NIST GCR 03-844
Low-Cost Manufacturing Process Technology for Amorphous Silicon Detector Panels: Applications in Digital Mammography and Radiography

2. Development of Low-Cost Process Technology

HOW DOES IT WORK?

In medical X-ray imaging, photosensitive detectors are a functional alternative to photographic film. Detector integrated circuits are fabricated with amorphous silicon (a-Si) semiconductor layers for full-field imaging of the entire breast or chest. Unlike single crystal silicon, a-Si layers can cover large areas for full-field imaging.

As indicated in Figure 1, the a-Si detector consists of a scintillator layer that converts incident X-ray energy to light and a photosensitive array that converts light into electrical charges. The photosensitive array is made up of picture elements (pixels) sized at 100–200 microns. Each pixel contains a photodiode that absorbs light from the scintillator and generates electrical charges and a field-effect transistor (FET) that serves to isolate each pixel element and acts as a switch to convey electrical charges to external electronics for read-out and image processing. The entire array of more than a million pixels can be read and converted to a composite digital representation in less than a second.

Figure 1. Principles of Amorphous Silicon Detector

GE was an early pioneer of digital imaging for medical applications and has developed an FDA-approved digital mammography system (Senographe 2000D) and an FDA-approved digital radiography system (Revolution XR/d). A key component in both systems is a full-field photosensitive detector plate large enough to capture the entire breast or chest area in a single image. To fabricate photosensitive a-Si detectors for GE Senographe digital mammography units and GE Revolution digital chest radiography units, GE and PKI currently use a manufacturing process that GE developed prior to ATP funding. This original manufacturing process has come to be known as the baseline process and provides a point of reference for ATP-funded process improvements.

In the baseline process, multiple layers of a-Si thin film are deposited on a glass plate and photolithography is used for pattern formation. The design approach (order and thickness of successive layers) starts with the buildup of FET layers, followed by thebuildup of diode layers, and completed with the deposition of the scintillator layer.

The baseline process consists of approximately 300 process steps and 11 photolithographic masks. The complexity of this process contributes to the high manufacturing cost of digital imaging devices and limits the widespread clinical utilization of digital mammography and radiography.

In response to GEGR’s proposal for federal funding to reduce process complexity and the high cost of photosensitive detectors and thereby enable increased clinical utilization of digital imaging, ATP cost shared the development of an LCM process for the fabrication of thin-film a-Si detectors.

LCM is an advanced manufacturing process with only seven photolithography masks and significantly reduced fabrication complexity. LCM reverses fabrication process order with the deposition of the diode island preceding the deposition of the FET.

Several a-Si fabrication steps, originally kept separate to optimize different aspects of photodiode and FET performance, are combined into dual-use process layers.

ATP PROJECT HISTORY

Beginning in the early 1970s, medical imaging has been moving toward replacing film-based technologies with digital imaging technologies. Computed tomography (CT) and magnetic resonance imaging (MRI) were originally developed as digital technologies tied to computers. Ultrasound and nuclear medicine transitioned to a digital format, and doctors and radiologists became increasingly familiar and experienced with computerized medical image processing.

Digital mammography, chest radiography, and cardiac imaging represent a more recent initiative. Starting in 1990, GEGR led the development of thin-film a-Si digital detectors. The Defense Advanced Research Project Agency (DARPA) participated in funding early demonstration projects for detector fabrication. GE’s initial target market for utilizing a-Si digital detectors was cardiac imaging, a field not particularly sensitive to the higher equipment costs of digital technology.

In the early 1990s, GE developed digital detectors for the more cost-sensitive mammography and chest radiography applications. During this second phase, GE obtained federal support from the U.S. Army Medical Research and Material Command, the U.S. Navy, Department of Radiology and Nuclear Medicine, and the Office of Women’s Health in the Department of Health and Human Services to develop and field test detectors for the GE Senographe 2000D digital mammography and Revolution XR/d digital radiography units.

Looking for ways to increase Senographe 2000D and Revolution XR/d market penetration, GE research scientists and engineers actively pursued ways to reduce the fabrication costs of digital detectors. GEGR proposed an innovative technical approach, the LCM process, for internal funding. LCM would reduce process complexity and eliminate several photolithographic masks. However, “getting internal support for the LCM proposal was hard. At the time, it was difficult for some executives to see the advantages of taking digital imaging beyond cardiac applications. Also, combining process steps in the fabrication of a-Si integrated circuits was considered to be a high-risk proposition” (Edelheit, 2001), including the following uncertainties:

• Could processing be developed to substantially reduce the number of process steps without compromising detector performance?
• Could the photodiode device be deposited at the same temperature as the FET without degrading detector performance?
• Could data lines be properly insulated to facilitate low noise readout?

Business risks were also significant. The President’s Task Force on National Healthcare Reform had recently been appointed. The task force was perceived by GE as advocating major reform for the healthcare industry, creating widespread market uncertainties. As a result of the technical risks and market uncertainties, GE was reluctant to approve the LCM process proposal for internal funding. Subsequently, GE Global Research, together with PKI, its strategic manufacturing partner for digital detector fabrication, approached the ATP for a grant to cost share the development of this high-risk, enabling process technology. “Given technical and other risks, had the ATP turned down the GEGR/PKI proposal, the promising lowcost manufacturing process initiative would have been shelved” (Edelheit, 2001).

