*Article* **SmS**/**EuS**/**SmS Tri-Layer Thin Films: The Role of Di**ff**usion in the Pressure Triggered Semiconductor-Metal Transition**

#### **Andreas Sousanis, Dirk Poelman and Philippe F. Smet \***

Lumilab, Department of Solid State Sciences, Ghent University, Krijgslaan 281/S1, 9000 Gent, Belgium; andreas.sousanis@ugent.be (A.S.); dirk.poelman@ugent.be (D.P.)

**\*** Correspondence: philippe.smet@ugent.be

Received: 16 September 2019; Accepted: 20 October 2019; Published: 24 October 2019

**Abstract:** While SmS thin films show an irreversible semiconductor-metal transition upon application of pressure, the switching characteristics can be modified by alloying with other elements, such as europium. This manuscript reports on the resistance response of tri-layer SmS/EuS/SmS thin films upon applying pressure and on the correlation between the resistance response and the interdiffusion between the layers. SmS thin films were deposited by e-beam sublimation of Sm in an H2S atmosphere, while EuS was directly sublimated by e-beam from EuS. Structural properties of the separate thin films were first studied before the deposition of the final nanocomposite tri-layer system. Piezoresistance measurements demonstrated two sharp resistance drops. The first drop, at lower pressure, corresponds to the switching characteristic of SmS. The second drop, at higher pressure, is attributed to EuS, partially mixed with SmS. This behavior provides either a well-defined three or two states system, depending on the degree of mixing. Depth profiling using x-ray photoelectron spectroscopy (XPS) revealed partial diffusion between the compounds upon deposition at a substrate temperature of 400 ◦C. Thinner tri-layer systems were also deposited to provide more interdiffusion. A higher EuS concentration led to a continuous transition as a function of pressure. This study shows that EuS-modified SmS thin films are possible systems for piezo-electronic devices, such as memory devices, RF (radio frequency) switches and piezoresistive sensors.

**Keywords:** SmS; EuS; semiconductor-metal transition; structural properties; piezoresistivity; interdiffusion; rare earths; thin films

#### **1. Introduction**

Materials science plays a significant role in the fabrication of new devices based on chemical compounds with specific properties, providing unprecedented device capabilities. Such a device is the piezoelectronic transistor (PET) [1–4], which can exploit the pressure-induced semiconductor to metal transition (SMT) of certain compounds. Examples are Sm chalcogenides [5,6], Mott insulators, as well as other material oxides (e.g., Sr2IrO4) [7]. Specifically, SmS features a hysteretic pressure-induced SMT [6,8,9] at around 0.65 GPa, where the resistance drops significantly. In single crystals, the system returns to the semiconducting state upon release of the pressure, while for thin films thermal annealing or a tensile force is needed. Although this behavior was mainly studied in bulk crystals [6,10], we recently managed to observe hysteretic resistance loops in SmS thin films [11]. Another approach to provide a tunable, hysteretic piezoresistive behavior could be the use of alloyed systems [12,13] that shift the energy bands of the primary material (SmS), without inducing the metallic state, leading to reversible switching characteristics upon releasing force. In order to achieve this, SmS can be alloyed with similar materials that have a somewhat wider band gap. This slightly opens the band gap and shifts the SMT to higher pressures, in comparison to pure SmS. In Figure 1, we see a qualitative representation of the hysteretic discontinuous resistance change in SmS as a function of pressure, as observed in single crystals. This drop can be explained as follows. At atmospheric pressure, the material possesses its semiconducting high resistance state (HRS), which changes to a metallic state (low resistance state, or LRS) when pressure is applied. The required pressure strongly depends on the gap between the *4f* states of Sm ions and the *5d(t2g)* degenerate conduction band (CB) [14]. SmS shows a smaller gap (~0.15 eV), than SmSe (transition pressure = 2 GPa) and SmTe (transition pressure = 4.5 GPa), resulting in an isostructural transition at lower pressures (at around 0.65 GPa) [15]. The substitution of SmS with, for instance, wide band gap rare earth-based materials (e.g., EuS or YbS) [16,17], promotes a shift of the *5d* band of the alloyed system to higher energies, with the increase of the substituent further increasing the Sm *4f-5d* band gap. Nevertheless, there are other elements substituting for Sm, which directly promote a chemically triggered transition to the metallic state at ambient conditions. Such elements, for example, are Gd and Y [12,13,17,18], both decreasing the phase transition to even lower pressure. This tunability of the piezoresistive response promotes SmS and alloyed SmS as possible candidates for memory and RF (radio frequency) switching devices [1–4,19].

**Figure 1.** Schematic representation of the piezoresistive response of single crystal SmS. At the critical pressure, a change from the high-resistive to the low-resistive state occurs, due to the pressure induced shift of the *5d* conduction band towards the *4f* states of Sm2<sup>+</sup> (inset). The top inset shows the gap between the *4f* states and *5d* conduction band at atmospheric pressure. The bottom inset demonstrates the closing of the gap, upon application of pressure. Upon release of the pressure, the SmS returns to the high-resistive state.

In this work, we report on the piezoresistance response of a SmS/EuS/SmS tri-layer thin film system. This tri-layer is an experimentally easy and controlled way to study the alloyed system Sm1-xEuxS. Since the individual films are very thin, interdiffusion between the layers is expected, especially at elevated temperatures during deposition or after post-deposition thermal annealing. We first present some of the basic properties of the separate compounds before showing the structural properties of the tri-layer system. We used two different substrate temperatures to study the diffusion between the layers upon deposition. Thermal post-annealing was performed to further investigate diffusion. The resistance drop, and its subsequent rise after pressure release, confirms that this material system is a promising candidate for several strain-based sensing devices. The resistance response strongly depends on the mixing between the layers. The as-deposited tri-layers at 250 ◦C showed a three states system behavior, while deposition at 400 ◦C can lead to a conventional system with two states.

#### **2. Materials and Methods**

An e-beam evaporator (model: Leybold Univex 450, Leybold GmbH, Germany) was used to deposit SmS on Corning (1737F) glass and 6 inch Si (100) wafers. Sm metal (Smart Elements GmbH, Vienna, Austria, 99.99%) was used as the target material and deposited at a rate of typically 0.8 nm/s under a reactive H2S (Praxair Inc., Danbury, CT, USA, 99.8%) flow, leading to a pressure of 1 <sup>×</sup> 10−<sup>5</sup> mbar during the deposition. The base pressure of the system was approximately 2 <sup>×</sup> <sup>10</sup>−<sup>6</sup> mbar. Details on the deposition conditions required for obtaining stoichiometric and well-crystallized SmS are described elsewhere [20]. EuS was first synthesized starting from Eu2O3 (Alfa Aesar, Thermo Fisher GmbH, Germany, 99.99%) powder. Eu2O3 was placed in a tube furnace for 2 h under H2S flow, at 1000 ◦C. The produced sulfurized powder consisted of EuS, as confirmed by XRD (not shown). Then, by using a hydraulic press, EuS pellets were prepared as target material for the deposition of EuS thin films by e-beam evaporation under identical H2S flow as for the deposition of SmS. Resistivity values were calculated from the sheet resistance measured using a four probe setup. In order to electrically insulate the thin films from the Si substrate, dedicated samples were deposited on top of a 600 nm Al2O3 (Alfa Aesar, Thermo Fisher GmbH, Germany, 99.99%) thin film, prepared by e-beam evaporation. For electrical measurements where the resistivity was measured across the thin film, iridium bottom and top electrodes (thickness 50 nm) were deposited by e-beam evaporation, starting from an Ir slug (Aldrich Chemistry BVBA, Belgium, 99.9%), with the substrate at room temperature. The bottom electrodes were blanket layers deposited over the entire substrate, whereas the top electrodes were deposited through a mask, yielding circular electrodes with a diameter of 1.5 mm. The SMT was introduced in SmS by rubbing the thin film surface with a round-shaped metal tip, without visibly damaging the thin film surface. The reverse MST (metal to semiconductor transition) was induced by thermal annealing at 400 ◦C in vacuum. The semiconducting and metallic SmS state will be referred to as S-SmS/HRS (high resistance state) and M-SmS/LRS (low resistance state), respectively.

Structural characterization of the fabricated thin films was carried out via X-ray diffraction (XRD) using a standard powder diffractometer (D8 with Ni-filtered CuKα<sup>1</sup> radiation, λ = 0.154059 nm, Bruker AXS GmbH, Karlsruhe, Germany). In situ high temperature X-ray diffraction (HTXRD) patterns were also measured using a Bruker D8 Discover system (equally using CuKα<sup>1</sup> radiation) with an integrated annealing chamber, able to support several atmospheric conditions. In the latter case, a linear detector was used, which allowed for collecting diffraction patterns in seconds, without moving the sample or detector. SEM analysis was carried out in an FEI electron microscope (Quanta FEG 200, Hillsboro, Oregon, USA), with a point resolution of 1.7 nm at 20 kV. Ultraviolet-visible (UV-Vis) spectra were recorded at room temperature in the specular reflectance geometry (V-W method) with a Varian Cary 500 UV-Vis spectrophotometer (Agilent, Santa Clara, CA, USA) in the wavelength range of 200–800 nm. Piezoresistance measurements were performed using a homemade device, similar to reported devices [21]. Finally, we used X-ray photoelectron spectroscopy (XPS) to detect and confirm the presence and diffusion of the components in the multi-layered structures. The used set-up was an ESCA S-probe VG (Thermo Fisher GmbH, Germany) with an Al(Ka) source (1486.6 eV). The base pressure of the system was 5 <sup>×</sup> <sup>10</sup>−<sup>10</sup> mbar, while the pressure during Ar-sputtering increased up to 2 <sup>×</sup> <sup>10</sup>−<sup>7</sup> mbar. The sputter time was 25 s per step and the measurement time after each step was 5 min, unless mentioned otherwise. In all studies of the tri-layers, the Sm *3d5*/*2*, S *2p*, O *1s*, and Eu *3d5*/*<sup>2</sup>* peaks were used for the calculation of elemental concentrations.

