KTA single crystals serving as the core material were grown by our research group. The samples measured 6 × 6 × 2 mm³, and no obvious defects were observed. The samples were numbered KTA-X, KTA-Y, and KTA-Z according to the crystal axis perpendicular to the test plane.

Au electrodes were used for detector fabrication. A physical mask was closely bonded to the KTA crystal, and an Au target with 99.999% purity was used as the sputtering source. The KTA crystal was placed in a vacuum chamber, and sputtering was initiated once the pressure fell below 10 Pa. Using ion sputtering and a physical mask, a 100-nm-thick, 2 × 2 mm² gold electrode was deposited onto the crystal surface to fabricate a single-crystal detector. The detector exhibits an Au/KTA/Au structure. This fabrication method is simple to perform and offers good reproducibility.

The X-ray source was an X-ray tube (E7252X, Toshiba, Japan) operated at 40-150 keV; the tube current was adjusted from 10 mA to 100 mA to vary the X-ray dose rate. Current-voltage (I-V) and current-time (I-t) curves were recorded using a 6517B high-precision electrometer (Keithley, USA). Measurements were conducted under optically isolated conditions (in the dark) to ensure that the recorded current originated entirely from X-ray excitation rather than from photo-generated carriers. In X-ray detection, the effective device area (4 mm²) is defined as the metal area on the device surface in contact with the probe. All detector performance tests were conducted on unpackaged devices at room temperature.

The X-ray imaging performance of KTA X-ray detectors was measured on a custom-built platform. The X-ray detector sample used for imaging was coated with a gold electrode with an area of 6 × 6 mm² on one side and a planar array electrode on the other side. The area of each small electrode was 0.5 × 0.5 mm², and the distance between the electrodes was 0.1 mm. The imaging object was placed between a planar array X-ray detector and an X-ray beam, and the electrical signals of the detector were recorded at different positions. The imaging object was chosen to be smaller than the planar array detector so that the whole object could be imaged in a single acquisition without moving the detector, thereby avoiding unwanted errors.

The KTA crystal belongs to the orthorhombic crystal system with space group Pna2₁, and its polar axis extends along the crystallographic Z-axis. Figure 1A shows KTA samples (polished on both sides) along three principal axis directions, which were used for the fabrication of X-ray detectors. The device configuration is displayed in Figure 1B. We further compared the X-ray absorption coefficients of KTA with those of some previously reported X-ray detection materials [Figure 1C]. Within the typical detection range, the coefficient of KTA crystals is lower than that of Se, Ga₂O₃, and CdZnTe due to the smaller atomic numbers of K, As, Ti, and O. Specifically, at 40 keV, a 2-mm-thick KTA crystal absorbs nearly 100% of the incident photons [Figure 1D]. Resistivity is a key parameter affecting X-ray detection performance. It determines the dark current level of the material, which in turn influences the detector’s sensitivity. To evaluate the resistivity of KTA crystals, we employed the two-probe method to measure the resistivity of KTA samples along different crystal orientations at room temperature. The results reveal that the resistivities along the X, Y, and Z axes are 4.02 × 10⁹, 7.69 × 10⁹, and 5.15 × 10⁶ Ω·cm, respectively, exhibiting significant anisotropy [Figure 1E and F]. This anisotropy stems from differences in atomic bonding patterns within the crystal structure and the directionality of carrier transport paths. Specifically, the resistivity along the Z-axis (polar axis) is approximately three orders of magnitude lower than that along the X- and Y-axes, indicating that the migration resistance of charge carriers in the Z-axis is lower. However, when the resistivity is too low, it leads to excessively high dark current, and photocurrent shielding effects arise, significantly reducing the detector’s sensitivity and signal-to-noise ratio. This hinders accurate measurements and practical applications [Supplementary Figure 1]. Therefore, polar-axis orientation is not an ideal choice for X-ray detection. To meet the requirements for high signal-to-noise ratio and high sensitivity in practical applications, the non-polar axis directions are the most appropriate choice.

