Infrared Photoluminescence of Nd-Doped Sesquioxide and Fluoride Nanocrystals: A Comparative Study

Abstract

Lanthanide ions possess various emission channels in the near-infrared region that are well known in bulk crystals but are far less studied in samples with nanometric size. In this work, we present the infrared spectroscopic characterization of various Nd-doped fluoride and sesquioxide nanocrystals, namely Nd:Y₂O₃, Nd:Lu₂O₃, Nd:Sc₂O₃, Nd:YF₃, and Nd:LuF₃. Emissions from the three main emission bands in the near-infrared region have been observed and the emission cross-sections have been calculated. Moreover, another decay channel at around 2 μm has been observed and ascribed to the ⁴F_3/2→⁴I_15/2 transition. The lifetime of the ⁴F_3/2 level has been measured under LED pumping. Emission cross-sections for the various compounds are calculated in the 1 μm, 900 nm, and 1.3 μm regions and are of the order of 10⁻²⁰ cm² in agreement with the literature results. Those in the 2 μm region are of the order of 10⁻²¹ cm².

1. Introduction

Lanthanide-doped nanocrystals are widely studied systems for their visible emission features thanks to their unparalleled advantages over other types of materials such as their excellent thermomechanical properties and chemical stability, the large Stokes shift and sharp emission lines, and their long emission lifetimes. In particular, upconverting nanocrystals have received great attention for many different applications in the biomedical field, such as biomedical imaging, drug delivery, and photodynamic therapy [1,2,3], as well as for thermometric measurements [4,5] and for security applications [6], just to name a few.

Lanthanide ions also possess many efficient near-infrared emission transitions that have been exploited for laser emission in bulk crystals [7]. The possibility to exploit the infrared emission of lanthanide-activated nanocrystals can determine a paradigm shift for some applications and can also open the way to a lot of new types of applications, for example, for deep tissue imaging [8], image-guided surgery [9], and forensic science [10]. For example, nanocrystals with infrared emission have added values for biomedical applications such as the reduction of tissue absorption, light scattering, and autofluorescence. Among the various proposed materials, lanthanide nanocrystals with their intriguing emission properties are among the most promising materials. Moreover, Nd shows some very intense emissions in various infrared regions at around 900 nm, 1064 nm, and 1300 nm. All these emissions come from the decay from the ⁴F_3/2 to the lower-lying ⁴I_9/2, ⁴I_11/2, and ⁴I_13/2 and have been widely exploited even for laser emission, but Nd ions also possess a weaker emission band at around 2 μm that has rarely been observed even in bulk crystals.

Sesquioxides are an important class of oxide crystals that possess good thermal and physical properties, have relatively low phonon energy compared with other oxides, and can be grown to good quality [11]. Unfortunately, the high temperature required for the growth of this class of materials as single crystals (around 2400 °C) makes this process quite demanding [12]. For this reason, the same compositions have been produced in fiber, ceramic, or nanopowder form. Y₂O₃ is probably the most widely studied sesquioxide when doped with Nd as bulk crystal [13], single crystal fiber [14], ceramic [15,16,17], and nanocrystals [18,19], but also other isomorphs such as Lu₂O₃ [20,21] and Sc₂O₃ [22,23] have shown very interesting emission properties when doped with Nd. In general, the focus of the spectroscopic investigations of these materials is limited to the visible absorption bands and to the main emission channel at around 1 micron, for which there is some inconsistency among the published values of the stimulated emission cross-section, especially when estimated with different techniques. Moreover, Nd also possesses other interesting emission channels at around 900 nm and 1300 nm from which even laser emission has been obtained [14], but very few reports of the emission cross-sections in these regions can be found in the literature. Last but not least, the emission at around 2 μm has never been reported to the best of the authors’ knowledge.

Fluoride crystals are considered the preferred choice for emissions in the near-infrared, thanks to their good thermomechanical properties combined with low-phonon energy values, but the bulk crystal growth of this class of materials is complicated due to the high purity needed both for the starting chemicals and for the growth atmosphere.

Synthesis of these materials in nanometric form is accomplished by a polymer complex solution technique (oxides) and a low-temperature, solid-state method (fluorides) to study the infrared emission properties of these materials.

