1. Introduction

Use of luminescent systems for in vivo imaging is of great interest to investigate pathologies in animal models [1]. Such non-invasive visualization tools offer great advantages in terms of cost and simplicity over expensive techniques at the laboratory scale like micro-Positron Emission Tomography or micro-Magnetic Resonance Imaging. It was recently shown that optical imaging could be advantageously carried out by using as luminescent probe a persistent phosphor emitting in the red/near-infrared part of the spectrum [2]. A red/near-infrared emission is necessary for luminescence to pass through the animal tissues since the main components of the tissues strongly absorb ultraviolet (UV)-blue-green and far-infrared parts of the spectrum [3]. By using persistent luminescence nanoparticle the illumination of the animal becomes unnecessary and autofluorescence of the tissues is avoided. UV excitation of the particle is carried out before its injection and therefore is not attenuated by the tissues. The technique was first demonstrated in 2007 with Ca_0.2Zn_0.9Mg_0.9Si₂O₆:Eu²⁺,Dy³⁺,Mn²⁺ (CZMSO) nanoparticles, from which persistent luminescence was detected in a small animal vasculature for up to one hour after their injection [2,4].

In this context we report here on a new red long-lasting phosphorescence (LLP) material potentially suitable for in vivo imaging. ZnGa₂O₄ is a wide band gap semiconductor which allows doping by luminescent ions such as transition metal ions. Due to its excellent chemical and thermal stability it has been considered as a phosphor of choice for plasma and field emission displays [5]. ZnGa₂O₄ shows blue luminescence attributed to self-activated centres when undoped [6,7], intense green emission when doped with Mn²⁺ [8] and red luminescence with Cr³⁺ doping [9]. It crystallizes in the normal spinel structure (space group O_h ⁷ (Fd3m)) with Zn²⁺ ions occupying tetrahedral sites (T_d) and Ga³⁺ ions occupying octahedral sites (D_3d) [10]. ZnGa₂O₄ is an ideal host lattice for Cr³⁺ since the latter presents an ionic radius of 0.62 Å identical to the one of Ga³⁺ in octahedral coordination [11]. When excited, Cr³⁺ emits via its ²E → ⁴A₂ transition which gives rise to a far red luminescence with a maximum at 696 nm in the zinc gallate host [9,12]. This wavelength range is totally suitable for in vivo imaging as it corresponds to a transmission maximum for the biological tissues [3].

In this paper we show that ZnGa₂O₄:Cr³⁺ powder compares better than CZMSO used up to now for in vivo imaging with UV excitation [2]. Mechanism of LLP is investigated by varying the nominal Zn/Ga stoichiometry and by using UV and X-ray excitation to record wavelength-resolved persistent luminescence and thermally stimulated luminescence (TSL).

2. Experimental

The ZnGa₂O₄:Cr powders were synthesized by a solid state method. Ga₂O₃ powder (Aldrich 99,99 + %), ZnO powder (Loba-Chemie GR 99%) and CrO₃ red flakes (Sisco Research Laboratory AR 99% min.) were mixed in an agate mortar with propan-2-ol. Pellets were prepared from the powder mixtures and fired in air at 1300°C for 6h. All the compounds were nominally doped with 0.5% mol chromium relative to gallium. Three compounds were prepared with various Zn/Ga stoichiometry, their nominal compositions being Zn_0.99Ga_1.99Cr_0.01O₄ (Zn-deficient compound, noted Zn-d) ZnGa_1.99Cr_0.01O₄ (stoichiometric compound, noted Zn-s) and Zn_1.01Ga_1.99Cr_0.01O₄ (Zn-excess compound, noted Zn-e) respectively. X-ray diffraction of the three compounds evidenced pure ZnGa₂O₄ crystalline phase with cubic spinel structure (not shown here).

