www.advhealthmat.de www.MaterialsViews.com

COMMUNICATION

Small-Dose-Sensitive X-Ray Image Pixel with HgI2 Photoconductor and Amorphous Oxide Thin-Film Transistor Jae Chul Park, Pyo Jin Jeon, Jin Sung Kim, and Seongil Im* Active matrix flat panel X-ray imager (AMFPI) has undergone extensive research since it has many advantages such as realtime operation, improved image quality, computer-aided diagnosis, and electronic filing over conventional analog X-ray imaging technologies.[1–4] The AMFPIs have thus been successfully introduced to such a variety of medical X-ray imaging applications as radiography, mammography, fluoroscopy, angiography, and radiotherapy.[5–10] It is classified into indirect conversion and direct conversion types by X-ray detection principle. In the indirect conversion, energetic X-rays are captured and converted to visible light through scintillator layer such as cesium iodide.[1,2,5,6] An array of amorphous silicon p-i-n photodiodes located below the scintillator then converts visible photons to electronic signal, so that amorphous silicon thin-film transistors (a-Si TFTs) transports the signal charge carriers to external readout integrated circuit (IC). On the other hand, the direct conversion uses photoconductive materials to directly convert X-rays to electrical signals.[7–10] Although indirect conversion type may take advantage of the excellent photodetection capabilities of photodiode, it has such essential inherent drawbacks as spatial resolution degradation and process complexity, which, respectively, originate from light spreading of scintillator and from building photodiode on each pixel of the TFT array. Hence, the direct conversion type device, which displays rather superior spatial resolution and an improved signal-tonoise ratio (SNR) due to direct charge conversion without photodiodes, has drawn much attention from many researchers. Conventional direct type flat panel X-ray image sensor is currently composed of amorphous selenium (a-Se) and a-Si TFT, respectively, for a photoconductor and a switch; this combination is classified as passive pixel sensor (PPS) for AMFPI system.[11,12] However, the direct type has a lot of issues to be improved for modern medical X-ray imaging applications, which generally involve fast image motion but still require highly resolved image quality at a low X-ray dose. Indeed, it is important to produce high quality images even with low X-ray dose, particularly because high-dose X-ray exposure would be harmful in any cases. For the purpose of low-dose

J. C. Park, P. J. Jeon, J. S. Kim, Prof. S. Im Institute of Physics and Applied Physics Yonsei University 262 Seongsanno Seodaemun-gu, Seoul 120–749, Korea E-mail: [email protected]

DOI: 10.1002/adhm.201400077

Adv. Healthcare Mater. 2014, DOI: 10.1002/adhm.201400077

X-ray/high-resolution image, enhancing the sensitivity of image sensor pixel becomes crucial; selective combination of more sensitive photoconductive material and advanced switching TFT for a sensor pixel would eventually improve the image resolution. The advanced photoconductive materials should have a low electron-hole pair creation energy (W) and a high mobility × carrier life time product (µτ) in general. Switching TFTs with a high field-effect mobility, low leakage current, and a low channel trapping noise may then be required for an ideal pixel formation. In the present study, we selected mercuric iodide (HgI2) as an advanced photoconductive material, which has lower cost and higher X-ray sensitivity than commercial a-Se or other materials such as cadmium telluride (CdTe), lead iodide (PbI2), and lead oxide (PbO)[13–15] as discussed and shown in Table S1 (Supporting Information). According to Table S1, our polycrystalline HgI2 material is better than others in overall properties such as W and µτ except those of single crystal Si and polycrystalline Cd(Zn)Te. However, those two materials need very thick film or high process temperature over 600 °C). For an advanced X-ray pixel switch array, we for the first time adopted amorphous oxide-based TFTs, which have incomparably higher channel field-effect mobility along with lower leakage current than those of a-Si TFTs.[16–18] The a-Si TFT has conventionally been used for the last 20 years since flat panel X-ray sensor was developed.[1,2] Its poor electrical properties could not cope with the modern X-ray image sensors, which either require fast moving pictures or require low charge losses during charge integration. Replacing conventional a-Se photoconductor/a-Si TFT switch array with HgI2/a-HfInZnO TFT combination, our novel X-ray image sensor thus demonstrated a state-of-the art properties in X-ray imaging applications. A HgI2 paste was fabricated by a particle-in-binder (PIB) method, which uses mercuric iodide powders incorporated into a polymer binder.[19,20] When a 100-µm-thick HgI2 film was coated by a screen-print process on the glass substrate, it clearly showed crystalline phase unlike a-Se film, due to the HgI2 particles embedded in the polymer binder, as respectively observed from X-ray diffraction (XRD) analysis of Figure 1a and scanning electron microscopy results of Figure 1b,c. The HgI2 film appears red in color as shown in the inset of Figure 1b. Based on the PIB-fabricated HgI2 photoconductor, we formed image sensor pixels. Figure 2a,b, respectively, display top and cross-section view schemes of our PPS architecture, prepared with an optical microscopy and a transmission electron microscopy image. But more detailed cross section of a pixel is schematically shown in Figure 2c. When X-ray was received by the

