US 4,600,853 AGrant
Saw-CTD Serial to Parallel Imager and Waveform Recorder
Issue Date:1986-07-15
•13 Claims
•4 Drawing Sheets
Abstract
A device for the high speed recording of photon images and nonrepetitive electrical waveforms which comprises a waveform recorder wherein surface acoustic waves excited in a can be GaAs, not layered piezoelectric-insulator-semiconductor layered structure produce a traveling electric field in the semiconductor substrate. Charges stored in the traveling potential wells and representing the instantaenous amplitude of a waveform to be recorded are transferred at static wells when a gate is dropped. Because each successive traveling well represents a different time instant of the waveform, the different spatial locations of the static wells correspond to different times. The output signal from the static wells can be selectively delayed before application to a display oscilloscope to enable display of the waveform at a rate many times slower than the actual frequency of the signal waveform.
Metadata
Assignee
- The United States of America as represented by the Secretary of the Navy
Inventors
- Robert R. Whitlock
- Nicolas A. Papanicolaou
Application Information
Application Number:US 768651&
Filing Date:1985-08-23
Priority Date:1985-08-23
Art Unit:212
Classifications
IPC:
H01L 4108
Field of Search:
310346367357365358330324333250235343313 R;313 B;313 C;3221;158;162726103;114;118201;213;2964.6;5.5111;113;102;109;77 R;77 A;78 E154;193301;370;39518117.1;100 CL
Patent Drawings (4 sheets)
Description
Background of the Invention
The present invention relates to a system for recording electrical waveforms and more particularly to such a system employing a charge transfer device.
The high speed recording of photon images involves both mature and developing camera technologies which span the electromagnetic spectrum from infrared to x-rays, with temporal resolutions of micro- to femto-seconds. Often, the cameras, like streak cameras, are very expensive, bulky or involve precision machinery, such as rotating mirrors.
For many applications, it would be advantageous to have small, solid state imaging devices manufactured and replicated by the methods of the semiconductor circuit industry. A solid state imager necessarily includes three elements: (1) an input comprised of an array of image sensors, to transduce the incident photon energy into charge carriers in the device; (2) a charge transfer technique, to move the charges from input to an output in an organized fashion; and (3) the output, which may involve storage, display, or further processing. The arrival of alternate technologies for any of these three elements can enable the development of a new class of imagers. In particular, alternate technologies for solid state charge transfer devices (CTDs) are now available.
The first CTD, originally conceived by N. Wiener (Cybernetics, John Wiley and Sons, Inc., New York, 1948, p.144), was the Bucket Brigade Device (BBD), in which charge is stored in a serial array of capacitors and transferred by proper activation of switches placed between them. Several versions of the BBD were later developed utilizing mechanical switches, vacuum tubes, or bipolar transistors as the switching elements. The fully integrated circuit version came about in 1970, when the switches were replaced by MOS transistors. See F. L. J. Sangster, "Integrated Bucket Brigade Delay Line Using MOS Tetrodes," Phillips Tech. Review 31;266.
It was in 1970 when W. S. Boyle and G. E. Smith, "Charge Coupled Semiconductor Devices," Bell Syst. Tech. Jour. 49, 1970, pp.587-593, conceived and demonstrated the first Charge Coupled Device (CCD). In its original version the CCD consisted of closely spaced electrodes on an isolated surface of a semiconductor. With a proper sequence of pulses applied to these electrodes, packets of minority carriers were transferred along the surface of the semiconductor. The impact of the CCD was immense and swift. Almost immediately dozens of schemes and variations of the device were under study and many potential applications were in the making.
Charge Coupled Devices play an important role as self scanned light sensors. See C. H. Sequin, "Image Sensors Using Surface Charge Coupled Devices," in Solid State Imaging, edited by P. G. Gespers, F. Van der Wiele and M. H. White, Noordhoff-Leyden, 1976, pp. 305-329. Since silicon is a photosensitive material, charge packets may be created by a light pattern directed onto the surface of the semiconductor. These packets are transported to the output port of the CCD and read out sequentially. The CCD can also perform analog signal processing functions such as high density storage (e.g., of an image), analog delay and transversal filtering.
Initial interest in the CCD was heightened by the striking simplicity of its structure and the range of potential applications. However, as the CCD technology progressed the device performance improved on the one hand while its complexity increased on the other hand. In practical devices the electrodes are driven by multiphase clocks (up to four phases) requiring complex peripheral circuitry. Multilevel electrodes and buried channels are required to improve the efficiency and speed. For imagers, the clock busses occupy a large fraction of the available area. These drawbacks limit the speed, density and simplicity of the device.
In the realm of nonrepetitive electrical waveforms, the fastest recorders are oscilloscopes, which operate on the principle of deflecting an electron beam. A common way to achieve fast response is to compromise the sensitivity of the oscilloscope. Compensation for the loss in sensitivity can sometimes be made by employing amplifiction in the electronics, which is difficult and expensive for fast pulses.
Digital recorders of nonrepetitive waveforms are even more expensive and generally have less bandwidth than state-of-the-art oscilloscopes.
For repetitive signals, sampling scopes offer better temporal resolution than oscilloscopes. One reason sampling scopes are used, is to obtain temporal resolution of the waveform, even if that requires the added inconvenience of sampling many repetitions of the waveform. If superior temporal resolution were available, many applications of the expensive sampling scopes would be obviated.
U.S. Pat. No. 4,389,590; "System for recording waveforms using spatial dispersion", R. R. Whitlock (1983) discloses devices that allow for the recordation of electrical waveforms, be it from an electrical source or a photon source, and allows for bandwidths of greater than 1 GHZ. However, one major limitation of this device is that the transfer time across a gate is no faster than one period of the wave. Another major limitation is the requirement for two waves.
Summary of the Invention
Accordingly, it is an object of the present invention to record electrical or optical waveforms.
Another object of the present invention is to record electrical or optical waveforms using a charge transfer device.