As a key parameter characterizing charge transport capability and carrier extraction efficiency, the carrier mobility-lifetime product (μτ) directly determines the ultimate responsivity and quantum efficiency of photodetectors. To quantitatively evaluate the carrier transport properties of KTA crystals, we systematically investigated their voltage-dependent photoconductive behavior by applying a linearly swept DC bias from 0 to 100 V at an X-ray dose rate of 55.0374 μGy_air·s^-1 (120 keV). Additionally, the dynamic photocurrent response was simultaneously monitored under dark and X-ray irradiation conditions to determine the field-induced charge collection efficiency. The photocurrent-voltage relationship was fitted using the Hecht equation^[29]:

(1) $$ \begin{equation} \begin{aligned} I=\frac{I_{0} \mu \tau V}{L^{2}}\left(1-\exp \left(\frac{-L^{2}}{\mu \tau V}\right)\right) \end{aligned} \end{equation} $$

where I₀ is the saturation current, µ is the carrier mobility, τ represents the carrier lifetime, V is the bias, and L is the thickness of the sample. The experimentally measured carrier mobility-lifetime product is shown in Figure 1G, Note that for a photoresistor-type device with symmetric planar metal-semiconductor-metal contacts, the single-carrier Hecht equation yields an effective μτ value that represents a weighted average of both electron and hole contributions, rather than the μτ of an individual carrier type^[29-31]. This is because, under uniform X-ray excitation throughout the crystal thickness, both electrons and holes are generated and contribute to the measured photocurrent, and the symmetric contact geometry does not provide the carrier selectivity needed to isolate one type^[32]. Therefore, the μτ values reported in this paper should be understood as an effective transport parameter characterizing the overall charge collection efficiency of the KTA detector in a photoresistor configuration. To measure the μτ values for electrons and holes independently, it is necessary to employ techniques such as α-particle excitation, which can selectively generate carriers near a specific electrode.

The high ionic charge density of Ti⁴⁺ and As⁵⁺ in KTA leads to strong Coulombic interactions and large ionic radii, thereby suppressing ion mobility and enhancing lattice stability. Temperature-dependent transient reverse currents of KTA (303-333 K) reveal ion migration dynamics [Supplementary Figure 2]. Under applied bias, field-driven ion migration establishes space-charge polarization until the current reaches steady state. Abrupt voltage removal initiates ion redistribution through vacancy diffusion, producing a concentration-gradient-induced reverse dark current that relaxes to a steady state. Ion activation energies for KTA crystals, fitted along the (100), (010), and (001) planes [Supplementary Figure 3], together with the dark current drifts measured along the polar axis direction [Supplementary Figure 4] collectively demonstrate weak ionic mobility. For comparison, halide perovskite detectors such as MAPbBr₃ exhibit ion activation energies as low as ~73 meV, leading to pronounced dark current drift and performance degradation under applied bias^[18]. In contrast, the substantially higher activation energy of KTA (> 200 meV) effectively suppresses field-driven ion migration.

Sensitivity, as a core performance metric for direct-type X-ray detectors, characterizes the detector’s ability to respond to X-ray signals. It is typically quantified by the photocurrent density or charge collection efficiency generated per unit dose of X-ray irradiation. For KTA detectors, their high carrier mobility-lifetime product (μτ) effectively enhances charge collection efficiency and thus improves the conversion of X-ray photons into electrical signals. Furthermore, when oriented along a non-polar axis (such as the X-axis or Y-axis), the high resistivity can significantly suppress the dark current, reducing background noise interference and thereby optimizing the sensitivity. To systematically evaluate the X-ray detection performance of KTA crystals, we measured the current-time (I-T) response curves of KTA detectors over a bias voltage range of 0 to 200 V. Figure 2A shows the I-T response curves under different bias voltages, and Figure 2B shows the variation in photocurrent as the X-ray dose rate increases from 2.94 to 7.43 μGy_air·s^-1. The results reveal that the carrier transport and collection efficiency increase significantly with rising bias voltage, leading to a corresponding enhancement of the photocurrent response. However, as the voltage rises further, the dark current increases along with the photocurrent. Sensitivity therefore reaches a saturation point, beyond which it no longer improves. Based on the photocurrent data, the detector’s sensitivity can be calculated using the following formula:

(2) $$ \begin{equation} \begin{aligned} S=\frac{\Delta I}{(A \times \Delta D)} \end{aligned} \end{equation} $$

where ΔI represents the difference between photocurrent and dark current, A represents the Au electrode area, and ΔD represents the changes in X-ray dose rate. A linear correlation exists between the current density generated under X-ray irradiation and the X-ray dose rate [Figure 2B and Supplementary Figure 5]. The detector's sensitivity was determined by fitting the slopes of these current density versus dose rate plots at various bias voltages [Figure 2C and D]. The experimental data show that the detector exhibits higher sensitivity under lower radiation dose rates. This phenomenon is primarily because the atomic numbers of the elements in KTA are not particularly high, resulting in limited X-ray absorption capacity. Meanwhile, excessively high dose rates lead to premature carrier recombination, which in turn affects the detector’s sensitivity. The maximum sensitivity of the KTA directional X-ray detector is 192.96 μC Gy_air^-1·cm^-2 (200 V, 70 keV). Under an applied bias voltage of 200 V and an X-ray dose rate range of 17.4432-34.7328 μGy_air·s^-1, this value is 9.6 times higher than that of the α-Se detector (20 μC·Gy_air^-1 cm^-2, measured at 20 keV). Furthermore, the sensitivity exhibits strong voltage dependence. However, when the bias voltage is too high, leakage current increases significantly and carrier recombination becomes severe. At the same time, factors such as breakdown or thermal effects significantly reduce the material’s charge collection efficiency and signal generation capability, so the sensitivity no longer increases with rising voltage.