© 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

wileyonlinelibrary.com

1

www.advhealthmat.de

COMMUNICATION

www.MaterialsViews.com

Figure 1. Physical properties of PIB-coated HgI2. a) The X-ray diffraction (XRD) spectra of amorphous selenium (Se) deposited by thermal evaporation and the spectra of mercuric iodide (HgI2) printed by particle-in-binder (PIB) method. b) The plan view and c) cross sectional SEM images of PIB-coated HgI2. The inset shows the photograph of screen-printed 100-µm-thick HgI2 film (red) on glass substrate. Polycrystalline phases are observed from b) and c) SEM images.

thick HgI2 photoconductor, generated holes and electrons are, respectively, collected to the top (TE) and bottom electrode (BE), while BE is connected to storage capacitor (CST or interdielectric) and simultaneously to the source electrode of a-HfInZnO TFT. A pixel size is 139 µm × 139 µm and its fill factor is 53% while the channel width/length (W/L) ratio of TFT is 10/10 µm. According to the pixel scheme and circuits of Figure 2c,d, the TE of photoconductor (PTE) is connected to a bias line of over −50 to −100 V (Figure 2d). The common ground line, which is the metal field shield (FS), is formed on the back channel of TFT to protect the back channel from being turned on by the high voltage applied to the photoconductor (Figure 2c). The drain of TFT is connected through a sense line to the charge amplifier, which is, in turn, connected to readout circuits. Right side of Figure 2d shows the timing diagram of our PPS for X-ray imaging. The CST is positively precharged to have a reference voltage (VREF) when the switching TFT (TSW) was on. After TSW is turned off, the Mo target-generated X-ray with 30 keV is then exposed to the HgI2 under VPTE ( = −50 to −100 V). During the radiation, the VREF in CST decreases to signal voltage (VSIG) due to the electron signal charges generated in the photoconductor. After the termination of the exposure, the TSW is again turned on to transfer the charges of CST into sense line connected to the amplifier. Here, the amplitude of VSIG is important as a practical signal in the pixel, which is closely related to the photoconductor sensitivity (S) and the performance of the switching TFT (See Figure S1, Supporting Information for the full circuit operation of our AMFPI.) Figure 3a shows the X-ray sensitivity (S) of HgI2 photoconductors (two samples A and B) as a function of electric field (F). (Dark current density was only ≈3 nA cm−2 as maximum as shown in Figure S2, Supporting Information.) The sensitivity of sample B is two times higher than that of sample A, since the sample B was more adequately prepared in consideration of the following three process factors: the viscosity of HgI2 paste that was carefully controlled with an optimum binder-tosolvent ratio and optimum particle dispersion, a smooth surface of HgI2 layer as achieved without remaining solvent pores, which cause leakage current, and O2 plasma treatment that was implemented to improve the contact between top electrode and HgI2 powder. As a result of the device processes, the electrical performances of photoconductors such as the mobility (µ) and

2

wileyonlinelibrary.com

lifetime (τ) of free charges generated by X-ray are also determined. The µτ value is extracted through the following wellknown equation, which is supposed to fit the relation between measured charge Q and applied electric field, F.[21] Q = Q0

