1. Introduction

Bismuth-activated near-infrared (NIR) luminescent materials have attracted considerable attention in recent years owing to their huge potential for the applications in the fields of telecommunications and biomedicine [1]. Until now, NIR photoluminescence (PL) has been observed in a diversity of matrices including glasses [2–10], crystals [11–19], ionic liquids [20], and molecular crystals [21], and applications of these systems for lasers, amplifiers and bioimaging have been demonstrated [22,23]. Obviously, we can view these reported systems almost as bulk materials or powders with sizes larger than 100 nm. For some applications such as bioimaging, realizing high-efficiency NIR PL from nanocrystals (NCs) is of paramount importance, but remains challenging owing to the difficulty in the simultaneous control over the size and Bi-related NIR active center (BRACs).

Over the past several years, lead halide perovskites have emerged as a new generation of low-cost, solution-processed semiconducting materials with favorable optoelectronic properties [24]. MAPbI₃ (MA = methylammonium), one of representative hybrid organic-inorganic lead halide perovskites, has been extensively investigated. Recently, we showed that Bi-doped MAPbI₃ demonstrates ultrabroad NIR emission [25,26], and electroluminescence has been successfully realized using Bi-doped MAPbI₃ as an emissive component. Nevertheless, MAPbI₃ has intrinsic drawbacks as follows. Firstly, MAPbI₃ contains organic group methylammonium, which has a strong quenching effect on BRACs, thus resulting in an extremely low PL quantum yield (PLQY). Secondly, MAPbI₃ has a poor thermal stability compared to inorganic cousins such as CsPbI₃. Considering that solid-state reaction can yield Bi-doped non-perovskite CsPbI₃ powders and that CsPbI3 NCs can be readily produced through a wet-chemistry route [27–30], we hypothesize that synthesizing bismuth-doped perovskite CsPbI₃ NCs could provide a chance for the attainment of a new class of NIR luminescent, nanometer-sized systems.

Here, we report on the synthesis of Bi-doped CsPbI₃ NCs with cubic shape and cubic perovskite crystal structure. To the best of our knowledge, this represents the first work on the attainment of NIR-luminescent, Bi-activated materials with an average size less than 10 nm. The PL properties of these NCs have been investigated by steady-state and time-resolved PL measurements. In particular, we find that, by optimizing the dopant concentration, a PLQY of 7.17% can be obtained, which is much higher than that of bismuth-doped MAPbI₃, suggesting huge potential for applications in optoelectronic devices or as biomarkers in the second biological window.

2. Experimental

Both Bi-doped and undoped CsPbI₃ NCs were synthesized following the hot-injection method reported by Protesescu et al. with some modifications [30]. Cesium oleate (Cs-oleate) was prepared by reacting cesium carbonate (0.4073 g) with oleic acid (2mL) in octadecene (20 ml) solvent. The mixture was under vacuum conditions and heated at 120 °C for 1h, then maintained purging of N₂ gas with continuous stirring at 150 °C until all solute had completely dissolved. Next, PbI₂ (0.0867 g) and octadecene (5 ml) were loaded in a three-neck flask and maintained under vacuum condition for 1h at 120 °C, followed by addition of oleylamine (500 µL) and oleic acid (500 µL). After complete dissolution of the precursor, the temperature was raised to 150 °C. The Cs-oleate (408 µL) prepared in the above step was then injected into the contents of the three-necked flask, which was immediately transferred to an ice-water bath. The CsPbI₃ NCs that formed in the octadecene were centrifuged at 3000 rpm for 5 min and the sedimentation was discarded. The supernatant was centrifuged at 12000 rpm for 5 min and then the pellet was dispersed in toluene after washed twice with hexane and once with toluene. The NCs were doped by adding BiI₃ into Cs-oleate with designed Bi to Cs molar ratios of 0.1%, 0.5%, 1%, and 2%. The fresh samples were drop-casted on glass substrates for the characterizations of X-ray diffraction (XRD), NIR PL, time-resolved NIR-PL, and PLQYs. Other characterizations were obtained using the freshly prepared colloidal NCs dispersed in toluene. XRD patterns were taken at room temperature using a Bruker D8 ADVANCE diffractometer with Cu Ka radiation (λ = 1.54056 Å). The transmission electron microscope (TEM) images of the samples were collected using a FEI Tecnai G20 electron microscope operating at 200 kV. The absorption spectra were taken by a double-beam UV-Vis-NIR spectrophotometer (Cary 5000, Agilent). The PL spectra were analyzed with an FLS 980 spectrofluorometer (Edinburgh Instruments Ltd). Time-resolved PL measurements monitored at 680 nm were acquired on a Lifespec II setup (Edinburgh Instrument, UK) with the excitation of a picosecond-pulsed 507 nm laser. The time-resolved PL measurements at 1145 nm, under the excitation of nanosecond-pulsed light from the second harmonic (532 nm, 10 Hz, pulse duration: 15 ns) of a Nd:YAG laser, were performed by detecting the modulated luminescence signal with a photomultiplier tube (Hamamatsu, H10330-75), and then analyzing the signal using a photon-counting multichannel scaler.

