2D amorphous silicon arrays can be sued for medical imaging, non-destructive testing, and high-speed document scanning. We have built a 200 spi imaging system with an active area containing 2304 X 3200 pixels, the largest amorphous silicon imaging system described to data. Packaged with the array are peripheral electronics which include active matrix drivers, charge sensitive amplifiers, two 12 bit A/D converters, and control logic. Digital data travel via fiber to a frame grabber in a personal computer. Software includes gain/offset corrections, line and pixel corrections, window and level controls, and a user interface. Through a combination of layout optimization, amplifier design, and system timing, we have demonstrated a noise level of 1.5 Ke RMS and a signal to noise ratio of 1900.
Today's imaging systems utilize fast operation to increase their throughput. At high line rates the illumination required to collect a reasonable image becomes prohibitive. Time delay and integration (TDI) offers greatly enhanced responsivity to allow faster operation in terms of line rates. This combination of sensitivity and speed is unmatched in other sensor architectures. The standard multi- stage source follower output amplifier usually involves a trade off between speed and sensitivity through sizing of the first FET. We present a high bandwidth and sensitivity, scalable architecture for readout of TDI sensors. A key component of this architecture is the minimization of output amplifier load and parasitic capacitance. The methodologies used in the design and modeling of the output structure will be presented. This basic model has been confirmed over a range of device dimensions. A 4096 element, multi-tap TDI image sensor incorporating this architecture has been fabricated using a standard CCD process. Discrete and in- camera measurements will be presented demonstrating operation at > 100 kHz line rates and with > 300 V/((mu) J/cm2) peak responsivity. Methods of controlling and reducing the stray loading on the sensor output will also be discussed.
Photodiode devices, in which the photosite consists of a reverse biased pn diode, have excellent quantum efficiencies at visible wavelengths and in the UV. However, they display high levels of dark and bright image lag, and high levels of fixed pattern noise (FPN) when operated with electronic shuttering. We have addressed these performance issues by replacing the photodiode photosites with pinned photodiode (PPD) photosites. In the PPD the n+ region of the conventional photodiode is replaced by a n region and a shallow highly doped p region - the surface potential in the photosite is pinned such that the photosite behaves as an ungated buried channel well. The high quantum efficiencies associated with photodiodes are maintained while allowing for large reductions in image lag and fixed pattern noise. We have developed PPD processes for two different photosite architectures. In the first architecture, charge is generated in the PPD and immediately spills to an adjacent gated integration well. In the second architecture, the charge is generated and stored in the PPD. Each of the architectures can be configured to allow for antiblooming/electronic shuttering. Both of the PPD processes and their associated architectures have been characterized, and order of magnitude reductions in image lag have been observed for PPD photosites relative to conventional photodiodes. No degradation in QE, PRNU, or well capacity has been observed. One of the PPD processes has been implemented in a family of high sped, quad output, linear sensors with 200 MHz data rates. Performance results are presented.
The concept and performance of the fully depleted pn- junction CCD system, developed for the European XMM- and the German ABRIXAS-satellite missions for soft x-ray imaging and spectroscopy in the 0.1 keV to 15 keV photon range, is presented. The 58 mm X 60 mm large pn-CCD array uses pn- junctions for registers and for the backside instead of MOS registers. This concept naturally allows to fully deplete the detector volume to make it an efficient detector to photons with energies up to 15 keV. For high detection efficiency in the soft x-ray region down to 100 eV, an ultrathin pn-CCD backside deadlayer has been realized. Each pn-CCD-channel is equipped with an on-chip JFET amplifier which, in combination with the CAMEX-amplifier and multiplexing chip, facilitates parallel readout with a pixel read rate of 3 MHz and an electronic noise floor of ENC < e-. With the complete parallel readout, very fast pn-CCD readout modi can be implemented in the system which allow for high resolution photon spectroscopy of even the brightest x-ray sources in the sky.
A bouwblok concept is described which allows one to fabricate several large area CCD image sensors from a single mask set. The size of the various imagers can differ both horizontally as well as vertically. The new method drastically reduces the development time and the associated cost of a new sensor. Because all images use of same basic pixel structure, the characteristics of new configurations can be fairly well predicted.
