An aging study of the signal and noise characteristics in large-area CMOS detectors

Transcription

1 An aging study of the signal and noise characteristics in large-area CMOS detectors Jong Chul Han a, Seungman Yun a, Chang Hwy Lim a, Tae Gyun Youm a, Sung Kyn Heo b, Tae Woo Kim b, Ian Cunningham c, Ho Kyung Kim a* a School of Mechanical Engineering, Pusan National University, Busan , South Korea b Sensor Business Division, E-WOO Technology, Co., Ltd., Bora, Giheung, Yongin , South Korea c Imaging Research Labs, Robarts Research Institute, 100 Perth Drive, London, Ontario N6A 5K8, Canada ABSTRACT For a detector consisting of a phosphor screen and a photodiode array made by complementary metal-oxidesemiconductor (CMOS) process, we have experimentally re-investigated the long-term stability of the signal and noise characteristics as a function of the accumulated dose at the entrance surface of the detector in addition to the previous study [IEEE Trans. Nucl. Sci. 56(3) 1121 (2009)]. The irradiation and analysis were more systematically performed. We report the aging effect in image quality in terms of dark pixel signal, dynamic range, modulation-transfer function (MTF), and noise-power spectrum (NPS). Unlike the previous study, the electronic noise was dominantly increased with the total dose and the other statistical and structural noise sources were nearly independent on the cumulative dose. Similarly, the increase of dark pixel signal and the related noise gradually reduces the dynamic range as the total dose increases. While MTF was almost insensitive to the total dose, degradation in NPS was observed. Therefore, preprocessing without properly updated offset and gain images would underestimate the detective quantum efficiency when performing quality control of a detector in the field. Restoration of degraded dark signals due to aging is demonstrated by annealing the aged detector with thermal activation energy. This study provides a motivation that the periodic monitoring of the imagequality degradation is of great importance for the long-term and healthy use of digital x-ray imaging detectors. Keywords: Aging, CMOS, DQE, image quality, MTF, NPS, radiation damage, stability 1. INTRODUCTION Large-area, flat-panel type detectors are now widely used not only for digital radiography in radiology but also for volumetric computed tomography in image-guided radiotherapy and pre-clinical small-animal imaging. Among various detector configurations, indirect-conversion detectors, which employ a scintillator to detect incident x-rays and convert into optical photons, may be advantageous against the radiation damage on the underlying active device elements consisting of an electronic panel although the scintillator itself would be damaged either. Nevertheless, we cannot avoid deposition of a certain amount of radiation dose in the underlying panel because of the unattenuated (or direct) x-rays, characteristic x-rays, and Compton-scattered x-rays from a scintillator generated during irradiation. 1,2 Compared with the signal and noise generated from primary x-ray interactions in a scintillator, the magnitude of signal and noise induced by interactions of direct x-rays or secondary x-rays within a photodiode is much higher, once they occur, because the effective W-value, the average ionization energy to create a single electron-hole pair, of the silicon is much smaller than that of a scintillator. Therefore, investigation of radiation damage in the detector hardware as well as image quality is an important issue for the reliable use of detectors in clinic, even if the detector adopts indirect-conversion method. While a large amount of effort has been invested in evaluating the performance degradation of the pixel-element devices themselves due to the radiation damage, 3,4 however, there has not been sufficient attention paid to the evaluation of performance degradation of an imaging detector that is an important issue affecting the quality control and assurance of detectors over their lifetime. * hokyung@pnu.edu; phone ; fax Medical Imaging 2010: Physics of Medical Imaging, edited by Ehsan Samei, Norbert J. Pelc, Proc. of SPIE Vol. 7622, 76223Y 2010 SPIE CCC code: /10/$18 doi: / Proc. of SPIE Vol Y-1