3. Results and discussion

Figure 1(a) shows the XRD pattern of the undoped CsPbI₃ NCs and Bi-doped CsPbI₃ NCs, both of which can be indexed to the cubic phase, indicating that the crystalline structure of CsPbI₃ remains well after Bi incorporation. Furthermore, the crystallinity of the CsPbI₃ NCs does not show a noticeable change upon Bi doping. Notably, the positions of two main diffraction peaks at ca. 15 ° and 29 ° for the 2% samples undergoes a shift of 0.3° to a higher value, indicating that the CsPbI₃ lattice contracts due to the incorporation of Bi³⁺ ions into matrices. This can be attributed to the smaller size of Bi³⁺ with respect to Pb²⁺. From the transmission electron microscopy (TEM) images and statistical distributions of the edge length in Figs. 1(b) and 1(d) and Figs. 1(c) and 1(e), we can conclude that the NCs do maintain their sizes and shape after Bi³⁺ doping. The actual Bi to Pb molar ratios in the CsPbI₃ NCs, as determined by inductively coupled plasma mass spectrometry (ICP-MS), were 0.07, 0.36, 0.58, and 0.86% for the 0.1, 0.5, 1, and 2% samples, respectively. This evidences that a higher Bi concentration in the precursor solution favor the incorporation of Bi ions into the CsPbI₃ lattice. Note that hereafter we denote the samples by their nominal concentrations. We note that Bi doping does not improve the stability of CsPbI₃ NCs, and one day exposure in air can destroy the samples.

 

Fig. 1 Characterization of doped and undoped CsPbI₃ NCs. (a) XRD patterns of undoped and Bi-doped CsPbI₃ NCs with different Bi doping concentrations. The red vertical lines at the bottom correspond to the diffraction peaks of the cubic CsPbI₃. TEM images and statistical distributions of the edge length of undoped (b/c) and 2%-doped (d/e) CsPbI₃ NCs.

Download Full Size | PDF

With the increasing of Bi concentration, the color of NC solution changes from red to brown. To examine the photophysical characteristics of these samples, the UV-vis absorption spectra were taken. As displayed in Fig. 2(a), the absorption spectra exhibit systematic red shift with the increase of Bi concentration, indicating that doped Bi ions could influence the electronic structure of the host. Next, we compared the PL spectra of Bi-doped and undoped NCs upon excitation at 450 nm. The CsPbI₃ NCs display a narrow red emission centered at about 683 nm, which can be attributed to band-edge emission of CsPbI₃ NCs. A redshift from 680 nm to 685 nm was followed by a blue shift with increasing Bi/Pb ratios. It is widely accepted that as dopant content increases, impurity levels interact with each other to form a band-like structure within the band gap of the host [31]. Given the slight shift and the substitutional doping of Bi, we thus attribute them to filling of the conduction band with extra electrons donated by Bi [31].

 

Fig. 2 (a) Absorption and PL spectra of Bi doped and undoped CsPbI₃ NCs. The PL spectra were obtained under 450 nm excitation. (b) Visible PL decay curves of Bi-doped and undoped CsPbI₃ NCs.

Download Full Size | PDF

As can be seen, the intensity of the band-edge emission monotonously decreases with the increase of Bi doping concentration. Deep insight into the luminescence properties of these NCs come from time-resolved PL measurements. As shown in Fig. 2(b), the visible PL decays of the NCs monitored at 680 nm become progressively faster with increasing Bi concentration. The average lifetimes of the undoped, 0.1%, 0.5%, 1% and 2% samples are 124, 113, 87, 52, and 39 ns, respectively, which were calculated by the equation:

Interestingly, we can detect an ultrawide emission band from 800 nm to 1600 nm peaked at 1145 nm from Bi-doped CsPbI₃ NCs, as shown in Fig. 3(a). We stress that changing the excitation wavelengths cannot affect the emission lineshape. It is necessary to point out that the undoped CsPbI₃ NCs do not show any NIR PL. This suggests that the NIR emission in Bi-doped CsPbI₃ NCs originates from BRACs. It is noted that the lineshape and peak wavelength of the NIR PL are similar to the hybrid cousin [25, 26]. Combined with the results as shown in Fig. 2, we conclude that energy transfer occurs between the CsPbI₃ host and BRACs. Assuming that the Bi doping-induced, shortened lifetime are merely due to the energy transfer from CsPbI₃ host to BRACs, the energy transfer efficiencies can be roughly estimated by (τ_un-τ_x)/τ_un, where τ_un and τ_x are average lifetimes of undoped and Bi-doped CsPbI₃ NCs, respectively. The calculated energy transfer efficiencies are 8.9%, 29.8%, 58.1%, and 68.5% for the 0.1%, 0.5%, 1% and 2% samples, respectively.