⎛ −d ⎞ ⎞ μτ F ⎛ 1 − exp ⎜ ⎝ μτ F ⎟⎠ ⎟⎠ d ⎜⎝

(1)

where Q0 is the total charge and d is the thickness of the photoconductor. The µτ products of A and B were, respectively, determined to be 1.3 × 10−6 cm2 V−1 and 1.5 × 10−5 cm2 V−1 for electrons by fitting the curves for measured S values (=Q/Q0) as shown in Figure 3a. The µτ of the photoconductor sample B was estimated to be one order of magnitude higher than previous data (for holes) from commercial a-Se and that (for electrons) of screen-printed HgI2 photoconductor (sample A).[22–26] Figure 3b shows the plot of relative sensitivity (final S/initial S) as a measure of instability, which is evaluated with the exposed X-ray dosage. Here, the inset shows X-ray exposure time and interval respectively as 0.6 and 3 s. In our experimentation, X-ray dose deposited into the photoconductor was 0.63 Gy per single shot. According to the plots, the relative sensitivity of our HgI2 (sample B) was little varied till the accumulated dosage reaches to 100 Gy whereas that of a-Se rapidly degraded by 25% of initial S only after 4 Gy. (The 100 Gy is a general number of accumulated X-ray dose for commercial detectors when those are used for a year in a hospital.) This is because thick a-Se has much higher density of charge traps inside. Moreover, the initial S of a-Se was in fact only 0.15 nC mR−1 cm−2 even under a much higher electric field (F) of 10 V µm−1, and in fact, it is eight times smaller than that (1.2 nC mR−1 cm−2) of our HgI2, which was obtained under 10 times smaller F, 1 V µm−1. These results indicate that our HgI2 is incomparably excellent over a-Se in terms of basic properties. The long-term stability of HgI2 against humidity-temperature stress was also characterized and plotted with aging time as shown in Figure 3c. Our HgI2 appeared quite resistant against the temperature stress at 50 °C when it was exposed to dry ambient without passivation for 500 h, however, it revealed a definite weakness under a high relative humidity of 90% showing 50% decrease of initial S after 500 h. This could be effectively solved by polymer

© 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Adv. Healthcare Mater. 2014, DOI: 10.1002/adhm.201400077

www.advhealthmat.de www.MaterialsViews.com

COMMUNICATION

in the midgap states of polycrystalline HgI2 particle and HgI2/binder interface; the trapinduced current is counted as the loss of signal charges, which may play as the main cause of image lag in dynamic imaging applications such as fluoroscopy and angiography. The relative amount of first frame lag (NLAG) can be expressed by[27]

N LAG =

N FD N LI

(2)

where NFD is the pixel signal of first dark frame after termination of X-ray exposure and NLI is the average pixel signal of last ten image frames. In the present study, the image lag was about 0.9%, which is actually quite good compared with previous reports of 10%.[14,27] (In practical application the lag should be less than 5%). Together with sensitivity of photoconductive material, signal to noise ratio (SNR) is another important performance parameter representing the image quality of imaging devices. The SNR is generally defined as the ratio of the mean signal value per pixel (Mp) to the total standard deviation (σ) of the pixel values across the image sensor.

SNR =

Mp σ

σ = σ P2 + σ I2

Figure 2. Schematic and operation of pixel for X-ray image sensor. a) A plan view of optical microscope image of a pixel before HgI2 screen-printing, where an area indicated by a blue solid arrow is magnified and converted to a cross sectional images shown in b) by TEM. c) A schematic cross-sectional image of a pixel after coating the photoconductor on the TFT structure;blue dashed-line arrow in a) corresponds to c). A red-rectangle part of the c) cross section image belongs to b). ILD represents interlayer dielectric, PC is photoconductor, FS is field shield line, TSW is switch TFT, CST is storage capacitor, and PTE is the top electrode of photoconductor. d) X-ray image sensor pixel circuit (left) and the timing diagram for the sensor operation (right).

encapsulation layer such as parylene. Figure 3d shows the time domain plot of image lag, according to which a temporal decay of photocurrent is observed as a function of read-out frame number after the termination of X-ray exposure. (Here, the frame number was counted for time sequence when a-Si TFT back panel reads out the signals coming from our sample B photoconductor, whether X-ray shot is ON or OFF). This persistent photoconductivity phenomenon is attributed to the gradual release (temporal decay) of charge carriers trapped

Adv. Healthcare Mater. 2014, DOI: 10.1002/adhm.201400077

(3)

(4)

where σP is the standard deviation of incident X-ray photons (to be received by the photoconductor) and σI is that of photoelectric signals from image sensor pixel array. Supposing that image sensor is perfect, the SNR is only influenced by incident X-ray photons with Poisson distribution, to be expressed below.

