Three- and Two-Photon NIR-to-Vis (Yb,Er) Upconversion from ALD/MLD-Fabricated Molecular Hybrid Thin Films

We report blue, green, and red upconversion emissions with strongly angular-dependent intensities for a new type of hybrid (Y,Yb,Er)–pyrazine thin films realized using the atomic/molecular layer deposition thin-film fabrication technology. The luminescence emissions in our amorphous (Y,Yb,Er)–pyrazine thin films of a controllable nanothickness originate from three- and two-photon NIR-to-vis excitation processes. In addition to shielding the lanthanide ions from nonradiative de-excitation, the network of interconnected organic molecules serves as an excellent matrix for the Yb3+-to-Er3+ excitation energy transfer. This suggests a new approach to achieve efficient upconverting molecular materials with the potential to be used for next-generation medical diagnostics, waveguides, and surface-sensitive detectors.


INTRODUCTION
Photon upconversion is a phenomenon where lower energy photons are piled up to create higher energy photons. 1 This can be achieved by using lanthanide ions with suitable energy levels, that is, those that possess energy level ladders with appropriate steps. For upconversion from near-infrared (NIR) to the visible light the most efficient ion is Er 3+ , which alone is capable of upconversion luminescence. When Yb 3+ sensitization for the Er 3+ emission is used, even higher emission efficiencies may be obtained. Another possibility to achieve upconversion is by triplet−triplet annihilation in organic molecules, but then, the conversion is limited to the visible range only. 2 Upconversion materials are currently intensely studied in particular for application in medical diagnostics and imaging to replace the traditional Eu 3+ -and Tb 3+ -based photoluminescence markers. 3−5 Upconversion is preferred over the regular photoluminescence because it allows excitation in the optical transparency window of the tissue, thus preventing the undesirable autofluorescence. Once the autofluorescence can be avoided, there is no need for the time-resolved measurements that are required for the Eu 3+ and Tb 3+ markers.
The current state-of-the-art upconverters are inorganic nanocrystals. Unfortunately, application of such nanocrystals in diagnostics is strongly limited by the availability of right-sized nanocrystals as well as the difficulties in their dispersibility to liquid media and surface functionalization. In comparison with the present inorganic upconverters, molecular metal complexes could offer much higher sensitivities in bioapplications. This is because in the upconverting molecular materials all the optically active sites can be brought to the close proximity with the biologically active ligands, whereas in the inorganic nanocrystals this is the case only for the active sites on the very surface of the crystals. However, a nearly surmountable challenge with the molecular upconverters has been the remarkably strong nonradiative relaxation of excitation in metal complexes caused by the high-energy phonons of the ligands. 6 There are some successful works though, reporting the direct NIR-to-vis upconversion in lanthanide complexes. In these relatively rare examples, the nearly impossible has been made possible by ensuring that (i) each emitter has enough sensitizers in the close vicinity, (ii) the sensitizers' excited states have long enough lifetimes for energy transfer, and (iii) the emitter is protected from the high-energy vibrations. 7 Aboshyan-Sorgho et al. 7 reported Cr 3+ -sensitized NIR-to-vis upconversion from Er 3+ in a molecular trinuclear complex. However, the emission was observed only at low temperatures, where the impact of multiphonon relaxation could be minimized. On the other hand, Suffren et al. 8 as well as Zare et al. 9 realized Cr 3+ -sensitized NIR-to-green Er 3+ upconversion from multinuclear complexes even at room temperature; according to the authors the successive excitation photons were stored by Cr 3+ before transferring to Er 3+ . Blackburn et al. 10 reported room-temperature two-photon upconversion for Tm 3+ -based complexes in d 6 -DMSO solutions, and Xiao et al. 11 for Nd(EDTA) 2 5− , Er(DPA) 3 3− , and Tm(DPA) 3 3− complexes, but only with a very high-power (100 kW) excitation. Hyppanen et al. 12 showed NIR-to-green upconversion for ion-associated Er(TTA) 4 in the presence of the IR-806 molecule as an absorber. Suzuki et al. 13 obtained green-to-UV upconversion in Er(III)tris(8-hydroxyquinoline) at room temperature, and