Large format CID x-ray image sensor J. Carbone*, Z. Alam, C. Borman, S. Czebiniak and H. Ziegler

CIDTEC 101 Commerce Blvd., Liverpool, NY 13088

ABSTRACT A large format (3 1 x 23 mm2 display) CID imager module capitalizes on CID large well capacity and radiation resistance to image dental x-rays. The module, which consists ofthe 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 RFIIEMI 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 6OO format. Each pixel is square and 38.5 microns/side. The imager is coated using a proprietary phosphor deposition process that results 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. Keywords: CIDs, dental X-ray, electronic imaging, radiation-hardened, charge transfer device

1. INTRODUCTION Digital imaging is a rapidly expanding technology that has advanced the performance of sophisticated and high-performance X-ray imaging systems'. Until now, these systems incorporated large-format intensified tubes or used film as the key detector element, and images were often generated eventually using off-line video processing or developing techniques. High-yielding integrated circuit processes are demonstrating large, high-performance charge transfer devices (CTDs) can be fabricated economically and transplanted into these formerly "exclusive" applications.

1.1. Market More recently, CTDs are displacing film in dental imaging applications. In most dental offices, images are created exposing a patient' s dental structure onto coated X-ray sensitive film. Using harmful chemicals that require storage and disposal, the film is then processed to yield the typical hard-copy film negative that is interpreted by the dentist, and stored for future reference. CTDs are capable of providing superior real-time X-ray images on monitor displays and/or hard copy photographs without the need to store conventional film and chemicals. Additionally, the information can be stored in electronic format.

1.2. CID compatibility CIDs are naturally compatible for X-ray imaging applications because of their efficient pixel geometry, large pixel well capacity and resistance to ionizing radiation. Their generic conceptlstructure has been described in many publications2' 3and will not be repeated herein. During readout, signal charge is transferred once within the pixel to be read out. Hence, CID pixels do not need to incorporate a large charge transport register that detracts from the pixel quantum efficiency and causes unwanted aliasing affects to the imager modulation transfer function (MTF). As a secondary consequence of this efficient pixel structure, CID pixels offer proportionately larger well capacities than other CTDs including charge coupled devices (CCDs), and therefore provide quantum-limited operation and higher dynamic range under high illumination conditions. Most importantly, CIDs are extremely resistant to ionizing radiation effects4. CCDs suffer extreme loss of charge transfer efficiency (CTh) and exhibit very noisy performance in the presence of high radiation flux rates. They even cease to operate following exposure to a few tens of kilorads total dose silicon from a cobalt source. CIDs have demonstrated5 the capability

90

SPIE Vol. 3301 • 0277-786X/98/$1O.OO

to provide ualitv video i lagcs in the presence of flux rates exceeding 50() Krads!hr. and. the continue to operate hcond total dosc exposures greater than I nicearad. N—ray imaging is a natural fit for (.]I)s where hlgh—cnerg\ N—rays generate several thousand light photons and quantum—

united operation is necessary to provide a good idco analog of the N-ray imaoe .Adlitionall\ the Cl!) is robust and will continue to operate afier several hundred thousand X-ra exposures (the cquialcnt of several hundred thousand rads total dose front an X—ray source).

1.3. ('11) dental imaging module A large forniat Cl 1)—based imager module Fig. I . has been fabricated for use in pulsed dental X—ray detection and imaging applications. Ihe module features a 3! s. 23 mnr optical format that is comparable with standard dental fi tin, and a S I 2 x 607 6rntat CII) pixel matrix that is compatible with SVCA standard television presentation format AC ID imager. conversion phosphor and ancillar\ electronics are encapsulated within the 40 x 28 x 5 mm robust, hut benign. lightseight package.

Fig. 1. Dental imaging module

Fig. 2. CII)41 imager

'ftc module is self-contained: i.e. is capable of detecting and reading out the X—ray image without the need for an external host. llndcr power. the module remains in a reset mode. Following detection of an X-ray event, the imager places itself into an integration niode where it remains until the X-ray exposure is complete. Readout begins asynchronously unmediately upon completion of the X-ray exposure.

flie CI[) imagcr (a.k.a. CID4I - Fig. 2.) is self-scanned and possesses the large pixel geometry and well capacit that are needed for quantuni—limited X—ray detection and imaging. It consists of a pixel matrix that incorporates on—board

preamplifiers with each row, and individual ro's and column multiplexers that provide connection and br voltage referencing for each ross or column. Sequential operation row and column shift registers specify the row/column cooi'dinates of the pixel to he read out. and on-hoard logic circuits provide the timing signals necessary to facilitate pixel charge readout and reset charge Injection). 1'Iic mager is coated with an X-ray cons ersion phosphor (Fig. 3i and is mounted on the hack-side of a small ancillary electronics circuit hoard (Fig. 4. (that provides rudimentary signal amplification and signal buffering between the iniager and an external system interface (typically a TV up-eonvcrterL The complete assembly is encapsulated in a black package that is robust and resistant to electromagnetic interference and many harsh chemicals. The assembly is light proof. moisture-proof. and meets FDA and RFI/EMI standards.