These and other objects of the present invention are realized by a waveform recorder comprising: a semiconductor body; a planar piezo-electric body having a main surface overlying and in proximity to the semiconductor body; wave propagation means for propagating acoustic wave signals on the main surface of the piezoelectric body to create traveling potential wells in the underlying semiconductor body; a traveling potential well path located in the semiconductor body; charge injection means for injecting into the potential wells traveling past at least one first location in the semiconductor body a quantity of minority charge carriers whose magnitude tracks with time the waveform of a signal to be recorded, the first location of the charge injection means being located near the start of and in the traveling potential well path; charge collection means for collecting the injected charge carriers, the charge collection means being located in the semiconductor body at a spatially separated second location and disposed so that the traveling potential well path starts at a position directly below the wave propagation means and continues straight to the charge collection means; semiconductor depletion means for depleting the semiconductor of majority carriers along the traveling potential well path; a gate located on the semiconductor body and alongside and parallel to the traveling potential well path, and adjoining the semiconductor depletion means, a plurality of static wells in the semiconductor body aligned alongside each other and running parallel to, alongside and overlapping the gate so the charge carriers injected into the traveling potential wells at each of a plurality of instants in time will fill the plurality of static wells when the gate is dropped, the gate having length at least equal to the length of the plurality of static wells; whereby the pattern of charge collected at the plurality of static wells represents the waveform of the signal to be recorded.
The advantages of this invention include its small size and economy of cost. Its ease of replacability. The ability for triggering after the event to be recorded occurs. The potential for high band width (high frequency--high temporal resolution). A tolerence for low transfer efficiency at the gate controlling charge transfer. And, the ability for the tranfer time of charge carriers across a gate to be below one period of the SAW.
Brief Description of the Drawings
A more complete appreciation of the invention and many of the attendant advantages thereof will be readily obtained as the same becomes better understood by reference to the following detailed description when considered in connection with the accompanying drawings, wherein:
FIG. 1 is a schematic drawing of a SAW-CTD oscilloscope.
FIG. 2 is a schematic drawing of a basic SAW device that emphasizes the SAW propogation path.
FIG. 3 is a side view schematic drawing of a basic SAW device that emphasises the SAW charge transfer function.
FIG. 4 is a graph of the charge carrying capacity as a function of peak electric potential of SAW wells.
FIG. 5 is a graph of the charge carrying capacity as a function of SAW frequency, for various SAW potentials.
FIG. 6 is a graph of charge carrying capacity for low values of well potential.
FIG. 7 is overhead schematic representation of an SAW-CTD streak camera.
FIG. 8 is an overhead schematic representation of SAW-CTD continuouly operating recorder.
FIG. 9 is a representation of the filling of potential wells with charge carriers.
FIG. 10 is a graph comparing CCD and SAW devices as a function of processing time versus bandwidth.
Description of the Preferred Embodiments
Referring now to the drawings, wherein like reference numerals designate identical or corresponding parts throughout the several views, and more particularly to FIGS. 1, 2 and 3 thereof, there is schematically shown a surface acoustic wave--charge transfer device (SAW-CTD) oscilloscope 10. Charge carriers injected into silicon layer 12 of semiconductor body 14 by carrier injector 16 are transported into a charge transfer (CT) zone 18 by traveling SAW wells 20 produced by a SAW launcher 26. As each SAW well 20 passes injector 16 in succession, it picks up the charges generated in P-N junction 30 since the passage of its predecessor. The train of filled or partially filled wells in the CT zone 18 is then a spatial representation of the original charge-generating waveform at the injector 16. When a gate 22 located between CT zone 18 and static wells 24 is dropped, the charges in wells 20 conduct laterally along the troughs of the SAW wells 20 and into the static wells 24, where they will remain until read out. Organization of oscilloscope 10 is serial input to parallel output. Depending on what the static wells are made of, readout is possible in Charge-Coupled-Device (CCD) fashion. The output signal gives a history of the input waveform (whether light or electrical), which is the primary function of an oscilloscope.
More specifically, the structure of a Surface Acoustic Wave Charge-Transfer Device (SAW-CTD), shown in FIGS. 2 and 3 and discussed in greater detail below, is similar to a CCD structure with the difference that the electrodes and clock busses are now replaced by the surface acoustic waves. A surface acoustic wave travelling in a piezoelectric medium 27 creates electric potential wells 20 which travel with the acoustic wave and extend into the nearby semiconductor 14. Minority carriers in the semiconductor may be injected into these potential wells and transferred synchronously with the wave, i.e., with the same velocity as the wave.
FIG. 2 shows monolithic structure 13 consisting of an n+ doped contact layer 15 on the bottom surface of an n doped silicon wafer 12, a thermally grown SiO.sub.2 insulating layer 29, and a sputter-coated ZnO piezoelectric film 27. The perimeter of the device is covered with acoustic absorber 19 to attenuate reflections at the boundaries. The input interdigitated aluminum contacts 26 atop the ZnO transduce rf power from an input lead into SAWs which are launched on the surface and propagate in both directions away from the input transducer 26. An identical transducer 36 is used for output tests and diagnosis. A metal field plate 28, to which a potential is applied, is located on the ZnO 27. The field plate serves to deplete the Si of majority carriers which otherwise would recombine with the minority carriers being transported in the travelling SAW wells, and to increase the depth of the travelling potential wells in the silicon. The input and output diodes are shown in FIG. 3.
FIG. 3 shows a side view of the device of FIG. 2, but here also illustrating the diodes 30 and charge transfer process. The diodes 30 are p+ regions on the surface of the n- Si substrate 12, with an electrical connection 21 for bias and signal. Since ZnO and SiO.sub.2 are optically transparent, the diodes may be operated as photodiodes. The charges 25 are injected into the Si 12 at the input diode 30 and are swept up into a travelling SAW potential well which, at that moment, is at the diode. The well is one of a train of potential wells (illustrated to approximate a sine wave) travelling with the acoustic speed, and transports the captured charges 25 along with it. (The wavelength of the sine wave, in the actual case, would equal the periodicity of the fingers of one phase of the two-phase transducer, and, for best resolution, the diode size in the progagation direction would be no larger than the wavelength.) Wells appearing at the output diode 30 at different times contain differing amounts of charge, in accordance with the input waveform. Upon reaching the output diode 30, the charges are conducted out of the Si and along the output lead 23. The diodes 30 should be placed directly under the transducers 26,36 to obtain a delay time equal to the SAW input-output delay time; for devices used in actual applications, the diodes could both be on the same side of the input transducer.
