Semiconductor device manufacturing: process – Making device or circuit responsive to nonelectrical signal – Responsive to electromagnetic radiation
Reexamination Certificate
2001-03-13
2004-01-13
Nelms, David (Department: 2818)
Semiconductor device manufacturing: process
Making device or circuit responsive to nonelectrical signal
Responsive to electromagnetic radiation
Reexamination Certificate
active
06677178
ABSTRACT:
FIELD OF THE INVENTION
This invention pertains to semiconductor devices and their manufacturing methods, in particular to rear surface incident-type light receptacle devices, which are useful for detection of ultraviolet and other short wavelength energy beams, and their manufacturing methods.
BACKGROUND OF THE INVENTION
In image-pickup devices based on charge-coupled device (CCD) architecture, incident photons are received by an array of pixel units (CMOS capacitors) on a silicon substrate. Each pixel comprises an electrode (sometimes called a “gate”) mounted on the silicon substrate with an intervening thin dielectric layer. In the silicon beneath each electrode, a depletion region (“charge well”) is formed. As a photon having an energy greater than the energy gap of the respective MOS capacitor enters the depletion region of a pixel, if the photon is absorbed in the depletion region, then the photon produces an electron-hole pair. The electron stays within the charge well to contribute to a charge accumulated there over a defined time interval. After the time interval (i.e., periodically), the respective charge wells of the pixels are “read out” in a controlled manner to downstream processing electronics that convert the charge data into corresponding image data. Several conventional schemes have been devised for outputting the charges from the pixels. One scheme is termed the “full-frame transfer” (FFT) scheme. With FFT, the optics used to direct incident light to the light-detecting surface can be at maximal aperture.
In certain types of conventional FFT-type CCD arrays, a representative pixel includes an electrode (“gate”) situated over the respective depletion region (charge well) of the pixel. For a CCD array sensitive to visible light, the electrode usually is made of ITO (indium tin oxide) or other suitable material that is transparent to certain wavelengths of incident light. Specifically, “long-wavelength” light (e.g., visible light), to which the electrode has a relatively low absorption coefficient, passes from an upstream direction through the electrode to the respective charge well. In each pixel, the electrode is situated proximally to the respective charge well (but separated from the charge well by the thin dielectric layer). The charge well accumulates charge from photons of the long-wavelength light transmitted through the electrode. The electrode has large respective absorption coefficients for light wavelengths that are relatively short (e.g., ultraviolet light) as well as for certain particulate radiation such as electrons. This absorption significantly reduces the sensitivity of the CCD array.
So-called “back-side-incidence” (BSI) CCD arrays have been proposed for detecting short-wavelength light and certain types of particulate radiation. In these arrays, with respect to each pixel, the light-incidence surface is on a “back” or “rear” surface of the silicon substrate, opposite a “front” surface on which the dielectric layer and the electrodes (gates) of the pixel are formed.
In BSI CCD arrays, as noted above, the substrate is usually silicon. The substrate typically is “thin” compared to the thickness of the silicon substrate of a conventional front-side-incidence CCD array (typically 300 &mgr;m to 500 &mgr;m). The “thin” substrate in a BSI CCD array has a thickness of approximately 10 &mgr;m to 20 &mgr;m. The substrate is thin in a BSI CCD because short-wavelength light has a relatively large absorption coefficient in silicon. Short-wavelength photons are absorbed near the incidence surface and converted to electron-hole pairs in the substrate proximal to the light-incidence surface. If the substrate were thicker, then the electron-hole pairs would recombine within the substrate before electrons could reach the respective depletion regions. This recombination substantially reduces the sensitivity of the device. Also, a thicker substrate in a BSI CCD array would tend to “mix” the electrons produced in various pixels, which would reduce the image resolution of the device.
The thin CCD substrate used in a BSI CCD array has low mechanical strength. To increase the strength, a “reinforcing substrate” (made of, e.g., silicon or glass) conventionally is bonded to the “front” surface of the CCD substrate (i.e., the surface on which the electrodes are formed). Such reinforcement prevents damage to the CCD substrate during various subsequent processing steps executed on the surface of the CCD substrate of the BSI CCD array, and facilitates handling during later fabrication processes (such as dicing).
According to conventional bonding methods, the reinforcing substrate is bonded to the CCD substrate using an adhesive having a post-cure hardness exceeding a certain hardness threshold. The adhesion is created by first applying a liquid silicate or other silicon oxide material to the reinforcing substrate, such as conventionally used when applying a layer of borophosphosilicate glass (BPSG) or spun-on glass (SOG). Then, an epoxy or other general-purpose resin adhesive is applied to adhere the CCD substrate to the silicon oxide surface of the reinforcing substrate.
Operationally, BSI CCD arrays must transfer photons efficiently from the light-incidence surface (back surface) of the CCD substrate to the respective depletion regions (charge wells) located on the front surface of the CCD substrate. To facilitate efficient diffusion of photons, ideally no crystal defects or metallic impurities should be present in the CCD substrate.
Conventional methods for fabricating BSI CCD arrays are directed to preventing formation of crystal defects and maintaining unwanted metallic impurities to insignificant concentrations in the CCD substrate. First, a semiconductor (silicon) “fabrication” substrate is prepared as a wafer on which the CCD substrate is formed subsequently by epitaxial growth. The fabrication substrate is formed with intrinsic gettering (IG) to minimize formation of crystal defects at least in a surficial layer of the fabrication substrate. The surficial layer is termed the “low-defect layer,” and its surface is termed the “low-defect surface.” The epitaxy is performed on the low-defect surface to form the CCD substrate. After forming the epitaxial layer, other layers and the pixel electrodes are formed on the epitaxial layer, followed by bonding of the reinforcing substrate. During a downstream step, the fabrication substrate is removed, such as by wet-etching, as discussed below.
Using the conventional process described above, it is difficult to obtain a BSI CCD array having a desired high sensitivity and resolution for short-wavelength light. There are several reasons for this difficulty.
First, as noted above, silicon oxide and a resin adhesive normally are used to bond the CCD substrate to the reinforcing substrate. Although silicon oxide adheres well to the reinforcing substrate, this bond is vulnerable to failure, resulting in the CCD substrate peeling away from the reinforcing substrate, especially during later fabrication steps (e.g., dicing). In addition, the surfaces to be bonded together must be mutually level during bonding. Consequently, the bonding process is complex inasmuch as it involves applying a silicon-based material (silicon oxide) onto a silicon-based reinforcing substrate, polishing the silicon-based material, applying the adhesive, and high-temperature processing to cure the adhesive.
Second, when using a general-purpose resin adhesive for bonding, stress arises in the adhesive interface after the adhesive is cured. The stress arises from differences in thermal-expansion coefficients of the substrates being bonded versus the adhesive itself. The stress is manifest as bending strain of the CCD substrate, causing operational problems such as disruptions in the transmission of photons through the CCD substrate to the respective diffusion layers and in the conduction of electrons from the diffusion layers to the respective electrodes. The stress also causes physical damage to the pixels.
Third, after forming the epitaxial layer on the low-
Klarquist & Sparkman, LLP
Le Thao P
Nelms David
Nikon Corporation
LandOfFree
Semiconductor devices including back-surface-incidence CCD... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Semiconductor devices including back-surface-incidence CCD..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Semiconductor devices including back-surface-incidence CCD... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3203287