Amorphous and polycrystalline photoconductors for direct conversion flat panel x-ray image sensors

Abstract

In the last ten to fifteen years there has been much research in using amorphous and polycrystalline semiconductors as x-ray photoconductors in various x-ray image sensor applications, most notably in flat panel x-ray imagers (FPXIs). We first outline the essential requirements for an ideal large area photoconductor for use in a FPXI, and discuss how some of the current amorphous and polycrystalline semiconductors fulfill these requirements. At present, only stabilized amorphous selenium (doped and alloyed a-Se) has been commercialized, and FPXIs based on a-Se are particularly suitable for mammography, operating at the ideal limit of high detective quantum efficiency (DQE). Further, these FPXIs can also be used in real-time, and have already been used in such applications as tomosynthesis. We discuss some of the important attributes of amorphous and polycrystalline x-ray photoconductors such as their large area deposition ability, charge collection efficiency, x-ray sensitivity, DQE, modulation transfer function (MTF) and the importance of the dark current. We show the importance of charge trapping in limiting not only the sensitivity but also the resolution of these detectors. Limitations on the maximum acceptable dark current and the corresponding charge collection efficiency jointly impose a practical constraint that many photoconductors fail to satisfy. We discuss the case of a-Se in which the dark current was brought down by three orders of magnitude by the use of special blocking layers to satisfy the dark current constraint. There are also a number of polycrystalline photoconductors, HgI(2) and PbO being good examples, that show potential for commercialization in the same way that multilayer stabilized a-Se x-ray photoconductors were developed for commercial applications. We highlight the unique nature of avalanche multiplication in a-Se and how it has led to the development of the commercial HARP video-tube. An all solid state version of the HARP has been recently demonstrated with excellent avalanche gains; the latter is expected to lead to a number of novel imaging device applications that would be quantum noise limited. While passive pixel sensors use one TFT (thin film transistor) as a switch at the pixel, active pixel sensors (APSs) have two or more transistors and provide gain at the pixel level. The advantages of APS based x-ray imagers are also discussed with examples.

Keywords: detector; direct conversion; x-ray image sensor; x-ray photoconductor.

Figure 1.

A simplified schematic illustration of a FPXI and its peripheral electronics that drive the sensor operation. At its base, the FPXI has a TFT-AMA with M × N number of pixels (e.g., 2,816 × 3,584) as a substrate. There is an x-ray photoconductor (a-Se) deposited on the AMA, and a top electrode to apply a voltage to the photoconductor. The x-rays absorbed in the a-Se layer above pixel (1,1) generate charges that drift and become collected and stored on the storage capacitor C₁₁ at this pixel. If a signal is applied to the gate G₁₁ of the TFT at pixel (1,1), by activating the gate line G₁, it conducts (it switches on) and the charge Q₁₁ on C₁₁ is transferred to the data line and hence to the external electronics. Data lines feed into charge amplifiers. The C₁₁ and TFT structures are not inside the glass substrate but on the surface of the glass substrate as indicated in Figure 2. The activation of the gate line G₁ allows the charges Q₁₁, Q₁₂, Q₁₃ etc. to be read at the same time to a multiplexer and then onto a digitizer etc. At the end the read-cycle for the 1st row, the 2nd row is addressed via G₂ and so on until all the rows are sequentially addressed, and hence the whole image is read out.

Figure 2.

A simplified schematic diagram of the cross section of a single pixel with a TFT. The charges generated by the absorption of x-rays drift towards their respective electrodes. The capacitor C₁ integrates the induced current due to the drift of the carriers which results in a stored charge Q₁ on C₁. The TFT is normally off and is turned on when the gate G₁ is addressed. (Not to scale).

Figure 3.

Anrad’s mammographic FPXI AXS-2430 (previously LMAM) is used in the USA and European mammography markets. The field of view is 24 cm × 30 cm. These FPXIs have a pixel pitch of 85 μm, high DQE, high MTF, high contrast, high dynamic range, high patient throughput, and are capable of tomosynthesis.

Figure 4.

Two typical x-ray images from an a-Se FPXI. Left, a typical x-ray image of a breast. Right, an x-ray image of a hand.

Figure 5.

Linear attenuation coefficient α and depth, 1/α vs. photon energy for various photoconductors of interest.

Figure 6.

(a) X-ray photons are incident on a central reference pixel C and are absorbed in the photoconductor over C. The x-ray generated electrons and holes drift respectively towards positive and negative electrodes, the latter being pixellated. There is no trapping and recombination and all the generated charges are collected. (b) Holes are trapped. These trapped charges result in a loss of sensitivity and resolution.

Figure 7.