Because of its simplicity and potential for high density and speed, the SAW-CTD concept shows promise for a variety of applications in signal procesing, memory and solid state imaging.
The first study of surface acoustic waves on a solid was done in 1885 by Lord Rayleigh, "On Waves Propagating Along the Plane Surface of an Elastic Solid," Proc. Lon. Math. Soc. 17, (1885-86) p.4, who investigated the case of waves at the surface of semi-infinite isotropic media. Stoneley, Geophys. Suppl. Monthly Notes Roy. Astron. Soc. 5, 1943 p. 343, extended the theory to anisotropic media and later other authors treated elastic waves in piezoelectric anisotropic semi-infinite media.
Layered media have proven to be of considerable interest with respect to piezoelectric surface acoustic waves because of the ability to derive benefits from the different materials involved. With the increasing sophistication of piezoelectric surface wave devices, the investigation of surface waves has been fully extended into the realm of anisotropic layered media. For example, various monolithic acoustoelectric and optoacoustic devices involving multilayered media have been recently developed. (Monolithic devices are structurally all one piece, as with a silicon substrate overlayered with a piezoelectric ZnO film, in distinction to "gap-coupled" or "separated media" devices in which the piezoelectric material is mechanically mounted very close to the semiconductor.) Piezoelectric overlayers are important in several applications such as filters, dispersive delay lines and guided wave structures.
Under certain crystal symmetry conditions or the assumption of weak piezoelectric coupling, the solutions describing the propagation of piezoelectric waves in a medium are decoupled into two independent modes: one set of solutions corresponding to transverse displacements called the "Love Modes," and a second set of solutions whose particle displacements are confined to the sagital plane and are referred to as "Rayleigh Modes." (The sagital plane is defined as the plane perpendicular to the surface of the structure and along the direction of propagation.)
The crystal symmetry of ZnO (hexagonal, class 6 mm crystal), a piezoelectric material 27, fulfills these conditions. The energy of the Rayleigh wave is concentrated at the surface of the medium; the particle motion and their associated potentials decay exponentially with depth, vanishing within a few SAW wavelengths of the free surface. When the wave equation solutions include exponentially dampened sinusoidal terms, the waves are called "generalized Rayleigh waves."
The most effective and commonly employed technique for the excitation of surface acoustic waves on piezoelectric media is by means of the metal interdigital transducer 26. This transducer consists of two sets of inter-leaved periodic metal electrodes deposited on the piezoelectric substrate 27 (see FIGS. 2 and 3). When an rf voltage waveform is applied across the transducer, fringing electric fields are set up between the metal fingers, inducing elastic stresses in the underlying substrate by virtue of the piezoelectric effect. These stresses propagate in the form of piezoelectric surface acoustic waves (Rayleigh waves) in both directions perpendicular to the transducer fingers. When the acoustic wave reaches an output transducer 36, as in FIG. 2, it is detected by means of the surface potential that accompanies the waves. That is, the input transducer 26 transforms electromagnetic energy to acoustic energy and the output transducer 36 converts it back to electromagnetic energy. The overall input-to-output efficiency with which these conversions occur is directly related to the electromechanical (or piezoelectric) coupling coefficient K.sup.2 of the material; insertion losses (power drop resulting from inserting the device into a circuit) as low as 12 db have been reported. See R. Wagers, G. Kino, P. Galle, and D. Winslow, "ZnO Acoustic Transducers Utilizing Gold Substrates," Ultasonics Symp. Proc. 1972, pp. 194-7.
The center to center distance d between the electrodes is equal to half the acoustic wavelength .lambda. and is related to the resonant angular frequency Wo of the transducer and the surface acoustic wave velocity Va by the expression: ##EQU1## Narrow electrodes imply higher frequencies. For example, a transducer with 10 micron finger width (and the same separation) corresponds to a wavelength of 40 microns and a frequency of 100 MHz if the surface wave velocity is assumed to be 4.0.times.10.sup.5 cm/sec.
Since the phase velocity of the acoustic waves is of the order of 10.sup.5 times slower than that of electromagnetic waves, the effect is to delay the signal by an amount equal to the acoustic transit time between the two transducers. For an acoustic wave with velocity 4.0.times.10.sup.5 cm/sec and transducer separation of 1 cm the corresponding delay would be 2.5 microsec. This SAW delay line constitutes the basic structure for many surface wave devices in signal processing and solid state imaging.
The SAW-CTD may be classified as an acoustoelectric device, since like other devices belonging to this family it is based on the principle of acoustoelectric interaction with carriers in a semiconductor. Such devices include the acoustoelectric amplifier, the acoustic convolver, and acoustoelectric imagers. However, the SAW-CTD device differs from the other acoustoelectric devices in two respects. First, it is a minority charge interaction device, and secondly it is a synchronous interaction device which means that the carriers are captured and transferred with a speed equal to that of the surface acoustic wave.
The principle for the operation of the SAW-CTD, in which charged particles are accelerated to the wave velocity in a solid, is analogous to that of the Travelling Wave Linear Accelerator (linac). See L. Brillouin, "Waves and Electrons Travelling Together--A Comparison Between Travelling Wave Tubes and Linear Accelerators," Physical Review, 74/1, July 1948, pp. 90-93, in which charged particles are accelerated and carried along on the crest of an electromagnetic wave travelling in free space, much like the surfer who rides on a single ocean wave. The simplest structure of the linac consists of the cylindrical waveguide, in which different modes of propagation can be excited. However, for charge transfer along the axis, a longitudinal electric field is required. Although a longitudinal travelling electric wave can be produced with the TM01 mode of the waveguide, such a device is not practical for particle acceleration in free space since the phase velocity Vp of the travelling wave always exceeds that which any particle can attain, no matter what its energy. As a consequence, the travelling wave pattern will rush past the particle alternately accelerating and decelerating it, without giving it a commulative energy gain. Nevertheless, by sharply corrugating the inside walls of the waveguides, inductance is added to the equivalent network of the guide and Vp is adjusted along the axis of the waveguide. A packet of charge may then be injected into the path of the travelling wave and be accelerated along the axis of the structure.