SNR =

N σP

σP = N

(5)

(6)

where N is average number of incident X-ray photons per pixel. Figure 4a shows the SNR plots of HgI2/a-Si TFT-coupled X-ray flat panel image sensors, where the sample A and sample B are, respectively, incorporated as HgI2 photoconductors but only the sample B was processed with three-roll mill for HgI2 binder (powder) mixing. The effects of particle dispersion by roll milling are illustrated with schemes of Figure 4b,c. In Figure 4a, the SNR of HgI2 with three-roll mill appears so significantly improved that it shows 20 times (or ≈2 times in dB) higher

© 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

wileyonlinelibrary.com

3

www.advhealthmat.de

COMMUNICATION

www.MaterialsViews.com

Figure 3. X-ray performances and stability of HgI2 photoconductive material. a) The X-ray sensitivity S of HgI2 photoconductor with respect to the electric field.The measured S data (Mea) were fitted by the curves (Cal) from Equation (1), into which the µτ value is intentionally plugged. In the unit of S ( = C mR−1 cm−2), R (Roentgen) means an X-ray amount. b) The instability of a-Se and HgI2 photoconductors as a function of the X-ray exposure, obtained from the accumulated X-ray dose; the input dose exposure per single pulse (0.6 s) was 0.63 Gy. c) The instability of HgI2 photoconductor in terms of stress time under the temperature (Temp)-humidity (Hum)-passivation (passi) condition. d) The transient time behavior of pixel signal as a function of frame. Inset shows a magnified view for NFD, the pixel signal of first dark frame after termination of X-ray exposure.

than the other case without roll mill process at the low X-ray dose of 10 mR. (Our HgI2 sample B is even superior to that of commercial product at the low dose as shown in Figure S3, Supporting Information.) According to international electrotechnical commission (IEC) 62220-1-2 regulation,[28] which defines the quality evaluation standards of digital X-ray imaging devices for mammography, a number of 32 217 X-ray photons per mGy enter into the each pixel of 70 µm × 70 µm size. Thus, maximum SNR of a perfect ideal image sensor is 46.8 dB according to Equation (5). Surprisingly, our HgI2 appears to

Figure 4. Signal to noise ratio (SNR) of HgI2/a-Si TFT-combined X-ray flat panel image sensor. a) The SNR curves with (w) and without (w/o) three-roll mill as a function of X-ray dose.The illustrations for dispersed HgI2particles in binder without b) and with c) three-roll mill process.

4

wileyonlinelibrary.com

show 43.5 dB, which is very close to the ideal SNR. This fact indicates that our pixels have a certain value of σI but its photoconductor portion is absolutely minimum. When we implemented X-ray fluorescence analysis on X-ray beam-scan area of 3 cm × 3 cm (not shown here), 17% of the area appeared to be occupied by Hg when processed with three-roll mill (sample B) while only 13% was occupied without the three-roll mill process (sample A). It is thus likely that the three-roll milling certainly enhances SNR because it contributes to a uniform distribution of HgI2 in each pixel as illustrated in Figure 4b,c. According to the PPS structure in Figure 2a,c, the device performance of switching TFT would additionally influence the practical VSIG value for an improved X-ray image although we have previously stressed the performances of photoconductor HgI2 (such as sensitivity S and photoconductor-related portion of SNR). It is because the field-effect mobility and gate swing (sub-threshold swing, S.S.) of TFT are closely related to how efficiently the signal charges can be transferred to external electronics from photoconductor. Therefore, the offcurrent of switching TFTs and the TFT-related portion of SNR become important as another remaining measure to determine the ratio of the maximum signal to the electrical noise (SNR); those measures are about how effectively we can save the photoconductor-collected signal charges, which are now stored in the CST. In Figure 5a, the drain current–gate/source voltage (ID–VGS) transfer characteristics of our a-HfInZnO TFT and typical a-Si TFT (courtesy from Samsung Electronics) are

