Single-electrode vertical gate multi-framing image pickup element
The branching image sensor design with a single vertical electrode and linearized potential distribution enhances the number of consecutive images captured at ultra-high speeds, addressing limitations in existing technologies.
Patent Information
- Application Number
- JP2023076972
- Authority / Receiving Office
- JP · JP
- Patent Type
- Applications
- Current Assignee / Owner
- Filing Date
- 2023-05-08
- Publication Date
- 2024-11-20
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
Branching image sensors are limited by the number of continuous images they can capture due to the number of transfer means and storage means, and there are challenges in increasing the speed and efficiency of signal transfer.
A branching image sensor design with one photoelectric conversion means and three or more transfer means arranged radially, utilizing a single vertical electrode and a repulsive force against electric charge, with a linearized potential distribution along the transfer path, and incorporating a photoelectric conversion layer made of germanium and storage means made of silicon.
The design allows for capturing a large number of consecutive images at ultra-high speeds with a temporal resolution of about 13 ps, significantly improving upon existing multi-framing cameras.
Smart Images

Figure 2024161860000001_ABST
Abstract
Description
[Technical field]
[0001] This patent relates to a solid-state imaging device that captures multiple images continuously with a time resolution of less than a nanosecond. [Background technology]
[0002] Branching Image Sensor
[0003] An ultra-high speed multi-framing image sensor is disclosed in Patent Document 1, Patent Document 2, and Non-Patent Document 1. Figure 1 shows the structure and wiring of the circuit diffusion layer of the pixel of this sensor. In this structure, at the center of the pixel, there are provided a photoelectric conversion means 100, eight transfer means 184 having two functions of collecting signal charges generated by incident light and sequentially transferring image signals in the radial direction, a storage means 102 that temporarily stores eight image signals within the pixel, and a readout means 103 that reads out the stored image signals outside the pixel after shooting.
[0004] FIG. 2 shows a model of the pixel of FIG. 1 so that the basic function of the pixel of FIG. 1 can be clearly seen.
[0005] In order to reduce crosstalk caused by the transfer of image signals in eight directions for the pixel in FIG. 1, FIG. 2 shows a charge collection means 179 placed at the center of the pixel, and the charge collection function and the image signal transfer function by transfer means 101 separated from each other.
[0006] Figure 3a shows a conceptual diagram of a cross section of the pixel in Figure 2. In the following cross-sectional views, it is assumed that light is incident from above. A photodiode (hereinafter, abbreviated as "PD") 100, which serves as photoelectric conversion means, sits on a circuit diffusion layer 178. However, the lower part of the PD and the charge collection means 179 in the center of the circuit diffusion layer are areas that share the two functions of generating and collecting signal charges. Hereafter, an image sensor model equipped with a pixel of this structure will be referred to as a "two-story type."
[0007] Figure 3b shows an inverted two-story type. In other words, the photoelectric conversion means is placed at the bottom and the circuit diffusion layer is placed at the top, which is the light receiving surface side. Conventional image sensors are of the inverted two-story type.
[0008] Recent image sensors use junction-type image sensors in which a junction chip for high performance is bonded underneath the sensor chip.
[0009] The joining chip part is not shown in Fig. 3, but an example is shown in Fig. 20. In this case, as shown in Fig. 3a, it is easier to electrically join the joining chip by forming a circuit diffusion layer on the bottom.
[0010] On the junction chip, there are arranged the subsequent stages of the readout circuit, such as the source follower circuit 136 in FIG. 1, pixel level digitization circuits shown in Patent Documents 3 and 4, additional analog memories, digital memories, and the like.
[0011] In particular, in the case of an ultra-high speed image sensor, which is the object of the present invention, it is desirable to place the driver circuit close to each pixel. Therefore, the present invention assumes a junction type image sensor having a sensor chip with the structure shown in Figure 3a, to which a junction chip carrying a driver circuit is bonded.
[0012] In this case, multiple functions are distributed between the sensor chip and the bonding chip, making the structure somewhat complicated.
[0013] When prototyping test sensors to check basic functions, it may be better to use an inverted two-story type, which can be realized by processing from the light-receiving surface side and does not require bonding.