#### **3. Results**

#### *3.1. Individual SmS and EuS Thin Films*

EuS thin films with a thickness of 25 nm are largely transparent in the visible region, although the UV-Vis spectra show a broad absorption band between 1.8 and 2.8 eV, caused by the transition from the *4f<sup>7</sup>* ground state to the *4f65d* configuration in Eu2<sup>+</sup> (Figure 2a). This is in line with EuS being a natural ferromagnetic semiconductor [22], showing an optical band gap of 1.65 eV for thin film EuS [23]. EuS thin films with a thickness of 25 nm deposited at 250 ◦C on Si (100) wafer show a good crystallinity with preferential growth orientation of the (200) planes parallel to the substrate. Notice that SmS and EuS have the same rock salt lattice structure, with almost identical lattice constants for S-SmS and EuS (5.970 Å and 5.968 Å respectively). Consequently, standard XRD analysis cannot be used to discriminate between both materials.

**Figure 2.** Basic properties of the deposited SmS and EuS thin films. (**a**) Optical (on glass) and structural (inset, on Si wafer) fingerprint of a 25 nm as-deposited EuS. (**b**) XRD patterns showing the structural behavior in the as-deposited (black line) S-SmS, as well as in the metallic state (red line), after rubbing. The lattice planes are indicated. In both thin film depositions, the substrate temperature was 250 ◦C.

Applying pressure to these EuS thin films does not introduce any transition. This is different for the SmS thin films, where moderate pressure can provide the SMT at room temperature. Usually, soft polishing is used to induce M-SmS [24,25]. Here, gently rubbing the sample surface leads to the SMT, with the accompanying reduction in lattice constant. This is demonstrated by the corresponding XRD peaks shifting to higher 2θ values (Figure 2b), in accordance with previous investigations [26]. The 100 nm S-SmS thin films show resistivity values in the range from 1.5 to 5 <sup>×</sup> <sup>10</sup>−<sup>1</sup> <sup>Ω</sup>cm (as derived from the sheet resistance), while the rubbed 100 nm M-SmS layers demonstrated 3–4 <sup>×</sup> 10−<sup>3</sup> <sup>Ω</sup>cm. These values are comparable to previous investigations [27]. The as-deposited EuS thin films showed a sheet resistance of about 1.7 MΩ (25 nm EuS thin film), which is almost one order of magnitude lower, relative to previous investigations on highly optimized insulating EuS thin films (the sheet resistance is higher than 20 MΩ for thicknesses between 20 and 200 nm) [28].

In the SmS/EuS/SmS tri-layers, we studied the diffusion process in this nanocomposite system in order to tune the piezoresistive behavior. The substrate temperature, post-deposition annealing, as well as the thickness of the diffusive layer (EuS) relative to the full tri-layer thickness were changed to influence the layer interdiffusion.

#### *3.2. Deposition and Annealing of Tri-Layers*

Figure 3 shows an XRD plot of a 35/30/35 nm SmS/EuS/SmS tri-layer deposited at 250 ◦C, before (black curve) and after rubbing (red curve). After rubbing the sample surface, we observe two peaks for the (200) lattice plane, one related to EuS and one to M-SmS. Hence, at a deposition temperature of 250 ◦C, the individual layers do not thoroughly mix. XPS analysis, discussed below, confirms that the thin films are not mixed when deposited at 250 ◦C.

**Figure 3.** XRD patterns of an as-deposited SmS/EuS/SmS (35/30/35 nm) tri-layer deposited at 250 ◦C (black curve), and after rubbing (red curve).

In order to probe the influence of temperature on the structural evolution of the tri-layer thin films, in situ HTXRD was performed up to 800 ◦C (with a heating rate of 10 ◦C/s). Figure 4a shows the in situ HTXRD patterns of an as-deposited (at 250 ◦C) 35/30/35 nm SmS/EuS/SmS tri-layer under 20% O2 and 80% He atmosphere. In this atmosphere, the tri-layer system remains stable up to 350 ◦C. As mentioned before, the peak at around 30◦ is composed of the reflection from the (200) lattice planes of SmS and EuS. Above 350 ◦C, the diffraction intensities decrease—although the (111) reflection of SmS and EuS remains visible—and at 500 ◦C the observed diffraction positions match with those of Sm2O2S. For comparison, Figure 4b shows the stable character of 25 nm EuS thin film up to 450 ◦C in the ambient-like atmosphere, in line with previous investigations [28,29].

**Figure 4.** (**a**) In situ XRD patterns for increasing temperature of an as-deposited 35/30/35 nm SmS/EuS/SmS tri-layer in an ambient-like atmosphere of 20% O2 and 80% He. (**b**) Same as in (**a**) for an as-deposited 25 nm EuS thin film on Si. Both thin films were deposited at 250 ◦C.

Then, we increased the substrate temperature to 400 ◦C to explore any changes in the switching properties and the degree of mixing in the tri-layer thin film stacks. In order to study the switching behavior, we rubbed the surface of a triple layer to induce the SMT, as shown earlier for the single SmS film (Figure 3). A clear SMT was still observed by following the shift of the (200) the diffraction peak (Figure 5a) from about 30.27◦ to 31.44◦ (corresponding to a change in d200 from 2.95 Å to 2.84 Å), although a smaller peak remained at the original position. While the former peak demonstrated a behavior typical for SmS, the latter peak is indicative of the presence of an EuS-like part in the tri-layer, which does not show switching. A similar picture is observed for the (111) peak. By annealing in vacuum at 400 ◦C, the semiconducting state can successfully be recovered (Figure 5a, blue curved), although the derived d200 lattice spacing of 2.92 Å is somewhat smaller than the original value.

**Figure 5.** (**a**) XRD of an as-deposited SmS/EuS/SmS (35/30/35 nm) tri-layer at 400 ◦C (black curve), after rubbing (red curve) and after annealing in vacuum at 400 ◦C, for 10 min, and then cooling down (blue curve). (**b**) Cumulative annealing in ambient air of a similarly rubbed sample, as in (**a**), to provide the thermally triggered metal to semiconductor transition. For comparison, XRD pattern of a 100 nm SmS thin film, after 30 min annealing at 350 ◦C, is also depicted with the purple dashed line.

Subsequently, we investigated the thermally induced metal-semiconductor reverse transition (in ambient atmosphere) of the tri-layer system deposited at 400 ◦C. The red curve on the bottom of Figure 5b corresponds to the rubbed (without any annealing) tri-layer initially deposited at 400 ◦C. This sample was fabricated following the same process as in Figure 5a for the red curve. Then, 10 min of post-annealing at 300 ◦C in air was performed, which did not yield any significant structural change. Prolonging the accumulated annealing time to 25 or 60 min did not switch back the M-SmS part of the thin film. The back switching of the tri-layer to the semiconducting state (smaller 2θ value) was observed, however, after only 10 min of annealing at a slightly higher temperature of 350 ◦C. In contrast, a single SmS film is quite stable, without showing any back switching behavior, when it is annealed at 350 ◦C, for 30 min in air (purple dashed curve in Figure 5b).

#### *3.3. Piezoresistive Behavior*

Figure 6a shows the piezoresistance of the tri-layer system of SmS/EuS/SmS, deposited at 250 ◦C, with layer thicknesses of 35/30/35 nm, using Ir bottom and top electrodes. It is important to notice that the application of pressure with the indenter leads to a relatively small area (of the order of tens of μm2) where the pressure is applied, as compared to the total top electrode area (1.8 mm2). Hence, the measured resistance across the thin film is essentially determined by two parallel resistors, one with variable resistance (where the pressure is applied) and one with fixed resistance (outside the indenter area). The values given below are for the combined resistance, as it is difficult to estimate the pressed area, given that it is a function of the applied force and thus the contact area of the indenter. When the indenter just makes contact with the electrode, the resistance is 76 kΩ (HRS). Application of pressure first leads to a limited, gradual decrease in the resistance, until a first sudden drop to 5.2 kΩ appears at a force of about 0.5 N. We define this as an intermediate resistance state (IRS). Note that the actual change in resistance from the semiconducting to the metallic state is higher, taking into account the effect of the parallel resistance. The force threshold is similar to the behavior of single SmS thin films, leading to the conclusion that the SmS-like parts in the tri-layer switch first. Looking at higher force values, we see a second sharp resistance drop at around 1.55 N (to a resistance of 260 Ω), which is likely related to a partly mixed alloy of SmS and EuS. The resistance gradually drops further upon increasing force, to 160 Ω at 2 N, which can be related to the piezoresistive behavior of Eu-rich SmS, where the pressure leads to a narrowing of the *4f* –*5d* gap [16]. Upon gradual release of the pressure, the resistance increases slowly again, showing a major, rather discontinuous, increase between 0.8 and 0.5 N. This is likely related to the switching back of the alloyed part of the tri-layer, which had

switched to an LRS around 1.5 N. Just before full release of the force (when the contact between the indenter and the top electrode is lost), the resistance is still an order of magnitude below the initial resistance of 76 kΩ. When contact is made again with the top electrode, the high resistance value is restored, indicating that the switching back of the relatively pure SmS part of the tri-layer (which had switched around 0.5 N during the loading) occurs very close to ambient pressure. As demonstrated in the inset of Figure 6a, the same piezoresistive behavior is found during 5 consecutive cycles of loading and unloading, showing that effectively a three state system, between an HRS, IRS and LRS, is obtained. It should be mentioned that the pressure needed for the second drop slightly changes upon cycling, with a variation between 1.3 and 1.6 N. A more integrated measurement approach where pressure is applied more uniformly is currently under development in order to further characterize the piezoresistive behavior.