Photoluminescence (PL) spectra were measured at room temperature (RT) on powder pellets with a Varian Cary Eclipse spectrofluorimeter. X-ray excited radioluminescence (XRL), LLP and TSL measurements were run on 1 mm-thick pellets of the powdered samples. For all three experiments light is collected via an optical fiber by a Scientific Pixis 100 CCD camera cooled at −65°C coupled with an Acton SpectraPro 2150i spectrometer for spectral analysis. X-ray excitation is produced by a molybdenum tube operated at 50 kV and 20 mA. UV excitation is produced by a 350 W xenon lamp filtered with a FG UV11 filter to remove visible excitation (bandpass 250-380 nm). In TSL experiments the 1 mm-thick pellet is fixed with a silver glue on the cold finger of a closed cycle helium cryogenerator. UV irradiation is performed at 30 K through a quartz window of the cryostat at 45° angle from the pellet surface and luminescence is detected through a different quartz window at 45° angle from the pellet surface. A heating rate of 10 K/min is applied. Prior to all LLP and TSL experiments the samples were bleached for 20 minutes at 200°C and kept in the dark before the experiment.

3. Results and discussion

Figure 1 shows the PL spectra of the Cr³⁺-doped ZnGa₂O₄ powders. The spectra display several narrow lines that constitute recognizable features of Cr³⁺ ion emission in a strong field with octahedral coordination. PL features of ZnGa₂O₄:Cr³⁺ at 13K as reported by Zhang et al. [13] are shown in Fig. 1 as dotted lines. At 13K, these authors report two zero phonon (ZP) lines known as R2 and R1 for the ²E → ⁴A₂ transition of Cr³⁺. A weak trigonal distortion around Cr³⁺ splits the ²E level into ²E (Ē) and ²E (2Ā) separated by 40 cm⁻¹ [12] giving rise to the two R lines. R2 and R1 have also been observed at 300K by Kahan et al. [12] with a high-resolution set-up at 14556 cm⁻¹ (687.0 nm) and 14518.8 cm⁻¹ (688.8 nm), respectively. Their position is in total agreement with the feature here observed at 688 nm which we attribute to the unresolved (R2,R1) doublet since our set-up does not allow resolution of these two lines. Since all lines should be red-shifted from 13K (Zhang et al. [12]) to RT (our experiment), the phonon side bands (PSB) of the R lines were identified in the spectra of Fig. 1 at 708 nm, 715 nm and 722 nm for the Stokes PSB and at 670 nm and 680 nm for the Anti-Stokes PSB. The PSB positions are consistent with data from infrared and Raman spectroscopies [13].

 

Fig. 1 PL spectra excited at 247 nm of Zn-d, Zn-s and Zn-e ZnGa₂O₄:Cr³⁺ compounds. Dotted lines show the attribution of Cr³⁺ lines in ZnGa₂O₄:Cr³⁺ at 13K according to Zhang et al. [13].

Download Full Size | PDF

An additional line observed at 695 nm in Fig. 1 is not accounted for by the above mentioned transitions. It therefore originates from another type of Cr³⁺ ion with perturbed short-range crystalline order relative to the ideal octahedral coordination of the normal spinel. The feature was here observed at 695 nm, i.e. at 147 cm⁻¹ lower energy than the (R2,R1) doublet. According to Mikenda et al. [14] a structure-dependent line named N2 for Cr³⁺ in ZnGa₂O₄ lies at 134 cm⁻¹ lower energy than R1. We therefore attribute the 695 nm line to the N2 line of Cr³⁺. After extensive investigation, particularly by Mikenda et al. [15–17] and Nie et al. [18], the N2 line of Cr³⁺ in ZnGa₂O₄ spinel was established to originate from Cr³⁺ ions possessing an environment distorted by an antisite defect, located in the first cationic neighbors of Cr³⁺. Our chromium-doped zinc gallates therefore present two types of doping ions: (i) Cr³⁺ in an ideal normal spinel environment noted $\text{C}{\text{r}}_n^{3+}$ and (ii) Cr³⁺ in a slightly distorted environment noted $\text{C}{\text{r}}_{dis}^{3+}$. ZnGa₂O₄ is known to crystallize in an almost normal spinel structure with a few percent inversion [19]. The inversion percentage varies with synthesis conditions and is favored by high temperature synthesis [6,20]. The comparison of the three compounds spectra shows that the intensity of the N2 line relative to the R lines increases as the stoichiometry evolves from a nominal zinc deficiency towards a zinc excess. More antisite defects seem therefore present as the Zn/Ga nominal ratio is increased. Considering the two possible antisite defects of opposite charge that can distort the coordination polyhedron of Cr³⁺, namely Zn_Ga’ (negatively charged) and Ga_Zn° (positively charged), the position of the N2 line at a photon energy 147 cm⁻¹ below the (R1,R2) line of undistorted $\text{C}{\text{r}}_n^{3+}$indicates that the crystal field is smaller for $\text{C}{\text{r}}_{dis}^{3+}$than for $\text{C}{\text{r}}_n^{3+}$. This suggests that the neighboring defect is positively charged, as is a Ga³⁺ ion in Zn²⁺ site (Ga_Zn°).