Ancillary electronics

Fig. 4.

Fig. 3. Phosphor-coated C11)41 imager

2. CID4I IMA(;ER 2.1. Description

A block diatrani of the CID4 1 imacr is illustrated in Fie .5. The pixel array consists of 12 colunins and 607 rows. The first 4 rows and the first 12 columns arc covered with aluminum arid arc used to provide optical hack references that arc used for video DC restoration and on—chip subtraction of column—oriented fixed pattern noise.

To External Amplifier

Source -

Drain

_.Q_ —— — Co

C)

Pixel Array 0

E

—— C)

Source

Black Reference Rows

From reference

I 1 Li LJ IJJ ColumnI Muxer LJH:T11111111HJ1H Column Scanner I

row

Drain

On-Chip

Logic —

I

I

Photodetector surrounds the entire chip

Fig. 5. CID4I Block diagram These areas are also pre-injected and cleared of possible charge generated and collected from scattered direct X-ray photons. The pixel area is fabricated on a precise epitaxial layer that acts as a fast charge coilcctor' for injected charge and for charge that might he generated deep within the epitaxial aver via direct X—ray detection (Direct X-ray detection is undesirable due to its objectionable noise character).

Readout of the picl arra Is achieved via placement of two scanners orthogonally located with respect to each other on two sides of the pixel array. 'l'he vertical scanner is a tour phase shift register and addresses individual rows. niultiple xing tie addressed ross to the appropriate voltage for charge injection or integration, or to the ross preamplifier for signal readout. 'l'he horiiontal scanner is also a four-phase shift register that provides sequential column addressing Scanner outputs select and operate the column niultiplexers which drise the columns to effect charge transfer froni column to ross or place the column into a charge integration or niection mode. Pixel readout is progressive, that is. all the elements in one ross f'ollossed all elements in the next row. Ura fl

T .i..

or

-

., 4

SouIc

C

--I.'

--

r —4 Li-

L,ra?rr

Column Drive Logic

] Fig. 7. Response to pulsed visible illumination source

Fig. ô. CII) preamplifiers and readout

Video output is Iron) two ports: one containing the actual signal and the other output is from the reference row as illustrated in Fig. 6. ('11) Preampli fiers and Readout. The latter output is subtracted from the actual video signal in real time to yield high qual it video signal shown in Fig. 7. — Response to pulsed light source. I muting resolution ol I) LP/mm. is illustrated in Fig. . - Limiting resolution.

I I

III

2

III

- __

:EIII vElli

iii

a

___

III III r-s

''

——— Fig. 8. Limiting resolution

Fig. 9. CID41 Pixel layout

Large geometry, P-channel preamplifiers that feature low Johnson and 1/f noise are mounted on-chip at the ends of each pixel row. The CID pixel layout is illustrated in Fig. 9. The "two-capacitor" concept is maintained but deviates from a conventional "crossed cell" electrode configuration in order to enhance the contrast ratio at limiting resolution. Relevant CID41 parameters (less phosphor characteristics) are contained in Table I.

Table I. CID41 Parameters (less phosphor)

Parameter

—

Value

Units

Pixel matrix:

Total (H X V) size (H X V) displayed (H X V) size (H X V)

Pixel size (H X V) capacity On-board Preamplifier FETs: Size (WJL) Transconductance (@ ima) " Johnson noise " 1/f noise " noise Limiting resolution

812 X 607

pixels

3 1 .3 X 23.4

microns2

800 X 600

pixels

30.8 X 23. 1

microns2

38.5 X 38.5 >106

electrons

800/3.5 3.5 1 .8

microns2

microns/microns ma/Volt

21

nV/Hz112 nV/Hz 1/2 100 Hz

< 400 > 900

electrons LP/MM

Additional logic and photodetector circuits are placed on-chip in order to minimize cabling and timing interface complexity between the imager, electronics module and system interface. These circuits include: clock drivers; clamp generation; start of X-ray (SOX) pulse; pixel rate clock output; and special photodetectors that monitor the occurrence and duration of X-ray pulses.

