Introduction

High-sensitivity, large-size neutron detectors play vital roles in industrial and biomedical applications1,2,3,4,5. 3He-based gaseous, scintillator, and semiconductor-based detectors dominate the commercial market6. Compared with gas and scintillators, compact structured semiconductor detectors exhibit high spatial resolution and fast time response due to the high energy resolution and wide linear range of semiconductors; these detectors are suitable for preparing an efficient neutron detector7,8.

LiInSe2 is a new type of semiconductor with a chalcogenide structure and has a large nonlinear optical coefficient and a suitable transparent wavelength range9,10. Importantly, 6LiInSe2 crystals display a remarkable ability to reflect ionizing radiation via the nuclear reaction of 6Li(n,α)3H and direct charge transport11,12. A high density of 6Li cations in the 6LiInSe2 lattice can capture > 95% of the neutrons in a wafer with a thickness of ~ 3.4 mm13. The facile preparation process and high crystal quality of LiInSe2 can ensure the high performance of the 6LiInSe2-based neutron detectors.

Although many studies have been devoted to the optimization of the crystal growth to obtain high-quality crystals, the optical and electrical efficiencies are still inadequate for high-resolution detection due to the presence of defects14,15,16,17. In general, the optical and electrical behaviors are highly dependent on the crystal defects because the formed defects can distort the energy band structure, decrease the carrier concentration and scatter the charge. Theoretical calculations indicate that defect-free LiInSe2 is a yellow crystal with a band gap of ~ 2.86 eV at room temperature18,19,20,21. As the antisite defects of In-substituted Li (InLi2+) and Li vacancies (VLi−) form due to a lack of Li, the strong shift in the absorption edge results in the transformation of the crystal color to red16. In addition, crystals containing defects exhibit a decrease in the charge mobility-lifetime product (μτ) of the detectors to 3–7.6 × 10−6 cm2/V due to the charge scattering by the defects18. As a result, yellow LiInSe2 crystals tend to exhibit better optical and electrical properties.

During LiInSe2 synthesis and growth, red crystals with defects are prone to deviation from the stoichiometric ratios due to the loss of elements10. Some optimization techniques for yellow LiInSe2 crystal growth have been carried out and include annealing, regulation of growth parameters, and stoichiometric ratio modulation; among these, stoichiometric ratio modulation is considered the most effective strategy22. In the Siemek study, the color of the LiInSe2 crystals varied from green to yellow to pink and further to red during the process of annealing in Se vapor23. Vijayakumar attempted to optimize the synthesis process to prepare LiInSe2 single crystals by controlling the stoichiometric ratio24. However, the proper adjustment of stoichiometry to suppress defects during melt growth remains a major challenge. Therefore, further exploration of the stoichiometric ratios for control of the defects inside the LiInSe2 crystals of different colors is necessary for clarifying the structure–performance relationship, optimizing crystal growth, and guiding LiInSe2 neutron detector fabrication.

In this study, LiInSe2 single crystals with different stoichiometric ratios of Li0.87In0.98Se2, Li0.94In0.99Se2, Li0.99In1.02Se2, and Li1.01In1Se2 were grown by the vertical Bridgman method. A yellow single crystal was obtained from Li1.01In1Se2 when excess Li was used. The band gap of the resulting yellow crystals reached 2.83 eV at room temperature; this value was consistent with the theoretical result of 2.86 eV. In addition, the relationship between the defect concentration inside LiInSe2 and the neutron detection performance was studied by photoluminescence spectroscopy. This mechanism analysis for the defect structure impacts of the LiInSe2 semiconductor on the neutron detection performance can promote its application in neutron detection.