In the operation of the SAW-CTD, the linac waveguide is now replaced by a transfer channel 18 defined on the surface of a semiconductor 14. The force needed for the acceleration and transfer of the charge is provided by a travelling acoustoelectric potential generated by a piezoelectric surface acoustic wave. The acoustic wave has a velocity slow enough to capture thermal carriers with velocities opposite to the direction of propagation (in marked contrast to the requirements of a linac). When SAWs propagate in a piezoelectric layer, they are accompanied by electric fields which extend beyond the surfaces of the layer. By placing such a piezoelectric layer over a semiconductor it is possible to generate travelling potential wells at or below the semiconductor surface. In the monolithic devices discussed below, field energy normally occupying free space can be forced below the surface and into the bulk by means of a metal field plate 28 atop the piezoelectric layer, thus strengthening the well. The simplest technique for injecting the signal charge into the travelling wells is by direct electrical injection of a p-n junction 20: injected charge is proportional to the signal applied to the input diode. See N. A. Papanicolaou, "A Monolithic Surface Acoustic Wave Charge Transfer Device," Ph.D. Thesis, Electrical Engineering Department, University of Maryland (1979). Another method is photoelectric injection in the silicon: charge injected into a well is a function of the incident light flux, and the device can thus be used for image sensing purposes. See R. J. Schwartz, S. D. Gaalema, R. L. Gunshor, "A surface wave interaction charged coupled device," IEEE 1976 Ultrasonics Symposium, 1976, pp. 197-200. The charges injected by the input diode are stored in these propagating wells and transferred synchronously with the wave. The delayed signal is collected a SAW transit time later at the output diode. See FIGS. 2 and 3. The field electrode 28 over the propagation region 18, in addition to enhancing the acoustoelectric potential at the silicon surface, also provides the bias for the depletion of unwanted majority charge carriers from the semiconductor surface, which is required for minority charge transfer, and prevents lateral migration of the minority charges to outside the transfer region. The interdigitated output tranducer 36, though not needed for charge transfer operations, is useful for monitoring and evaluating the surface acoustic wave.
The distribution and quantity of charge carried by a SAW potential well is largely determined by the parameter A. ##EQU2## where Va=the acoustic speed of the SAW, 2.pi.f=the angular frequency of the SAW, .mu..sub.s =the surface mobility of the charge carriers in the semiconductor, and .phi..sub.o =the maximum amplitude at the Si-SiO.sub.2 interface of the electric potential .phi..sub.a associated with the acoustic wave. For A<1, charge packets can be captured in the wells, and the distribution of charges in the well is as depicted in FIG. 9b. FIG. 9 shows the process of filling SAW potential wells 20. Potential wells 20 are depicted as spatial dimension (horizontal axis) versus potential (vertical axis). The filling of SAW wells 20 with charge carriers is controlled largely by the parameter A (see Equation 1). (a) High frequencies provide high fields but low charge capacities, due to the decreased size of the well. (b) Low values of A permit a greater capacity of the well without spillage of charges into an adjacent well. In the figure, phase angles .theta. with an m subscript refer to maximum charge capacity conditions. The angular relationships are:
and
Further analysis yields the result that the maximum charge carrying capacity Q.sub.m of a well is given by: ##EQU3## where w=the width of the well (perpendicular to SAW propagation), and S=the elastance of the structure. Since the angles are determined by A, the value of A controls the quantity in braces, which is always less than unity. For low A, near zero, the charge occupies nearly the whole well, whereas for A near but less than unity the charges only occupy the highest field portion of the well. Increasing W or .phi..sub.o increases Q.sub.m, as does decreasing S. Through the value A, increasing f increases Q.sub.m (by making the electric field associated with the well stronger) and increasing Va decreases Q.sub.m (by spilling charges out of the well, e.g. when the charges are initially injected into and accelerated by the well). However, these effects of f and Va are in competition with their relationships to Qm in the leading quantity of Equation 4.
Values of Qm were computed and are given in FIGS. 4-6 for a range of parameters, under the conditions Va=4.4.times.10.sup.5 cm/s, w=20 microns, S=4.6.times.10.sup.17 cm.sup.2 /Farad and .mu..sub.s =750 cm.sup.2 /V-s. The general decrease of Qm for higher frequencies can clearly be identified.
Especially noteworthy is the prediction that the order of 10.sup.4 electrons may be transported in a single well of width w=20 micrometers at 1 GHz frequencies. Noise levels of under 100 electrons/pixel are achievable in solid state devices, and a signal as low as 50 electrons/pixel can render a discernible pattern. A 1 GHz device would require 1 micrometer wide fingers for the interdigitated transducer, which are technologically fesible. Higher frequencies may ulitmately be limited by acoustic diffraction, e.g. from thousand Angstrom wide grains of ZnO; other materials and device architectures may permit yet higher frequencies.
In discussing the application of solid state devices to imaging, it is convenient to set forth some of the terminology presently in use. A still camera (temporally integrating, with respect to the event of interest) is a multichannel recording device (one channel per resolvable picture element or "pixel"), while a temporally resolving camera is a type of multichannel waveform recorder. A temporally resolving camera with one dimension of spatial imaging is known as a streak camera, while two dimensions of spatial imaging with temporal resolution result in a framing camera. Framing cameras, like a movie camera, normally record the whole 2D image in integral frames, a given frame being exposed during the same interval over its entire area. Other cameras, like a scanning TV camera, record different times at different locations in the image, and may be referred to as image dissecting cameras, particularly if the event of interest is changing on a time scale comparable to or less than the time spread of one frame.
Solid state imaging devices may be built as multichannel arrays of sensing and recording elements (with a charge transfer region between the sensing elements and recording elements). While many solid state sensors have been developed, designs shall be described herein in terms of the basic structure of FIG. 2, with silicon diodes being the sensor pixels 24 or static wells 24. It is to be understood that the injector diode 16 itself may be the photon imaging sensor, or the injector diode 16 may serve as an electrical input with signal supplied by a detecting sensor located elsewhere. The output elements 24 may also be silicon diodes (if the output signal is time-varying), or a Metal/Oxide/Semiconductor (MOS) capaciter (if each channel of output represents a single value).