Yang et al. 14 15 showed two-photon NIR-to-green and NIR-to-red upconversions in dinuclear (Yb,Er)-based metal−organic frameworks, but the red emission was highly weakened by the presence of OH − groups. Sun et al. 16 reported even upconversion from NIR to blue in single crystals of [(Y:Er−Yb)(oba)-(ox) 0.5 (H 2 O) 2 ] n . Unfortunately, neither Weng et al. 15 nor Sun et al. 16 gave details of the measurement conditions such as the excitation power or temperature. However, recently Nonat et al. 17 reported NIR-to-green and NIR-to-red room-temperature upconversion from fluorinated dinuclear Er 3+ complexes. They used D 2 O as the solvent and a rather low-power (5 W) laser. There are also a few reports of lanthanide luminescence after upconversion in organic ligands (see, for example, Wong et al. 18 and Luo et al. 19 ) An attractive new way of synthesizing inorganic−organic hybrid materials is provided by the currently strongly emerging atomic/molecular layer deposition (ALD/MLD) thin-film technique. 20−24 The technique has the capacity to yield novel hybrid materials with unique bonding schemes and molecular assemblies not readily accessible with any other synthesis technique. There are already exciting examples of functional properties discovered for such ALD/MLD-fabricated thin films relevant to next-generation thermoelectric, 25,26 Li-ion microbattery, 27 encapsulation, 28,29 and metal−organic framework 30,31 applications. Most importantly, the ALD/MLD fabrication of lanthanide-based inorganic−organic photoluminescence thin films has also been demonstrated. 32,33 The ALD/MLD technique is derived from the state-of-theart gas-phase thin-film deposition technique of simple inorganic materials, that is ALD, developed for industrial-scale applications already in 1970s. 34 An example of the industrial feasibility of ALD is the fabrication of ultrathin HfO 2 gate dielectric coatings for microprocessors. For the inorganic− organic thin films, ALD cycles are combined with MLD cycles based on an organic precursor. This enables the atomic/ molecular layer-by-layer growth of inorganic−organic hybrid thin films through sequential self-limiting gas−surface reactions of gaseous precursor pulses with high precision for the film thickness and composition. The ALD/MLD technology has several other inherent advantages highly relevant to advanced upconversion applications in e.g. medical diagnostics, such as the possibility to use the technique for (sub)nanoscale surface functionalization of various complex surface architectures, 35,36 and low-temperature deposition of flexible hybrid coatings on sensitive or demanding substrates such as polymers, textiles, 37 or biomaterials. 38 Very recently, we reported the ALD growth of highly homogeneous upconverting inorganic (Yb,Er) 2 O 3 thin films. 39 In the present work, we demonstrate for the first time the NIR-to-vis lanthanide upconversion emission for ALD/MLD-fabricated molecular inorganic−organic thin films. We chose pyrazine as the ligand for the (Y,Yb,Er) complex because it has been shown to form stable complexes 40 and upconverting metal−organic frameworks 41 with lanthanides. Most importantly it is revealed that (i) the emission shows all the three red-green-blue (RGB) main colors including the most-difficult-to-obtain blue color, (ii) these are achieved with both three-and two-photon processes, and (iii) the emission intensity is highly angle-dependent. The latter observation is extremely intriguing for application in ultrasensitive fluorescence assays and surface sensors for point-of-care diagnostics. 42 The thin-film samples were fabricated in a commercial ALD reactor (F-120 by ASM Microchemistry Ltd.) from our in-house synthesized 30,47 mixed (Y 0.92 Yb 0.04 Er 0.04 )(thd) 3 metal precursor and commercial 2,3-pyrazinedicarboxylic acid powder (Sigma-Aldrich) as the organic precursor. During the depositions, the (Y,Yb,Er)(thd) 3 and pyrazinedicarboxylic acid precursor powders were kept in glass crucibles inside the reactor at 130 and 145°C, respectively. Nitrogen (>99.999%; Schmidilin UHPN 3000 N 2 generator) was used as a carrier and purging gas, and a pressure of 2−4 mbar was maintained in the reactor during the film deposition. The following ALD/MLD precursor pulse/purge cycle was repeated for 150 times: 1.5 s (Y,Yb,Er)(thd) 3 /2 s N 2 /2 s pyrazine/4 s N 2 . The depositions were carried out at different temperatures in the range of 160−275°C on various substrates: silicon (100), DuPont polyimide, quartz, and nanocellulose.