2. Materials and Methods

For syntheses of materials, the following chemicals were used: Y₂O₃ (Alfa Aeser, 99.99%), Sc₂O₃ (Alfa Aeser, 99.99%), Lu₂O₃ (Alfa Aeser, 99.99%), Nd₂O₃ (Alfa Aeser, 99.9%), polyethylene glycol (molecular weight 200, Alfa Aeser), nitric acid (HNO₃, Macron, 65%), and ammonium hydrogen difluoride (NH₄HF₂, Sigma−Aldrich, 98.5%). Nd-doped sesquioxide nanocrystals were prepared by the polymer complex solution method as previously described [24,25]. In brief, the stoichiometric ratio of oxide precursors was dissolved in a hot nitric acid at 130 °C until reaching the completely transparent solution. Then, the polyethylene glycol was added to the solution at a mass ratio of 1:1 to the mass of oxides. The solution was stirred at 80 °C until the nitrate gasses dissipated and a clear gel was formed. The gel was pre-sintered for 2 h at 800 °C in a ceramic crucible to produce a voluminous white powder, which was subsequently formed into pellets and calcined for 24 h at 1100 °C. Nd-doped fluorides were prepared by a low-temperature, solid-state synthesis accompanied by fluorination, as previously described [26]. In brief, the appropriate amounts of oxides were mixed with NH₄HF₂, thoroughly ground in an agate mortar to ensure homogeneity, and then heated in two steps, in the air at 170 °C for 20 h and in the reducing atmosphere (Ar−10% H₂) at 500 °C for 3 h.

The structure of the obtained nanomaterials was checked by X-ray powder diffraction (XRD) using the Rigaku SmartLab device (measurement settings: Cu-Kα_1,2 radiation, λ = 0.1540 nm, ambient temperature, 2θ range 10–90°, measurement step 0.02°, and counting time 1 min/°). Scanning electron images were acquired by a field emission TESCAN MIRA3 microscope. Diffuse spectral reflectance measurements were performed on the FEI TECNAI G2 X-TWIN microscope. Measurements of diffuse reflection spectra were performed on a Thermo Evolution 600 spectrometer equipped with an integrating sphere and using the BaSO₄ spectrum as a white standard.

For infrared emission measurements, the sample was pumped by an 808 nm diode laser with about 400 mW output power. The emitted luminescence was collected by a parabolic mirror and was sent to an FTIR spectrometer (Magna860, Nicodom Ltd., Praha, Czech Republic) equipped with an MCT cooled detector. The resolution of the emission measurements was set to 1 cm⁻¹. All the spectra were corrected for the spectral response of the system using a blackbody source. Lifetimes of excited states were acquired after LED pumping at around 520 nm. The emission was collected by a lens, filtered by suitable filters to cut spurious pump light, and then sent to a fiber-coupled Si detector (OE-200-UV, Femto, Berlin, Germany). The amplification factor of the detector was 10⁹ in high-speed mode, so that the response time of the system was 17 μs.

3. Results

XRD patterns shown in Figure 1a confirm that the crystal structures of prepared sesquioxides are cubic bixbyite, space group Ia-3, and for prepared fluoride nanocrystals, it is orthorhombic, space group P_nma. No reflections belonging to impurity phases were observed. The average particle sizes of sesquioxides are around 350 nm (Figure 1b) and around 500 nm in fluorides (Figure 1c).

3.1. Visible and Near-Infrared Spectroscopy

Diffuse reflection spectra of Nd-doped sesquioxides and fluorides are shown in Figure 2a,b, respectively. Measurements reveal typical absorptions of trivalent Nd located in low-energy phonon hosts, among which the strongest absorption around 800 nm is due to electronic transitions to ⁴F_5/2 and ²H_9/2 from the ground state.