LLP was measured after UV excitation to compare the Cr³⁺-doped zinc gallates to the phosphor previously used for the demonstration of imaging (referred to as CZMSO). This reference phosphor is a silicate of nominal formula Ca_0.2Zn_0.9Mg_0.9Si₂O₆:Eu²⁺,Mn²⁺,Dy³⁺ described in [2]. Both phosphors were excited for 5 minutes with a filtered xenon lamp before their LLP was recorded for up to one hour at an ambient temperature of 30°C. The normalized LLP spectra of the ZnGa₂O₄:Cr³⁺ compounds recorded 45 s after the end of the excitation are shown in the inset of Fig. 2 . Due to wide monochromator slits the spectra do not show as resolved lines as in Fig. 1 but they still display the ²E → ⁴A₂ luminescence of Cr³⁺. LLP decays at the wavelength of emission maximum are compared in the main plot of Fig. 2. The Cr³⁺-doped zinc gallates showed a highly enhanced intensity relative to CZMSO. As its emission lies in the highest transmission spectral range of the biological tissues, ZnGa₂O₄:Cr³⁺ is an excellent new candidate for LLP imaging. Amongst the three ZnGa₂O₄:Cr³⁺ materials more intense LLP was obtained as the compound is nominally more Zn-deficient.

 

Fig. 2 LLP decays of Zn-d, Zn-s and Zn-e ZnGa₂O₄:Cr³⁺ compounds at 705 nm and of the silicate reference at 695 nm recorded at 303 K after 5 minutes UV excitation. Inset: normalized LLP spectra of ZnGa₂O₄:Cr³⁺ compounds 45 s after the end of the excitation.

Download Full Size | PDF

The compound with the most intense LLP, i.e. Zn-d, was investigated further. Both immediate and delayed (i.e. “persistent” or “long-lasting”) luminescence spectra of Zn-d were recorded under X-ray excitation. The ZnGa₂O₄:Cr³⁺ pellet was excited by X-rays for two minutes. During this time an XRL spectrum was recorded and is presented in Fig. 3 . After two minutes, X-ray irradiation was switched off and an X-ray induced long-lasting phosphorescence (XLLP) spectrum was recorded 20 seconds after the excitation had ended. The XRL spectrum displayed the same lines as shown in the PL spectrum of Fig. 1. An intense XLLP was observed when the excitation is switched off, and its normalized spectrum shown in Fig. 3 appeared very different from the XRL spectrum, as the N2 line was then highly predominant. One can distinguish the weak PSB identified above as the PSB of the R lines. The emission spectrum of $\text{C}{\text{r}}_n^{3+}$ is therefore still present but with a strongly reduced intensity. The striking difference between the XRL and the XLLP spectra confirms the existence of two distinct types of Cr³⁺ ions in the gallate. Undistorted $\text{C}{\text{r}}_n^{3+}$ shows very little XLLP whereas $\text{C}{\text{r}}_{dis}^{3+}$ is related to an intense delayed emission. Note that in the XRL spectrum the PSB of the R lines appear quite intense relative to the ZP lines. This is expected for a centrosymmetric environment around undistorted $\text{C}{\text{r}}_n^{3+}$. Purely electronic transitions (R lines) remain forbidden relative to phonon coupling transitions in a centrosymmetric environment. On the contrary the PSB of the N2 line are expected to present a smaller intensity relative to the ZP line since the symmetry of the crystal field is distorted by a closeby antisite defect. This is why the N2 ZP line appears very distinct in the XLLP spectrum.