**Figure 6.** (**a**) Resistance across the SmS/EuS/SmS (35/30/35 nm, deposited at 250 ◦C) tri-layer, upon applying force up to 2 N. Red arrow for increasing force, the navy arrow when unloading. (**b**) Resistance value for five consecutive cycles of loading and unloading, at a force of 0.05, 0.9 and 1.6 N (corresponding to the high (HRS), intermediate (IRS) and low (LRS) resistance state) during the loading phase of the cycles. (**c**) Resistance of a thin film tri-layer with the same composition, deposited at 400 ◦C.

In Figure 6c, we demonstrate the piezoresistance of a sample with identical composition and electrode configuration, deposited at a substrate temperature of 400 ◦C. A significant difference with the previous case (sample deposited at 250 ◦C, Figure 6a) is that no sudden resistance drops are observed, at least not with large changes in the resistance. As will be demonstrated below by means of XPS depth profiling, the tri-layers deposited at 400 ◦C show a higher degree of mixing between the individual SmS and EuS layers. The high resistance state shows values of about 29 kΩ, smaller than the corresponding value (76 kΩ) for the tri-layer deposited at 250 ◦C, while at higher force values (~1.6 N) the resistance reaches a value of 421 Ω.

A pure EuS layer does not show significant changes in the resistance up to pressures similar to those where the SMT appears in single SmS films. This is due to a larger gap between the *4f* and *5d* bands for EuS compared to SmS. A structural change does occur, from fcc to bcc, at around 20 GPa [30], while 36 GPa is needed for the complete structural change, accompanied by a change in the valence state [31]. For a 100 nm EuS layer in our experiment, with bottom and top electrodes, the resistance across the EuS thin film was 2.2 <sup>×</sup> 107 <sup>Ω</sup> at close to ambient pressure (when the indenter just made contact with the top electrode), while it remained at a high value of 1 <sup>×</sup> 107 <sup>Ω</sup>, at 1.5 N. Measuring across the 100 nm EuS layer, the observed values were similar to those recorded for high quality EuS layers, via sheet resistance measurements, as mentioned above. The value of 2.2 <sup>×</sup> 107 <sup>Ω</sup> is slightly higher than expected from the sheet resistance for 25 nm EuS.

#### *3.4. XPS Analysis*

To correlate the piezoresistance properties of the studied tri-layer system to the extent of inter-diffusion between the three individual layers, XPS depth profiles were recorded for tri-layers with different compositions and annealing conditions. Photoelectrons from the four main elements (Sm, S, O, Eu) were recorded at the distinct spectral regions for each component: Sm *3d5*/*2*, S *2p*, O *1s*, and Eu *3d5*/*2*. The presented results cover the total thickness of the tri-layer system, while the Si substrate is not included.

An as-deposited composite thin film, at 250 ◦C, does not yield a fully alloyed system (see Figure 7a). The Sm concentration decreases from approximately 60% in the outer parts of the stack to 24% in the mixed region, where the Eu concentration reaches 40%, while the S concentration remains relatively constant throughout the sample. Care must be taken in the interpretation of these XPS results, since the concentration distribution can be influenced by knock-on sputtering, obscuring the interfaces between films. In addition, it was found by the authors of [32] that any exposed SmS surface oxidizes, even in UHV (ultrahigh vacuum) conditions. Nevertheless, the oxidation is surface limited, with the O remaining at very low concentration in the main volume of the tri-layer, even after 4 h of annealing at 400 ◦C (blue curve in Figure 7c). However, the main conclusion from Figure 7a remains, namely that there is incomplete mixing of the layers after deposition at a substrate temperature of 250 ◦C.

In this case, the tri-layer system is mainly consisting of the SmS-like parts that show a pressure-triggered resistance and color change at low applied pressure, and the partly mixed EuS/SmS intermediate thin film. Before the application of pressure, both materials (SmS, EuS) demonstrate semiconducting properties. For these materials, the *4f<sup>6</sup>* states of the Sm2<sup>+</sup> and Eu *4f<sup>7</sup>* states lie between the valence band (formed by the *3p* orbitals of S) and the CB (constituted of the *5d* and *6s* orbitals of the lanthanides). In the case of SmS, the gap between the *4f* state and the bottom of the CB collapses at a pressure of about 0.65 GPa, providing metallic properties, or a low resistance state.

Based on previous investigation on bulk crystals [6], it can be concluded that a mixture between the layers leads to a shift of the pressure threshold value of resistance drop. This could be related either to the size effects or the electronic structure, with the occurrence of a Eu *4f7* level deep in the energy gap that is less important in the SMT of the mixed system [33]. Eu is among the rare-earth elements which do not induce the valence transition in SmS, so external pressure is needed to induce the SMT in those mixed systems. Initial studies have indicated that the size factor is the most significant reason for the SMT, especially for substituent elements with smaller ion sizes than Sm2+. In that case, the

bottom of the *5d* band lowers in energy, due to the local compression by the crystal lattice of the Sm ions. Nevertheless, other elements, such as Ca and Yb, though they show a smaller ion radius than Sm2+, do not manage to induce the SMT under ambient conditions. This means that the ionic radius is not the only factor determining the pressure threshold, but also the electronic structure plays a significant role. In our case of Eu-substituted SmS, the Eu ions are all divalent, as the trivalent charge state is not stabilized [34]. In contrast, for the semiconducting SmS, there is a mixture of Sm2<sup>+</sup> and Sm3<sup>+</sup> ions in the thin films, as witnessed from the derived lattice constant and XPS analysis. Hence the introduction of Eu2<sup>+</sup> in a SmS thin film forces the (alloyed) SmS to acquire a slightly larger lattice constant [35]. Indeed, in case of the 35/30/35 nm SmS/EuS/SmS tri-layer deposited at 400 ◦C, the (200) peak position is located at about 30.2◦, slightly lower than the corresponding value (~30.4◦) in a 100 nm SmS film. Consequently, the larger lattice constant leads to a larger energy spacing for the *4f*-*5d* gap [6]. The latter is likely the main physical reason for the second drop at higher force values (~1.5 N). Increasing the amount of Eu, we notice the change from a discontinuous to a continuous transition (see Figure 8), which is in accordance with previous results [6]. As a matter of fact, the three states system presented in Figure 6a is a consequence of the occurrence of more than one material system in the entire thin film stack.

As an increase in temperature can enhance the interdiffusion of SmS and EuS, the tri-layer thin film shows a better mixing upon deposition at 400 ◦C, with Eu more equally spread out towards the outer parts of the stack (Figure 7b). Additionally, the outer regions show a more stoichiometric composition (in terms of the relative concentration of Sm and S) in comparison with the deposition at 250 ◦C. Also, prolonged post-annealing of 4 h at 400 ◦C (Figure 7c) did not show any reliable progress on mixing. Besides, the oxygen relative concentration increases only close to the material´s surface. Deposition at considerably higher substrate temperature than 400 ◦C would be needed for further investigate the mixing in the SmS/EuS/SmS system. Another option would also be the post-annealing at temperatures higher than 400 ◦C. Nevertheless, any attempt to provide further mixing should seriously take into consideration the corresponding HTXRD results (Figure 4a), which indicate the oxidation of the system at elevated temperature.

Figure 7d,e show representative Sm 3d5/<sup>2</sup> and Eu3d5/<sup>2</sup> photoelectron spectra. The different traces correspond to measurements after 25 s (red curve), 150 s (navy curve) and 275 s (green curve) of argon ion sputtering. The XPS depth profiling process and the high surface reactivity of SmS can drastically change the ratio of the measured valence states of the lanthanide ions in the thin films [32]. Nevertheless, for the calculations of the element concentrations throughout the samples thickness, this is not an issue. For a further and detailed discussion of the valence state evaluation of Sm ions, as well as the surface oxidation of SmS thin films and the impact of depth profiling, we refer to the work done in [32].

Obviously, there is an intricate relation between the diffusion process (Figure 7) and the electrical changes (Figure 6) in the studied devices. For instance, in an inhomogeneous strongly correlated system, consisting of a semiconducting (EuS-like) and two metallic (M-SmS-like) phases, the measured property (e.g., resistance change) is an average value related to the entire system [36]. Also, the higher temperature deposition, at 400 ◦C, can provide an electronic state rearrangement, triggered by temperature [37]. This rearrangement can boost electrons either from the *4f* states of Sm or impurity levels within the energy gap to the CB. The measured electrical resistance, in the semiconducting state, can thus be lower upon deposition at 400 ◦C compared to films deposited at lower temperature, in accordance with our results (Figure 6a,c). Future work will focus on studying the influence of (measurement) temperature on the electrical behavior of these alloyed thin films, as this will yield insight in the dynamics of the metal-insulator transition [38].