© 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Adv. Healthcare Mater. 2014, DOI: 10.1002/adhm.201400077

www.advhealthmat.de www.MaterialsViews.com

COMMUNICATION

where SN is noise power spectral density, ID is drain current, e is the electron charge, αH is the Hooge’’s parameter, Ci is gate dielectric capacitance per unit area, W and L are the gate width and length of a-HfInZnO TFT, f is frequency. The αH of a-HfInZnO TFT was calculated to be 7.8 × 10−4 ≈ 1.9 × 10−3, which is an order of magnitude low than that of a-Si TFT (8.7 × 10−3–2.4 × 10−2).[29,30] This is also lower than any other values reported for the polycrystalline silicon, a-GaInZnO, Figure 5. Electrcial and SNR performances:a switching a-HIZO TFT with etch-stop structure pentacene, and poly(thienylene vinylene) TFTs.[31–34] Such a low αH value is attributed vs a-Si TFT. a) The transfer curves of a-Si TFT and a-HfInZnO TFT as a function of gate voltage (VGS) at the drain voltage (VDS) of 10 V. b) The plots of normalized noise power spectral density to the etch-stop structure TFT process, which (SN/ID2) as a function of gate overdrive voltage (VGS–VTH) at the frequency of 40 Hz and the was designed for relatively less damage and drain voltage of 1 V. The width/length of TFTs used was 25/10 µm. contamination.[33,34] X-ray source was composed of a molybdenum target, a Varian-Rad 85 X-ray tube, and an HTM-200 thus compared. According to the figure, the a-Si TFT appeared high-frequency generator, which are fully shielded by 1-cmto have a field-effect mobility of 0.6 cm2 V s−1, a sub-threshold thick lead. The distance from the X-ray source to the imager was swing of 0.8 V dec-1, a threshold voltage (VT) of 1.2 V, and off60 cm. (See Figure S6, Supporting Information for details on current of 5 × 10−14 A, while the a-HfInZnO TFT shows much the X-ray system and red-rimmed HgI2-coated imager module higher field-effect mobility of 27.7 cm2 V s−1, a sub-threshold swing of 0.2 V dec−1, a threshold voltage of 0.8 V and a off-curwith top electrode.) As our first display, Figure 6a,b show the image quality of a computer mouse taken by our HgI2/arent of 5 × 10−15 A. (See Figure S4, Supporting Information, for details on additional transfer and output curves of a-Si and HfInZnO-based and commercial a-Se/a-Si-based flat panel X-ray a-HfInZnO TFTs.) The device performances of a-HfInZnO image sensors, respectively. The used X-ray dosage for the image TFT are mostly superior to those of a-Si TFT in both respects of computer mouse is 10 mR, which is 10 times lower than of mobility and off-current by an order of magnitude. Along normal X-ray exposure used for mammography. The image of with the off-current, the low-frequency noise (LFN) generated the former appears much superior to that of the latter under the in the electronic devices is now the TFT-controlled portion of same low-dose X-ray exposure. As the second demonstrations, SNR performance since it sets a limit on how small signals the images of artificial human breast textile were exhibited (see can be detected and processed. The LFN cannot be completely the inset of Figure 6c). According to the X-ray image results of eliminated, but can be substantially reduced by an efficient Figure 6c,d, which were, respectively, obtained from HgI2/adesign of the devices. Our a-HfInZnO TFT is carefully fabHfInZnO TFT- and a-Se/a-Si TFT-adopting AMFPI systems ricated with an etch-stop structure to reduce the both etching under a low-dose X-ray exposure of 1 mR, the former appears damage and back channel contamination as previously shown more highly resolved than the latter, so that the glandular text and marked in Figure 2b,c. Figure 5b shows the plots of norcan be clearly distinguished from adipose (fat) background in malized noise power spectral density (SN/ID2), which is the the artificial textile. In general, the clear resolution between glandular and adipose is supposed to be very hard to obtain. function of the gate overdrive voltage (VGS–VT) as applied to For the last analysis for our imaging system, we also attempted the two switching TFTs with a-Si and a-HfInZnO (SN means detective quantum efficiency (DQE) measurements, since it is a noise power spectral density). We here put the drain/source related to the SNR ratio of detected X-ray image in overall estivoltage to the linear (Ohmic) regime to effectively find the LFN mation, which includes aforementioned discussion on SNR mechanism, since TFTs in the PPS circuit practically operate portions. Figure S7 (Supporting Information) shows the DQE at a linear regime. (See Figure S5, Supporting Information results from HgI2/a-HfInZnO TFT- and a-Se/a-Si TFT-adopting for details on the noise property measurements using SR570 low-noise current amplifier and Agilent 35670A dynamic AMFPI systems. According to the results, our new pixel panel signal analyzer.) In the present study, the log10 SN/ID2 versus exhibits much superior DQE values to those of conventional a-Si TFT-based panel in the whole spatial frequency (mm−1) regime, log10(VGS – VTH) plot turned out to have the slopes of ≈−1 at a fixed frequency of 40 Hz for both TFTs, showing a good particularly at the low X-ray dosages. This means that our new agreement with the prediction of Hooge’s empirical law.[23] system can provide better resolution benefits in both large and detailed parts of image, more sensitively at small dose X-ray It does confirm that the dominant LFN is due to the bulk exposure. In this regard, our AMFPI imager with the HgI2/amobility fluctuation by electron-phonon scattering in the intrinsic channel region. Therefore, Hooge’s parameter αH, HfInZnO TFT combination is certainly promising to achieve highly resolved images for human body textile. (Additional the criterion of the LFN magnitude can be extracted from the discussion on image quality and pixel performance is found in following equation.[29] Supporting Information where four combinations of photoconductor/switching TFT of a-Se/a-Si, a-Se/a-HfInZnO, HgI2/a-Si, SN eα H and HgI2/a-HfInZnO TFT are compared one another as also = (7) I D2 CiWL(VGS − VT ) f briefly summarized in Table S2, Supporting Information).