 

Fig. 3 XRL and XLLP spectra of Zn-d ZnGa₂O₄:Cr³⁺ at 293 K.

Download Full Size | PDF

TSL of Zn-d was measured after 5 minutes UV irradiation by the filtered xenon lamp at 30K. The glow curve for the emission at 694 nm is shown in Fig. 4(a) . The spectra at selected points of the glow curves (A, B, C and D) are displayed in Fig. 4(b). Decaying intensity right at the beginning of the glow curve (point A) reflects a possible athermal tunnelling recombination process, weak in intensity relative to the thermally stimulated processes. However it is interesting to compare the shape of the spectrum at position A with spectra at positions B, C and D (see Fig. 4(b)). At point A, the spectrum was clearly composed of the (R2,R1) lines at 688 nm and of the N2 line at 694 nm similarly to the XRL spectrum presented in Fig. 3. The spectra only differed in terms of PSB since at 30K the Stokes PSB were present whereas, as expected at low temperature, no anti-Stokes PSB were observed. On the contrary the spectra at points B, C, and D (maxima of the TSL glow curve) displayed very much enhanced N2 line relative to the R and PSB lines as already observed in the XLLP spectrum of Fig. 3. This shows two different behaviors in terms of emitting ions and charge trapping and release in the material. On the one hand a minor part of the electrons and holes created by UV excitation get weakly trapped and then recombine via a tunnelling mechanism that leads to luminescence of all types of Cr³⁺ ions. On the other hand and more interestingly, $\text{C}{\text{r}}_{dis}^{3+}$is the main emitting centre for TSL peaks B, C and D, which indicates that it is directly involved in the recombination process. Let us note that no difference but the effect of temperature on the PSB could be noticed from the comparison of spectra at points B, C and D. Hence the same defect spatially related to $\text{C}{\text{r}}_{dis}^{3+}$ is at stake in B, C and D, i.e. all along the TSL glow curve.

 

Fig. 4 (a) TSL of Zn-d ZnGa₂O₄:Cr³⁺ at 694 nm after 5 minutes UV excitation. Inset: TSL spectrum at 318 K with reduced slitwidths. (b) TSL spectra at the beginning of the TSL curve (A) and at the main TSL peak maxima (B, C and D).

Download Full Size | PDF

As the TSL was recorded at a slow heating rate (10 K/min.) the LLP observed at RT is mainly related to peak D of the glow curve, while peaks B and C should only contribute to the initial decay of the LLP curve. For a short period of time the slitwidths of the monochromator were suddenly reduced to the minimum at 318K in the rise of peak D showing the spectrum presented in the inset of Fig. 4(a). Again the spectrum showed a clear predominance of the N2 line of $\text{C}{\text{r}}_{dis}^{3+}$ and is totally similar to the XLLP spectrum shown in Fig. 3. Therefore the mechanism of the LLP is the same whatever the excitation source (UV or X-rays).