[0014] Therefore, in preliminary studies using simulations and prototypes, the inverted two-story type shown in Figure 3b or Figure 3c is used as appropriate.
[0015] The structure shown in Figure 2 makes it possible to record eight consecutive image signals at extremely short intervals, which is necessary for transferring the image signals from the photoelectric conversion means to the storage means in the pixels. Also, since the readout can be performed at a sufficiently slow speed, it is possible to reduce the readout noise and obtain a recognizable image even in the case of insufficient incident light due to ultra-high speed shooting.
[0016] In Fig. 2, the image signal is branched in eight directions from the charge collecting means for transfer. This may be the first stage of transfer, and the second stage may be branched further in two directions to capture 16 images continuously.
[0017] Therefore, in Non-Patent Document 2, an image sensor having such a pixel structure is called a "branching image sensor."
[0018] Non-Patent Document 3 discloses a circuit diffusion layer 104 for one pixel of a solid-state imaging device shown in FIG.
[0019] Figure 5 shows a conceptual diagram of the cross section of each pixel in Figure 4. Hereinafter, this branching image sensor structure will be referred to as the "single-storey type."
[0020] The single-story type is characterized in that it has a photoelectric conversion means 105 and transfer means 106 extending substantially radially from the photoelectric conversion means in the same horizontal plane as the circuit diffusion layer, and each transfer means 106 is made of a vertical gate having two vertical electrode pairs 108 extending substantially radially from the center of the pixel in parallel with the cross section.
[0021] Between the pair of vertical electrodes is placed a horizontal transfer path 109 through which signal charges pass in the horizontal direction. In addition, a readout means 110 is provided for transferring image signals from a storage means 107 that temporarily stores the image signals to outside the pixel or onto the junction chip.
[0022] As shown in Figures 5a to 5c, the depth at which the image signal storage means and readout means are placed at the bottom (Figure 5a), center (Figure 5b), or light-receiving surface side (Figure 5c) of the circuit diffusion layer varies depending on the purpose.
[0023] When the storage means and readout means are on the light receiving surface side (Fig. 5c), most of the processes can be done from the light receiving surface side, so the process is relatively easy. When they are on the bottom (Fig. 5a), it is suitable for making the latter stage of the readout means on the bonding chip. When they are on the middle (Fig. 5b), it is suitable for speeding up the signal transfer after the signal charge passes through the transfer means, such as when transferring image signals from the storage means or FD to the bonding chip.
[0024] On the other hand, in FIG. 5d, the storage means are distributed in the form of elongated columns substantially throughout the entire depth of the circuit diffusion layer.
[0025] In Fig. 5d, the storage means also doubles as a floating diffusion (FD). In Fig. 5d, the FD is distributed throughout the depth of the circuit diffusion layer. Since the capacity of the FD is large, it is not suitable for detecting a small number of electrons, but it contributes to improving the imaging speed.
[0026] 3a, 3b, 5a to 5c are typical examples, and there are intermediate or modified types. For example, FIG. 3c shows an inverted two-story type having a vertical transfer means 106.
[0027] In the following, the signal charge will be electrons, and the holes will be transferred in the opposite direction to the electrons and continuously removed from the pixel and further from the sensor. Conversely, holes can be considered as signal charge. However, in this case, in the following explanation of the semiconductor structure, n-type and p-type will essentially be swapped and the polarity of the voltage will be inverted.
[0028] In the branching image sensor described in Figures 1 to 5, the number of transfer means is K (K ≥ 3). A higher voltage VH than the other electrodes is applied to the electrode of one of the transfer means (hereinafter referred to as the "transfer electrode"), and a lower voltage VL is applied to the other transfer electrodes. VH is applied to the other transfer electrodes in sequence at a time interval of the imaging time resolution.
[0029] Figure 6 shows an example of a simulation of the potential (electric potential) for the flat structure shown in Figure 4 and the flat structure shown in Figure 5d. The pixel size is 1 μm, and the vertical electrodes are arranged to a depth of 2.5 μm from the light receiving surface. The FD is also a bar-like shape that is distributed vertically from 0 μm to 2.5 μm. In addition, VL=0V, VH=3.3V.