**Figure 7.** XPS depth profiling analysis of an as-deposited 35/30/35 nm SmS/EuS/SmS tri-layer system, deposited 250 ◦C (**a**), deposited at 400 ◦C (**b**), and after 4 h post-deposition annealing at 400 ◦C (**c**). (**d**) Sm 3d5/<sup>2</sup> photoelectron peaks for the fabricated samples in Figure 7a,c, from top to bottom, respectively. (**e**) Eu 3d5/<sup>2</sup> photoelectron peaks, as in (**d**). In both (**d**) and (**e**), the red curve corresponds to the photoelectron spectra after 25 s of sputtering, the navy one to 150 s, and the green curve to the sputtering time of 275 s.

In order to overcome the limited diffusion lengths, tri-layers consisting of thinner SmS and EuS layers were also deposited. Substrate temperature was chosen at 400 ◦C, in order to maximize the interdiffusion. Based on literature about Eu-doped SmS bulk crystals, we attempted to deposit two types of compositions, thus aiming at a different piezoresistive behavior. For a fully mixed tri-layer system that would show a discontinuous resistance change, the Eu concentration should be below the critical value of about 25% substitution [6]. For a second tri-layer, a higher Eu amount was chosen, in order to obtain a continuous resistance change. This is because only at lower Eu concentration is the resistance change due to the first-order valence change in the Sm ions.

For the first type, we deposited a tri-layer SmS/EuS/SmS with thicknesses of 18/4/18 nm, yielding an overall Eu fraction of 10%. Despite the small thickness, the thin film is well crystallized with both (111) and (200) reflections prominently visible (Figure 8a). For the corresponding piezoresistance behavior (Figure 8b), only one drop is observed. This is probably a result of the small thickness of the intermediate EuS layer, which is now more homogeneously diffused into the outer SmS layers. In the case of using 50% of EuS (6/12/6 nm SmS/EuS/SmS), only the (200) XRD peak appeared (Figure 8a), while the resistance response to the applied force is almost perfectly continuous (Figure 8c). Nevertheless, around 1.1 N, a sudden drop in the resistance can be observed (Figure 8c), which points to the mixing not being fully completed. In both cases there is a change of roughly one order of magnitude in the resistance, when increasing the applied force from about 0 N to 1.5 N. Taking into consideration that we used the same top electrode material (Ir) with the same area as in the devices in Figure 6, the moderate resistance change should be attributed to either the lower structural quality, the smaller thickness or the occurrence of EuS-rich films, which show a rather limited change in resistance.

**Figure 8.** (**a**) XRD patterns of the thinner tri-layer systems (18/4/18 nm in blue dots and 6/12/6 nm in purple stars). Piezoresistance behavior for 18/4/18 nm tri-layer (**b**) and 6/12/6 nm tri-layer (**c**) deposited in between Ir electrodes. For clarity, in (**b**) red arrow represents the loading process, while the navy arrow shows the unloading process.

To shed light on the diffusion evolution, we used XPS depth profiling (with a sputter time of 3 s per step) for the case of 6/12/6 nm SmS/EuS/SmS (Figure 9). The two compounds homogeneously interdiffuse throughout the entire volume of the deposited tri-layer system, although some local variations could still occur. These results are in line with the previous analysis related to the resistive behavior.

**Figure 9.** XPS depth profiling for the 6/12/6 nm SmS/EuS/SmS tri-layer system deposited at 400 ◦C.

#### **4. Conclusions**

In this manuscript, we reported on high-quality SmS/EuS/SmS tri-layer thin films deposited by e-beam evaporation. The structural properties were determined, as well as measurements of their resistance response to the applied force. This work demonstrates a well-defined hysteretic pressure-triggered semiconductor to metal transition (SMT) in SmS/EuS/SmS thin films. Depending on the substrate temperature and thus the degree of interdiffusion, we were able to demonstrate either a three or a two state piezoresistive system. Three state piezoresistive systems are not common and can for instance be used in nano-sensors, where they can selectively operate at different regimes of force, providing either a continuous or discontinuous change of electrical properties. In case of a 35/30/35 nm stack deposited at a substrate temperature of 400 ◦C, the resistance change tends to become continuous. A post-annealing at 400 ◦C up to 4 h did not lead to significant additional diffusion between the layers. HTXRD results on the nanocomposite system demonstrated that an oxidation process begins at 500 ◦C. Thinner nanocomposite layers were also deposited to evaluate the influence of thickness to the diffusion, showing improved mixing between the layers. This resulted in a more continuous piezoresistive response. The change in resistance for the studied pressure range is more limited as in the case of pure SmS thin films, since no semiconductor to metal transition occurs. This work shows promising experimental results on the piezoresistance response of SmS/EuS/SmS tri-layer nanocomposites. On the one hand, this triggers the further exploration of this system, where the degree of interdiffusion could be further controlled to arrive at specific piezoresistive responses. On the other hand, these developments support the future application in contemporary integrated piezo-based electronic devices, such as piezo-electronic memories and RF switches. Last but not least, studying the behavior of SmS thin films, using other substituting lanthanides, like Gd or Y, could also be future avenues.

**Author Contributions:** Conceptualization, A.S., D.P. and P.F.S.; Investigation, A.S.; Methodology, A.S., D.P. and P.F.S.; Resources, D.P. and P.F.S.; Supervision, D.P. and P.F.S.; Writing—original draft, A.S.; Writing—review and editing, A.S., D.P. and P.F.S.

**Funding:** The authors are grateful to the European Union for the funding within Horizon 2020 research and innovation program under grant agreement No 688282. P.F.S. acknowledges the BOF GOA Project ENCLOSE at Ghent University for financial support.

**Acknowledgments:** A.S. would like to cordially thank Nico De Roo for the XPS depth profiling measurements, as well as Jonas Joos, for assistance with the fabrication of the EuS powder. Lastly, we would also like to thank the Draft and CoCoon research groups (both at Ghent University) for the set-up for measuring piezoresistance and HTXRD, respectively.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## *Article* **Investigation of Pyrophosphates KYP2O7Co-Doped with Lanthanide Ions Useful for Theranostics**

**Adam Watras 1,\*, Marta Wujczyk 1, Michael Roecken 2, Katarzyna Kucharczyk 3,4, Krzysztof Marycz 3,4,5 and Rafal J. Wiglusz <sup>1</sup>**


Received: 7 October 2019; Accepted: 7 November 2019; Published: 11 November 2019

**Abstract:** Diphosphate compounds (KYP2O7) co-doped with Yb<sup>3</sup><sup>+</sup> and Er3<sup>+</sup> ions were obtained by one step urea assisted combustion synthesis. The experimental parameters of synthesis were optimized using an experimental design approach related to co-dopants concentration and heattreatment as well as annealing time. The obtained materials were studied with theinitial requirements showing appropriate morphological (X-Ray Diffraction (XRD), Scanning Electron Microscopy (SEM)) and spectroscopic properties (emission, luminescence kinetics). Moreover, the effect of Er3<sup>+</sup> and Yb3<sup>+</sup> ions doped KYP2O7 on morphology, proliferative and metabolic activity and apoptosis in MC3T3-E1 osteoblast cell line and 4B12osteoclasts cell line was investigated. Furthermore, the expression of the common pro-osteogenic markers in MC3T3-E1 osteoblast as well as osteoclastogenesis related markers in 4B12 osteoclasts was evaluated. The extensive in vitro studies showed that KYP2O7 doped with 1 mol% Er3<sup>+</sup> and 20 mol% Yb3<sup>+</sup> ions positively affected the MC3T3-E1 and 4B12 cells activity without triggering their apoptosis. Moreover, it was shown that an activation of mTOR and Pi3k signaling pathways with 1 mol% Er3<sup>+</sup>, 20 mol% Yb3<sup>+</sup>: KYP2O7 can promote the MC3T3-E1 cells expression of late osteogenic markers including RUNX and BMP-2. The obtained data shed a promising light for KYP2O7 doped with Er3<sup>+</sup> and Yb3<sup>+</sup> ions as a potential factors improving bone fracture healing as well as in bioimaging (so-called in theranostics).

**Keywords:** diphosphates; up-conversion; theranostics

#### **1. Introduction**

In recent years, much attention has been paid to rare earth phosphate phosphors due to their appealing features, such as chemical stability and diversity in crystallographic structure [1]. The phosphates could be used as a matrix for doping with optically active ions, such as the rare earth metals. Potential application of the rare earth phosphates could be related to such areas as: cell bioimaging [2–4], light-emitting diodes [5–8], solar cells [9–11] as well as regenerative medicine.

Potassium yttrium(III) diphosphate(V) KYP2O7 is a polymorphic compound. Depending on the annealing temperature so-called the low temperature phase (β-KYP2O7) or the high-temperature phase (α-KYP2O7) could be obtained. On the basis of ionic radius ratio (*r*K+/*r*Y3<sup>+</sup> = 1.68) value, polymorphism of KYP2O7 can be explained [12]. Three different synthesis routs were published for the KYP2O7: solid state reaction [13], one step urea-combustion synthesis [14] and boric acid flux method [15]. According to our knowledge the modern luminescent material KYP2O7 has never been employed as a matrix for investigation of up-conversion processes in biomedical applications.