2.2. Imager operation

In the dormant mode (time zero), the imager is in injection (CID rows and columns are at the same potential as the epitaxial layer and do not collect charge). The occurrence of an X-ray (or visible) photo event generates a SOX pulse that initiates the charge integration, readout and return to reset (dormant mode) sequence within the imager. The rising edge of the SOX pulse puts the imager into an integration mode; i.e. the columns are raised above threshold, which results in creation of potential wells at all pixel locations and allows the storage of charge within each pixel site. The SOX pulse remains "high" until the X-ray or photo exposure is complete (An on-board intelligent discriminator anticipates the frequency which these pulses occur). Following exposure, the SOX returns to a "low" state and pixel readout begins immediately. During the readout of the first row, the reference row is pre-injected to clear any charge that may have accumulated during the integration period from visible or spurious direct X-ray photons. Every subsequent row is read simultaneously together with the reference row, and subtracted on-chip to produce the output video signal. Once the entire array has been read, the imager automatically returns into injection (or reset mode) and awaits the next SOX pulse. This timing sequence is illustrated in Figure 10. — Global timing.

3. X-RAY CONVERSION The conversion phosphor is a commercial gadolinium oxysulfide activated with terbium. The emission peak for this material is at 545 nm, which is a reasonably good match to the spectral response of the CID41 . The phosphor thickness, 85 microns, was chosen to maximize X-ray absorption without compromising the limiting resolution of the pixel, measured slightly greater than 9 LP/mm.

94

''

CID Rows

-4

2tj

C

so x Cu urnu Drv-.

Co U m[

Row Reset

-

Reduut

SeUUr.c

Reset

Tmu

Expuse SettHn Time

Fig. 10. Global Timing Diagram l'he x—ray enereiv hsorption of the phosphor was calculated to he approximately 3O for a typical dental source X—ra\ is shix n the calculated incident x-ray photon fluence for a self-rectified 7() kVp x-ray generator with .spectrunl.5 In Fig. 2—mm aluminum filtration. .-\lso shown in Foz. I I is the calculated x—ray tluence on the detector using a model for patient anatoniv. lhe mode! sitiiulates patient anatomy with a 2—cm thick slab of muscle and a 2—cm thick slab of hone. lhc third curve in Fig. I I is the calculated x-ray absorption by the phosphor. I

900

r incident tuence 800

.

700

.

8uenco on detector

.

phosphor absorption

600

x Q. 500

0 0

.

400

C

.

0 300

r /

200

100

0

-.

- -- - -

010

20

30

40

-S. -------

50

60

70

80

X-ray Energy (key)

Fig. 11. Incident X-ray fluence. fluence on detector after passing through patient anatomy, and Xray absorption of fluence on detector by phosphor. X-ray transmission measurements made for several different phosphor thickness using an ionization chamber detector are he I 3 Each absorbed X-ray photon

shown in Fig. I 2. The energy emission elTiciencv of the phosphor is estimated to generates 2(XR)-40()O visible photons.

95

90 85 80 75

0

70

Cl)

E C

65

I.- 60 55 50

45 40

0

20

40

60

80

100

120

140

160

180

Phosphor Thickness (microns)

Fig. 12. X-ray transmission as a function of phosphor thickness

3.1. X-ray imaging results

An unpn)cessed X-ray image of a "dental phantom' was taken with the CID4I X-ray module and is shown in Fig. 13. The image was taken using a nominally 60 keV conventional dental X-ray source.

Fig. 13. Dental X-ra' image using 60 ke\ source

4. SIGNAL PROCESSING A block diagram ofthe DIM Ancillary Electronics is shown in Fig. 14. The DIM module Ancillary Electronics consists of the SOX (Start of X-ray) detection circuitry, video processing circuitry, a crystal oscillator and pixel clock output driver.