Charge-transfer-efficiency (CTE) is a parameter that is associated with the optical performance and radiometric accuracy of a charge-coupled device (CCD). While modern CCDs are typically quoted as having CTE > 0.99999, we show that a) this efficiency can be degraded by exposure to energetic protons and b) the measured efficiency, at least in the case of irradiated devices, is dependent on the size of the signal being transferred. A comparison of two techniques of CTE measurement is presented, with emphasis on a 55Fe x-ray charge-generation and collection technique tailored especially for use with linear CCDs. While our technique follows the same general x-ray method widely used to characterize area CCDs, its implementation, including the processing of the resulting experimental data, is somewhat novel. The x-ray technique requires no special device circuitry or equipment and, in our configuration, may be used on virtually any linear or bilinear CCD; the same general technique is used on area arrays. It is shown that the technique is appropriate for characterizing small-signal CTE with or without a high background. It is also shown that the electrical injection CTE, while requiring the injection circuitry to be designed and built-in to the CCD structure, is more appropriate for large-signal CTE measurement. Data are presented showing experimental pre- and post-irradiation CTE measured as a function of signal level and background level for a two-phase linear CCD. Additionally, it is shown that in our tests, the CTE degraded even at very small accelerated doses, but further degradation was at least partly compensated by an enhanced background due to increasing dark currents.
Charge coupled devices (CCDs) are the detector of choice for instruments that detect low levels of light for wavelengths 300 nm <EQ (lambda) <EQ 1100 nm. Contemporary devices have read noise level equivalents of a few electrons, and the ability to store over 100,000 electrons per pixel. In order to take full advantage of these characteristics, while ensuring that digital quantization noise will not dominate over readout noise, the dynamic range of the readout system must exceed 18 bits. We present a simple scheme that exploits the fact that Poisson noise dominates the error budget in most contemporary CCD systems in order to achieve an effective dynamic range of over 20 bits by suing dual 16 bit A/D converters of different sensitivity.
A large format CID imager module capitalizes on CID large well capacity and radiation resistance to image dental x- rays. The model, which consists of the imager, conversion phosphor and ancillary electronics, is encapsulated in a 40 X 28 X 5 mm3 robust package that is lightproof, moisture-proof and meets FDA and RFI/EMI standards. Data exposure and readout is simple. The imager normally exists in an active reset mode until x-ray application automatically places the imager into a charge integration mode. Readout begins immediately upon completion of the x- ray exposure or manual application of an external trigger source. The imager returns to the reset mode once the data read out is complete. Pixels are arranged in an SVGA compatible 800H X 600V format. Each pixel is square and 38.5 microns/side. The imager is coated using a propriety phosphor deposition process that result in a limiting resolution of 9 LP/mm from an x-ray illumination source. Better than 2,000:1 dynamic range and shot-noise limited operation is achieved. Direct x-ray detection and attendant noise is minimized via the phosphor and epitaxial layer that lies beneath the pixel array. The imager/module architecture and electro-optical performance are described in detail here in.
We describe experimental techniques for characterizing the absolute response of charge-coupled devices (CCD) to incident hard x-rays using the high energy x-ray source at the Lawrence Livermore National Laboratory. We present responsivity and quantum detection efficiency measurements for a standard, front-illuminated, scientific CCD to monoenergetic 8-98 keV K-(alpha) x-rays. This systematic study out to high energies reveals the contribution of different absorption processes to the CCD detection efficiency. For lower energies below 20 keV the CCD behaves like an ideal photoelectric detector as expected. Increasingly above 40 keV the photoelectric effect in the CCD epitaxial region is augmented by incoherent or Compton scattering where a fraction of the energy from the photon scattering event is transferred to the electrons and subsequently detected. The Compton scattering mechanism dominates the photoelectric effect above 100 keV giving the CCD a predicted detection efficiency which remains constant from 150 keV to 1 MeV assuming that the scattered electrons finally come to rest within the active region. These physics issues will be briefly discussed and are particularly relevant to deep active region solid-state detectors with application for hard x-ray detection above 40 keV.