2.2. Characterization. Thickness, density, and roughness of the films were determined by X-ray reflectivity (XRR; PANalytical X'Pert MPD PRO Alpha 1) measurements using X'Pert HighScore Plusreflectivity. Grazing incidence X-ray diffraction (GIXRD) measurements were done using the same instrument. The organic content was confirmed and the bonding scheme was investigated using Fourier transform infrared spectroscopy (FTIR; Nicolet Magna-IR spectrometer 750), where an average of 32 scans with 4 cm −1 resolution was applied for each sample. The samples were also characterized with UV−visible absorption spectroscopy (PerkinElmer LAMBDA 950 UV/vis/NIR absorption spectrophotometer). Surface morphology was investigated with atomic force microscopy (AFM; Veeco Dimension 51000 scanning probe microscope; NanoScope Controller, Digital Instruments, Inc.; NanoScope Analysis 1.5 software). The Y:Yb:Er content was confirmed from X-ray fluorescence spectrometer (PANalytical AxiosmaX with a Rh tube, 3 kW model) data.
The upconversion measurements were carried out using AvaSpec-HS-TEC CCD spectrometer and Optical Fiber Systems IFC-975-008 NIR laser (6 W, 974 nm). In the excitation path, a 900 nm long-pass filter (Edmund Optics) was used to cut off wavelengths lower than NIR. The emitted light was collected at 90°angle to the excitation and directed through a 900 nm filter (Newport, 10SWF-900-B) to exclude the scattered excitation radiation. An optical fiber with 600 μm diameter was used as an emission light path between the sample compartment and the detector. All spectra were measured at room temperature.

RESULTS AND DISCUSSION
First, we demonstrate that our ALD/MLD process for the (Y,Yb,Er)−pyrazine thin films from (Y,Yb,Er)(thd) 3 and pyrazinedicarboxylic acid precursors works in the expected ideal atomic/molecular layer-by-layer deposition mode. This can be seen from Figure 1a,c, where we show that the growth per cycle (GPC) remains essentially constant independent of the precursor pulse lengths, and the film thickness increases in a linear manner with increasing number of ALD/MLD cycles. These experiments were carried out on silicon substrates at 160°C . We also investigated higher deposition temperatures up to 275°C. From Figure 1c, it is seen that the GPC remains essentially constant up to 225°C, and then, it slightly declines

ACS Applied Materials & Interfaces
Research Article for 250−275°C possibly due to (partial) decomposition of the pyrazinedicarboxylic acid precursor. At all the deposition temperatures, the resultant thin films were visually homogeneous. Thus, we may conclude that our (Y,Yb,Er)(thd) 3 + pyrazinedicarboxylic acid ALD/MLD process yields highquality thin films through self-limiting gas−surface reactions of the precursors with the selected ALD/MLD parameters in the deposition temperature range of 160−225°C. For the rest of the depositions, we selected the following precursor/purge sequence: 1.5 s (Y,Yb,Er)(thd) 3 /2 s N 2 /2 s pyrazine/4 s N 2 and fixed the deposition temperature to 160°C, which is low enough for the temperature-sensitive substrates as well. With these parameters, the GPC value was found to be 3.4 Å/cycle such that, for example, 60 and 150 ALD/MLD cycles resulted in ca. 20 and 50 nm thick films, respectively; these 20−50 nm thick films were investigated in all our further experiments. From XRR and AFM measurements (not shown here), the roughness of the films was determined to be 0.2 nm; the density value calculated from the XRR data was 2.0 g/cm 3 . From GIXRD, the films were found amorphous up to 250°C; only raising the deposition temperature to 275°C resulted in some crystallinity, see Figure 1d.
Next, we discuss the chemical state of our (Y,Yb,Er)− pyrazine thin films based on an FTIR investigation (see Figure  2a,b, where the spectrum recorded for the hybrid thin film is compared to that recorded for the pyrazinedicarboxylic acid precursor). The features seen in the latter spectrum are all wellexplained by the literature data for 2,3-pyrazinedicarboxylic acid; 48 the assignment of the major features is shown in Figure  2a 51 Another important observation is that instead of the CN bands, observed in the pyrazinedicarboxylic acid precursor at 1577− 1445 cm −1 , only one band is observed at 1448 cm −1 for the hybrid thin film, suggesting that the metal cations most likely are bonded to the pyrazine molecule via nitrogen atoms as well. Moreover, the stretching mode v(CH) at 3096 cm −1 as well as the bending modes β(CH) of the aromatic ring observed at 1208, 1182, 994, and 868 cm −1 in the spectrum of the 2,3pyrazinedicarboxylic acid are missing in the spectrum of the hybrid thin film. These changes in the charge distribution of the aromatic ring are due to 2,3-pyrazinedicarboxylic acid−metal interactions. 52 On the basis of all the aforementioned observations, we present a possible 2D-bonding scheme for our hybrid (Y,Yb,Er)−pyrazine thin films in Figure 2d.