Experiment

Synthesis and crystal growth

LiInSe2 was synthesized using pure Li (4N), In (6N), and Se (6N). In this experiment, polycrystalline LiInSe2 was directly prepared by a solid-state reaction with high-purity Li, In and Se under a pressure of 5 × 10−5 Pa. and 925 °C. To prevent the evaporation of Se and the reaction between Li and the quartz ampoule bottle, both Li and Se were added in excess. During the growth process, the LiInSe2 single crystal was grown by the vertical Bridgman method with quartz with a carbon layer and graphite crucibles inside. The polycrystalline material was placed in a growth furnace, and the furnace temperature was increased from room temperature to 950 °C; afterward, the material was soaked at 950 °C for 2 h. The crucible was then moved to the starting position before being controlled to descend at a rate of 0.5 mm/h. A temperature gradient of 10 °C/cm was achieved in the growing area of the furnace. After growth, the LiInSe2 crystal was cooled to room temperature. The chemical compositions of the as-grown crystals were quantified using an inductively coupled plasma optical emission spectrometer (ICP‒OES: Optima 8300). The resulting compositions were identified as Li0.87In0.98Se2, Li0.94In0.99Se2, Li0.99In1.02Se2 and Li1.01In1Se2; these were very close to the stoichiometric ratio of LiInSe2. The detailed growth parameters of the above crystals are given in Table 1. The uncertainty of element determination in the experiment is less than ± 0.005.

Table 1 Synthesis methods for four stoichiometric ratios of different colored crystals.
Full size table

Characterization

The transmission spectrum of LiInSe2 was obtained by an ultraviolet-vis-near infrared spectrometer (UV‒Vis–NIR spectrometer) and a Fourier transform infrared spectrometer. The band gap of LiInSe2 and the defects in the crystal were analyzed by diffuse reflectance spectroscopy and photoluminescence spectroscopy, respectively. The UV‒vis–NIR transmission spectra and diffuse reflectance spectra were recorded in the wavelength range of 200–2200 nm with a UV‒Vis–NIR spectrometer, and BaSO4 was used as the 100% reflectance standard. Infrared transmission spectra were recorded in the wavelength range of 2.5–20 μm with a Nicolet Nexus Fourier transform infrared spectrometer. The photoluminescence spectra were measured using a He-Cd laser with an output wavelength of 325 nm and a maximum output power of 25 mW. A TRIAX 550 spectrometer was used to record the photoluminescence (PL) emission with an error range of ± 0.3 nm. A low-temperature sample compartment was used to hold the sample with liquid nitrogen as the freezing medium, and the variable temperature range was 10 K to room temperature.

Results and discussion

UV‒vis–NIR transmission and diffuse reflectance measurements

Figure 1 shows LiInSe2 wafers with four different colors: the colors of Li0.87In0.98Se, Li0.94In0.99Se, Li0.99In1.02Se2 and Li1.01In1Se2 are dark red, deep red, light red, and yellow, respectively. Specifically, a higher Li content in the crystals contributes to a lighter color. To investigate more clearly the effect of the Li content on the optical properties of LiInSe2 wafers, we analyzed the IR and UV‒vis–NIR transmission spectra of the LiInSe2 wafers shown in Fig. 2a and b, respectively. The transmittance of the Li0.94In0.99Se2, Li0.99In1.02Se2 and Li1.01In1Se2 crystals reached 70% at wavelengths of 0.8–9 μm, while the Li0.87In0.98Se2 crystal displayed a narrower transmission range and inferior transmission performance. From the transmission spectrum of LiInSe2 in the ultraviolet‒visible band (Fig. 2b), the absorption edges of the transmission spectra of the four chips vary considerably in the UV direction, with 610 nm for the Li0.87In0.98Se2 crystal, 538 nm for the Li0.94In0.99Se2 crystal, 565 nm for the Li0.99In1.02Se2 crystal to 436 nm for the Li1.01In1Se2 crystal. Similarly, the resulting optical band gaps of the Li0.87In0.98Se2, Li0.94In0.99Se2, Li0.99In1.02Se2, and Li1.01In1Se2 crystals are ~ 1.98 eV, 2.05 eV, 2.09 eV and 2.83 eV, respectively, by using a Tauc plot (see Fig. 2c)25. Importantly, our resulting band gap of ~ 2.83 eV for the Li1.01In1Se2 crystals is highly consistent with the given band gap of ~ 2.86 eV for the theoretical defect-free LiInSe2 yellow crystals at room temperature20,21,22,23. This result confirms that the yellow Li1.01In1Se2 crystals we obtained are high-quality single crystals that closely match to theory.