To evaluate the performance stability of the KTA detector under continuous irradiation conditions, we subjected the device to frequent X-ray irradiation at a dose rate of 7.43 μGy_air·s^-1 and recorded the I-T curves of the switching process over time [Figure 2E]. Notably, the KTA X-ray detector exhibited high photocurrent and stable dark current characteristics. The unencapsulated device retained 95.5% of its initial photocurrent even after being exposed to air for over 1,000 s. The 4.5% loss can be attributed mainly to electrode damage caused by prolonged probe contact, suggesting that operational stability can be further enhanced by optimizing the contact design. Subsequently, after the device was left at room temperature (25 ± 2 °C in ambient air) for 30 days and retested, the photocurrent density showed no significant decrease [Figure 2F], confirming its excellent stability under routine conditions. Admittedly, this stability assessment was limited to intermittent irradiation tests. To verify its practical applicability in clinical and industrial settings, a systematic evaluation of long-term operational stability under continuous X-ray irradiation is still required. However, under standard test conditions described above, the photocurrent values remained highly stable throughout the observation period. Based on this, we can cautiously conclude that the KTA detector has the potential for long-term service under standard operating conditions. Extreme or continuous irradiation scenarios remain within the scope of future optimization and validation and do not undermine the validity of the current conclusion.

The detection limits of KTA detectors are derived from current-time (I-T) curves measured under X-ray exposure. Specifically, the minimum detection limit is defined as the dose rate at which the signal-to-noise ratio (SNR) equals 3^[30,31]. This SNR is calculated using the following equation:

(3) $$ \begin{equation} \begin{aligned} S N R=\frac{\left(I_{{photo }}-I_{{dark }}\right)}{\sqrt{\frac{1}{N} \sum_{i}^{n}\left(I_{i}-I_{{photo }}\right)^{2}}} \end{aligned} \end{equation} $$

Where I_photo represents the average photocurrent, I_dark represents the average dark current without X-ray irradiation, and I_i represents the photocurrent. Figure 2G shows the signal-to-noise ratio of KTA single-crystal detectors. The lowest detection limit of the KTA detector reaches 0.09 μGy_air·s^-1 (50 V, 40 keV). For comparison, conventional X-ray detectors exhibit detection limits of 5 μGy_air·s^-1 for α-Se (20 keV)^[32], 69.5 μGy_air·s^-1 for β-Ga₂O₃:Mg SC (50 keV)^[34], and 9.8 μGy_air·s^-1 for β-Ga₂O₃:Al (50 keV)^[35], and the standard medical test result is 5.5 μGy_air·s^-1[29]. These lower detection limits demonstrate the potential of KTA X-ray detectors for medical examinations. A comparison of the sensitivity and detection limits of the KTA detector with other reported materials is shown in Figure 2H and Table 1.

By utilizing the intrinsic electric field inherent in KTA, we developed a self-powered X-ray detection technique. The experimental setup is shown in Supplementary Figure 6. This approach efficiently collects and transports charge carriers generated by X-ray excitation, eliminating the need for an external voltage. As shown in Figure 3A, the dark current is less than 1 pA at 0 V bias, significantly lower than that under an applied bias. Furthermore, at 0 V bias, the detector exhibits both a substantial photocurrent response [Figure 3B] and excellent stability [Figure 3C]. As the applied X-ray dose rate was gradually increased from 2.94 to 7.43 μGy_air·s^-1, the photocurrent density along the Y-axis increased from 0.24 to 0.74 nA·cm^-2, while that along the X-axis increased from 0.40 to 0.73 nA·cm^-2 [Figure 3D]. By fitting the slopes, the sensitivity of the KTA detector along the X-axis direction was determined to be 68.84 μC·Gy_air^-1 cm^-2 at 0 V, and along the Y-axis direction, it was 110.72 μC·Gy_air^-1 cm^-2 [Figure 3E]. This performance improvement is attributed to the unique crystal structure of KTA: under zero-bias conditions, this structure exhibits high carrier mobility and low recombination rates, enabling efficient collection of X-ray-induced carriers. The combination of low dark current and high sensitivity gives KTA detectors an excellent signal-to-noise ratio [Figure 3F]. The measured detection limits of the KTA detector are 0.113 μGy_air·s^-1 along the Y-axis direction and 0.061 μGy_air·s^-1 along the X-axis direction, both well below the dose rates required for clear medical imaging. Furthermore, the self-powered mode eliminates the risk of electrical breakdown caused by high external bias, significantly enhancing the detector’s long-term stability and safety for clinical use. Consequently, KTA-based self-powered detectors, with their compact structure, low power consumption, and high performance, hold significant application potential in the field of low-dose medical imaging.