3.2. Sesquioxides

All sesquioxide samples show four emission bands that are composed of a series of well-separated peaks, as expected by the strong crystal field of these crystal matrixes [27]. The first band extends from 11,500 cm⁻¹ to 10,000 cm⁻¹ and corresponds to the ⁴F_3/2 → ⁴I_9/2 transition, the second extends from 9600 cm⁻¹ to 8600 cm⁻¹ and corresponds to the ⁴F_3/2 → ⁴I_11/2 transition, and the third extends from 7800 cm⁻¹ to 6700 cm⁻¹ and corresponds to the ⁴F_3/2 → ⁴I_13/2 transition. The peak position agrees with the energy level position reported in the literature [27]. Spectra are very similar among the various compositions. In fact, we can notice a strong similarity in the shape of these emission spectra, with only a small shift of the emission features and small differences in the relative emission intensity among the three compounds. This is not unexpected since Y₂O₃, Sc₂O₃, and Lu₂O₃ are isomorphs. When going from Y₂O₃ to Lu₂O₃ and to Sc₂O₃, the emission features experience a tendency to redshift that is more pronounced for the longest wavelength emission peaks within each band. This can be ascribed to the increasing crystal field strength in the three compounds [20]. The strongest peaks of the first band are located at 10,560 cm⁻¹ (947 nm) in Y₂O₃, 10,240 cm⁻¹ (977 nm) in Lu₂O₃, and 10,350 cm⁻¹ (966 nm) in Sc₂O₃. As usual for Nd-doped compounds, the strongest emission band is the one located at around 1 micron with maxima at 9265 cm⁻¹ (1079 nm) for Y₂O₃, 9253 cm⁻¹ (1081 nm) for Lu₂O₃, and 9237 cm⁻¹ (1083 nm) for Sc₂O₃. The maxima of the 1.3 μm band are located at 7363 cm⁻¹ (1358 nm) for Y₂O₃, 7352 cm⁻¹ (1360 nm) for Lu₂O₃, and 7311 cm⁻¹ (1368 nm) for Sc₂O₃. Moreover, in all cases, we were able to observe a fourth emission band in the 2 μm region that extends from about 4500 cm⁻¹ to about 6000 cm⁻¹. This band is usually considered very weak, and the emission has rarely been reported in the literature, even in bulk crystals. As for the other bands, also in this region, the shapes of the spectra look very similar for the three compounds with a tendency to red-shifting when passing from Y₂O₃ to Lu₂O₃ and to Sc₂O₃. The highest peaks are located at 4800 cm⁻¹ (2083 nm) for Y₂O₃, 4760 cm⁻¹ (2101 nm) for Lu₂O₃, and 4632 cm⁻¹ (2159 nm) for Sc₂O₃.

From the emission spectra, we calculated the emission cross-section of the ⁴F_3/2 → ⁴I_i (i = 9/2, 11/2, 13/2, 15/2) emission bands with the following equation [28]:

$${\sigma }_{em}\left(\nu \right)=\frac{c^{2}I\left(\nu \right)}{8\pi \tau n^{2}h{\nu }^3\int \frac{I\left(\nu \right)}{h\nu }d\nu }$$

(1)

where c is the speed of light in vacuum, $h$ is Planck’s constant, $I\left(\nu \right)$ is the fluorescence signal, and $n$ and $\tau$ are the crystal refractive indexes at 1 μm wavelength and the radiative lifetime, respectively, both taken from the literature as reported in

Table 1. Parameters used for cross-section calculation.

Compound	n	Ref.	τ (μs)	Ref.
Nd:Y2O3	1.90	[29]	318	[18]
Nd:Lu2O3	1.91	[30]	300	[31]
Nd:Sc2O3	1.97	[32]	230	[33]
Nd:YF3	1.45	[34]	783	[35]

for the various compounds. For LuF₃, we could not find proper references to published values; therefore, we used the values of the isomorph compound YF₃. In Equation (1), the integral is over the whole emission region of the ⁴F_3/2 decay channels, including the 2 μm emission band. It is worth mentioning that we performed all the calculations in the frequency domain using Equation (1), because the experimental data were acquired with an FTIR that works at fixed wavenumber intervals, instead of using the equivalent expression in wavelength, as reported in Equation (14) of ref [28] that must be used when working with grating spectrometers.

Figure 3a–c show the emission cross-sections of all the Nd-doped sesquioxides in the 11,500 cm⁻¹–6500 cm⁻¹ region measured at 1%Nd doping level for oxides and 5%Nd doping level for fluorides because these were the samples with the highest emission intensities. In this region, we can distinguish the three main emission bands. The shape and peak position of the various bands qualitatively agree with published results, when available, and the cross-section peak intensities we obtained are compared with the literature results in Table 2, Table 3, Table 4, Table 5 and Table 6. It is evident that large discrepancies are present among the literature results, especially for the most studied of these compounds, such as Nd:Y₂O₃, where many different estimates are present. Our results compare well with the variation interval of published values. In all cases, the highest emission cross-section is that of the ⁴F_3/2 → ⁴I_11/2 transition, and our calculations for this band are in good agreement with published results. The emission cross-section of the other decay channels is not always known in the literature, and when present, our results compare well with published values.

It may be worth noting that these results are similar or slightly lower than the emission cross-section of well-known laser crystals. For example, the maximum emission cross-section of YLF is about 2 × 10⁻²⁰, 18 × 10⁻²⁰, and 3 × 10⁻²⁰ cm² for the ⁴F_3/2 → ⁴I_9/2, ⁴F_3/2 → ⁴I_11/2, and ⁴F_3/2 → ⁴I_13/2 transitions, respectively [36].

Table 2. Emission cross-sections of Nd:Y₂O₃.