Recently, tissue engineering together with regenerative medicine has become amore and more powerful tool in the field of bone regeneration as well as theranostics [16–19]. There are serious requirements for developing a strategy that could improve bone fracture regeneration, especially for elderly patients suffering from osteoporosis [20,21]. Bone fracture naturally involves two opposite processes, i.e., osteogenesis and osteoclastogenesis. The balance between these two processes ensures a new bone formation and finally bone regeneration. In the osteogenesis process the bone tissue formation is directly mediated by osteoblasts. This process is regulated on gene expression level by several transcripts including collagen type II, bone morphegenic protein 2 (BMP2), osteocalcin (OCL), osteopontin (OPN) and alkaline phospotase (ALP). The dynamic process thatis bone formation is mediated by several signaling pathways including mammalian target of rapamycin (mTOR) and phosphoinositide 3-kinase (Pi3k) regulating osteoblastogenesis and osteoclastogenesis. The activation of osteoclastis required for proper bone shaping and providing access to bone-stored minerals [22]. Thus, the induction of osteoblasts as well osteoclast activity and maintaining proper balance between them ensures proper bone remodeling and fracture regeneration, since over activity of osteoclast will lead to bone resorption. This phenomenon is well-known for several disorders including osteoporosis. The ability to improve osteoblasts viability with simultaneous inhibition of osteoclastogenesis seems to be a real challenge for novel materials. Moreover, the modern materials that serve additional functionality i.e., bioluminescence, which allows visualizing regenerative processes in a non-invasive way, are strongly required. The bioluminescent agents including Er3<sup>+</sup> and Yb3<sup>+</sup> besides their physical functions might additionally promote osteoblast activity, which can serve as their additional benefit.

In this paper, samples of KYP2O7 co-doped with Er3<sup>+</sup> and Yb3<sup>+</sup> ions, were obtained using the one step urea-combustion method. Moreover, the spectroscopic investigation of occurring up-conversion processes into KYP2O7 matrix doped with Er3<sup>+</sup> and Yb3<sup>+</sup> ions was presented. Furthermore, the effects of KYP2O7 doped with 1 mol% Er3<sup>+</sup> and 20 mol% Yb3<sup>+</sup> on MC3T3-E1 osteoblasts and 4B12 osteoclasts were investigated paying special attention to viability, apoptosis, mitochondrial activity as well as an expression of common osteogenic and osteoclastogenesis related markers on mRNA levels.

#### **2. Materials and Methods**

The *x* mol% Er3<sup>+</sup>, *y* mol% Yb3<sup>+</sup>:KYP2O7 (where *x* = 0.25, 0.50, 0.75, 1, 2, 5; *y* = 1, 2, 5, 10, 15, 20) powders were obtained by one step urea assisted combustion synthesis on the grounds of the synthesis route described elsewhere by R. P ˛azik et. al [14]. Reactants weight was calculated in stoichiometric manner with an exception to 10% excess for K2CO3·1.5H2O as well as to 20% excess of CH4N2O in reference to metal cations. The raw materials used for the synthesis purpose are: Y2O3 (Alfa Aesar GmbH & Co KG, Karlsruhe, Germany, 99.99%), Er2O3 (Alfa Aesar GmbH & Co KG, 99.99%), Yb2O3 (Alfa Aesar GmbH & Co KG, 99.99%), K2CO3·1.5H2O (Chempur, Piekary Slaskie, Poland, 99.0%), CH4N2O (PPH "POCh" S.A. Gliwice, Poland, 99.5%), (NH4)2HPO4 (Carl Roth GmbH + Co. KG, Karlsruhe, Germany, 99.999%), HNO3 (POCH S.A., Gliwice, Poland, 65%, ultrapure). Each of the final mixtures was dried for 24 h at 90 ◦C, later annealed at series of temperature ranging from 600 ◦C up to 800 ◦C for 4, 8 and 12 h.

The X-ray diffraction patterns were obtained by the use of X'Pert Pro PANalytical diffractometer (Cu, Kα1: 1.54060 Å) (Malvern Panalytical Ltd., Malvern, UK) in a 2θ range of 10◦–50◦, with a scan rate of 1.3◦/min for 30 min at a room temperature. Investigation of morphology was performed using scanning microscope, specifically the FEI Nova NanoSEM 230 microscope (FEI Company, Hillsboro, OR, USA) equipped with the EDS spectrometer (EDAX PegasusXM4). Hydrodynamic size of the particles dispersed in water was determined by the use of dynamic light scattering technique supported by Zetasizer Nano-ZS (Malvern Panalytical Ltd., Malvern, UK) that is equipped with the He-Ne 633

nm laser (see Figure S1). Also, zeta potential was distinguished (see Figure S1). The emission spectra, as well as power dependence functions were recorded using the laser diode (λexc = 980 nm), with regulated power ranging from 0 to 4 W (Changchun New Industries Optoelectronics Tech. Co. Ltd., Jilin, China). For measurements the KG5 Schott filter was applied and as an optical detector the Hamamatsu PMA-12 photonic multichannel analyzer (Hamamatsu Photonics K.K., Hamamatsu City, Japan) was used. Furthermore, the obtained emission spectra are the result of averaged measurements, where the fixed parameters are the exposure time (200 ms) and the cumulative amount of measurements (15). Decay curves were collected using the tunable Ti:Sapphire laser (LOTIS TII, Minsk, Belarus) (λexc = 980 nm) pumped by the second harmonic of the YAG:Nd3<sup>+</sup> pulse laser (ƒ = 10 Hz, *t* < 10 ns).

Mice osteoblasts MC3T3-E1 and osteoclasts 4B12 were used in this study. The cells were cultured in Minimum Essential Medium (MEM) Alpha w/o ascorbic acid (Gibco A10490-01) supplemented with 10% of Fetal Bovine Serum (FBS) (SigmaAldrich, Lenexa, KS, USA) with addition of 1% Penicillin/Streptomycin (P/S) (Sigma Aldrich, USA). In turn 4B12 cells were cultured in EMEM Alpha (Sigma M0200) supplemented with 10% of FBS and 30% of calvaria-derived stromal cell conditioned media (CSCM) without addition of antibiotic. The MC3T3-E1 were cultivated at 80% of confluence and they were passaged every 5 days by enzymatic dissociation using Trypsin-EDTA solution (SigmaAldrich, Saint Louis, MO, USA). Cells were incubated at 37 ◦C in a humidified atmosphere with 5% CO2.

Cell metabolic activity was measured by means of TOX-8 resazurin-based method using in vitro toxicology assay kit. MC3T3-E1 and 4B12 cells were plated into 96-well plates (3 <sup>×</sup> 103 cells per well) in 4 replicates. Next metabolic activity of MC3T3-E1 were measured when cells were exposed to *x* = 10, 15 and 20 mol% of KYP2O7:1 mol% Er<sup>3</sup>+, *x* Yb3<sup>+</sup>, which was incorporated into the culture medium. For examination, compound was diluted in phosphate-buffered saline (PBS). First 10 mg of compound was suspended in 1 mL of PBS. Next this solution was added directly to cell culture medium in the proper concentration. The MC3T3-E1 and 4B12 cells were cultured in the presence and absence of tested materials for 120 h. After 24 and 120 h, culture medium was replaced with 10% solution of resazurin in fresh complete medium and incubated at 37 ◦C for 2h in CO2 cell culture incubator. Reduction of the dye was measured spectrophotometrically at a wavelength of 600 and 690 nm reference length (Epoch, Biotek, Bad Friedrichshall, Germany).

For analysis of genes expression, MC3T3-E1 and 4B12 cells were cultured 120 h onto KYP2O7:1 mol% Er3+, 20 mol% Yb3+. Then, cells were lysed in TRI Reagent and next total RNA was isolated using phenol-chloroform method described previously by Chomczynski and Sacchi [23]. To perform cDNA synthesis gDNA was digested with RNase-free (ThermoScientificTM, Whaltam, MA, USA), DNase I and next cDNA was synthesized using Tetro cDNA Synthesis Kit (Bioline, London, UK). qRT-PCR was performed using CFX ConnectTM Real-Time PCR Detection System (Bio-Rad, Hercules, CA, USA) for gene expression analysis. Reaction mixture contained 1 μL of cDNA in a total volume of 10 μL using SensiFAST SYBR & Fluorescein Kit (Bioline, London, UK). The concentration of primers in each reaction was equal to 500 nM; primer sequences used in individual reactions are listed in Table 1. The algorithm used for quantitative expression of the investigated genes was performed using the 2−ΔΔCT method in relation to housekeeping gene (GAPDH).

To visualize theactin cytoskeleton and location of mitochondria the epifluorescent microscope (Olympus Fluoview FV1200, Tokyo, Japan) was used. Cells cultured onto KYP2O7:1 mol% Er3<sup>+</sup>, 20 mol% Yb3<sup>+</sup> were stained with PhalloidinAtto 488 staining for F-actin visualization. For this purpose, the cells were fixed in 4% paraformaldehyde (PFA) (Sigma Aldrich) for 45 min at RT, then washed with phosphate-buffered saline (PBS) (SigmaAldrich) three times and permeabilized using 0.3% Tween 20 (SigmaAldrich) in PBS for 15 min. For nuclei visualization PhalloidinAtto 488 in PBS (dilution 1:700) (SigmaAldrich) staining for 45 min was performed. Obtained pictures were analyzed using ImageJ software 1.51j version (NIH, Bethesda, MD, USA). For the visualization of the mitochondria network, staining with the MitoRed was performed. For this purpose the culture medium was removed and cells were washed twice with PBS. After that the culture medium with MitoRed (1:1000) was added

to cells in an amount equivalent to 350 μL per well. Cells were incubated for 30 min, after that they were washed three times with PBS. Later 4% PFA was added in an amount equal to 300 μL per well for 45 min, then cells were washed three times with PBS and put on DAPI. To visualize the cells morphology the contrast phase photos was taken (Zeiss, Oberkochen, Germany).