In self-powered detectors, no bias voltage is applied, so the dark current is reduced to the lowest level inherent to the material. This characteristic resolves the significant problem of conventional X-ray detectors where both photocurrent and dark current increase simultaneously, and it also reduces photon noise to some extent, thereby lowering the detection limit. In addition, the performance of the KTA detector in self-powered mode surpasses that of many previously reported self-powered X-ray detectors such as GaN^[43], MAPbBr₃^[44], and MhyPbBr₃ (Mhy = methylhydrazine)^[45], in the balance of sensitivity and detection limit. It can thus be concluded that polar oxide crystals exhibit good self-powered detection performance. Table 2 compare the detection limits of some previously reported self-powered X-ray detectors with those of the KTA self-powered detector. In practical applications, reducing the response time of X-ray detectors not only significantly lowers radiation exposure but also improves imaging efficiency, which fully meets safety requirements in fields such as radiation protection and clinical practice. To this end, we tested the response time of the KTA crystal X-ray detector. Figures 3G and H show the photocurrent response time of the KTA X-ray detector at an X-ray energy of 40 keV. In self-powered mode, the response time was approximately 320 ms, indicating a rapid signal response. This ability to effectively capture rapidly changing processes and provide precise time-resolved information for clinical diagnosis or real-time imaging further validates its application potential. Meanwhile, we compared its performance with that of self-powered detectors reported in the literature [Figure 3I], finding that the KTA detector achieves an effective balance between sensitivity and detection limit.

In self-powered mode, the outstanding X-ray detection performance of the KTA detector is attributed to the built-in electric field of the polar crystal, which enables a synergistic combination of high sensitivity, fast response, high stability, and low detection limits. To elucidate the microscopic mechanism underlying the polarity of the KTA crystal, we systematically analyzed the atomic arrangement within the unit cell. The data reveal that the KTA crystal is a polar oxide crystal composed of distorted TiO₆ octahedra and AsO₄ tetrahedra. Its crystallographic parameters, obtained via computational analysis, are detailed in Supplementary Table 1. The distorted TiO₆ octahedra within the unit cell generate intrinsic dipole moments. By performing a vector sum of these dipole moments, we obtained the net dipole moment of a single KTA unit cell. This method directly characterizes the strength of the built-in electric field induced by crystal polarization. The significant dipole moment observed in the KTA crystal [Figure 4A] provides direct experimental evidence for the presence of a strong built-in electric field, with a measured dipole moment of 25.93 × 10^-4 esu·cm·A^-3. To quantitatively characterize the built-in electric field strength, we employed a series of positive and negative bias voltage gradients to induce an asymmetric photocurrent response in the device.

Under experimental conditions, the detector is simultaneously exposed to X-ray irradiation and an applied bias voltage. Gradual voltage tuning enables the identification of the null bias condition for zero photocurrent output. At this equilibrium state, the applied voltage completely counteracts the built-in electric field and eliminates the photocurrent response. The strength of the built-in electric field can then be calculated using the following formula:

(4) $$ \begin{equation} \begin{aligned} E_{{built-in }}=\frac{V_{{null }}}{d} \end{aligned} \end{equation} $$

where V_null is the voltage at which the photocurrent vanishes, and d is the thickness of the active layer.

As shown in Figure 4B-D, the photocurrent response gradually weakens and eventually disappears as the applied negative voltage increases. Ultimately, the built-in electric field strength of KTA was measured to be 35 V·mm^-1 in the X-axis, 40 V·mm^-1 in the Y-axis, and 4 V·mm^-1 in the Z-axis. This is because KTA crystals exhibit a pronounced multi-domain structure at room temperature; the 180° antiparallel domains along the Z-axis suppress the macroscopic polarization field. Since the crystals were not subjected to deliberate single-domain treatment, the presence of domain walls significantly weakens the built-in electric field along the Z direction^[50].