The comparison of A_Q (QE) and η_CC (CCE) contributions to the x-ray sensitivity for two different attenuations, i.e., for two different photon energies. Full colors are for Δ = 1/4 (δ = L/4) and hatched colors are for Δ = 1 (δ = L). HCCE and ECCE are the hole and electron collection efficiencies respectively. The radiation receiving electrode is positively biased. For a unipolar semiconductor either HCCE or ECCE would be zero.

Figure 8.

An example of one simple linear cascaded systems model recently used in modeling the DQE of a PbO FPXI, which neglects K-fluorescence reabsorption. After [101].

Figure 9.

DQE(f) vs. spatial frequency f for a negatively biased PbO photoconductive x-ray sensor at three different applied fields, F = 0.5 V μm⁻¹, 1 V μm⁻¹, and 2 V μm⁻¹. The symbols are the experimental points reported by Simon and coworkers and the solid line is the calculations based on a cascaded linear system model; further details may be found in [60,101].

Figure 10.

The effects of charge trapping on the resolution (MTF) depends on the type of carriers that have been trapped; whether carriers were drifting to the top or bottom electrode. C is the central (reference) pixel and L and R are the neighboring left and right pixels. The transient currents flowing into the pixels are integrated and eventually yield the collected charges at the pixels.

Figure 11.

Measured presampling MTF vs. f of a polycrystalline CdZnTe detector in comparison with a calculated MTF in which deep trapping of charge carriers is included in the model. Blurring due to charge carrier trapping in the bulk of the photoconductor cannot be neglected. The detector thickness is 300 μm and the pixe pitch is 150 μm. After [103]. Data from [104].

Figure 12.

(a) In a single layer of a-Se sandwiched between two electrodes, the dark current is due to the injection of holes (most dominant) and electrons from the positive and negative contacts respectively as well as some thermal generation of electron and hole pairs or hole emission from defect states. (b) In the case of an x-ray photoconductor, there are two thin layers between the a-Se and the electrodes. The hole-trapping layer traps holes and allows electron transport (an n-like layer) and the electron-trapping layer traps electrons and allows hole transport (a p-like layer). The structure is often referred to as an nip type a-Se photoconductor. The radiation receiving side is the positive electrode.

Figure 13.

The best reported values to date of dark current density for a-Se and polycrystalline photoconductive layers. Note that most of these are measured at relatively low applied electric fields where it is questionable that the charge collection efficiency is adequate. It is not possible to scale these to the same field as the field dependence of the dark current is rarely linear and in general is unknown. All polycrystalline layers are labeled as deposited by physical vapour deposition (PVD), screen printing (SP) or close space sublimation (CSS). Solid colors represent values obtained from films that have not yet been used to obtain x-ray images, hashed bars represent values from demonstrated x-ray imagers. The grey hashed area represents the acceptable range for dark current in an FPXI. Data have been taken from various sources, including the following: a-Se (i-layer and n-i-p) from [15], a-Se (n-i) from [118], HgI₂ (PVD at 0.25 V/μm and SP) from [51], HgI₂ (PVD at 0.4 V/μm) from [122], PbI₂ (PVD) from [56], PbI₂ (SP) from [48], Cd_0.95Zn_0.05Te from [58], PbO (PVD) from [59], PbO (SP) from [123], PbBr₂ and HgBr₂ from [124] and BiI₃ from [125].

Figure 14.

Top left. A HARP tube is a TV pick-up (video) tube with avalanche gain; it is called a Harpicon. Top right. A schematic illustration of the HARP and its operation under avalanche. Bottom. A snap-shot image from a real time movie of a rainbow formed under moonlight at night at Iguazu Falls, Brazil, taken by a HDTV-Harpicon. (Courtesy of Dr. Kenkichi Tanioka, NHK, Japan).

Figure 15.

(a) A modified-HARP structure with an RIL (resistive interface layer) and the development of a fully electroded image sensor, which can be used in direct and indirect conversion detectors. (b) Experimentally measured field dependence of avalanche gain for 15 μm thick modified-HARP structure (open circles) in comparison with the theoretical field dependence of the avalanche gain (crosses) for the same layer thickness.

Figure 16.

TOF signal from a-Se HARP blocking structure with a RIL in avalanche regime for F = 100 V μm⁻¹. The inset shows the dependence of the drift mobility on the applied field.

Figure 17.

Three transistor APS pixel circuit and the timing diagram.

Figure 18.

A direct conversion sensor based on coating a 64 × 64 pixel array with a-Se.

Figure 19.

X-ray resolution image test from (a) a commercial FPD14 PPS array at 1.6 lp/mm shown in center of image (b) prototype 64 × 64 APS array at 2.0 lp/mm shown in center of image and (c) the resolution target.