F: forward; R: reverse; p53: tumor suppressor p53; BCl-2: B-cell lymphoma; BAX: Bcl-2 associated X protein; p21: cyclin dependent kinase inhibitor 1A; Cas-9: Caspase-9; GAPDH: Glyceraldehyde 3-phosphate dehydrogenase.

#### **3. Results**

#### *3.1. Structural Analysis*

The β-KYP2O7 crystallizes in monoclinic system that belongs to the *P21*/*c* space group and the α-phase crystallizes in orthorhombic system that can be assigned to the *Cmcm* space group. In the matrix Y3<sup>+</sup> ions are substituted by the selected optically active ions RE3<sup>+</sup>, herein meaning Er3<sup>+</sup> and Yb3<sup>+</sup> Figure 1.

**Figure 1.** Projection of the β-KYP2O7 unit cell (**a**) and super cell (**b**) indicating the Y3<sup>+</sup> and P5<sup>+</sup> coordination polyhedra.

The X-ray diffraction patterns were collected for all of the samples (see Figure 2 and Figures S2 and S3). Independently of the dopant concentration and the annealing time, each collected X-ray diffraction pattern shows a match to the theoretical pattern no. 160190 from ICSD. Up to the annealing temperature of 700 ◦C the sample crystalizes in the low temperature phase β-KYP2O7 (see Figure 2b).

**Figure 2.** Representative XRD patterns of 1 mol% Er3<sup>+</sup>, 1 mol% Yb3<sup>+</sup>:KYP2O7 annealed at 600 ◦C for 4, 8 and 12 h (**a**) as well as annealed at 600–800 ◦C for 4 h (**b**).

Above the annealing temperature of 750 ◦C, the high-temperature phase α-KYP2O7 can be observed, matched with the pattern no. 75171 from ICSD. In the case of the annealing temperature from 750 to 800 ◦C, a decrease in the amount of β-KYP2O7 phase can be observed in favor of the high-temperature α-KYP2O7 phase. XRD patterns show presence of the YPO4 phase. Although the peaks from the YPO4 phase (ICSD no. 184543) overlap with β-KYP2O7, one could be noticed that this additional phase manifests itself in an increased intensity of certain peaks, when compared to the β-KYP2O7 theoretical pattern.

Obtained, representative SEM images of the1 mol% Er3+, 1 mol% Yb3<sup>+</sup>: KYP2O7 material, annealed at 600 ◦C for 12 h have been shown in Figure 3 in two different magnifications.

**Figure 3.** Representative SEM images of the 1 mol% Er3<sup>+</sup>, 1 mol% Yb3<sup>+</sup>:KYP2O7, annealed at 600 ◦C for 12 h with different magnifications (**a**) with 3 um scale bar and (**b**) with 1 um scale bar.

#### *3.2. Luminescence Properties*

The emission spectra were measured at room temperature (300 K) with excitation wavelength λexc = 980 nm of the continuous wave (CW) laser power of 1.56 W (see Figure 4 and Figures S4–S6). Measurements were carried out for the samples annealed at two temperatures, 600 and 650 ◦C for 12 h with varying content of the co-dopants. Each of the spectrum consists of five bands, three of them can be assigned as Er3<sup>+</sup> transitions 2H11/2→4I15/2, 4S3/2→4I15/2, 4F9/<sup>2</sup> <sup>→</sup>4I15/<sup>2</sup> observed respectively at 522, 540 and 650 nm. Samples annealed at 650 ◦C globally show more intense emission in comparison to those annealed at 600 ◦C. Within samples with varying content of Er3<sup>+</sup> ions and fixed at 15 mol% concentration of Yb3<sup>+</sup> ions, the highest emission intensity shows the one doped as follows: 1 mol% Er3<sup>+</sup>, 15 mol% Yb3<sup>+</sup>. Among all samples with varying content of Yb3<sup>+</sup> and concentration of Er3<sup>+</sup> fixed at 1 mol%, the most intense emission can be ascribed to the sample with 20 mol% of Yb3+, regardless of the annealing temperature.

**Figure 4.** Representative emission spectra of KYP2O7 doped with *x* mol% Yb3<sup>+</sup> ions and co-doped with 1 mol% Er3<sup>+</sup>under the excitation wavelength λ = 980 nm, *P* = 1.56 W, heat-treated at 650 ◦C for 12 h.

In emission spectra, additional bands at 470 and 480 nm can be observed. These bands can be assigned to transitions occurring in Tm3<sup>+</sup> ions, respectively 1D2→3F4 and 1G4→3H6.

In addition, in the emission spectra a band at the wavelength of 490 nm is noticed and marked with asterisks in Figure 4 and Figures S4–S6. The emission corresponds to the second harmonic generation (SHG) from the excitation source, which is a diode laser λexc = 980 nm.

Decay profiles were measured for the 4S3/2→4I15/<sup>2</sup> transition prominent at 547 nm wavelength. Measurements were employed at room temperature (300 K) for the samples with varying concentration of the dopants, annealed 650 ◦C for 12 h. Each of decay curves was fitted with the double exponential function in Figure 5.

**Figure 5.** Decay time measured for 15 mol% Yb3+, 1 mol% Er3+:β-KYP2O7 annealed at 650 ◦C for 12 h.

Values of the two decay times, the fast component τ<sup>1</sup> and the slow component τ2, are listed in Table 2. Due to the low emission intensity determination of the decay times for the samples with the lowest concentration of dopants were impossible. In case of the fixed value of erbium concentration, the increase in concentration of ytterbium is followed by the elongation of decay times. For higher concentrations of the dopants, reduction in the decay times can be observed. In addition, there is a correlation between the decay times and the intensity of the emission spectra. Those samples with the high intensity of emission can be assigned to the long decay times as well.


**Table 2.** Luminescence decay times for β-KYP2O7 samples annealed at 650 ◦C for 12 h.

*3.3. Metabolic Activity, Morphology and Apoptosis of MC3T3-E1 Osteoblasts and 4B12 Osteoclast Cultured onto KYP2O7:1 mol% Er3*<sup>+</sup>*, x mol% Yb3*<sup>+</sup>

The viability and proliferative rate analysis of osteoblasts cultured onto KYP2O7:1 mol% Er3<sup>+</sup>, 20 mol% Yb3<sup>+</sup> materials showed their beneficial effect on MC3T3-E1 number of cells (Figure 6). The highest proliferative activity of MC3T3-E1 cells was observed when they were exposed to KYP2O7: 1 mol% Er3+doped with 20 mol% of Yb3+in dose 500 μg/mL. Incorporation of 20 mol% Yb3<sup>+</sup> in KYP2O7: 1 mol% Er3<sup>+</sup> resulted in constant metabolic improvement through 120 h culture test. Similar effect

was observed in 4B12 osteoclast cells, which reached the highest metabolic activity after 120 h, when cultured onto KYP2O7:1 mol% Er3<sup>+</sup> doped with 20 mol% Yb3<sup>+</sup> in dose 500 μg/mL. On the basis of mentioned results KYP2O7:1 mol% Er3<sup>+</sup>doped with 20 mol% of Yb3<sup>+</sup>in dose 500 μg/mL was used in further experiments.

**Figure 6.** The viability and proliferative activity of MC3T3-E1osteoblasts(I) and 4B12osteoclast(II) cultured onto (**A**,**D**) 1 mol% Er3<sup>+</sup>, 10 mol% Yb3<sup>+</sup>:KYP2O7; (**B**,**E**) 1 mol% Er3<sup>+</sup>, 15 mol% Yb3<sup>+</sup>:KYP2O7 and (**C**,**F**) 1 mol% Er3+, 20 mol% Yb3<sup>+</sup>:KYP2O7addition after 24 and 120 h.

For analysis of cells morphology the contrast phase pictures was taken (Figure 7). It was found that Yb3<sup>+</sup> in the 500 ug/mL dosage positively affects morphology of both osteoblasts as well as osteoclasts. For analysis of the mitochondrial and actin network of MC3T3-E1 osteoblasts and 4B12 osteoclast, KYP2O7:1 mol% Er3<sup>+</sup> co-doped with 20 mol% Yb3<sup>+</sup> was proceeded. The creation of abundant actin network was observed in MC3T3-E1 osteoblasts when cultured onto KYP2O7:1 mol% Er3<sup>+</sup> co-doped with 20 mol% Yb3<sup>+</sup> when compared to the control group. The cells presented typical for osteoblast round-like shape morphology with a well visible nuclei. Moreover, cells communicated with each other and created a well-developed cell-to-cell network, as shown by the well-developed actin network (Figure 8). Actin network also testifies of increased adhesion of osteoblast when cultured onto KYP2O7:1 mol% Er3<sup>+</sup>, 20 mol% Yb3<sup>+</sup>. In the case of the osteoclasts, we also observed more developed cytoskeleton and actin staining showed well develop actin network when the cells were cultured with KYP2O7:1 mol% Er3<sup>+</sup>, 20 mol% Yb3<sup>+</sup>. Mitochondrial staining revealed that in both cells type i.e., osteoblasts and osteoclasts, dense network around nuclei was created, when the cells were cultured onto KYP2O7:1 mol% Er<sup>3</sup>+, 20 mol% Yb3+. It might suggest thata 500 μg/mL dose significantly promotes mitochondrial biogenesis, which resulted in creation of well-developed mitochondrial network (Figure 8). Moreover, it seems that this dose induces slight apoptosis in MC3T3-E1 osteoblasts while no prominent effect was observed in osteoclast cells. Incorporation of KYP2O7:1 mol% Er3<sup>+</sup>, 20 mol% Yb3<sup>+</sup> into osteoblasts culture resulted in significant up regulation of p21 and Cas-9 mRNA level since BAX transcript was significantly down regulated (Figure 9). It was found that p21 transcript was considerably down regulated in osteoclast cells when cultured onto KYP2O7:1 mol% Er3<sup>+</sup>, 20 mol% Yb3<sup>+</sup> in comparison for control culture.