Figure 1
Figure 1
Full size image

LiInSe2 wafers with different color from dark red (Li0.87In0.98Se2), deep red (Li0.94In0.99Se2), light red (Li0.99In1.02Se2) and yellow (Li1.01In1Se2).

Figure 2
Figure 2
Full size image

IR (a) and UV‒vis–NIR (b) transmittance spectra of the LiInSe2 wafers with different stoichiometric ratios. (c) Tauc plot of (αhν)2 vs. hν.

According to the Kubelka–Munk formula19,20,21, diffuse reflectance spectral data were processed to obtain the powder absorption spectra of the Li0.87In0.98Se2, Li0.94In0.99Se2, Li0.99In1.02Se2 and Li1.01In1Se2 crystals (Fig. 3a). The strongest absorption for all samples occurred in region “a”; this is the real absorption edge of the LiInSe2 crystals since the photon energy can be strongly absorbed as it approaches the band gap. Combining the Tauc equation25 with absorption spectra, the spectra of (ahν)2 with respect to hν for the Li0.87In0.98Se2, Li0.94In0.99Se2, Li0.99In1.02Se2 and Li1.01In1Se2 crystals are shown in Fig. 3b. The spectral lines of all the samples overlap in region, and the absorption coefficients reaches the maximum values at the high-energy edge of region “a.” Therefore, the straight line in region “a” is the real band gap absorption edge of the LiInSe2 crystal. As depicted in Fig. 3b, the intrinsic band gap of LiInSe2 can be estimated to be 2.83 eV; this corresponds to the sample Li1.01In1Se2 crystals derived from transmission spectra.

Figure 3
Figure 3
Full size image

(a) Absorption spectra of the as-grown LiInSe2 crystals with different stoichiometric ratios and (b) Kubelka–Munk plot of (αhν)2 vs. hν.

Note that internal defects have a significant influence on the results of transmission spectra as the light passes through the crystal. In contrast, since only light reflected at the crystal surface can be collected in the reflection spectra, the defects within the crystal have a negligible effect on the spectrum. Therefore, the transmittance spectra showing the strong absorption in the red region due to the high concentration of defects inside the crystal likely leads to a redshift of the red crystal.

Photoluminescence spectra

The PL spectra of the Li0.87In0.98Se2, Li0.94In0.99Se2, Li0.99In1.02Se2 and Li1.01In1Se2 wafers are shown in Fig. 4a. Clearly, the Li1.01In1Se2 crystal shows a wide resulting luminous band in the PL spectrum from 1.5 to 3.1 eV. To analyze the photoluminescence type of the peak at ~ 3 eV, we measured the luminescence of the Li1.01In1Se2 crystal using a laser power ranging from 1 to 8 mW, as shown in Fig. 4b. Gaussian bimodal fitting was carried out for the photoluminescence peaks between 3.07 and 3.11 eV. The Gaussian peak centers of P1 and P2 are 3.08 eV and 3.09 eV, respectively, according to the deconvolution analysis of the PL spectrum (Fig. 4b). The photoluminescence types of the two peaks are determined by the relationship between the laser power and the PL intensity, as follows:

$$ I \propto L^{k} $$
(1)

where I is the PL intensity, L is the laser power, and k is the index associated with the photoluminescence type. When 0 < k < 1, donor–acceptor pair (DAP) emission occurs. When 1 < k < 2, free exciton (FE) emission occurs. The resulting k values of P1 (k1) ~ 1.19 and values of P2 (k2) ~ 1.28 are shown in Fig. 4c, indicating that the two emission peaks P1 and P2 both correspond to FE emission.