We have performed precise measurements of x-ray absorption constants for all the thin films comprising CCD gate structure, namely, phosphorous doped polysilicon, silicon dioxide, and silicon nitride. X-ray absorption of these films shows large oscillations around the corresponding absorption edges: nitrogen K, oxygen K, silicon L and K. As a result, quantum efficiency of a CCD in the soft x-ray range deviates significantly from the generally assumed simple model predictions. In order to cover the range of energies from 60 eV to 3000 eV transmission measurements were performed at several synchrotron beamlines at ALS, PTB BESSY, SRC. A model of the CCD response with near edge x-ray absorption structure taken into account predicts a very complicated shape of the energy dependence of the quantum efficiency around silicon and oxygen absorption edges. Experimental measurements of CCD quantum efficiency relative to a calibrated detector were performed at BESSY for both frontside illuminated and backside illuminated CCDs for energies around the oxygen absorption edge. Experimental results were found to be in a good agreement with our model.
Linear solid-state detectors are nowadays a widespread media in industrial and medical x-ray imaging. The resolution reached with this system has been largely improved in these past years, but is still too poor for some high resolution applications. We first have carried out an optimization of the detector characteristics through a behavioral simulation using a hardware description language. Furthermore, our work concerned the resolution enhancement for this kind of detectors via signal processing. Our approach takes into account the modeled point spread function (PSF) of the system. This modeled PSF is obtained with a new edge technique. The knowledge about the system response is used in a restoration scheme in order to improve the response of the detector to the high frequencies in the digital image. The restoration problem is an ill posed problem ad uses an inverse Wiener filtering. Another intrinsic limitation of solid-state detectors is the spatial sampling step. In order to overcome this problem, we also tested the feasibility of a finer sampling of the acquired image, buy interlacing several slightly shifted acquisitions of the same test object. The restoration applied to this finer sampled signal results in a resolution enhancement that is theoretically impossible to reach with a single detector acquisition. Some experimental results obtained on a variable bar-space pattern phantom are presented. This kind of phantom allows for a precise evaluation of the modulation transfer function on the acquired and processed images. The contribution of the image processing to the restoration enhancement can thus be quantified.
Study of the possible use of a-SiN: H thin films for 2D direct x-ray sensor arrays lead to the development of a simple prototype. The sensor array is a 100 X 100 array of simple cross-over silicon rich a-SiN:H thin film diodes with sizes 200 X 200 micrometers formed on a 2 inch glass substrate. There are neither switching elements nor x-ray conversion layer involved which leads to extremely simple 2 mask processing for the whole array. Specific behavior of the a-SiN:H sensing diodes under x-ray irradiation requires special attention to be paid to the driving strategy and read out electronics. Experimental results obtained with the help of the prototype provide solid base for the discussion on both sensor properties and electronic components used for the sensor array control.
Life characteristics of an EB-CCD incorporating a full frame transfer CCD have been evaluated. Applying -8kV to the photocathode, the accelerated life test corresponding to 3,500 hours of operation was carried out. As the result, there are no degradation of the photocathode sensitivity and the gain. However, an offset is increased linearly proportional to the operation period, including the upper and lower of the irradiated area. The increment after the life test is approximately 700 at 20 degree centigrade with a multi-pinned phase (MPP) operation. From the comparison between the MP and the non-MPP operation of the CCD, the original of the offset is concluded to be the increased dark current (DC) from the Si-SiO2 interface, which is damaged by the Bremsstrahlung x-ray. The DC slightly affects the signal as the offset when positive voltage to gate electrodes depletes the Si-SiO2 interface during the vertical transfer. The DC does not affect the signal at all with the MPP operation in the exposure time. The lifetime of the EB-CCD is calculated to be much longer than 10,000 hours, if the lifetime is defined by the period that the DC fulfills the half of the full well capacity.
