Origami-inspired perovskite X-ray detector by printing and folding

The study addresses the challenge of efficiently detecting hard X-ray photons, which require thick absorber layers that hinder charge collection in conventional planar direct-conversion detectors. Metal-halide perovskites offer strong X-ray absorption, efficient charge generation, and good transport, and can be processed on flexible substrates by low-cost printing. The authors propose an origami-inspired folded architecture that reorients thin perovskite absorber pixels parallel to the incident X-ray beam, increasing effective absorption length without thick films, and physically separating pixels to mitigate charge sharing. The research aims to design, fabricate, and characterize a foldable perovskite sensor array operating from 50 to 150 kVp, quantify sensitivity gains from folding, assess spatial resolution potential via edge methods, and examine operational stability under continuous irradiation.

Recent work has established perovskite semiconductors as promising direct X-ray absorbers and photodetectors with solution/printing processability (refs 7–14). Prior edge-on detector concepts in materials like CdZnTe and silicon-strip detectors motivated alternative geometries to increase absorption path length (refs 15–16). The authors previously simulated folded perovskite detectors and demonstrated flexible printed perovskite X-ray detectors (refs 17–18). Reported spatial resolutions for other absorber systems (e.g., MAPbBr3, CsFAGA:Sr, a-Se, Cs2AgBiBr6, MA3Bi2I9, MAPbI3) range from ~3 to 10 lp/mm at MTF=0.2, providing benchmarks that the folded perovskite design aims to surpass.

Device design and fabrication: A foldable perovskite sensor array was fabricated on a 25-µm-thick polyethylene naphthalate (PEN) substrate. Laser-engraved folding lines (LPKF ProtoLaser R4) guided precise folding. The planar pixel stack comprised: 75 nm Au bottom electrode (thermal evaporation); 15 nm sputtered NiOx hole transport layer (Pro Line PVD75, Ar, 1.3×10−3 mbar, 100 W rf, base pressure <1×10−7 mbar); a ~6 µm inkjet-printed triple cation perovskite (TCP) absorber with composition CS0.1(FA0.83MA0.17)0.9Pb(Br0.17I0.83)3 (2–5 kHz, 2000 dpi; additive 2.4 vol% L-α-phosphatidylcholine in DMSO for processing); electron transport layers of 25 nm C60 and 3 nm BCP (thermal evaporation); and 75 nm Au top electrodes (thermal evaporation). Pixel effective area in planar configuration was ~2.21 mm2 from mean electrode widths Wt×Wb = 1.02 mm × 2.17 mm. Interconnects to external readout used aerosoljet-printed silver lines (Optomec Aerosol Jet 5X; silver dispersion ink 736481; Ultrasonic Decathlon; 600 µm nozzle; 4 printed layers; sheath gas 200 sccm; atomizer 100 sccm; TGME addback; temperatures 29 °C; ultrasonic power 530 mA; platen 100 °C), followed by photonic sintering (Pulseforge 1200, 350 V, 200 µs pulses, 150 repeats, 2 Hz). A second PEN foil was added as a protective cover before folding. Folded architecture: The detector foil is folded so that the perovskite absorber is reoriented parallel to the X-ray beam. The absorption efficiency then depends on lateral folding length (several mm) rather than film thickness, increasing effective absorption. Physical separation of folded pixels reduces charge sharing. Laser-engraved lines ensure alignment; printed silver lines and a shared bottom electrode allow simple connection through a flat cable. Electrical and X-ray characterization: I–V characteristics were measured with a Keithley 2450 under dark and X-ray irradiation (50 and 150 kVp). Time-resolved current at zero bias (short circuit) was recorded while varying dose rate via X-ray tube current steps. Dose-rate calibration used a MagicMaX dosimeter (IBA) with RQA semiconductor detector. Two X-ray sources were used: XT9160-DED (Viscom, W target, 0.4 mm Al filtration) for sensitivity; XWT-225 microfocus (X-RAY WorX) for spatial resolution and stability tests. Sensitivity modeling: Theoretical sensitivity Stheo = Jtheo/Dair,theo was computed via spectral integration over photon energy using SPEKTR 3.0 to simulate spectra and a prior simulation framework for detector efficiency and air mass energy-absorption coefficients. Assumptions included: built-in field 0.13 V across 6 µm (0.13/6 V/µm), 75 nm Au top electrode (planar), active folding length 2.1656 mm, TCP density 3.75 g/cm3, electron–hole pair creation energy 4.0–5.5 eV, and mobility–lifetime product 2×10−6 cm2/V for both carriers. Spatial resolution (MTF): Presampled MTF was derived from presampled edge spread function (ESF) measurements of a single folded pixel using a 2 mm tungsten edge under 150 kVp. ESF(x) was acquired for various edge positions, differentiated numerically to obtain LSF(x), then Fourier transformed (FFT in MATLAB) to get MTF(u). An analytical ESF/MTF model (Boone & Seibert, 1994) was also fit to data for validation. Pixel pitch x0 ≈ 4dPEN + dTCP determined sampling cutoff frequency uc = 1/(2x0). Operational stability: Continuous 150 kVp irradiation at Dair ≈ 1.8 mGy/s was applied for >19 h while monitoring detector current response under ambient conditions without encapsulation.