Adv. Healthcare Mater. 2014, DOI: 10.1002/adhm.201400077

© 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

wileyonlinelibrary.com

5

www.advhealthmat.de

COMMUNICATION

www.MaterialsViews.com

Figure 6. X-ray images taken by HgI2/a-HfInZnO-based flat panel X-ray image sensors. The X-ray image of a computer mouse taken by a) our HgI2/a-HfInZnO-based and b) a-Se/a-Si-based flat panel X-ray image sensor. The X-ray images of phantom taken by c) HgI2/a-HfInZnO-based and d) a-Se/a-Si-based flat panel X-ray image sensors. The inset of figure c) is the photograph of artificial human textile (phantom), which is a substitute designed to match the X-ray beam attenuation in a human breast.

In conclusion, a high-performance direct-conversion flatpanel X-ray image sensor was developed with a-HfInZnO TFT switch array and HgI2 photoconductor, aiming at replacing the commercial sensor composed of a-Si TFT and a-Se photoconductor. Compared to the commercial sensor, our X-ray image sensor appeared to have much higher detection sensitivity, the sensitivity-sustainability, and minimized noise level under a low electric-field, demonstrating highly resolved images of artificial human breast textile. We thus conclude that our HgI2/aHfInZnO TFT-combined X-ray-detecting flat panel is very promising as the state-of-the-art X-ray image sensor for nextgeneration medical imaging systems.

HgI2 Photoconductive Material: The mercuric iodide (HgI2) paste was coated on the indium tin oxide (ITO) bottom electrode (prepared for TFT panel) by screen-print process in favor of low cost, large area, and mass production. Compared to photoconductive materials such as a-Se, PbO2, and PbI2, the HgI2 is active enough to chemically react with conventional metals used in flat panel X-ray sensor systems. In order to minimize such a reaction, we chose ITO as a collection electrode to contact HgI2. The HgI2 paste made by particle in binder (PIB) method consists of polycrystalline HgI2 powder, polymeric binder (polyvinyl butyral; 1010 Ω cm) and solvent.

wileyonlinelibrary.com

Supporting Information Supporting Information is available from the Wiley Online Library or from the author.

