) for 2 h. They showed that the bandgap varied with the sulfur content and that ZnO 1−x S x nanoparticles can photocatalyze the degradation of methyl orange. In 2017, Zhang et al. underlined the importance of doped zinc oxide to understand the intriguing ferromagnetic properties of certain d 0 components and used the same combustion method than Pandey . As a consequence of oxidation by air, they measured a significant amount of sulfate groups at the surface of their nanoparticles using XPS. Abdullah et al. (2017) synthesized ZnO 1−x S x nanoparticles using zinc(II) acetate and thioacetamide for the photocatalysis of the hydrogen evolution reaction. Gultom et al. (2017) in the same group also showed that nickel-doped ZnO 1−x S x nanoparticles were suitable for hydrogen production.

Other Proposed Transition Metal Oxysulfides
Crystalline nano-aggregates of cobalt nickel oxysulfides (CoNi)O y S z were claimed by Liu (2013). However, the author noticed that sulfur only existed in oxidized species (S IV , S VI ) in the material using XPS. By definition, it cannot be called "oxysulfide" but should rather be named "oxysulfate." In 2017, Liu et al. reported "fullerene-like oxysulfide hollow nanospheres" . The name is also abusively employed in this case, as the authors demonstrated that their nanoparticles are composed of crystalline nickel sulfide mixed with amorphous nickel oxide.

GENERAL CONCLUSION
The oxysulfide family is full of surprises. Most of its members are strictly synthetic because of size and electronegativity differences of oxygen and sulfur. It explains why these compounds were obtained as pure materials and identified relatively late in the history of Chemistry. However, since the 1950's, many compositions have been synthesized, from the simplest ternary compounds to oxysulfide materials containing five or more different atoms.
Among this family, Ln 2 O 2 S must be pointed out. It was the first discovered phase, and represents one of the simplest oxysulfide compositions. It is nearly the only one for which the scientific community proposed various applications especially in the domain of imaging. Consequently, it led to the development of numerous synthetic approaches for Ln 2 O 2 S nanoparticles. Size, morphology, composition, and reactivity of Ln 2 O 2 S nanoparticles are more and more controlled and understood. However, there is still an open door for the study of novel Ln 2 O 2 S nanoparticles. Surprisingly, nanoscaled Ln 2 O 2 S with heavy lanthanides remain hard to obtain. To the best of our knowledge, Tm 2 O 2 S nanoparticles were never prepared. Also, most of the articles focus on the luminescence applications. Yet, in the domain of catalysis for instance, europium, cerium, or ytterbium redox properties have only been poorly explored in such compounds. Lastly, at this point there is no general method to control the size and shape of these nanoparticles on a broad range: only slight adjustments are proposed so far. Such control would enable understanding the influence of the nanoscale on the electric, magnetic, and optical properties. In particular, controlling the nanoparticles thickness could allow to identify the transition between a regime of indirect bandgap (as in the bulk) to the direct one.
Regarding transition metals, their integration in quaternary oxysulfides at bulk scale is successful. Recently, lanthanidefree compositions were obtained. Still, crystalline ternary oxysulfides are very rare, and the characterization of the amorphous products is still incomplete, even if molybdenum or titanium oxysulfide thin films for instance were explored in many studies. One of the challenges is to identify preparation route that lead to crystalline compounds that are easier analyzed by structural techniques. The synthesis of nanoscaled materials with accelerated diffusion processes and possible metastable phases, the developments of characterization techniques and the promising field of applications of transition metals oxysulfides should start an unprecedented area of novel oxysulfide syntheses, identifications and applications.