2 Fig. 1. The photograph (left) shows the experimental layout. Half of the CMOS photodiode array was covered by the commercial phosphor screen, which is the typical configuration as an x-ray imaging detector, and the other region was kept as bare. To monitor environmental temperature during irradiation, Si wafer attached onto the printed circuit board was contained in the detector housing. The piece of Si wafer mimics the substrate of the CMOS photodiode array. The photograph (right) shows the inside of vacuum furnace, which was used for the annealing test of damaged CMOS photodiode arrays with a thermal activation energy. Recently, an active-pixel sensor based on the CMOS (complementary metal-oxide-semiconductor) technology is greatly investigated as digital x-ray imaging detectors, because of its unique advantages, such as very low image lag and larger pixel fill factor compared to conventional flat-panel detectors based upon amorphous silicon thin-film transistor technology.5,6 In our previous study,7 we investigated the radiation damage and its effects on the performance of a CMOS photodiode array in conjunction with a phosphor screen in terms of dark signal, noise, dynamic range, modulation-transfer function (MTF), noise-power spectrum (NPS), and detective quantum efficiency (DQE). In this study, we re-investigated the long-term stability of the signal and noise characteristics as a function of the accumulated dose at the entrance surface of the CMOS detector. Bare CMOS sensor without a phosphor screen was included in the experiment. The irradiation and analysis of the signal and noise characteristics were more systematically performed. 2. MATERIALS AND METHODS 2.1 Experimental We configured a sample detector with a commercial phosphor screen (Min-R2000TM, Carestream Healthcare, Inc., USA) and a CMOS photodiode array (RadEyeTM, Rad-icon Imaging, Corp., USA). The Gd2O2S:Tb-based phosphor screen has a thickness of 84 μm and a density of ~4 g/cm3. The CMOS photodiode array has a format of pixels with a pitch of 48 μm. Detailed physical specifications and characteristics of the phosphor screen and the CMOS photodiode array can be found in references [5, 7, 8]. As shown in Fig. 1, half of the CMOS photodiode array was covered by the phosphor screen, which is the typical configuration as an x-ray imaging detector, and the other region was kept as bare. This configuration can give us the effect of the phosphor screen on the radiation damage in the signal and noise characteristics. It is noted that, in order to monitor environmental temperature during irradiation, a piece of silicon wafer attached onto a printed circuit board, as shown in Fig. 1, was contained in the detector housing. Two-different x-ray sources were employed for the irradiation. To provide a heavy irradiation dose for the CMOS detector, a relatively high-power source (max. 810 watts; EXG-6TM, E-Woo Technology, Co., Ltd., Korea) was used with a source-to-detector distance (SDD) of 300 mm. On the contrary, images for the analysis of signal and noise characteristics were obtained with a small-power x-ray source (max. 80 watts; UltrabrightTM, Oxford Instruments X-ray Technologies, Inc., USA) with a SDD of 400 mm. The applied energy was set to 50 kvp for both x-ray sources. In general, the radiation effect on a device is evaluated in terms of total ionizing dose (or total dose), which is the accumulated energy per unit mass of the device. It is, however, difficult to directly measure the actual energy absorbed in the device. Instead, we measured the air kerma rate in the unit of mgy/s at the detector entrance surface by using a calibrated ion chamber and electrometer (Piranha R&F/M 605, RTI Electronics AB, Sweden). For the irradiation with Proc. of SPIE Vol Y-2

