Semiconductor device manufacturing: process – Gettering of substrate
Reexamination Certificate
2001-04-20
2003-12-30
Eckert, George (Department: 2815)
Semiconductor device manufacturing: process
Gettering of substrate
C438S058000, C438S143000, C438S476000, C438S473000, C257S292000, C257S447000, C257S448000
Reexamination Certificate
active
06670258
ABSTRACT:
TECHNICAL FIELD
The present invention relates to electromagnetic radiation detectors, and more specifically to backside-illuminated semiconductor photodiode arrays.
BACKGROUND
A typical photodiode array includes a semiconductor substrate of a first conductivity type, having a front side formed with an array of doped regions of a second, opposite conductivity type, and an opposing back side that includes a heavily-doped bias electrode layer of the first conductivity type. For simplicity, the frontside doped regions are referred to below as gates, independent of their function as anodes or cathodes. Similarly, the abbreviation BEL is used to denote the backside bias electrode layer.
To provide a framework for the discussion below, an example of a photodiode array is used below in which the frontside doped gate regions have p-type conductivity, the substrate is n-type, and the backside bias electrode layer is, accordingly, a heavily-doped n-type layer. All statements and claims herein are equally true if the conductivity types of all the layers are reversed and the corresponding changes are made to the polarities of the charge carriers, applied voltages and electric fields.
Typically, the gate and bias electrode layers are formed internally to the crystalline semiconductor substrate. They are therefore native to and homostructural with the substrate. In most implementations, an external gate contact, formed from one or more non-native, heterostructural, conducting layers external to the substrate, is formed over a portion of each of the frontside gates. Similarly, one or more non-native, heterostructural, external back contacts may be formed over all, or a portion of, the backside bias electrode layer. In the case of silicon substrates, the gate contacts are usually formed from one or more metals, metal-silicon intermetallic compounds, or deposited, heavily-doped polysilicon, or a combination of a plurality of these materials. In the present context, polysilicon is considered to be both non-native and heterostructural to the crystalline silicon substrate. Similarly, silicon dioxide (SiO
2
), the amorphous “native oxide” of silicon, is both non-native and heterostructural to the substrate in this context. Back contacts to silicon photodiode arrays may use the same materials or one or more transparent conducting materials such as indium-tin oxide, which is an amorphous typically non-stoichiometric mixture of indium oxide and tin oxide. In many applications, an array of readout circuits is also formed on the front surface of the substrate.
A potential difference, referred to as a reverse bias, can be applied between the gate and the bias electrode layer to produce a depletion region within the substrate extending into the substrate from the p-n junction between the gate on the front side and the substrate. Hence, a photodiode is effected by the gate, the substrate and the BEL. External gate contacts or back contacts are ancillary elements provided to facilitate electrical connections to the photodiode array, not essential components thereof.
Such a photodiode array may be configured either in a frontside-illuminated mode to receive photons from the front side or in a backside-illuminated mode to receive photons from the backside. The frontside-illuminated mode, however, usually results in a lower external quantum efficiency (ratio of photocarriers collected to incident photons) than the backside-illuminated mode, because the conducting elements of the gate contacts and the readout circuits (if provided) reduce the active photosensitive area of the array on the front side. In comparison, the entire back side can be used to collect incoming radiation when properly configured. All other factors being equal, enhanced photosensitivity results in increased signal-to-noise ratio. In single-particle radiation detection applications using either direct (intrinsic) detection in the substrate or indirect detection (e.g., using scintillators as discussed below), enhanced photosensitivity results in improved particle energy resolution. In addition, conducting lines and other physical features such as steps in dielectric thickness on the front side can scatter light into the photosensitive areas of adjacent photodiodes, thereby reducing image contrast. Contrast degradation modifies the modulation transfer function of the array and can reduce the useful spatial resolution of the array. Therefore, backside illuminated photodiode arrays are frequently used in imaging applications to improve photosensitivity, signal-to-noise ratio, particle energy resolution and spatial resolution.
In a backside-illuminated photodiode, photocurrent is typically generated by band-to-band absorption. Photons with energy greater than the bandgap of the semiconductor substrate enter the back of the substrate and are absorbed, producing electron-hole pairs. If an electron-hole pair is generated outside the depletion region of a gate, the minority carrier (a hole in the example above) may diffuse to the edge of the depletion region beneath one of the gates. The electric field within the depletion region “collects” the hole by accelerating it towards the gate. If, however, a photon is absorbed within the depletion region of a gate, the electric field “collects” the hole as above, but accelerates the electron towards the undepleted substrate, or, if the substrate is fully depleted below a gate, towards the backside bias electrode layer. In either case, the photocurrent will flow through the photodiode and the external circuitry that maintains the bias between the gate and the bias electrode layer. If readout circuitry is provided on same semiconductor substrate, the circuit elements associated with each gate will produce a signal that represents a mathematical function of the photocurrent, the quantity of charge caused by the photon absorption, or a combination of both.
In low light-level imaging applications such as night photography, nuclear medical imaging, photon medical imaging, x-ray computed tomography and ballistic photon detection, it is critical for photodiode arrays simultaneously to exhibit high external photon conversion efficiency (defined as the ratio of photocarriers collected to photons incident on the back surface of the substrate) and extremely low reverse-bias leakage currents. For brevity, “quantum efficiency” is used below to denote external photon conversion efficiency, and “leakage current” is used to denote reverse-bias leakage current.
In low light-level imaging systems, the input optical signal often is in the form of short pulses, a few nanoseconds to a few microseconds in duration. For these applications, it is highly beneficial for the photodiodes to have short pulse response times, often referred to collectively as transition times or, singly, as rise and fall times, in the rough order of magnitude range of 10 to 100 ns or less. The signal processing system associated with the photodiode arrays typically “shapes” the output pulse by integrating the photocurrent generated within a time window of fixed duration in the same order of magnitude as the length of the photopulse.
Photodiode arrays with one or more long transition times produce output photocurrent pulses significantly longer than the input photopulse. These long photocurrent pulses produce smaller output pulses from the signal processing electronics than do those from faster photodiode arrays. Therefore, slow photodiode arrays may result in low output signals from the signal processing electronics and therefore degrade the signal-to-noise ratio of the image. In applications with higher photon flux rates, e.g., x-ray computed tomography, the signal processing electronics may respond to the average photocurrent collected by a pixel over a time interval of fixed duration, longer than the duration of a single optical pulse. For these applications, longer transition times than those required for efficient single pulse detection may be acceptable.
To achieve high quantum efficiency, the bias electrode layer should be thin enough to be transp
Carlson Lars S.
Mollet Alan
Sheridan John
Zhao Shulai
Díaz José R
Digirad Corporation
Eckert George
Fish & Richardson P.C.
LandOfFree
Fabrication of low leakage-current backside illuminated... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Fabrication of low leakage-current backside illuminated..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Fabrication of low leakage-current backside illuminated... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3146489