The UV−vis absorption spectrum obtained for the films is shown in Figure 2c together with that for the pyrazinedicarboxylic acid precursor. The pure precursor spectrum shows the well-known transitions at 212 nm due to H-2-L (68%) and at 268 and 318 nm due to H−L (96%) and (H−L + 1 (96%)) transitions, respectively. 48 Upon the (Y,Yb,Er)−pyrazine complex formation, the absorption wavelength shifts from 268 to 290 nm; similar shifts have been seen in related metal complexes and ascribed to ligand-to-metal charge transfer. 53−56 Additionally, new weak peaks appear at 1384 and 2210 nm; these most probably originate from the metal−ligand interaction. As for the characterization of the upconversion properties, we first investigated several hybrid thin films grown on silicon, which is a rigid substrate thus allowing easy upconversion characterization. We initially prepared a series of thin films with different thicknesses from 20 to 50 nm to check the possible effect of the film thickness on the upconversion; within this thickness range the upconversion intensity was confirmed to be essentially independent of the film thickness. The films were excited with a 974 nm radiation, and they showed green and red upconversion emissions that are due to the 2 H 11/2 , 4 S 3/2 → 4 I 15/2 and 4 F 9/2 → 4 I 15/2 transitions of Er 3+ , respectively ( Figure   3a). The fact that the films show excellent upconversion emission intensity indicates that the ALD/MLD process has enabled the build-up of a structure with close Yb 3+ −Er 3+ pairs

ACS Applied Materials & Interfaces
Research Article enabling the efficient energy transfer and a shielding of the energy transfer and relaxation processes from the vibrations of the ligands. As the (Y,Yb,Er)−pyrazine thin films were found to exhibit a weak absorption band in the NIR range (Figure 2b), we tentatively believe that the ligands may absorb the 974 nm excitation and transfer it to Yb 3+ and Er 3+ , thus enhancing the upconversion process further.
A striking new feature for the (Yb,Er) upconversion in molecular compounds is that the present films also show blue emission of Er 3+ due to the 2 H 9/2 → 4 I 15/2 transition ( Figure  3a). To the best of our knowledge, there are no reports on NIR-to-blue emissions in amorphous molecular compounds yet. This is because the blue lines require a pile-up of three photons in the upconversion process (Figure 3b), and its probability is much lower than that of the two-phonon processes. In molecular complexes, the probability is made even lower because of the vibrations of the ligands that act as efficient pathways for multiphonon de-excitation.
The upconversion emission intensity is known to be dependent on the pump excitation power, 57 and therefore, we carried out the pump power measurements. Moreover, the laser wavelength is dependent on the temperature of the laser, and therefore, the temperature affects the effectiveness of excitation. To account for this, we measured a few spectra per each pump power using different cooling powers for the laser. Upconversion is a multiphoton absorption process, which means that the upconversion intensity and pump power have an exponential dependence. 58 At low excitation powers, a linear fitting of the log−log dependence between the intensity and pump power will give a slope, which corresponds to the number of photons needed for the upconversion process. As shown in Figure 3c, for our (Y,Yb,Er)−pyrazine thin film, the slope of the blue emission is 2.7, which indicates a three-photon process, as expected. For the green emission, the slope is 1.7, which is in agreement with a two-photon process (see, for example, ref 59). For red emission, the slope is 2.3 which may indicate a three-photon process with some leaking pathways during the photon pile-up, or it may be a two-photon process as discussed in the literature. 60−63 The plot of NIR-to-blue emission also shows a decrease of slope at the highest pump excitation powers. This is caused by an excitation powerinduced change in the net excited-state dynamics because of increasing excitation photon density. 64 The blue emission can be thought to be more strongly affected by this than the green or red ones, because it requires more energy-climbing steps to succeed, that is, small changes in the first two steps accumulate to bigger ones in the third step. The saturation of sensitized upconversion emission has been reported and discussed in more detail in earlier literature, (see, for example, refs 64,65 ). All in all, as proven by Suyver et al., 62 the saturation indicates that the upconversion emission in our films is not due to the direct excitation of Er 3+ but with a process involving sensitization via Yb 3+ ions.