• Planar vs folded sensitivity: Under 50–150 kVp X-ray radiation, planar pixels achieved 25–35 µC/(Gyair cm2). The folded detector achieved several hundred µC/(Gyair cm2), with a record 1409 µC/(Gyair cm2) at 150 kVp, without photoconductive gain and at zero external bias.
• Electrical behavior: I–V curves exhibited diode-like blocking in reverse and transmission in forward bias. Folding caused only slight changes attributed to mechanical stress on electrodes/interconnects; device functionality remained intact. Hysteresis consistent with ionic motion was observed, mitigated by zero-bias operation during X-ray tests.
• Spatial resolution: Presampled MTF exceeds 20 lp/mm at MTF=0.2 under 150 kVp, indicating exceptional potential spatial resolution. Current prototype pixel pitch x0 ≈ 106 µm yields uc ≈ 4.7 lp/mm, limiting realized resolution due to sampling. With thinner PEN (dPEN = 1.5 µm) and similar absorber thickness (dTCP), uc could reach ~41.7 lp/mm to exploit the >20 lp/mm presampled capability.
• Fill factor considerations: With dPEN = 1.5 µm, estimated fill factor could approach ~50%; improvements are possible by increasing absorber thickness and reducing substrate thickness.
• Operational stability: Stable response for >19 h continuous 150 kVp irradiation at Dair = 1.8 mGy/s, totaling Dair,cum = 126.8 Gyair, with no degradation under ambient, unencapsulated conditions. This dose is equivalent to >422,000 chest radiographs (0.3 mGyair per exam).
• Architecture benefits: Folded design increases effective absorption length without thick films, avoids photolithography, physically separates pixels to reduce charge sharing, and simplifies scaling/connection versus stacked edge-on assemblies.

The folded architecture directly addresses the trade-off between absorber thickness (for X-ray absorption) and charge collection efficiency in planar detectors. By reorienting a thin, printed perovskite absorber parallel to the X-ray beam, the effective interaction length is set by the folding length, boosting absorption and sensitivity without thick films that exacerbate trapping and recombination. The observed increase from 25–35 µC/(Gyair cm2) in planar to up to 1409 µC/(Gyair cm2) at 150 kVp in folded mode validates the concept. Physical pixel separation in the folded geometry mitigates charge sharing, contributing to high presampled MTF (>20 lp/mm). The ability to operate at zero bias without photoconductive gain reduces noise and power demands, advantageous for large-area systems. Stability under high cumulative dose and ambient conditions demonstrates robustness of the TCP composition and transport layers, suggesting practical viability. While the prototype’s realized resolution is limited by sampling (pixel pitch), straightforward materials and mechanical optimizations (thinner substrates, optimized folding) can unlock the full spatial-resolution potential. Compared with stacked edge-on designs, the folded sheet offers simpler alignment, connectivity, and scalability using printed interconnects and a single flat cable, making it promising for cost-effective, high-performance X-ray imaging.

This work demonstrates a proof-of-concept origami-inspired folded perovskite X-ray detector fabricated entirely by printing and vacuum deposition without photolithography. The device delivers markedly enhanced X-ray sensitivity in folded configuration (up to 1409 µC/(Gyair cm2) at 150 kVp) at zero bias and exhibits presampled MTF >20 lp/mm under 150 kVp, alongside excellent operational stability (>19 h continuous irradiation; 126.8 Gyair cumulative dose) under ambient, unencapsulated conditions. The folded architecture resolves the absorber-thickness vs. collection-efficiency trade-off and facilitates high spatial resolution by geometric reorientation and pixel separation. Future work should: (i) optimize fill factor via thicker absorbers and thinner substrates; (ii) reduce sampling limitations by minimizing pixel pitch; (iii) characterize and minimize detector noise; (iv) integrate printed thin-film transistors and suitable patterning for 2D arrays (e.g., stripe-like electrodes); and (v) scale to large-area, low-cost detector systems.

• Realized spatial resolution is limited by sampling (pixel pitch); current uc ≈ 4.7 lp/mm prevents exploiting the >20 lp/mm presampled MTF.
• Fill factor and pixel pitch estimates neglect thicknesses of electrodes, transport layers, and interconnects; surface roughness and air gaps also affect performance.
• Presampled MTF reflects potential resolution; actual imaging will be impacted by aliasing and system sampling.
• K-edge fluorescence and Compton scattering persist and can degrade spatial resolution.
• Mechanical folding introduces slight resistance changes in electrodes/interconnects.
• Noise performance was not comprehensively assessed and is identified as an important topic for future research.