3 Acquire dark, white & edge knife images for DQE evaluation Irradiation 0 Gy 10 Gy 20 Gy 30 Gy 35 Gy 40 Gy 45 Gy 48.5 Gy Acquire dark images every 1 Gy irradiation step Acquire dark images every 0.5 Gy irradiation step Fig. 2. Schematic illustration of data acquisition during irradiation. To monitor the effect of irradiation on the CMOS detector, dark signal was repeatedly checked. For given milestones, DQE evaluation was performed. the high-power source, the air kerma rate of ~2.1 mgy/s was measured at the surface of the detector. The attenuation through the phosphor screen was estimated to ~58% in the air kerma rate. Time schedule describing the irradiation for the intended radiation damage and the acquisition of images for the analysis of the signal and noise characteristics with the total dose is illustrated in Fig. 2. The total irradiated dose is about 50 Gy, which is larger than the previous experiment by a factor of 1.6. Up to the total dose of 30 Gy, dark images were acquired at every 1-Gy irradiation, while white images for NPS and edge-phantom images for MTF were acquired at every 10-Gy irradiation. After irradiation of the total dose of 30 Gy, all the images were acquired at half the periods of 1 Gy or 10 Gy. To avoid the over-load in x-ray tube operation, the irradiation cycle was composed of 15-s irradiation and 45-s cooling period. This timing sequence of irradiation of the x-ray tube was realized with field programmable gate arrays. To investigate the restoration of degraded dark signals, we performed annealing test of an aged detector with thermal activation energy. CMOS detectors, provided by the same manufacturer for the detector used in the above irradiation test, have been implemented into the microtomography application. 9 Long-time and intense use over 5 years has failed the normal operation of CMOS detectors because of the increased dark signal level up to full-well capacity. We have annealed these in a vacuum furnace with temperature as shown in Fig. 1 (right picture). The applied cycles of temperature were computer-controlled. 2.2 Analysis The dark signal was analyzed with 20 images acquired at the given time (or irradiation dose) interval. The analyzing image size is pixels locating at the respective central regions, where the phosphor is covered or not, in a dark image. The degradation of spatial resolution with the total dose was assessed by measuring MTF for an edge-knife impulse response. To avoid aliasing, a slanted edge method was employed. 10 MTF result at the given irradiation dose intervals is the average of 15-MTF curves. NPS was determined by two-dimensional Fourier analysis of white images. 8 For a single white image, 13 sub-images with a region-of-interest (ROI) of pixels were extracted. The NPS at the given irradiation dose intervals is the average for 130 ROIs considering the obtained 10 images. The DQE was assessed from the measured MTF, NPS, and the estimated photon fluence q as 2 qgmtf ( f ) DQE( f ) =, (1) NPS( f ) where f denotes the spatial frequency and G is the detector system gain. The fluence was calculated using the experimentally measured exposure and half-value layer, and the computational program for x-ray spectral analysis. 8 The dynamic range of detector in this study is defined as S S ( X ) ( X ) max Γ = dark, (2) σ dark ( X ) Proc. of SPIE Vol Y-3

4 Fig. 3. (a) Dark signal levels as a function of total irradiated dose for two different detector configurations. In the case of the detector employing the phosphor screen, the total dose can be considered as the measured dose at the entrance surface of the phosphor screen or the estimated dose at the surface of the CMOS sensor considering the attenuation through the phosphor screen. (b) Standard deviation of dark signal values as a function of mean pixel value. where X denotes the accumulated total dose, and S max (= 4095 ADU in this study) and S dark (X) are the maximum signal and the dark signal at the given total dose, respectively. σ dark (X) is the standard deviation of the dark signal at the given total dose X. The upper bar denotes the mean value. 3. RESULTS The effect of accumulated total dose on the signal and noise characteristics is summarized in Fig. 3. The error bars indicate the standard deviations of the average dark values calculated for 20 images. As expected, the level of dark signal is increased as the total dose increases as shown in Fig. 3(a). The increase of dark signal follows quadratic-dependence on the total dose, which implies that the accumulated dose accelerates the increase of dark signal. In other words, the dark current becomes stronger with increasing absorbed dose within a photodiode. As mentioned in the previous study, 7 the dark current signal is the result of combined degradation mechanisms among the photodiode, the transistors, and the interconnectivity in a pixel. And it is believed that the buildup of positive charge in the silicon dioxide (SiO 2 ) layer mainly increases the leakage current across the p-n junction of the photodiode. Bare region in the CMOS sensor without an overlying phosphor screen gives a larger increase in dark signal with respect to the total dose accumulation. This situation would be expected intuitively. More interestingly, if we replot the result of the CMOS sensor having a phosphor screen considering the total dose at the surface of CMOS photodiode array instead of the entrance surface of the phosphor screen, the trend of increase of dark signal is higher than that of bare region in the CMOS sensor. This phenomenon may be explained by the beam hardening through the phosphor screen because the dark signal and current are strongly x-ray energy-dependent. The noise property in the CMOS sensor during irradiation is different from the previous observation. 7 In the previous study, three distinct dependencies of noise characteristics were observed as the dark signal level increases; signalindependent, square-root proportional, and linear proportional regimes. Although the noise characteristic quadratically depends on the dark signal level, the dependency is gradually reduced as the dark signal increases as shown in Fig. 3(b). This result implies that the fixed-pattern noise (FPN), which increases the standard deviation linearly with the signal, is negligible in this investigation. The negligible FPN in this investigation might be due to the use of different x-ray tube, the better electrical shield of the detector housing, and the employment of a CMOS sensor having a better quality, but the reason is not clear at this moment. Detailed investigation would be needed. The increase of dark signal level for the total dose definitely consumes a significant part of the total dynamic range. The calculated dynamic range in decibels is plotted in Fig. 4 based on the measured dark signal and noise for two-different detector configurations. The dynamic ranges in decibels of the CMOS sensor with and without the overlying phosphor Proc. of SPIE Vol Y-4