We also investigated the red/green upconversion emission intensity ratios for the power series of our (Y,Yb,Er)−pyrazine films. The results show that the red emission intensity increases relatively more than the green emission when the excitation power increases, that is, we observed a linear rise from the 0.43 red/green ratio at 0.23 W to 0.90 at 0.79 W with a slope of 0.85. This is essential, as it is known that the red emission is more important than the green one in bioapplications owing to its better ability to penetrate tissues. 66,67 Because we observed no excitation power-dependent saturation for neither emission, this observed increase in the red/green ratio is due to the increase of the temperature along the increasing excitation power; as can be seen from the energy level diagram depicted in Figure 2d, the energy differences between the 2 H 11/2 , 4 S 3/2 , 4 F 9/2 , and 4 I 11/2 levels are much smaller than those between 4 I 11/2 , 4 I 13/2 , and 4 I 15/2 . Therefore, increasing temperature will increase the multiphonon de-excitation probability of the greenemitting 2 H 11/2 and 4 S 3/2 levels much more efficiently than that of the red-emitting 4 I 11/2 level.
We also tested the dependence of the incident angle of the laser beam on the upconversion emission intensity. Because of the finite thickness of thin films, they will show an excitation angle dependence of emission intensity because of effects such as excitation photon density and penetration depth that have no influence in bulk luminescent materials. Angular dependencies for thin film luminescence have been presented in refs 68−70 and typically the intensity varies with angle, but there is still emission from all angles. However, for our film, the upconversion emission is obtained only with incident angles between ca. 20 and 50°, as demonstrated in Figure 3d. Such a narrow angular range is something that is typical of surface plasmon-coupled emission. 71 Thanks to this narrow angular range with or without possible surface plasmon contribution as we envision that our films could be used as waveguides and surface-sensitive detectors.
Finally, inspired by the excellent results for the films deposited on silicon, we fabricated (Y,Yb,Er)−pyrazine thin films on transparent and flexible polyimide and nanocellulose substrates. Also, these films show the green and red upconversion emissions (Figure 4). The log−log dependence of the intensity and pump excitation power present slopes of 1.4 (green) and 1.5 (red) for the films on the polyimide substrate and 1.3 (green) and 2.1 (red) for those on the nanocellulose substrate (not shown here). Thus, both upconversion processes can be taken to be two-photon

ACS Applied Materials & Interfaces
Research Article processes. Accordingly, the red/green ratios for the cases of the polyimide and nanocellulose substrates are somewhat lower than those for the films grown on silicon. Tentatively, we believe that the small difference could be just due to slightly different random distributions of the Yb 3+ and Er 3+ cations. Nevertheless, the results confirm the excellent upconversion performance of our molecular (Y,Yb,Er)−pyrazine upconverting thin films on different surfaces enabled by the ALD/MLD fabrication protocol.

CONCLUSIONS
Molecular metal complex upconverters would be exciting material candidates for application in medical diagnostics to replace the traditional photoluminescence markers as they allow excitation in the optical transparency window of the tissue such that the problematic autofluorescence of the tissue can be avoided. However, the nonradiative relaxation of excitation caused by the high-energy phonons of the ligands in typical metal complexes so far investigated has almost remained a surmountable challenge for the realization of such otherwise ideal upconverting materials.
In this work, we have demonstrated that our new type of (Y,Yb,Er)-based hybrid inorganic−organic thin-film materials synthesized using the strongly emerging ALD/MLD technique can overcome the aforementioned challenge. We showed that these hybrid thin films allow both two-and three-photon processes to yield upconversion luminescence emissions with all the three RGB main colors including the most-difficult-toobtain blue color. Also very interestingly, the emission intensities were found to be highly angle-dependent.
Although further studies are definitely required, the novel inorganic−organic thin-film materials precisely fabricated using the ALD/MLD technique could open up interesting application possibilities in bioanalytics and medical diagnostics; note that the ALD/MLD technique inherently brings several advantageous properties for the films including the excellent integratibility with various substrate surfaces as demonstrated in the present work. Our proof-of-the-concept results are for (Y,Yb,Er)−pyrazine thin films. We, however, envision that further tuning/optimization of the upconversion characteristics could be achieved by systematically mapping various organic constituents in the (Y,Yb,Er)−organic complex.