The simplest single channel SAW-CTD is shown in FIG. 3. A photodiode or electrical charge-injection diode 16 (as the figure is drawn) injects charge into the SAW train. The charge is transferred to the output diode, which collects the charge from the SAW wells and conducts it away on the output conductor. This device has the input and output in direct line along the direction of SAW propagation. The net effect is to delay the signal between the input and output diodes by the propagation time of the SAW. Multiple channels of this device could be placed side by side to record a 1D spatial image, but such a device would be functionally equivalent to a simple line of sensor diodes, except for the charge transfer and resultant delay.
The single pixel version of a SAW-CTD Series-to-Parallel Imager is depicted schematically in FIG. 1. The input diode 16 is located at the beginning of a CT zone 18 defined by a Field Plate 28 atop, and a transfer gate 22 overlapping and along the field plate 28. The gate 22 also overlaps the static wells 24 so when the gate 22 is dropped, charges 25 in the traveling wells 20 flow from the CT zone 18 into the static wells 24.
Charges produced in the input diode 16, whether by electrical or incident photon waveform, are swept along with the SAW emanating from the launcher 26. Each travelling SAW well carries one charge packet, giving a temporal resolution of one period of the SAW wave. The physical dimension of the charge injection port 30 of the input diode 16 must be somewhat smaller than the SAW wavelength to maintain this resolution. As the photon-producing event progresses, the history of the event is mapped out in the CT zone by the charges in the train of SAW wells. Thus far, this device is like the SAW delay line. However, the output for the present device is different: after the wave train is extended out onto the surface of the device, the transfer gate 22 is dropped for a time interval generally not to exceed one period of the SAW, during which time charges from the travelling SAW wells are captured by static wells 24 and stored. See FIG. 1. (For long wavelength, low frequency SAWs, raising the gate back up within one period may not be required, since a long period can permit a very high transfer efficiency, thus totally emptying the CT zone of charge carriers. If the CT zone is totally empty, there is no need to raise the gate.) Readout from the static wells can be accomplished by charge coupling (CCD) or by parallel processing, perhaps directly into a computer memory. The input waveform is temporally serial information, but spreading this information out onto a surface in dispersed fashion permits the storage of the information to proceed temporally in parallel into many static wells 24 simultaneously. To put it another way, a fast input signal can be read out with slow electronics, which is very desirable. This device, with an electrical waveform as input, can be made the basis of a SAW-CT oscilloscope or storage scope. The long signal delays inherent in SAW-CTDs permits gate triggers to arrive after the waveform to be recorded has been injected into the SAW wells. The relationship of static wells 24 to potential wells is 1 to 1 in the region of interest. There can be more than 1 static well 24 to each potential well 20 but may result in lower sensitivity with no gain in bandwidth. If fewer static wells 24 than potential wells are used, then a loss of bandwidth results. For a more complete discussion of the design of the static wells (and the injection p-n junction 30) see M. J. Howes and D. V. Morgan, eds., "Charge-Coupled Devices and Systems" Wiley-Interscience, N.Y., 1979. For such an oscilloscope, a convenenient output format could be achieved with available video monitors and related equipment, to produce a display having as a vertical axis the charge contained in a pixel, versus a horizontal axis signifying an ordinal numbering of the static wells.
The barriers and gates may be physically fabricated in different ways (see M. J. Howes and P. U. Morgan, eds., "Charge-coupled devices and Systems", Wiley-Interscience, N.Y., 1979), all having the common characteristic that they modify the lateral potential profile in the semiconductor to prevent unwanted migration of the charges. Fixed barriers or "channel stops" may be made by increasing the doping concentration at the location of the barrier. A variable barrier, or gate, takes the form of an insulating layer on the Si and a metal electrode over top, with a potential applied to the metal; the potential profile in the Si is architectually determined by the thickness of the insulator (which may be fabricated to different thicknesses in a predetermined areal pattern), and may be varied in level according to the voltage applied to the metal electrode.
A multipixel, series-to-parallel SAW-CTD of this variety becomes a streak camera. The history of the incoming waveform at each pixel is recorded in a respective line of static wells, as in FIG. 7. Trigger delays, even to the end of the photon-emitting event, are possible here also. The total recorded information may be displayed in video as intensity in a graph having axes of one spatial dimension versus time, as is the case for conventional streak camera images. Such a device, with electrical waveforms as inputs to the diodes, becomes a multichannel oscilloscope; this device brings in another significant advance to the state of the art. Fixed barriers should be placed between each pixel.
A variant of this device that may be operable as a continuously cycling recorder, having a larger input diode is shown in FIG. 8. The carriers are injected into the SAW wells in a wide CT zone which bifurcates into two branches. After a SAW well passes this bifurcation, the carriers in one branch cannot laterally move to the other branch. The result is two wave trains in two distinct CT zones containing the same waveform. A propagation delay between the two CT zones is achieved by requiring the wave in one zone to propagate farther before it is stored in static wells. The transfer gates 22 are timed to store successive time intervals of the original waveform in the static wells of the two branches, which may then be read out independently. Again, post-event gate triggering is feasible.
The SAW-CTD of FIG. 1 can also be emulated by a CCD in which a single input diode injects charge directly into a CCD charge transfer zone. The input waveform would be spatially spread out in the CCD-CT zone just as in the SAW-CTD; the CCD could comprise its own output wells (rather than having output wells to the said of the CT zone, as in the SAW-CTD). A multi-channel version of this CCD would, like the SAW-CTD, perform as a multichannel waveform recorder or as a streak camera.
The charge transfer zone in a silicon SAW-CTD will be, or can generally be made, photosensitive. Electron-hole pairs may be generated in the Si CT zone by photoelectric absorption of incident light. For most of the devices, the presence of a metal field plate 28 above the CT zone 18 is an effective block to optical photons. A polysilicon gate, on the other hand, would be transmissive. Shorter wavelength photons, e.g. x-rays, would not be blocked by a thin film field plate. Significant levels of photo-induced carriers in the CT zone (compared to those from the intended sensor) would be deleterious (except perhaps as a means to control carrier traps). The most effective way to reduce such photoinduced noise is to shield against it, as with the metal field plate 28.