5 Fig. 4. Estimated dynamic range of two detector configurations as a function of total irradiated dose. Fig. 5. Measured MTFs (a) and NPSs (b) of the CMOS detector with respect to various total irradiated dose levels. In the measurements of NPS, gain-offset correction was not applied. screen drop to ~40% and 17%, respectively, of the starting dark signal status of detector over the course of the experiment. These values correspond to only ~0.8% and 0.1% in terms of full-well capacity-to-noise ratio. As mentioned in the previous study, 7 the reduction in dynamic range during the use of detectors is significant for any clinical imaging application for the obvious reason of loss of range available for the expected range of x-ray intensities after attenuation by human anatomy. Fig. 5 shows the measured results of MTF and NPS of the detector with respect to total dose. While there is no significant change in MTF performance with total dose, the noise-power spectral densities are gradually increased as the accumulation of total dose increases. In the NPS measurements as shown in Fig. 5(b), however, we cannot observe any sign of the existence of FPN components in spatial frequencies even with no application of the gain-offset correction procedure. These measured NPSs would largely degrade the DQE performances. Therefore, gain-offset correction is essential to avoid the effect of the increase in dark signal and noise on the DQE evaluation. The NPS results after applying the gain-offset correction procedures are plotted in Fig. 6(a), in which the low-frequency trend has completely been removed and the NPS is almost unchanged over entire spatial frequency region as the total dose increases. It is noted that the dependence of NPS on total dose has been reversed after applying gain-offset correction. Proc. of SPIE Vol Y-5

6 Fig. 6. Measured NPSs (a) and DQEs (b) of the CMOS detector with respect to various total irradiated dose levels. In the measurements of NPS, gain-offset correction was applied. -Measured dark signals /-MJPIIu LlIIJIdLLJIt, LItb U ]U lb U Zb ju Elapsed time (day) Fig. 7. Dark signal change of the damaged CMOS sensors, which were annealed in the vacuum furnace, as a function of time. The applied temperature cycle is also plotted. The resultant DQE performances are plotted in Fig. 6(b). Although slight differences in MTF and NPS results with total dose as shown in Fig. 5(a) and Fig. 6(a) affect the DQE results, the DQE results are almost independent on total dose. We have annealed the aged detectors in a vacuum furnace with temperature. We found that the application of thermal energy of 100 C over 50 hours worked on the reduction of dark signal level (see the hatched region of Fig. 7). This recovery is based on the hypothesis that positive ions built in the SiO 2 layer by irradiation are neutralized with electrons, which have enough energy to jump back into the SiO 2 layer, from the silicon substrate. During this experiment, however, we observed the increase of malfunctioned or dead lines in the CMOS detectors, which may be due to the disconnection between photodiode pixel arrays and the peripheral control/readout electronics at higher temperature. 4. DISCUSSION AND CONCLUSION As described in the previous section, large discrepancy of this investigation compared with the previous study 7 is the noise characteristic. The noise components contributing to dark images consist of electronic noise, statistical quantum noise, and structured fixed-pattern noise. Therefore, for an image obtained at a certain total dose level the noise characteristic can be analyzed as Proc. of SPIE Vol Y-6