From the above results and due to the property of Cr³⁺ to constitute a very sensitive local probe, we can propose a possible mechanism for the LLP in ZnGa₂O₄:Cr³⁺. Previous works showed that antisite defects of the type Ga_Zn° and Zn_Ga’ are present in the compound [6,19]. Further because ZnO is known to evaporate at lower temperature than Ga₂O₃ and because the compounds were synthesized at 1300°C it is reasonable to assume that Zn vacancies, noted as V_Zn ^” are also present. The fact that both LLP and TSL are mainly due to $\text{C}{\text{r}}_{dis}^{3+}$ indicates that this species is a recombination centre. It is thus tempting to assign the nearby antisite defect to a trap. Following our hypothesis that the latter is a Ga_Zn° defect (which we note Ga_Zn°{Cr³⁺}), one may propose a mechanism explaining the dominant contribution of $\text{C}{\text{r}}_{dis}^{3+}$ to LLP and TSL. The UV irradiation creates electrons (e’) and holes (h°) that are trapped by defects. Electrons are trapped around $\text{C}{\text{r}}_{dis}^{3+}$ ions according to the reaction: Ga_Zn°{Cr³⁺} + e’ → Ga_Zn ^x{Cr³⁺}, while holes are trapped at negatively charged defect sites such as Zn_Ga ^’ or V_Zn”. Holes released from these traps by thermal activation then recombine at $\text{C}{\text{r}}_{dis}^{3+}$site, giving the N2 emission line according to the reaction: Ga_Zn ^x{Cr³⁺} + h° → [Ga_Zn°{Cr³⁺}]* → Ga_Zn°{(Cr³⁺)*} → Ga_Zn°{Cr³⁺} + hν where * represents excited states.

With the comparison of the three samples with different Zn/Ga ratio we found that more antisite defects were present close to Cr³⁺ with increasing Zn/Ga ratio, whereas the LLP was more intense with decreasing Zn/Ga ratio. Decreasing the Zn/Ga ratio favors more Zn vacancies (and thus more hole traps) which may then be responsible for the enhanced LLP. Within this hypothesis the TSL peak D as well as the LLP at RT are due to V_Zn ^” from which the holes get thermally detrapped to recombine at $\text{C}{\text{r}}_{dis}^{3+}$ site. Further experiments will be carried out in the future to constrain the proposed mechanism.

4. Conclusion

We showed in this work that ZnGa₂O₄:Cr³⁺ is a new high-performance red long-lasting phosphor, compared to CZMSO used for demonstrating the technique of LLP in vivo imaging [2]. On the contrary to other red LLP materials studied for this application [21–24] ZnGa₂O₄:Cr³⁺ could be conveniently excited through its bandgap by UV light. After 5 minutes excitation it emitted light around 695 nm via the ²E → ⁴A₂ transition of Cr³⁺ for a prolonged time. This emission range perfectly matched the optical window of biological tissues and as such may be readily detected across small animal tissues.

In addition to the exciting properties of ZnGa₂O₄:Cr³⁺ for imaging, the compound revealed new elements in the field of LLP. Firstly ZnGa₂O₄ may be an intrinsic persistent phosphor since intense green [25] and now red LLP were observed in the same host. Furthermore no codoping was necessary to obtain intense persistent luminescence which makes investigation of mechanisms easier and decreases the chances of non-radiative loss. The use of a small dopant concentration (0.5%) and of a similar size dopant as the substituted ion (the ionic radii of Cr³⁺ and Ga³⁺ are identical) also reduces the occurrence of defects that constitute potential loss centres. Secondly the use of Cr³⁺ as both luminescent centre and probe of the local environment appeared as of uttermost use in the investigation of the mechanisms. The very narrow emission lines of Cr³⁺ ions in Ga³⁺ sites are so sensitive to chromium environment that the coordination sphere of the luminescent centre Cr³⁺ can be elucidated. Hence we showed that antisite defects Ga_Zn of the ZnGa₂O₄ spinel structure were likely to be involved in the charge trapping and recombination processes responsible for LLP in ZnGa₂O₄:Cr³⁺. The LLP intensity was singularly improved by decreasing the nominal Zn/Ga ratio, i.e. probably by creating more Zn vacancies that constitute the thermally emptied traps in the LLP mechanism.