Experimental Section

6

The HgI2 powers were uniformly dispersed in the binder by three roll mill and power size is in the range of 10–30 µm. The binder has approximately the same electrical resistance as that of HgI2. After screen-printing the 100-µm-thick HgI2 film on the TFT panel, the solvent was evaporated by annealing at the temperature of 70 °C. Mo film as a top electrode was subsequently patterned through a shadow mask of the area of 1 cm × 1 cm by DC magnetron sputtering at room temperature. Finally, a parlyene passivation layer was formed on the photoconductive layer by thermal evaporation. The thickness of deposited parlyene was 18 µm and was highly moistureproof. In order to evaluate the characteristics of absorbed X-ray in the fabricated HgI2 film, the Mo target-generated X-ray with 30 keV was irradiated while the electric field of 1.0 V µm−1 was applied between the top and bottom electrodes of HgI2 film. Fabrication of a-HfInZnO Thin Film Transistor: The fabrication process of TFT panel is as follows. Before the fabrication of TFT array, the glass substrate was chemically cleaned using aqueous mixtures of H2SO4–H2O2 and then rinsed with deionized water. A 300-nm-thick SiOX as a buffer layer was first deposited on the substrate by plasma-enhanced chemical vapor deposition (PECVD). Next, a 200-nm-thick Mo layer as a gate was sputtered at room temperature and then chemically patterned by wet etching. A 200/50-nm-thick SiNX/SiOX as the gate insulator was subsequently deposited by PECVD at 300 °C. Then, a 50-nm-thick a-HfInZnO active layer was deposited by radio frequency (RF) magnetron sputtering at room temperature under the conditions of an Ar/O2 flow ratio of 10, the chamber pressure of 5 mTorr and the RF power of 80 W. The composition of the prepared targets was HfO2:In2O3:ZnO = 0.15:1:2 in mol%. A 100-nm-thick SiOX as a etch stop layer was deposited on the a-HfInZnO active layer at 150 °C by PECVD. The dry etching for contact hole formation and the wet etching for active pattern were sequentially performed. Then, a 200-nm-thick Mo layer for the sense line was sputterdeposited at room temperature and patterned by wet etching; here the drain of TFT was connected to the sense line through contact hole. A passivation layer is deposited twice to form a shield layer for preventing the back channel of TFT from turning on. A 100/100-nm-thick SiNX/SiOX was deposited as the first passivation layer and then a 100-nm-thick Mo layer for field shield (FS) layer is sputter-deposited at room temperature. After patterning FS layer, the second passivation layer is again deposited and a via etch is performed to provide for the electrical connection between pixel electrode and the source of TFT. Finally, all the TFT panels were then annealed at 300 °C in a tube furnace for 30 min. X-Ray Image Measurement: Imaging system consists of source and image sensor array. X-ray source consists of a molybdenum target, a Varian-Rad 85 X-ray tube, and an HTM-200 high-frequency generator, which are fully shielded by 1-cm-thick lead. The distance from the X-ray source to the image sensor was 60 cm. The amount of X-ray exposure was measured and calibrated by a Victoreen 06-524-3000 ion chamber connected to a Keithley 35050A dosimeter. The X-ray image was taken by direct-conversion flat-panel X-ray image sensor that combines a-HfInZnO TFT switch array with HgI2 photoconductor, under the following condition: an X-ray energy of 30 keV and a low-dose X-ray exposure of 0.6–10 mR.

Acknowledgements Authors acknowledge the financial support from NRF (NRL program, No. 2012–0000126), BK21 Project.

© 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Received: February 4, 2014 Revised: March 15, 2014 Published online:

Adv. Healthcare Mater. 2014, DOI: 10.1002/adhm.201400077

www.advhealthmat.de www.MaterialsViews.com

Adv. Healthcare Mater. 2014, DOI: 10.1002/adhm.201400077

COMMUNICATION

[1] R. A. Street, S. Nelson, L. E. Antonuk, V. P. Mendez, Mater. Res. Soc. Symp. Proc. 1990, 192, 441. [2] L. E. Antonuk, J. M. Boudry, C. W. Kim, M. Longo, E. J. Morton, J. Yorkston, R. A. Street, Proc. SPIE 1991, 1443, 108. [3] L. Peng, L. Hu, X. Fang, Adv. Mater. 2013, 25, 5321. [4] T. Liu, J. Hu, Z. Jin, F. Jin, S. Liu, Adv. Healthc. Mater. 2013, 2, 1576. [5] T. Yamazaki, T. Tamura, M. Nokita, S. Okada, S. Hayashida, Y. Ogawa, Proc. SPIE 2004, 5368, 379. [6] Y. El-Mohri, K.-W. Jee, L. E. Antonuk, M. Maolinbay, Q. Zhao, Med. Phys. 2001, 28, 2538. [7] E. Samei, M. J. Flynn, Med. Phys. 2003, 30, 608. [8] M. F. Stone, B. V. Jacak, W. Zhao, P. O’Connor, B. Yu, P. Rehak, Med. Phys. 2002, 29, 319. [9] W. Zhao, W. G. Ji, A. Debrie, J. A. Rowlands, Med. Phys. 2003, 30, 254. [10] D. C. Hunt, J. A. Rowlands, O. Tousignant, Med. Phys. 2004, 31, 1166. [11] F. Taghibakhsh, K. S. Karim, IEEE Trans. Electron Devices 2008, 55, 2121. [12] F. Taghibakhsh, K. S. Karim, IEDM Tech. Dig. 2007, 1011. [13] M. Simon, R. A. Ford, A. R. Franklin, S. P. Grabowski, B. Menser, G. Much, A. Nascetti, M. Overdick, M. J. Powell, D. U. Wiechert, IEEE Trans. Nucl. Sci. 2005, 52, 2035. [14] R. A. Street, S. E. Ready, K. Van Schuylenbergh, J. Ho, J. B. Boyce, P. Nylen, K. Shah, L. Melekhov, H. Hermon, J. Appl. Phys. 2002, 91, 3345. [15] S. Adachi, N. Hori, K. Sato, S. Tokuda, T. Sato, S. Yamada, K. Uehara, Y. Izumi, H. Nagata, Y. Yoshimura, Proc. SPIE 2000, 3977, 38. [16] K. Nomura, H. Ohta, A. Takagi, T. Kamiya, M. Hirano, H. Hosono, Nature 2004, 432, 488. [17] S.-H. K. Park, C.-S. Hwang, M. Ryu, S. Yang, C. Byun, J. Shin, J.-I. Lee, K. Lee, M. S. Oh, S. Im, Adv. Mater. 2009, 21, 678.