The discrepancy between the experimentally measured spontaneous polarization intensity and the calculated unit cell dipole moment may stem from a variety of factors, including local structural distortions, interfacial interactions, and the limitations of the experimental measurements. Despite discrepancies between measured values and theoretical calculations, the overall trend is consistent with the theoretical predictions. It can be concluded that the built-in electric field primarily originates from the material’s spontaneous polarization, while interfacial effects play a secondary role; together, these two factors enable efficient charge extraction and transport in the absence of an applied bias.

The above results demonstrate that the built-in electric field in polar oxide crystals can effectively promote carrier transport, thereby enabling excellent self-powered detection performance. Based on this outstanding detection performance, we further evaluated the imaging capabilities of the self-powered KTA detector. We successfully assembled a complete imaging system, whose structure is shown in Figure 5A. An arrow-shaped metal object (6 mm long × 4.5 mm wide) was selected as the imaging target and placed between the detector and the X-ray source. An X-ray energy of 40 keV was applied [Supplementary Table 2], and imaging tests were conducted in self-powered mode. Figure 5B presents the X-ray imaging of this arrow-shaped metal object. The image clearly reproduces the arrow’s complete outline and edge details, demonstrating that the KTA crystal self-powered detector possesses excellent spatial resolution. This result is consistent with the fast response and high sensitivity described earlier, indicating that the KTA detector is capable of resolving object features on the submillimeter scale in self-powered mode. However, the current spatial resolution is limited by various factors, including the pixel pitch of the electrodes and the crystal thickness. The pixel pitch determines the limiting resolution of the sampling, while the crystal thickness affects lateral diffusion during charge transport and crosstalk between adjacent pixels. By further optimizing the pixel geometry, adjusting the crystal thickness, and employing collimation techniques, it is expected that the resolution of this detector in practical imaging applications will be significantly improved.

As described above, we comprehensively analyzed the performance of the KTA detector under X-ray irradiation at various dose rates. The results indicate that the sensitivity of the KTA detector along the Y-axis is superior to that along the X-axis under all tested dose rates [Figure 6A-C]. This phenomenon can be understood by examining the photocurrent response at low dose rates. At the lowest measured dose rate of 2.94 μGy_air·s^-1, the photocurrent density along the X-axis (0.40 nA·cm^-2) was approximately 66.7% higher than the value along the Y-axis direction (0.24 nA·cm^-2). As the dose rate increased to 7.43 μGy_air·s^-1, the photocurrent along the Y-axis became comparable to that along the X-axis (0.74 and 0.73 nA·cm^-2), which is consistent with the higher overall sensitivity slope along the Y-axis (110.72 and 68.84 μC·Gy_air^-1·cm^-2). To explore the underlying causes, we analyzed the crystal structure of the KTA. This anisotropy can be attributed to structural differences along the crystallographic axes: specifically, the parallel, twisted TiO₆ chains along the Y-axis form a highly ordered in-plane network of channels, significantly enhancing charge transport. In contrast, the structure along the X-axis exhibits a more staggered arrangement [Figure 6D and E]. This difference directly determines the sensitivity of the detector along the different axes. Quantitatively, the Y-axis sensitivity (110.72 μC·Gy_air^-1·cm^-2) exceeds that of the X-axis (68.84 μC·Gy_air^-1·cm^-2) by approximately 60% in self-powered mode, consistent with the higher built-in electric field and the more continuous charge-transport pathways provided by the parallel TiO₆ chains.

Furthermore, the comprehensive analysis of the experimental data indicates that the lowest detection limit is achieved along the X-axis direction in self-powered mode. This is attributed to a better balance among the built-in electric field strength in the X-axis direction, the charge separation efficiency, and the dark current level. Compared with the stronger built-in electric field along the Y-axis (which may cause slightly higher dark current), the electric field strength along the X-axis ensures that the carrier drift velocity is sufficient to capture weak X-ray signals while effectively suppressing unwanted background current interference. At the same time, although the interlaced crystal structure along the X-axis direction reduces the overall efficiency of charge transport, it also decreases the probability of non-radiative carrier recombination, thereby improving the signal-to-noise ratio for weak signals. The combined effect of these factors enables X-ray detection at lower doses along the X-axis in self-powered mode, thereby achieving the lowest detection limit.

Experimental results show that the structural anisotropy of KTA crystals determines the anisotropic transport of charge carriers and the distribution of the built-in electric field, thereby influencing the detector’s sensitivity and detection limit. This correlation between structure and performance reveals a key physical principle: by controlling the crystal orientation and structure, the balance between sensitivity and detection limit can be systematically adjusted. This design principle applies not only to KTA but can also be extended to other polar oxide detector materials, providing theoretical guidance for the development of next-generation oriented X-ray detectors tailored to specific application scenarios.