The prospect of using SAW-CTDs for particle detection is not to be overlooked, whether detection is performed directly as drawn in the above figures or indirectly with detected electrical signals coming, as in FIG. 3, from particle detecting pixels located away from the SAW CT zone.
The low speed of surface acoustic waves is advantageous in many applications, while also imposing some limits on practicality. For instance, the device of FIG. 1 is constrained to operating above about 10 MHz, for which a 1 cm CT zone yields but 22 wavelengths (output wells) for Va=4.4.times.10.sup.5 cm/s. Thus, SAW devices find application at the higher frequencies, often significantly higher than typical CCD frequencies. This is illustrated in FIG. 10 for CCDs and for SAW devices not involving charge transfer. FIG. 10 is a comparision of CCDs and Surface Wave Devices (SWDs). The application of SWDs has been to frequencies as high as 1 GHz, and CCds to lower frequencies, with considerable overlap. T.sub.d is time delay, and T.sub.d w is a time-bandwidth product. Values are approximate, for actual devices other than SAW-CTDs, and do not represent theoretical limits. The figure summarizes devices in use, but it does not represent design upper limits or portray the fastest reported SAW-CTDs or CCDs, some reaching 1 GHz or more, which are under development for non-imaging applications. Higher SAW frequencies than those shown in the figure may be attained by taking the pains necessary to obtain submicron linewidths in transducer fingers, gates, injectors, barriers, etc. For a given linewidth, higher acoustic speeds result in higher frequencies. For faster waveform recording and imaging, the higher frequencies are essential.
One of the most fundamental limits on charge transfer and speed in SAW-CTDs is the dependence of well capacity on frequency and acoustoelectric potential. The frequency dependence is given in FIGS. 4-6 for a particular device structure, and may be summarized briefly as exhibiting a definite peak in charge capacity but falling only very slowly at high frequencies, with a nearly linear dependence on potential at the higher frequencies. The maximum achievable potential is set by numerous factors, including the high power limit and design of the SAW launcher (number of fingers, frequency, capacitance, shape, piezoelectric coupling coefficient, etc.), the materials constants of the constituents (carrier mobility, permittivity, piezoelectric constants, etc.), and the architecture of the device (boundary conditions, dimensions, etc.). The requirements to operate the device in depletion, rather than accumulation or inversion, sets a materials limit on .phi..sub.o ; notice however that inversion layers can be swept clear of charges by a travelling SAW train There are definite design tradeoffs to be evaluated in going to higher frequencies. For instance, at a given frequency, the potential has a maximum at a particular thickness of ZnO. A recent study of charge capacity in ZnO-Si SAW-CTDs predicts higher capacities to be achievable with the Sezawa mode of SAW propagation rather than with the lower order Rayleigh mode. Thus, a set of calculations such as given in FIGS. 4-6 may not represent the optimized attainable potential.
FIG. 4 is a graph of charge carring capacity as a function of peak electric potential of SAW wells, for a single 20 micron wide SAW well, under the conditions v.sub.a =4.4.times.10.sup.5 cm/s, w=20 microns, S=4.6.times.10.sup.17 cm.sup.2 /F, and .sub.s =750 cm.sup.2 /V-s for electron carriers (hole mobility is 300 cm.sup.2 /V-s). Charge is given in coulombs and in number of electrons, for various SAW frequencies.
FIG. 5 is a graph of charge carrying capacity as a function of SAW frequency, for various SAW potentials. Note the existence of a peak in capacity.
FIG. 6 is a graph of charge carrying capacity for low values of well potential. Even for low potentials, GHz frequencies, and narrow wells, significant numbers of electrons may be transported in a single well originating from a single pixel.
In addition to charge capacity, charge transfer efficiency (CTE) must also be considered. The transfer of charge in the present SAW-CTD devices occurs in several different operations: charge injection from an in-line diode 16 into the SAW wells, charge transport synchronously with the SAW velocity in the SAW wells (in the SAW propagation direction), charge extraction from the SAW wells by an in-line output diode 36, lateral transfer of charge along the trough of a SAW well through a transfer gate 22 and into a static well 24, and the readout of the static (CCD) wells 24. For each of these charge transfer operations there is a corresponding CTE: injection CTE, gate CTE, lateral CTE, synchronous CTE, extraction CTE, storage CTE, and CCD CTE. Some of these are discussed briefly below.
High speed charge transfer in certain types of CCDs (i.e., surface channel CCDs) is limited by inefficiencies introduced at the Si-SiO.sub.2 interface by fast-acting interface states. These states act as carrier traps which hold and later release charge. The effect of interface or surface states on SAW-CTDs has been experimentally and theoretically investigated. These traps may be dealt with in three ways: by keeping them filled (the "fat zero" in CCD jargon), by avoiding them (the CCD buried channel), or perhaps by operating at speeds faster than or much slower than the states.
Firstly, it would be easy to implement the fat zero into a SAW CT zone: a diode, placed in the CT zone 18 prior to the signal injecting diode, could bleed a steady state current into the SAW wells sufficient to keep most of the interface states filled. None of the wells would be left completely empty, and zero signal level would be represented by a finite but small quantity of charge called the fat zero. The fat zero approach, however, is not a complete solution to the problem. Since the deepest portion of the well fills first, nearly empty wells (the fat zero) occupy less of the well than wells full of signal charges; thus, the dwell time of the fat zero at the traps is less than the signal. Alternatively, the CT zone 18 could be photo-illuminated at a low irradiance. Secondly, buried channels in CCDs are a micrometer or more beneath the interface. For surface acoustic waves, the acoustic potential decays exponentially to nearly zero at a depth of a few wavelengths below the top surface of the ZnO, in the absence of a field plate 28. Nonetheless, a buried channel ZnO-Si SAW-CTD may be possible to construct, since the presence of the field plate 28 above the ZnO, as in FIGS. 2 and 3, alleviates this situation dramatically by increasing the potential at the SiO.sub.2 -Si interface and by decreasing the rate of decay with depth. In a GaAs SAW-CTD, on the other hand, a buried channel device has been demonstrated. And thirdly, as for outrunning the traps, this is done already with the very slow traps (release times of days). The crucial quantity in question is therefor the number density of traps located in the CT channel 18 with filling and releasing rates matched to the transit time of the signal charge past the trap. Since this number is energy dependent, analyzing the situation becomes more complicated. In addition, the traps may behave as scattering centers and thereby decrease mobility. Buried channels reduce the effectiveness of the traps by transferring the charges through a spatial minimum (or at least a reduced value) in the number distribution function of traps; there is also the possibility of transferring charges through a temporal minimum or energy minimum in the number distribution of traps as a function of trapping and releasing rate and cross section. SAW-CTDs may actually be of use in measuring these quantities, since charge capacity is dependent on mobility, energy varies with impressed potential, and temporal resolution is a function of trapping and releasing rates and cross sections. As a general rule, however, note that higher frequency SAWs have shorter wavelengths, shorter travel distances for a given number of pixels, and therefore intercept fewer surface traps.