In its 1995 General Competition, the ATP selected the proposed joint-venture project for a three-year award to develop an LCM process for large-area a-Si devices to be used in medical imaging systems and other applications. The award was later extended to five years. The core challenge was to combine manufacturing steps, which originally were kept separate to optimize different aspects of device performance without sacrificing any aspect. The ATP agreed to cost share $1.575 million of the $3.438 million project and GEGR and PKI committed to fund the balance.

The GEGR/PKI joint venture used a well-defined project structure to assign complementary roles to each strategic partner. At project inception, GEGR provided PKI access to its proprietary a-Si technology. Throughout the project, GEGR provided technology development, product specification, device testing, and evaluation for both 21 cm and 41 cm detector units. GEGR conducted additional process development for 21 cm detectors, and PKI provided process technology development and pilot production runs for 41 cm detectors. Once in full-scale production, GE and PKI fabrication departments will provide digital detectors to their “customer,” GE Medical Systems (GEMS), for use in Senographe digital mammography and Revolution digital radiography units.

MAJOR INNOVATIONS

The original GEGR baseline fabrication process was developed with the intent of optimizing the performance of each device (diode and FET). Given its complexity, the process is characterized by a large number of steps, long manufacturing cycle times, and high cost.

The goal of the ATP-funded low-cost process was a technology breakthrough in processing simplicity and lower cost. The project was successfully completed in 2000 and resulted in less complex fabrication with fewer masks (seven versus eleven) and fewer total process steps (200 versus 300). Fewer process steps increased production yield and reduced manufacturing cycle times. “Detailed modeling has verified that reductions in mask count closely scales with reduced process cost,” pointing to 25 percent cost reduction from low-cost process implementation (General Electric Global Research staff, 2001).

The major innovation of the low-cost process is the “interleaved fabrication of the light sensitive diode and the thin film transistor switching (FET) element without compromising either diode or FET performance” (Giambattista, 2001). Interleaved fabrication included the following dual and multiuse layers:

• Gate metal layer for scan line, FET gate, and bottom contact.
• FET dielectric layer for FET gate, diode sidewall passivation, and common electrode insulation.
• Barrier dielectric layer for FET sidewall passivation and protection barrier.
• Indium tin oxide (ITO) layer for diode and contact pads to drive electronics.

Additional innovations included:

• Elimination of labor-intensive test and repair steps for fabrication throughput advantage and improved data line repair capability intrinsic in device structure.
• Electronic noise reduction in data lines without additional process complexity.

Related technical accomplishments included:

• Reaching an acceptable compromise in FET and diode deposition temperatures.
• Identifying gate metals with acceptable sidewall slope after the etch.
• Optimizing FET island etch for selectivity to gate dielectric removal.
• Contact finger design for electronic bonding.

Technical accomplishments are reflected in patents issued to General Electric Company: US5838054 for Contact Pads for Radiation Imagers, US5648296 for Post Fabrication Repair Method for Thin Film Imager Device, and a patent filed for Gated Diodes for a Reduced Mask Imager Process. While the low-cost process was demonstrated to be technically feasible for producing medical quality X-ray imagers up to 41 cm2, the process is still considered to be developmental and “all risks have not been fully retired” (Giambattista, 2001). The need for additional effort beyond the ATP project was anticipated in the 1995 GEGR/PKI proposal to ATP: “Commercialization of program results will be accomplished by further, independent design and development to assure that ATPfunded R&D leads to a manufacturable low-cost process” (Giambattista, 2001).

Additional work is required to eliminate particle contamination that limits the yield impact of mask count reductions in fully optimizing the etch profile of bimetallic gate structures and to block light hitting the FET causing leakage in the data lines (Giambattista, 2001). Appendix A provides a more in-depth description of the LCM process technology and the technical accomplishments of the ATP-funded project.

LOW-COST PROCESS IMPLEMENTATION

The decision to implement the low-cost process, the timing of that decision, and the resolution of remaining technical issues will be made by GE and by PKI management.

It is expected that this business decision will be shaped by the following three factors:

• Demonstrated technical feasibility.
• Growing market demand for direct digital imaging in medical applications.
• Competitive pressures to reduce component (a-Si detector) costs.

These three factors are currently evolving and should effectively support a business decision to complete development and to implement the LCM process.

• Technical feasibility has been demonstrated and economic analysis indicates a 25 percent cost reduction.
• Independent market research points to long-term market demand for digital screening mammography and radiography.
• The FDA has approved GE digital mammography and digital radiography systems As other equipment vendors obtain FDA approval, there will be substantial downward pressure on equipment pricing. Even now, GE management is “very interested in component price reduction in the 5 to 10 percent range” (Giambattista, 2001).

A senior GE executive offered the following observation concerning the prospects of LCM implementation:

Return to Table of Contents or go to Section 3. Digital Mammography.

Date created: April 25, 2003
Last updated: August 2, 2005

ATP website comments: webmaster-atp@nist.gov  / Technical ATP inquiries: InfoCoord.ATP@nist.gov.

NIST is an agency of the U.S. Commerce Department
Privacy policy / Security Notice / Accessibility Statement / Disclaimer / Freedom of Information Act (FOIA) /
No Fear Act Policy / NIST Information Quallity Standards / ExpectMore.gov (performance of federal programs)