 

Fig. 3 (a) NIR PL excited by 450 nm and PLQY of Bi-doped CsPbI₃ NCs with different nominal Bi concentrations. (b) NIR PL decays of Bi-doped CsPbI₃ NCs. The monitored wavelength is 1145 nm. (c) Schematic illustration of the electronic transition in doped NCs. The thick and thin red lines represent the direct excitation of CsPbI₃ NCs and nonradiative decay from the conduction band to the in-gap state, respectively. The violet line represents the NIR emission.

Download Full Size | PDF

We further roughly evaluated the PLQYs of NIR emission in the range of 800-1600 nm for Bi-doped CsPbI₃ NCs films under the excitation of 450 nm laser with a power of 10 mW using an integration sphere method [32]. Because the lack of detectors that can simultaneously measure the excitation and emission signals, we thus measured the emission intensity at visible and NIR spectral ranges by a Thorlabs PM200 power meter equipped with highly sensitive S120VC and S132C power sensors, respectively. The separation of visible (i.e., band-edge) and NIR emission was realized by using different optical filters (FELHO500 and FELHO950, Thorlabs). Totally, four emission intensities for each sample were measured. Firstly, we used the silica substrate as a reference that was fixed in the holder of an integration sphere, and the excitation power, W₁, was measured by using S120VC power sensor. Secondly, the NCs film was placed in the sample holder of the integration sphere, and the power, W₂, was measured using the S120VC sensor. No filter was used for the W₁ and W₂ measurements. Thirdly, the FELHO500 filter was added after the W₂ measurement, and the power, W₃, was measured using the S120VC sensor. Finally, the FELHO500 filter was replaced by the FELHO950 filter, and the power, W₄, was measured using the S132C sensor. The PLQYs of doped NCs can be estimated by W₄/(W₁-W₂ + W₃). As shown in Fig. 3(a), the PLQY decreases from 7.17% to 5.16% along with the increase of Bi concentration, which are over two orders of magnitude larger than that of Bi-doped MAPbI₃. The NIR decays of all Bi-doped CsPbI₃ NCs films were measured upon nanosecond-pulsed photoexcitation at 532 nm. It is noted that the NCs films with a higher Bi concentration demonstrate relatively shorter lifetimes (Fig. 3(b)), which might be caused by the back energy transfer from the BRACs to the CsPbI₃ matrix or concentration quenching. It is noteworthy that the lifetimes observed here are much shorter than those in other matrices [1], probably as a result of allowed transition occurring in this system. We note that such fast decays require a fast excitation rate that can be achieved by 532 nm pulsed light with a duration of ca. 15 ns.

We note that the emission lineshapes of Bi-doped CsPbI₃ NCs are similar to that of Bi-doped MAPbI₃ [25]. In our previous work, we attributed such emission to polaron-stabilized luminescent structural defects induced by Bi doping [25–27]. Owing to similar ionic radii of Bi³⁺ and Pb²⁺, Bi preferentially occupies the Pb site. The direct consequence of such a substitution is the introduction of electrons required to maintain charge neutrality. Meanwhile, owing to the slight size mismatch between Bi³⁺ and Pb²⁺, the replacement inevitably gives rise of the local structure distortion of the [PbI₆] octahedra, thus creating NIR emission center. That is, the NIR PL can be attributed to the creation of in-gap trapping states in CsPbI₃, which provides nonradiative channels of excited charge carriers when the excitation energy is larger than the band gap; the radiative transition from the in-gap state to the valence band gives rise to the NIR emission (Fig. (3c)). In contrast to hybrid MAPbI₃ in which the strong C-H and N-H vibrations could cause nonradiative decays of emissive states, we surmise that the inorganic matrix of CsPbI₃ NCs, without any organic group inside the matrix and well-passivated by oleylammonium ligands in the surface, favors a high PLQY. We stress that more work is necessary to gain deeper insight into the emission mechanism of this system.

4. Conclusion

In conclusion, we have realized the synthesis of Bi-activated, NIR-luminescent NCs by using CsPbI₃ perovskite as a host. The synthetic method is simple and easy-to-control, and the samples demonstrate PL covering the telecommunication and second biological optical windows and with a high PLQY (7.17%). Additionally, efficient energy transfer from the semiconducting CsPbI₃ to BRACs can be achieved. To the best of our knowledge, this represents the first work on the synthesis of Bi-doped NIR-luminescent NCs. We anticipate that the material system developed here could find applications in the fields of optoelectronic and photonic devices as well as biological imaging. We also expect that our work could stimulate research interest for the synthesis of Bi-doped luminescent nanomaterials, which will speed up their practical applications in a diversity of areas.