[18] J. C. Park, S. Kim, S. Kim, C. Kim, I. Song, Y. Park, U.-I. Jung, D. H. Kim, J.-S. Lee, Adv. Mater. 2010, 22, 5512. [19] M. Schieber, A. Zuck, H. Gilboa, G. Zentai, IEEE Trans. Nucl. Sci. 2006, 53, 2385. [20] H. Jiang, Q. Zhao, L. E. Antonuk, Y. El-Mohri, T. Gupta, Phys. Med. Biol. 2013, 58, 703. [21] A. Zuck, M. Schieber, O. Khakhan, Z. Burshtein, IEEE Trans. Nucl. Sci. 2003, 50, 991. [22] M. Z. Kabir, S. O. Kasap, Appl. Phys. Lett. 2002, 80, 1664. [23] R. A. Street, S. E. Ready, L. Melekhov, J. Ho, A. Zuck, B. N. Breen, Proc. SPIE 2002, 4682, 414. [24] G. Belev, S. O. Kasap, J. Non-Cryst. Solids 2006, 352, 1616. [25] C. Allen, G. Belev, R. Johanson, S. O. Kasap, J. Non-Cryst. Solids 2008, 354, 2711. [26] K. Wang, F. Chen, G. Belev, S. O. Kasap, K. S. Karim, Appl. Phys. Lett. 2009, 95, 013505. [27] H. Du, L. E. Antonuk, Y. El-Mohri, Q. Zhao, Z. Su, J. Yamamoto, Y. Wang, Phys. Med. Biol. 2008, 53, 1325. [28] IEC 62220–1–2 (2007), Medical electrical equipment – characteristics of digital X-ray imaging devices: Part 1–2. Determination of the detective quantum efficiency – detectors used in mammography, International Electrotechnical Commission (IEC), Geneva, Switzerland. [29] J. Rhayem, M. Valenza, D. Rigaud, N. Szydlo, H. Lebrun, J. Appl. Phys. 1998, 83, 3660. [30] M. Valenza, C. Barros, M. Dumas, D. Rigaud, T. Ducourant, N. Szydlo, H. Lebrun, J. Appl. Phys. 1996, 79, 923. [31] A. Mercha, L. K. J. Vandamme, L. Pichon, R. Carin, O. Bonnaud, J. Appl. Phys. 2001, 90, 4019. [32] L. K. J. Vandamme, R. Feyaerts, G. Trefan, C. Detcheverry, J. Appl. Phys. 2002, 91, 719. [33] I.-T Cho, W.-S. Cheong, C.-S. Hwang, J.-M. Lee, H.-I Kwon, J.-H. Lee, IEEE Electron Device Lett. 2009, 30, 828. [34] J. C. Park, S. W. Kim, C. J. Kim, S. Kim, D. H. Kim, I.-T. Cho, H.-I Kwon, Appl. Phys. Lett. 2010, 97, 122104.

© 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

wileyonlinelibrary.com

7

Small-dose-sensitive X-ray image pixel with HgI2 photoconductor and amorphous oxide thin-film transistor.

A new X-ray image sensor is demonstrated with an oxide thin-film transistor backplane and HgI2 sensing material. It displays outstanding image quality...
1MB Sizes 0 Downloads 0 Views