**Figure 7.** The MC3T3-E1 osteoblasts (**A**) and 4B12 osteoclasts (**B**) morphology visualized in control cells and in cells cultured with KYP2O7 doped with 1 mol% Er3<sup>+</sup> ions co-doped with 20 mol% Yb3<sup>+</sup> ions in dose 500 μg/mL in MC3T3 cells (**C**) and 4B12 cells (**D**) by contrast phase microscope. Magnification ×100, scale bars: 100 μm.

**Figure 8.** The F-actin, DAPI and MitoRed staining. (**A**,**B**,**C)** in upper graphs presented F-actin and DAPI staining of MC3T3-E1 cells; (**D**,**E**,**F**) showed F-actin and DAPI staining of MC3T3-E1 cells with investigated material KYP2O7 doped with 1 mol% Er3<sup>+</sup> ions co-doped with 20 mol% Yb3<sup>+</sup> ions in dose 500 μg/mL; (**G**,**H**,**I**) in upper graphs presented F-actin and DAPI staining off 4B12 cells; (**J**,**K**,**L**) showed F-actin and DAPI staining of 4B12 cells with investigated material KYP2O7 doped with 1 mol% Er3<sup>+</sup> ions co-doped with 20 mol% Yb3<sup>+</sup> ions in dose 500 μg/mL; (**A**,**B**,**C**) in lower graphs presented MitoRed and DAPI staining off MC3T3-E1cells; (**D**,**E**,**F**) showed MitoRed and DAPI staining of MC3T3-E1 cells with investigated material KYP2O7 doped with 1 mol% Er3<sup>+</sup> ions co-doped with 20 mol% Yb3<sup>+</sup> ions in dose 500 μg/mL; (**G,H**,**I**) in lower graphs MitoRed and DAPI staining off 4B12 cells; (**J**,**K**,**L**) showed MitoRed and DAPI staining of 4B12 cells with investigated material KYP2O7 doped with 1 mol% Er3<sup>+</sup> ions co-doped with 20 mol% Yb3<sup>+</sup> ions in dose 500 μg/mL. Scale bars presented in the images obtained using epifluroescent microscope were equal 50 μm.

**Figure 9.** Evaluation of apoptosis in MC3T3-E1 osteoblasts and 4B12 osteoclast. To evaluate apoptosis in cells, the expression of (**A,G**) p21, (**B,H**) Bcl-2, (**C,I**) CAS-9, (**D,J**) p53 (**E,K**) BAX was analyzed. The (**I,L**) BAX:BCL-2 ratio was calculated using relative expression values of both BCL-2 and BAX.

#### *3.4. Expression of Osteogenic and Osteclastogenic Markers in MC3T3-E1 Osteoblasts and 4B12 Osteoclast Cultured onto KYP2O7: 1 mol% Er3*<sup>+</sup>*, 20 mol% Yb3*<sup>+</sup>*in Relation to mTOR and Pi3K Pathway*

Evaluation of the expression of pro-osteogenic markers on mRNA level in MC3T3-E1 osteoblasts showed beneficial effect of KYP2O7:1 mol% Er3<sup>+</sup>, 20 mol% Yb3<sup>+</sup> material on osteogenesis process (Figure 10). It was found that KYP2O7:1 mol% Er<sup>3</sup>+, 20 mol% Yb3<sup>+</sup> promotes in MC3T3-E1 cells expression of RUNX-2 as well as BMP-2 mRNA level, since reduces expression of Coll-1 and ALP transcripts. In turn, it was observed that KYP2O7:1 mol% Er3<sup>+</sup>, 20 mol% Yb3<sup>+</sup> promotes in 4B12 osteoclast expression of PU. 1, which is involved in regulation of beta(3) integrin expression during osteoclast differentiation. Moreover, elevated expression of INTA5 in 4B12 osteoclasts was observed.

**Figure 10.** Comparison of the expression levels of osteogenesis-related genes using quantitative real-time PCR analysis. The expression of (**A**) RUNX-2, (**B**) Coll-1, (**C**) BMP2, (**D**) ALP, and (**E**) OPN in MC3T3-E1 osteoblasts and the expression of (**A**) PU.1, (**B**) INTB3, (**C**) c-fos, (**D**)INTA5 and (**E**) MMP-9 in 4B12 osteoclast cultured onto KYP2O7 doped with 1 mol% Er3<sup>+</sup>, 20 mol% Yb3<sup>+</sup> ions. In lower graphs the expression of (**A,C**) mTOR and (**B,D**) PI3K and (**E**) AKT in MC3T3-E1 cells and 4B12 cells was presented.

The elevated expression of both mTOR as well as Pi3K in MC3T3-E1 osteoblasts was observed when cultured onto KYP2O7:1 mol% Er3<sup>+</sup>, 20 mol% Yb3<sup>+</sup> material in comparison to the control group (Figure 10). Moreover, in 4B12 osteoclast a significant reduction of MMP-9 expression was found together with up regulation of INTA5 transcript. There were no significant differences between mTOR and Pi3K expression in 4B12 osteoclast exposed to KYP2O7:1 mol% Er3<sup>+</sup>, 20 mol% Yb3<sup>+</sup> and control.

#### **4. Discussion**

Obtained X-ray patterns for samples annealed at variety of temperature show decreasing presence of β-KYP2O7 phase in favor of α-KYP2O7 phase beginning at 750 ◦C. Optimal heat treatment parameters were set to be: 600 ◦C, 12 h and 650 ◦C, 12 h. Given parameters allow gratifying emission properties, shown in latter section, for up-conversion process characterization. Further spectroscopic and biological analysis were employed for the samples annealed with aforementioned parameters. Our observations are in accordance with literature data. It has already been proven, by us and others that α-KYP2O7 phase dominates over β-KYP2O7 one above 700 ◦C [14,15]. In reference to annealing time and doping level of β-KYP2O7 no reports were found. However, for up-conversion processes in different matrices doping level of Yb3<sup>+</sup> and Er3<sup>+</sup> ions similar to concentrations stated as optimal in this paper [24,25]. Size and morphology of the KYP2O7:Er<sup>3</sup>+, Yb3<sup>+</sup> powders were estimated using SEM microscopy, in Figure 3 shown are agglomerates (≈ 3 μm) of elongated submicron particles with the shape of flat plates.

On the basis of the emission spectra, the highest intensity emission band can be assigned to the sample with concentration ratio of 1 mol% Er3<sup>+</sup> and 20 mol% Yb3+, when annealed at 650 ◦C for 12 h. For concentrations higher than 1 mol% Er3<sup>+</sup> a decrease in emission band intensity can be observed, due to concentration quenching of activators' emission [26]. Hence, the optimal doping concentration was chosen to be 1 mol% Er3<sup>+</sup>. The samples heavily doped with Yb3<sup>+</sup> ions, where the above-mentioned phenomenon is not observed, show a monotonic increasement of intensity within analyzed concentration range (1–20 mol% Yb3+). Therefore, the optimal doping concentration was chosen to be 20 mol% Yb3<sup>+</sup>. Decay times of analyzed samples show direct correlation with emission spectra. Emission's intensity increasement is followed with decay time elongation. The samples exhibiting concentration quenching deviate from the mentioned trend and consequently reduction in decay time is being observed. Lengthening of the decay time might refer to an occurrence of energy transfer between upconverting ions [27].

Measurements of power dependence (PD) (see Figure S7), shown as a double-logarithmic function of emission intensity versus laser pump power, allow for estimating several absorbed photons vital for up-conversion process occurrence [28]. Results assert a two-photon nature of the 2H11/2→4I15/<sup>2</sup> and the 4S3/2→4I15/<sup>2</sup> transitions at <sup>λ</sup> = 522–540 nm with *n* values varying from 1.8 to 2.0. Therefore, the anti-Stokes emission may occur via two routes: Energy Transfer Up-conversion (ETU) or Excited State Absorption (ESA). ETU is the most efficient one out of all UC processes, as a resemblance to the full resonance is the closest [29].These UC processes are not easy to distinguish by power dependence, owing to the fact that *n* value equals 2 for all cases. Short decay times for samples with minor content of co-dopants may indicate dominance of ESA, while highly doped samples might be favoring ETU, due to their longer decay times. Occurrence of the UC processes can be distinguished also with presence of arise and further prolongation of rise time in decay time function.