Figure 4
Figure 4
Full size image

(a) PL spectra of LiInSe2 wafers with different stoichiometric ratios at 10 K, (b) PL spectra of LISe-Y at ~ 3 eV at 10 K under different laser powers, and (c) plot of the PL intensity vs. laser power.

The emission peaks in the PL spectra of Li0.99In1.02Se2 to Li0.87In0.98Se2 redshift because of the presence of deeper defects [16,26). Considering the results in the transmittance spectra, the defect concentrations in Li0.99In1.02Se2, Li0.94In0.99Se2 and Li0.87In0.98Se2 successively increase, and the energy of the excitation peaks in the PL also continuously decreases. Due to the high concentration and diversity of deep defects, peak splitting fitting is difficult to perform, and the specific defect types cannot be confirmed. Thus, the red crystals are likely attributed to the high concentration of defects. High-concentration defects also have a strong secondary absorption effect on the intrinsic FE photoluminescence of LiInSe2.

To analyze the possible types of defects inside the LiInSe2 crystals, the photoluminescence (PL) spectra of the two yellow wafers, Y1 and Y2, from Li1.01In1Se2 (yellow crystal) are shown in Fig. 5. Both the Y1 and Y2 wafers show DAP emission at 2.08 eV and 1.98 eV and FE emission at 3.06 eV. Specifically, the peak (DAP)Y1 conforms to a Gaussian distribution, while (DAP)Y2 deviates from a Gaussian distribution.

Figure 5
Figure 5
Full size image

PL spectra of the Y1 and Y2 wafers at 10 K under illumination with a 325 nm laser.

Considering that E(DAP) = Eg − (EA + ED) and the defect depth of the resulting defects from Cui et al.26, the possible defect types are listed in Table 2.

Table 2 Values of the peak wavelength, E(DAP), (EA + ED) and corresponding defect type.
Full size table

For DAP(Y1) and DAP(Y2), the EA + ED values at 2.08 eV and 1.98 eV are approximately 0.98 eV and 1.08 eV; these values are consistent with the values of E (VSe+) + E (LiIn2−) and E (InLi+) + E (LiIn2−), respectively. In addition, the stoichiometric ratio analysis indicates a higher cation content in the yellow crystals, indicating that VSe+ and LiIn2− are more likely to form internal defects in the Li1.01In1Se2 crystal. Note that the inclusion of InLi2+ and VLi− defects in the crystal is the main reason for the red color. In our study, we artificially modulate the stoichiometric ratio to manipulate the type of defects. Due to the added excess Li, the InLi2+ and VLi− defects were suppressed, resulting in the formation of the more desirable yellow crystals. This study also provides a research and experience basis for the subsequent yellow crystal growth of 6LiInSe2.

Conclusion

In conclusion, we modulate the stoichiometric ratio of LiInSe2 to control the color and the defect type of crystals grown via the vertical Bridgman method. The resulting LiInSe2 crystals turn yellow as the stoichiometric ratio of Li increases. The band gap of the yellow Li1.01In1Se2 crystal reaches ~ 2.8 eV, as determined by UV‒vis–NIR transmission and diffuse reflectance spectroscopy; this result is consistent with the theoretical value for LiInSe2. The PL spectra of Li1.01In1Se2 indicate that VSe+ and LiIn2− have internal defects. The modulation of the stoichiometric ratio effectively regulates the types and colors of the defects. Specifically, InLi2+ + VLi− in the red crystals are converted to VSe+ + LiIn2− in the yellow crystals due to the addition of excess Li. This study on yellow 6LiInSe2 crystal growth and defect analysis can effectively promote the practical application of 6LiInSe2 neutron detectors.