Lateral charge transfer bears similarity to CCD charge transfer. Charge transfer through an open gate in CCDs proceeds by three mechanisms, self-induced fields, fringing fields from the gate electrodes, and thermal diffusion of the charge carriers. Self-induced fields, resulting from the electrostatic repulsion of charges in the wells, is the dominant impetus when the charge density in the well is high, e.g. when the well is more than about 10% occupied. For fewer charges in the well, fringing fields and thermal diffusion, a slow process, predominate. In CCD's very high transfer efficiencies are required. For an overall efficiency of 90%, a 100-stage three-phase CCD must pass a charge packet through hundreds of gates, each having a single transfer efficiency of 99.97%. This imposes a severe frequency limitation on CCDs. Nevertheless, GaAs CCDs have reportedly achieved above 1 GHz operation.
Although the SAW-CTDs do have a fast transfer gate 22, which is analogous to a CCD gate, there is only one such gate for the devices discussed here. Charge carriers remaining in the CT zone after the transfer gate is raised are discarded and do not become a source of noise. Thus, considerable transfer inefficiency can be tolerated without smearing charges between adjacent wells. Therefore, higher gate transfer frequencies should be attainable by SAW-CTDs.
The device architectures disclosed herein may also be operated with an amplitude modulated rf signal source as input to the SAW input transducer, and with the carrier injector supplying a quantity of charge sufficient to fill each travelling potential well. The signal input is then to the rf transducer, in distinction to the above operation of the device wherein the rf amplitude is constant. Waveform recording may be carried out in this fashion, although the signal bandwidth will not be sufficient to resolve a zero to 100% contrast within one period of the SAW.
Furthermore, the same concept described herein applies to gap-coupled devices or monolithic GaAs, where the GaAs is used in place of the monolithic structure 13.
Finally, it should be noted that it is obvious that electrical leads and connections run to and from the various elements of the invention but have been omitted.
Obviously, numerous (additional) modifications and variations of the present invention are possible in light of the above teachings. It is therefore to be understood that within the scope of the appended claims, the invention may be practiced otherwise than as specifically described herein.
Claims
What is claimed as new and desired to be secured by Letters Patent of the United States is:
1. A waveform recorder comprising: a semiconductor body; a planar piezoelectric body having a main surface overlying and in proximity to the semiconductor body; wave propogation means for propogating acoustic waves on the main surface of the piezo-electric body to create traveling potential wells in the underlying semiconductor body; a traveling potential well path located in the semiconductor body, the traveling potential well path beginning at the wave propagation means and extending straight away therefrom; charge injection means for injecting into the potential wells travelling past a first location in the semiconductor body a quantity of minority charge carriers whose magnitude tracks with time the waveform of a signal to be recorded, the first location of the charge injection means being located near the start of and in the travelling potential well path; semiconductor depletion means for depleting the semiconductor of majority charge carriers along the traveling potential well path, said depletion means located atop the piezoelectric body; a gate located on the semiconductor body and alongside and parallel to the traveling potential well path and overlapping the semiconductor depletion means; a plurality of static wells in the semiconductor body aligned next to each other and running parallel to, alongside and overlapping the gate so charge carriers injected into the traveling potential wells at each of a plurality of instants in time will proceed into their respective static wells of the plurality of static wells when the potential across the gate is lowered, the gate having length at least equal to the length of the plurality of static wells.
2. A waveform recorder as described in claim 1 wherein the semiconductor body is comprised of a planar insulating body upon which the piezoelectric body is overlying and in proximity therewith; an N doped silicon wafer underneath the planar insulating body; and an N+ doped contact layer underneath the N doped silicon; and wherein the piezoelectric body is a sputter-coated ZnO piezoelectric film.
3. A waveform recorder as described in claim 1 wherein the wave propagation means is a metal interdigital transducer.
4. A waveform recorder as described in claim 1 wherein the charge injection means is a P-N junction.
5. A waveform recorder as described in claim 1 wherein the semiconductor depletion means is a conductive plate.
6. A waveform recorder as described in claim 1 wherein the static wells are MOS capacitors.
7. A waveform recorder as described in claim 1 wherein the gate is made of metal.
8. A waveform recorder comprising: a semiconductor body having a planar insulating body, an N doped silicon wafer underneath the planar insulating body, and an N+ doped contact layer underneath the N doped silicon; a planar piezoelectric body made of sputter coated ZnO having a main surface overlying and in proximity to the planar insulating body of the semiconductor body; a metal interdigital transducer for propagating acoustic waves on the main surface of the piezo-electric body to create traveling potential wells in the underlying semiconductor body; a traveling potential well path located in the semiconductor body the traveling potential well path beginning at the wave propagation means and extending straight away therefrom; a P-N junction for injecting into the potential wells travelling past at least one first location in the semiconductor body a quantity of minority charge carriers whose magnitude tracks with time the waveform of a signal to be recorded, the first location of the charge injection means being located near the start of and in the travelling potential well path; a metal plate for depleting the semiconductor along the traveling potential well path; a gate made of metal located on the semiconductor body and alongside and parallel to the traveling potential well path and adjoining the metal plate; a plurality of MOS capacitors in the semiconductor body aligned next to each other and running parallel to, alongside and overlapping the gate so the charge carriers injected into the traveling potential wells at each of a plurality of instants in time will proceed into their respective MOS capacitors of the plurality of MOS capacitors when the gate is dropped, the gate has length at least equal to the length of the plurality of MOS capacitors.