7 Fig. 8. Extracted individual noise coefficients as a function of total irradiated dose for two different detector configurations; (a) CMOS sensor coupled with the phosphor screen and (b) bare CMOS sensor = ke + kq x + ks x σ, (3) where the coefficient k j describes the dependency between the level of dark signal x and the corresponding noise component, and the subscript j denotes each noise component; e, q, and s for electronic, statistical, and structural noise, respectively. As shown in Fig. 8, the electronic noise term is dependent on total dose, but the others are almost independent on the total dose. In addition, the structural noise term is negligibly related to the total dose compared with other noise terms, which may explain the observations in this study. Two-different detector configurations show the same trend. It should be noted that this analysis assumes that the response of detector is linear. However, most detectors based on the CMOS process show small-signal nonlinearity and the detector used in this study also shows nonlinear response for small-signal range (less than 10 μgy). Therefore, this analysis would not be applied. Since the detector shows linear response for reasonable x-ray irradiation range in medical application (but, the negative intercept in the characteristic curve measurements), however, the above analysis may be meaningful. In this study, we have systematically investigated the aging of signal and noise characteristics of CMOS photodiode array detectors. While the increase of dark signal showed quadratic-increasing form along the accumulated total dose, that of noise showed saturating form. In the noise characteristics, electronic noise was dominant and structural FPN was negligible in this study. These phenomena result in significant reduction of dynamic range as the total dose increases. With the proper gain-offset correction procedures, MTF, NPS, and DQE evaluations were not significantly changed at any total dose level. ACKNOWLEDGEMENTS This work was supported by a Grant-in-Aid for Strategy Technology Development Programs from the Korea Ministry of Knowledge Economy (No ) and the Korea Research Foundation (KRF) Grant funded by the Korea government (MEST) (KRF D01339). Proc. of SPIE Vol Y-7

8 REFERENCES [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] Kim, H. K., "Generalized cascaded model to assess noise transfer in scintillator-based x-ray imaging detectors," Appl. Phys. Lett. 89, (2006). Yun, S., Kim, H. K., Lim, C. H., Cho, M. K., Achterkirchen, T., and Cunningham, I., "Signal and noise characteristics induced by unattenuated x rays from a scintillator in indirect-conversion CMOS photodiode array detectors," IEEE Trans. Nucl. Sci. 56(3), (2009). Boudry, J. M. and Antonuk, L. E., "Radiation damage of amorphous silicon photodiode sensors," IEEE Trans. Nucl. Sci. 41(4), (1994). Boudry, J. M. and Antonuk, L. E., "Radiation damage of amorphous silicon, thin-film, field-effect transistors," Med. Phys. 23(5), (1994). Graeve, T. and Weckler, G., "High-resolution CMOS imaging detector," Proc. SPIE 4320, (2001). Farrier, M., Achterkirchen, T. G., Weckler, G. P., and Mrozack, A., "Very large area CMOS active-pixel sensor for digital radiography," IEEE Trans. Electron Dev. 56(11), (2009). Kim, H. K., Cho, M. K., Achterkirchen, T., and Lee, W., "Radiation effects on the performance of CMOS photodiode array detectors and the role of gain-offset corrections," IEEE Trans. Nucl. Sci. 56(3), (2009). Cho, M. K., Kim, H. K., Graeve, T., Yun, S. M., Lim, C. H., Cho, H., and Kim, J.-M., "Measurements of x-ray imaging performance of granular phosphors with direct-coupled CMOS sensors," IEEE Trans. Nucl. Sci. 55(3), (2008). Kim, H. K. et al., "Performance characterization of microtomography with complementary metal-oxidesemiconductor detectors for computer-aided defect inspection," J. Appl. Phys. 105(9), (2009). Samei, E., Flynn, M. J., and Reimann, D. A., "A method for measuring the presampled MTF of digital radiographic systems using an edge test device," Med. Phys. 25(1), (1998). Proc. of SPIE Vol Y-8