In the paper, we present experimental results from measurements on CMOS PAS imager designed by CIMI-SUPAERO on two different technologies. In both cases, pixels with photoMos and photodiode structures have been designed. The first circuit has been developed using a standard CMOS DLP/DLM 1.2 micrometers process from Austria Micro Systems. The detector array consists of 32 X 32 square pixels with 50 micrometers pixel pitch; fill factor is 75 percent for photodiode and 50 percent for photoMos. The circuit is also including row and column address decoders and the readout circuitry so as to perform on-chip correlated double-delta sampling to reduce column to column fixed pattern noise. Two other chips have been developed with a standard CMOS SLP/DLM process form MIETEC with 0.7 micrometers design rules, which includes a 128 X 128 pixels array, with 21 micrometers pixel pitch and analogue redout circuitry. Among 10 different arrays, no faulty one was observed for both circuits. In this paper, we compare both performances of 32 X 32 pixels and 128 X 128 pixels in terms of dark current, quantum efficiency, conversion gain, dynamic range, linearity and spatial uniformity.
In image sensors with passive pixels the column capacitance is large compared to the capacitance of the pixel. The charge-to-voltage conversion occurs in the column amplifier relatively far from the pixel. This may result in a high sensitivity to interference, especially in cases other electronic circuitry is located on the same chip. Two types of CIF CMOS imagers are presented that use different read- out options to counter this effect. Both designs use differential read-out as DRAM's do. This means that the pixel is compared to a reference cell. The first type uses a reference cell on the same row; the second type utilizes a fully symmetrical way of read-out, similar to digital memories by having this reference on the same column. Furthermore, two other means of image quality improvement are applied. A boost circuit is sued to generate a negative voltage for driving the selecting transistor to insure that it is completely switched on during pixel reset. By this, threshold differences between pixels do not affect the reset voltage. The second is a well thought-out column amplifier that calibrates its offset before reading the pixel information.
Shrinkage of pixel structures and layouts for CMOS active pixel image sensors are studied. Reduction of CMOS device design rule with the scaling-law can make the pixel size small, naturally. However, using minimum design rule, quarter micron rule or sub quarter micron rule, costs expensive. Therefore, pixel size shrinkage using relatively rough design rule have been studied for reduction of the chip cost. We have already reported about small pixel structure by replacement of row-select transistor by row- select capacitor, by omission of reset transistor with forward bias reset operation, and by omission of reset transistor with pinned-buried reset channel. We have also reported about small pixel by high packing density layout named 'I-shaped cell' and its zigzag layout. However, these pixel shrinkage have some disadvantages. In this paper, we propose a novel pixel structure driven by pulse operation of drain line for row select and reset. Conventional row select structure, row select transistor or row select capacitor, is omitted by the row-select channel that contains low impurity concentration and has no gate structure. Moreover, conventional reset transistor is also replaced by reset channel structure in like manner. These structures and triple level pulse operation of drain realize quite simple pixel structure in which amplification transistor is the only gate structure. A large fill factor of 37 percent is obtained by this structure, in 5.6 micrometers X 5.6 micrometers pixel designed by 0.7 micrometers rule.
Fixed pattern noise (FPN) for a CCD sensor is modeled as a sample of a spatial white noise process. This model is, however, not adequate for characterizing FPN in CMOS sensors, since the redout circuitry of CMOS sensors and CCDs are very different. The paper presents a model for CMOS FPN as the sum of two components: a column and a pixel component. Each component is modeled by a first order isotropic autoregressive random process, and each component. Each component is modeled by a first order isotropic autoregressive random process, and each component is assumed to be uncorrelated with the other. The parameters of the processes characterize each component of the FPN and the correlations between neighboring pixels and neighboring columns for a batch of sensor. We show how to estimate the model parameters from a set of measurements, and report estimates for 64 X 64 passive pixel sensor (PPS) and active pixel sensor (APS) test structures implemented in a 0.35 micron CMOS process. High spatial correlations between pixel components were measured for the PPS structures, and between the column components in both PPS and APS. The APS pixel components were uncorrelated.