9. A split-path waveform recorder comprising: a semiconductor body; a planar piezoelectric body having a main surface overlying and in proximity to the semiconductor body; wave propogation means for propogating acoustic waves on the main surface of the piezo-electric body to create traveling potential wells in the underlying semiconductor body; a traveling potential well path located in the semiconductor body, the traveling potential well path beginning at the wave propagation means and extending straight away therefrom; charge injection means for injecting into the potential wells travelling past a first location in the semiconductor body a quantity of minority charge carriers whose magnitude tracks with time the waveform of a signal to be recorded, the first location of the charge injection means being located near the start of and in the travelling potential well path; semiconductor depletion means for depleting the semiconductor of majority charge carriers along the traveling potential well path, said depletion means located atop the piezoelectric body; a first gate located on the semiconductor body along one side and parallel to the traveling potential well path and adjoining the semiconductor depletion means; a second gate located on the semiconductor body and on the opposite side of the traveling potential well path where the first gate is located, said second gate being parallel to the traveling potential well path and to the first gate, and overlapping the semiconductor depletion means; a first plurality of static wells in the semiconductor body aligned next to each other and running parallel to, alongside and overlapping the first gate so charge carriers injected into the traveling potential wells at each of a plurality of instants in time will proceed into their respective static wells of the plurality of static wells when the first gate is dropped, the first gate having length at least equal to the total length of the first plurality of static wells, the first plurality of static wells beginning near the start of the traveling potential well path; a second plurality of static wells in the semiconductor body aligned next to each other and running parallel to, alongside and overlapping the second gate so the charge carriers injected into the traveling potential wells at each of a plurality of static wells will proceed into their respective static wells of the plurality of static wells when the second gate is dropped, the second plurality of static wells beginning after the first plurality of static wells has ended, the second gate having length at least equal to the length of the second plurality of static wells; and a fixed potential barrier located in the center of the traveling potential well path and parallel to said first and second gates for causing the traveling potential wells to be split in half, one half of which can fill the first plurality but not the second plurality of static wells, the other half filling the second plurality but not one first plurality of static wells, the length of the fixed potential barrier being at least as long as the total length of the first and second plurality of static wells.
10. A waveform recorder as described in claim 9 wherein the semiconductor body is comprised of a planar insulating body upon which the piezoelectric body is overlying and in proximity therewith; an N doped silicon wafer underneath the planar insulating body; and an N+ doped contact layer underneath the N doped silicon; the planar piezoelectric body is made of sputter coated ZnO; the wave propogation means a metal interdigital transducer; the charge injection means is a first P-N junction; the semiconductor depletion means is a metal plate; the first gate is made of metal; the second gate is made of metal; and the first and second plurality of static wells are made of MOS capacitors.
11. A multichannel waveform recorder comprising: a semiconductor body; a planar piezoelectric body having a main surface overlying and in proximity to the semiconductor body; wave propogation means for propogating acoustic wave signals on the main surface of the piezo-electric body to create traveling potential wells in the underlying semiconductor body; a plurality of traveling potential well paths located in the semiconductor body, the traveling potential well paths being in parallel with each other and being of the same length, the traveling potential well paths beginning at the wave propagation means and extending straight away therefrom; a plurality of charge injection means each for injecting into the respective potential wells traveling past a plurality of first locations in the semiconductor body a quantity of minority charge carriers whose magnitude tracks with time the waveform of a signal to be recorded, one of the plurality of charge injection means and one of the plurality of first locations being located each in a respective one of the traveling potential well paths; a plurality of semiconductor depletion means for depleting the semiconductor along the respective traveling potential well paths, each depletion means located above a respective traveling potential well path; a plurality of fixed potential barriers separating adjacent traveling potential well paths, which prevent the migration of injected change carriers between traveling wells of adjacent pairs of traveling well paths; a plurality of gates each of which are located on the semiconductor and alongside and parallel to a respective traveling potential well path and overlapping the respective semiconductor depletion means, each gate having length at least equal to the respective distance of the plurality of static wells alongside the respective traveling potential well path; pluralities of static wells in the semiconductor body, the static wells making up each plurality aligned alongside each other and running parallel to and alongside each respective gate so the charge carriers injected into the respective traveling potential wells at each of a plurality of instants in time will fill the respective plurality of static wells when the gates are dropped.
12. A waveform recorder as described in claim 11 wherein: the semiconductor body is comprised of a planar insulating body upon which the piezoelectric body is ovelying and in poximity therewith; an N doped silicon wafer underneath the plasma insulating body; and an N+ doped contact layer underneath the N doped silicon; the planar piezoelectric body is made of sputter coated ZnO; the wave propogation means is a metal interdigital transducer; the charge injection means are P-N junction; the semiconductor depletion means are metal plates; the gates are made of metal; and the static wells are MOS capacitors.
13. A method for recording electrical waveforms comprising the steps of: creating SAW traveling potential wells in a semiconductor depleting majority charge carriers from a predetermined zone in the semiconductor in order to create a charge transfer zone, the SAW traveling potential wells traveling through the charge transfer zone injecting minority charge carriers into the SAW traveling potential wells at a plurality of instants in time, with the amount of minority charge carriers injected into the SAW traveling potential wells being proportional to the amplitude of the waveform being recorded; filling a plurity of static wells that are alongside the charge transfer zone with minority charge carriers that have been injected into the SAW traveling potential wells by changing the potential across a gate that overlaps and links the charge transfer zone and the static wells thusallowing the charges in the SAW traveling potential wells to flow into the static wells; changing back the potential of the gate to prevent any further flow of charges from the SAW traveling potential ells to the static wells before the SAW traveling potential well has time to travel to the next static well; and reading out the charge in the static wells.
Patent Citations (4)
Non-Patent Literature (1)
- Whitlock et al. "Streak and Framing Camera Designs Using Surface Acoustic ve Charge-Transfer Devices" Proceedings of the 16th Int'l. Congress on High Speed Photography and Photonics, Strasbourg, France, Aug. 27-31, 1984.