	Decay Channel	σem (10−20 cm2)
	4F3/2 →	This work	[14]	[18]	[16]	[15]	[17]	[13]
Nd:Y2O3	4I9/2	2.4	-	-	-	-	4.89	1.8
	4I11/2	7.3	6.9	1.73	7.24	5.13	6.35	6.8
	4I13/2	1.5	5.5	-	-	-	0.92	-
	4I15/2	0.07	-	-	-	-	-	-

Table 2. Emission cross-sections of Nd:Y₂O₃.

	Decay Channel	σem (10−20 cm2)
	4F3/2 →	This work	[14]	[18]	[16]	[15]	[17]	[13]
Nd:Y2O3	4I9/2	2.4	-	-	-	-	4.89	1.8
	4I11/2	7.3	6.9	1.73	7.24	5.13	6.35	6.8
	4I13/2	1.5	5.5	-	-	-	0.92	-
	4I15/2	0.07	-	-	-	-	-	-

Table 3. Emission cross-sections of Nd:Lu₂O₃.

	Decay Channel	σem (10−20 cm2)
	4F3/2 →	This work	[20]	[21]	[27]	[37]
Nd:Lu2O3	4I9/2	2.4	-	-	1.9	-
	4I11/2	5.9	8.49	6.5	5.0	6.5
	4I13/2	1.3	-	-	3.1	-
	4I15/2	0.04	-	-	-	-

Table 3. Emission cross-sections of Nd:Lu₂O₃.

	Decay Channel	σem (10−20 cm2)
	4F3/2 →	This work	[20]	[21]	[27]	[37]
Nd:Lu2O3	4I9/2	2.4	-	-	1.9	-
	4I11/2	5.9	8.49	6.5	5.0	6.5
	4I13/2	1.3	-	-	3.1	-
	4I15/2	0.04	-	-	-	-

Table 4. Emission cross-sections of Nd:Sc₂O₃.

	Decay Channel	σem (10−20 cm2)
	4F3/2 →	This work	[33]
Nd:Sc2O3	4I9/2	3.2	-
	4I11/2	9.7	9.5
	4I13/2	3	-
	4I15/2	0.09	-

Table 4. Emission cross-sections of Nd:Sc₂O₃.

	Decay Channel	σem (10−20 cm2)
	4F3/2 →	This work	[33]
Nd:Sc2O3	4I9/2	3.2	-
	4I11/2	9.7	9.5
	4I13/2	3	-
	4I15/2	0.09	-

Table 5. Emission cross-sections of Nd:YF₃.

	Decay Channel	σem (10−20 cm2)
	4F3/2 →	This work	[35] *
Nd:YF3	4I9/2	0.3	0.51
	4I11/2	4.9	0.74
	4I13/2	0.9	0.4
	4I15/2	0.06	0.032

* calculated.

Table 5. Emission cross-sections of Nd:YF₃.

	Decay Channel	σem (10−20 cm2)
	4F3/2 →	This work	[35] *
Nd:YF3	4I9/2	0.3	0.51
	4I11/2	4.9	0.74
	4I13/2	0.9	0.4
	4I15/2	0.06	0.032

* calculated.

Table 6. Emission cross-sections of Nd:LuF₃.

	Decay Channel	σem (10−20 cm2)
	4F3/2 →	This work
Nd:LuF3	4I9/2	0.2
	4I11/2	4.7
	4I13/2	1.1
	4I15/2	0.1

Table 6. Emission cross-sections of Nd:LuF₃.

	Decay Channel	σem (10−20 cm2)
	4F3/2 →	This work
Nd:LuF3	4I9/2	0.2
	4I11/2	4.7
	4I13/2	1.1
	4I15/2	0.1

The stimulated emission cross-section for the ⁴F_3/2→⁴I_15/2 decay is depicted in Figure 4a–c for all investigated compounds. This transition appears as a series of separated groups of peaks of increasing intensity. The highest emission cross-section is observed at around 2.1 μm in all compounds.

We also measured the ⁴F_3/2 decay time under LED pumping on 3% and 1% doped samples. The decay profile is always exponential and lifetime values measured on 1% doped samples are reported in

Table 7. Decay time of 1%Nd:Y₂O₃.

		τ (μs)
		This work	[13]	[18]	[16]	[15]	[14]	[17]
Nd:Y2O3		320	300		321	232	340	315
	Radiative		378	318	322			354

,

Table 8. Decay time of 1%Nd:Lu₂O₃.

		τ (μs)
		This work	[20]	[31]	[38]
Nd:Lu2O3		420	286	300
	Radiative		344		165

and

Table 9. Decay time of 1%Nd:Sc₂O₃.