The standard method for measuring QE for a CCD sensor is not adequate for CMOS APS since it does not take into consideration the random offset, gain variations, and nonlinearity introduced by the APS readout circuits. The paper presents a new method to accurately estimate QE of an APS. Instead of varying illumination as in the CCD method, illumination is kept constant and the pixel output is continuously observed - sampling at regular intervals. This makes it possible to eliminate random offset. The experiment is repeated multiple times to obtain good estimates of the pixel output mean and variance at each sample time. The sensor response is approximated by a piecewise linear function and using the Poisson statistics of shot noise gain, charge and read noise are estimated for each line segment. This procedure is repeated at no illumination so that dark charge may be estimated and subtracted from the total charge estimates. The method can also be used to estimate readout noise and gain FPN. Results from 64 X 64 pixel APS test structures implemented in a 0.35 micrometers CMOS process are reported. Using 6 different chips and 16 pixels per chip QE equals 0.37, gain FPN equals 2 percent, dark charge equals 832e-, and readout noise equals 40e-, are estimated.
Electron detector arrays are employed in numerous imaging applications, from low-light-level imaging to astronomy, electron microscopy, and nuclear instrumentation. The majority of these detectors are fabricated with dedicated processes, use the semiconductor as a stopping and detecting layer, and utilize CCD-type charge transfer and detection. We present a new detector, wherein electrons are stopped by an exposed metal layer, and are subsequently detected either through charge collection in a CCD-type well, or by a measurement of a potential drop across a capacitor which is discharged by these electrons. Spatial localization is achieved by use of two metal planes, one for protecting the underlying gate structures, and another, with metal pixel structures, for 2D detection. The new deice doe not suffer from semiconductor non-uniformities, and blooming effects are minimized. It is effective for electrons with energies of 2-6 keV. The unique structure makes it possible to achieve a high fill factor, and to incorporate on-chip processing. An imaging chip implementing several test structures incorporating the new detector has been fabricated using a 2 micron double-poly double-metal process, and has been tested inside a JEOL 6400 electron microscope.
Radiation exposure of CCD devices degrades the charge transfer inefficiency (CTI) by the creation of electron trap sights within the bulk silicon. The presence of electron traps tend to smear the signal of a point-like image. This affects CCDs used in star trackers where sub-pixel centroiding is required for accurate pointing knowledge. To explore the effects of radiation damage in CCD devices, we have developed a Monte-Carlo model for simulating charge transfer in buried channel CCDs. The model is based on the Shockley-Read-Hall generation-recombination theory. The CTI in CCD devices was measured before and after exposure to mono-energetic 61 MeV protons. Our data show that displacement damage in the bulk silicon increases the CTI of the CCD device. CTI was measure don irradiated CCD devices at various temperatures form -10 to -150 C, thus providing estimates of the electron trap energy levels created in the CCD silicon. The dominate post-radiation rap energy level was the silicon E-center found to be at an energy of 0.46 eV, which is in good agreement with other published values. To fit our data over the complete temperature range, we also required electron traps of 0.36 eV and 0.21 eV. Our model also includes the effects of charge cloud growth with signal volume and clocking rates of the CCD device. Determining the types and levels of radiation a CCD device will encounter during its operational life is very important for choosing CCD operating parameters.
The direct deposition of polycrystalline semiconductor HgI2 detectors on pre-deposited specially designed pixel electrodes is described, using two methods, the hot wall vapor deposition, HWVD, and thick film screen print (SP) methods. Some characterization results of the HgI2 material used to facilitate the detectors are described. The pre-deposited substrate is made by standard hybrid technology. The electrode pattern is a 16*16 pixel square pattern each with a size of 1.48 mm and with 0.1 mm spacing; the total area covered by the pixels is (25.28 mm)2 equals 639.078 mm2. In order to fan out the pixels to read-out electronics, holes were made through the ceramic thickness and connecting lines were drawn on the opposite side of the ceramic alumina substrate, where complicated patterns can be produced. The pixel detector is tested with beta particles, and data showing the leakage current vs. bias, are given showing a resistivity of about 2*1012 ohm cm. The current and the average charge signal are reported for three different HgI2 pixel detectors. The signal for one of the detectors is about 1100 electrons at 800 V bias voltage and for the second detector, the resistivity is in the same order of magnitude and the charge collection is somewhat better, reaching 1600 electrons at 700 V. One of the detectors was connected to a second hybrid designed for mounting of 8 castor 1.0 chips. CASTOR 1.0 is a VLSI circuit designed for imaging and the results are being evaluated.