sox Det

Photo Diode SOX Input

Ext

sox __________Video

CID 41

Imager Clamp

Osc

Clock In Pixel CIk Out

Pixel CIk Out /Pixel CIk Out

Fig. 14. Block diagram — Ancillary Electronics

The SOX detection circuit consists of a voltage comparator and a time delay circuit. The comparator is used to detect the magnitude of the output from the photo detector during X-ray radiation. The delay circuit is a time discriminator circuit that keeps the SOX signal to the imager asserted until the passing of the last X-ray pulse. An external SOX input is also provided for test purposes and to facilitate manual (or host controlled) operation of the module. The read out format is a continuous stream of pixels with no gaps between lines. The first four rows as well as the first twelve pixels in every row are opaque (to light and X-rays), and serve as optical black references for subsequent video processing. The video processing circuit consists of a post amplifier, clamp circuit, and output amplifier. The post amplifier magnifies the video signal and provides a low impedance drive for the clamp circuit. The Imager provides a clamp signal, which is asserted during the third through tenth black reference pixel of each line, and is used to DC restore and remove KTC noise from each video line. The output amplifier boosts the video to a one-volt level and provides a low impedance drive for the coax cable. The crystal oscillator provides a continuous clock to the imager, which remains operational while under power. The imager provides a pixel clock out with each pixel of video except during the clamp and dormant intervals.

4.1. Interface The following is a listing of the interface signals between the DIM module and host system: (Optional) External SOX; Pixel Clock; /Pixel Clock; -7.5 VDC; +7.5 VDC; Video; Video Return; Ground; and, Shield.

97

5. MODULE PACKAGE 5.1. Description The dental module (Fig. I .) consists of a circuit board populated with surface-mount components (Fig.4.) with the imager attached to the opposite side (Figs. 2. & 3.), all contained within a visible-blind encapsulation. An X-ray sensitive scintillator is deposited directly on the imager. The unit is fully encapsulated in medical grade (USP Class VI) epoxy and is immersible in common dental sterilizing agents. Full EMI shielding is integral to the unit.

An extremely flexible cable with flex relief is attached to the module and provides the electrical interface between the module and television display interface. Strain relief is accomplished via tensile members within the cable. The cable is jacketed with polyurethane to maintain flexibility and provide bite-through resistance. The scintillator is deposited using a proprietary technique that yields high quality films with good batch repeatability. Process parameters can be varied with minimal effort to alter the film spectral and resolution characteristics.

5.2. Tests and Certifications Modules have been tested, evaluated and certified to the levels shown in Table II.

5.3. Table II. Certifications Description Cable-to-module interface flex Cable pull Shock Extended soak

Cytotoxicity Radiation EMI shielding Dielectric (Hipot) ESD FDA

Criteria 22 mm bend radius, 180 degree arc Tension applied to failure >2000g drop-shock 3.4% glutaraldehyde sterilizing agent Non-toxic when ingested Accelerated lifetime No interaction with environment 4000 volt breakdown Eposure to 6000 volt discharge Clinical evaluation

Result >106 cycles @ 1 cycle/sec

Unit breaks before cable tensile members fail Passed Passed with no degradation

Passed independent evaluation Passed Passed

Passed, undamaged Passed, undamaged Passed

6. ACKNOWLEDGEMENTS The authors thank the many efforts of co-workers at CIDTEC for their responsive support, and David Breithaupt, Andy Beardslee (InfiMed Corporation), John Cox (Primex General Imaging) and Gerry Michon for their advice and technical suggestions during the course of this development program.

7. REFERENCES 1 B . Carison, Pushing the Medical and industrial Digital X-ray Image Envelope —At Last, Advanced Imaging pp. 24-29,

November (1997). 2 Carbone, J. Zarnowski, F. Arnold, J. Hutton, New low-noise, random access, radiation-resistant and large-format charge injection device (CiD) imagers, Proceedings SPIE vol.1900, pp-.l7O-18O (1993) 3 R.F.Wentink, J. Carbone, Charge injection Device(CJD) Technology — a solutionforphoton andparticle imaging applications, Nuclear Instruments and Methods in Physics Research A 37 pp. 522-528 (1994) 4 j. Carbone, J. Zarnowski, M. Pace, S. Czebiniak, R. Carta, Megarad and Scientific CiDs, Proceedings SPIE vol.2654 (1996). 5 J. Carbone, S. Czebiniak, R. Carta, New CID Detectors/Cainerasfor Use in ionizing Radiation Environments, American Nuclear Society (ANS) Proc. 43rd Conference on Robotics and Remote Systems, SanFrancisco, CA, pp. 43-50, (1995). 6 G. Michon, Method andapparatusfor Sensing Radiation and Providing Electrical Readout, United States Patent #3,786,263, Jan. 15, 1974. 7 G. Michon, CID Imager with improved sensitivity, SPSE Conference on Electronic Imaging, Oct. 13-17 (1996).

98

8 John Cox, Primex General Imaging, unpublished model.

Dave Breithaupt, Infimed, private communication.

'° J.A. DePooter and A. Bril, J. Electrochemical Soc. 122 1088 (1975).

99