Weak emission intensity of the 4F9/2→4I15/<sup>2</sup> transition at <sup>λ</sup> = 650 nm shows that metastable 4F9/<sup>2</sup> state is not being favorably populated. PD measurements, for aforementioned transition, show the *n* value equal to 1.0–1.3, letting us believe that the 4F9/2→4I15/<sup>2</sup> transition is influenced by non-linear, nonradiative process, such as cross relaxation.

Presence of transitions from Tm3<sup>+</sup> seen in emission spectra, may stem from contamination of reactants, herein especially erbium oxide. It is a well-known fact that Tm3<sup>+</sup> ion can play a role of an activator in UC processes, similarly to Er3<sup>+</sup> ions, if matrix is co-doped with Yb3<sup>+</sup> ions. Hence, thulium ions compete as an activator with erbium ions.

The materials dedicated for bone fracture regeneration require specific characteristics including stimulation of bone formation processes as well as inducing matrix formation. The phosphates are well-known for their pro-osteogenic ability; however, KYP2O7 doped with rare earths elements including Er3<sup>+</sup> and Yb3<sup>+</sup> ions were not previously investigated. In this study, we showed that KYP2O7 doped with 1 mol% of Er3<sup>+</sup> and 20 mol% Yb3<sup>+</sup>in dose 500 μg/mL promotes osteoblasts metabolic activity and induces their highest proliferative potential. In previous research using nanometric hydroxyapatites doped with Er3<sup>+</sup> we observed a similar effect; however on stem progenitor cells and olfactory ensheathing cells [30]. Moreover, we observed that KYP2O7:1 mol% Er<sup>3</sup>+, 20 mol% Yb3<sup>+</sup> promotes also proliferative and metabolic activity of osteoclast. Furthermore, the cytoskeleton development including actin formation was noted in MC3T3-E1 osteoblasts as well as 4B12 osteoclasts when they were exposed for KYP2O7:1 mol% Er<sup>3</sup><sup>+</sup>, 20 mol% Yb3<sup>+</sup>. The observed arrangement of actin fibers indicates about fully stretched of cells and this allows us to evaluate the material as biocompatible [31]. Additionally, we observed improved cell-to-cell contact and creation of a well-developed network suggesting beneficial effect of KYP2O7:1 mol% Er<sup>3</sup>+, 20 mol% Yb3<sup>+</sup> on matrix formation. Interestingly, similar to osteoblasts, osteoclasts presented a well-developed cytoskeleton and actin network. The beneficial effect of KYP2O7:1 mol% Er<sup>3</sup>+, 20 mol% Yb3<sup>+</sup> on osteoblasts activity might be associated with improved mitochondrial biogenesis and creation of dense mitochondrial network. The mitochondria morphology and especially their fission and fusion is one of the elements of assessment cells viability, senescence and metabolism [32]. We indicated that examined material improved mitochondria network and not causes their fission what evidence about positive influence of KYP2O7:1 mol% Er3<sup>+</sup>, 20 mol% Yb<sup>3</sup> on cells viability. Together with improved mitochondrial function, we observed that KYP2O7:1 mol% Er3<sup>+</sup>, 20 mol% Yb3<sup>+</sup> negatively affects expression of p21 and Cas-9 on mRNA level. Obtained data indicate on rather neutral role of KYP2O7:1 mol% Er3<sup>+</sup>, 20 mol% Yb3<sup>+</sup> on osteoblasts apoptosis although significant down regulation of BAX transcripts was observed. Moreover, the appearance of nuclei, after staining with DAPI showed that KYP2O7:1 mol% Er3<sup>+</sup>, 20 mol% Yb3<sup>+</sup> not implicates the chromatin condensation and DNA fragmentation, which indicates the lack of induce apoptosis by the examined material and well biocompatible of it [33]. What is more important, the beneficial effect of the material for pro-osteogenic genes expression including RUNX-2 as well as BMP-2 mRNA in MC3T3-E1 cells was observed. Interestingly, at the same time reduced expression of Coll-1 and ALP transcripts was noted. Obtained data clearly indicates on promotion of early markers of osteogenesis expression instead late markers expression. It suggests that KYP2O7:1 mol% Er3+, 20 mol% Yb3<sup>+</sup> might exert a beneficial effect on bone mineralization process and matrix formation. Observed pro-osteogenic effect of KYP2O7:1 mol% Er<sup>3</sup>+, 20 mol% Yb3<sup>+</sup> on MC3T3-E1 osteoblasts might be partially explained by the elevated expression of both mTOR as well as Pi3K signaling pathways. It was previously showed that both mTOR as well as Pi3K are positively associated with bone formation and bone remodeling effect [34]. What is more, we observed that KYP2O7:1 mol% Er3+, 20 mol% Yb3<sup>+</sup> enhanced the expression of BMP-2 and mTOR in MC3T3 osteoblasts. That fact indicates on pro-osteogenic properties of fabricated material as interplay between these two protein was shown to modulate and enhance osteogenesis [35]. Moreover, is worth adding that KYP2O7:1 mol% Er3<sup>+</sup>, 20 mol% Yb3<sup>+</sup>decreased expression of MMP-9. The increased level of this metalloproteinase is typical for osteoporotic bones. So the fact that modulation of the amount of transcripts MMP-9 by KYP2O7: 1 mol% Er3<sup>+</sup>, 20 mol% Yb3<sup>+</sup> affects the restoration of the balance between osteoblasts and osteoclasts in osteoporotic bones.

#### **5. Conclusions**

Optimization of the Potassium yttrium(III) diphosphate(V) synthesis parameters and a degree of doping was reached. Research shows a stable β-KYP2O7 crystallographic structure and gratifying spectroscopic properties, obtained by finding optimal synthesis conditions (such as annealing temperature, annealing time and degree of doping). The heating parameters of 600 and 650 ◦C as well as the heating time of 12 h were considered the best parameters of the synthesis process. Globally the most intense emission was obtained for samples co-doped with 1 mol% Er3<sup>+</sup> and 20 mol% Yb3<sup>+</sup> ions. In addition, the studies were carried out to consider KYP2O7 co-doped with erbium and ytterbium ions, as a future material used in biomedical applications, especially theranostics. Additionally, phosphate KYP2O7 doped with 1 mol% of Er3<sup>+</sup> and 20 mol% Yb3<sup>+</sup> positively affects MC3T3-E1 osteoblasts morphology, proliferative as well as metabolic activity. Although no positive effect in relation to apoptosis was found, KYP2O7:1 mol% Er<sup>3</sup><sup>+</sup>, 20 mol% Yb3<sup>+</sup> significantly promotes expression of early markers of osteogenesis via mTOR as well as Pi3K which sheds a promising light on that system as an agent promoting fracture bone regeneration. Moreover, observed inhibitory effect on osteoclastogenesis suggests the potential beneficial role of KYP2O7:1 mol% Er<sup>3</sup><sup>+</sup>, 20 mol% Yb3<sup>+</sup> in treatment of osteoclast related disorders.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2079-4991/9/11/1597/s1, Figure S1: Results of the dynamic light scattering (DLS) expressed via z-average size parameter and zeta potential measurements for the representative sample KYP2O7:1 mol% Er<sup>3</sup><sup>+</sup>, 1 mol% Yb3<sup>+</sup> heat-treated at 650 ◦C for 12 h; Figure S2: XRD patterns of β-KYP2O7 annealed at 600 ◦C for 12 h with varying content of Yb3<sup>+</sup> ions and fixed 1 mol% Er3<sup>+</sup> (a) as well as with varying content of Er3<sup>+</sup> and fixed 15 mol% Yb3<sup>+</sup> (b); Figure S3: XRD patterns of β-KYP2O7 annealed at 650 ◦C for 12 h with varying content of Yb3<sup>+</sup> ions and fixed 1 mol% Er3<sup>+</sup> (a) as well as with varying content of Er3<sup>+</sup> and fixed 15 mol% Yb3<sup>+</sup> (b); Figure S4: Emission spectra of KYP2O7 doped with *x* mol% Yb3<sup>+</sup> ions and co-doped with 1 mol% Er3<sup>+</sup> under the excitation wavelength λ = 980 nm, annealed at 600 ◦C for 12 h; Figure S5: Emission spectra of KYP2O7 doped with x mol% Er3<sup>+</sup> ions and co-doped with 15 mol% Yb3<sup>+</sup> under the excitation wavelength λ = 980 nm, annealed at 600 ◦C for 12 h.; Figure S6: Emission spectra of KYP2O7 doped with *x* mol% Er3<sup>+</sup> ions and co-doped with 15 mol% Yb3<sup>+</sup> under the excitation wavelength λ = 980 nm, annealed at 650 ◦C for 12 h; Figure S7: Power dependence measurements of the 4F9/2→4I15/<sup>2</sup> (a) and of the 2H11/2, 4S3/2→4I15/<sup>2</sup> (b) for samples KYP2O7 annealed at 650 ◦C.

**Author Contributions:** A.W. and R.J.W. conceived and designed the experiments as well as analyzed all data; M.W. performed the experiments as well as analyzed all data; M.R. and K.K. contributed reagents/materials/analysis tools; K.M. designed the experiments as well as analyzed all data; all authors contributed to the writing of the paper.

**Funding:** This research was funded by the National Science Centre (NCN), grant number 'Preparation and characterization of biocomposites based on nanoapatites for theranostics' (No. UMO-2015/19/B/ST5/01330).

**Acknowledgments:** The authors would like to thank E. Bukowska for performing XRD measurements and D. Szymanski for SEM images.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

*Article*