		τ (μs)
		This work	[33]	[39]	[40]
Nd:Sc2O3		335	180	224	260
	Radiative		344

and compared with the literature values on low concentration samples, whenever available. On higher doped samples, concentration quenching effects make the lifetime shorter than the radiative value; we measured 217 μs, 211 μs, and 324 μs in 3%Nd-doped Y₂O₃, Lu₂O₃, and Sc₂O₃, respectively. The product of quantum efficiency and the dopant concentration can be considered as a figure of merit of the material [17]. In the case of 3%Nd:Y₂O₃, for example, considering a radiative lifetime of 354 μs, this value is 1.8, about 2.7 times higher than that obtained by Kumar and co-workers for the same doping level [17] from which laser emission has been obtained. The values obtained for the other compounds at 3% doping level are 1.8 for Lu₂O₃ and 2.8 for Sc₂O₃.

For Sc₂O₃, we investigated the dependence of the emission intensity and of the lifetime as a function of the doping level from 0.5% to 7%. The results are shown in Figure 5. As expected, both the emission intensity and the lifetime decrease with the concentration. The low-doping level value of the lifetime is slightly lower, but consistent with the theoretical radiative lifetime reported in Table 1, but the high concentration values are typically much longer than those measured in Y₂O₃ with similar doping levels. These results indicate that concentration quenching in Sc₂O₃ is not very strong and confirm the high quality of our samples.

3.3. Fluorides

We also acquired the emission spectra from 5%Nd:YF₃ and 5%Nd:LuF₃ samples and calculated the emission cross-section with Equation (1), as for sesquioxides. Results are shown in Figure 3d,e for the 11,500 cm⁻¹–6500 cm⁻¹ region and in Figure 4d,e for the 6000 cm⁻¹–4000 cm⁻¹ region. Since the decay time of LuF₃ is not known in the literature, the value for YF₃ has been used, instead. The emission intensity of fluoride samples is, in general, much weaker than that of sesquioxide samples. This can be ascribed either to the higher Nd doping level of our fluoride samples that can cause concentration quenching effects, or to a worse matching of the emission wavelength of our pump diode that causes lower absorption. In all cases, the emission is dominated by the 1-micron band. The emission cross-sections of the two compounds have similar shapes and intensity, as expected from the fact that the two compounds are isomorphs, and are much different from that of sesquioxides. The Stark splitting of the energy levels is in general smaller, and single peaks usually merge into continuous bands. The maximum emission cross-section recorded in the 1 μm region is 5 × 10⁻²⁰ cm² and 4.7 × 10⁻²⁰ cm² for YF₃ and LuF₃, respectively. The emission cross-section in the 2-micron region follows the same features already described: the shape is very similar between FY₃ and LuF₃ and is composed of an almost featureless band with a few peaks with maximum intensity of about 1 × 10⁻²¹ cm².

Emission lifetimes of the ⁴F_3/2 level have been recorded under LED pumping, and results are reported in

Table 10. Decay time of Nd:YF₃.

		τ (μs)
		This work	[35]
Nd:YF3	5%	169
	Low C		588
	Radiative		783

and

Table 11. Decay time of Nd:LuF₃.

		τ (μs)
		This work
Nd:LuF3	5%	119

and compared with the literature for YF₃. Measured decay times are 170 μs and 120 μs for YF₃ and LuF₃, respectively. If compared to the radiative lifetime of YF₃ of 783 μs determined in [35], we can observe that concentration quenching at this high doping level is strong.

These results show that fluoride materials generally show broader and weaker emission features in all wavelength regions, although fluoride crystals have lower phonon energy. This is probably due to the longer radiative lifetime of fluoride materials, but we cannot rule out interaction with possible quenching centers that are known to severely affect the emission efficiency of lanthanide-doped fluoride materials. The highest emission cross-sections are obtained from Nd:Sc₂O₃ in all regions.

4. Conclusions

We have synthesized and characterized a set of different Nd-doped fluoride and oxide nanocrystals, namely Nd:Y₂O₃, Nd:Lu₂O₃, Nd:Sc₂O₃, Nd:YF₃, and Nd:LuF₃. Under 808 nm pumping, we observed the three main emission bands in the near-infrared region, and we measured the lifetime of the ⁴F_3/2 level under LED pumping. In all cases, we were able to detect the weak 2-micron emission from the ⁴F_3/2 → ⁴I_15/2. Using the emission and lifetime data, we calculated the emission cross-sections of the various emission bands for all the compounds. Oxide materials generally showed narrower emissions, higher emission cross-sections, and shorter lifetimes. The results are in good agreement with the literature data, whenever available.