[0030] In the two-story type of FIG. 1 to FIG. 3, signal electrons generated substantially instantaneously across the depth direction by light incident on the photoelectric conversion means 100 are transferred vertically within the photoelectric conversion means and then transferred to the adjacent storage means by the transfer means to which VH is applied from the central area of the circuit diffusion layer 178.
[0031] By sequentially adding VH to K transfer paths, K image signals are stored.
[0032] Whether it is a two-story or single-story type, once an electron has been transferred outside the kth transfer means with VH applied to its electrode, the voltage of this electrode immediately becomes VL and the electrode voltage of another transfer means changes to VH, so the electron cannot return from the kth transfer means to the photoelectric conversion means.
[0033] Therefore, the minimum time resolution of an image sensor having the two-level pixels of FIG. 3 is substantially defined by the time Δtmv required for a signal electron to move between the upper and lower ends of the photoelectric conversion means.
[0034] In the flat-story pixel of FIG. 5, the minimum time resolution is substantially defined by the time Δtmh required for a signal electron to pass through the photoelectric conversion means from one end to the opposite end in the horizontal direction.
[0035] Both Δtmv and Δtmh are defined in terms of (distance / velocity) in the direction of electron transport.
[0036] For the purpose of ultra-high speed imaging, it is desirable that the speed in the direction of electron transfer be as high as possible.
[0037] Figure 7a shows the relationship 111 between the electric field E and drift velocity V for intrinsic silicon. Figure 7a also shows the relationship 112 between the electric field and drift velocity for intrinsic germanium. As shown in the figure, 95% of the saturated drift velocity is defined as the saturated drift velocity Vs. The electric field relative to Vs is also defined as the saturated electric field Es.
[0038] Hereinafter, in contexts where it is obvious that the material is an intrinsic semiconductor, the word "intrinsic" will be omitted.
[0039] Figure 7b shows the diffusion coefficient D versus electric field for silicon and germanium. The diffusion coefficient is larger as the electric field is smaller.
[0040] Figure 8 shows the absorption coefficient (the reciprocal of the average penetration depth) 113 of silicon versus the wavelength of incident light. The absorption coefficient 114 of germanium is also shown.
[0041] To achieve the highest imaging speed, i.e., the shortest time resolution, it is necessary to make the electric field as close as possible to Es, minimizing the distance the electrons travel. If the shortest distance possible for a certain structure is Ls, then the shortest time resolution is represented by the value Ls / Vs.
[0042] At this time, the potential difference across the photoelectric conversion means is substantially Es×Ls.
[0043] In the two-story type, the minimum distance Ls is basically determined by the average vertical penetration depth Lsv of light.
[0044] In the single-storey type, the transfer distance of the signal electrons is represented by the time it takes for the signal electrons to move horizontally within the photoelectric conversion means. The corresponding minimum distance, i.e., the minimum aperture of the photoelectric conversion means, is basically determined by the diffraction limit Lsh of light.
[0045] In the silicon photoelectric conversion means, Vs is 0.091 μm / ps from Figure 7a. From Figure 8, Lsv is 1.74 μm for green light with a wavelength of 550 nm, which is often adopted as the standard wavelength of visible light.
[0046] Therefore, the minimum time resolution for the second-order type is on the order of Lsv / Vs = (1.74 / 0.091) = 19.1 ps.
[0047] However, as shown in Non-Patent Document 4, in a rigorous analysis that uses the fact that the intensity of incident light is exponentially distributed in the depth direction within the photoelectric conversion means and the condition that the concave portion of the apex of the distribution obtained by adding the time distributions of the two signal charges becomes flat and it is theoretically impossible to separate the two distributions, etc., the time resolution for the second-order type is 11.1 ps.
[0048] In the following analysis, we will compare orders of magnitude, so we will use Lsv / Vs, an easy-to-understand index, as a representative value of the time resolution for the second-order type.
[0049] Next, consider the case of a one-story type. The diffraction limit of 550 nm visible light is about 0.25 μm. In this case, the diffraction limit is the representative spatial size Lsh. Therefore, the minimum time resolution for the one-story type is on the order of Lsh / Vs = (0.25 / 0.091) = 2.75 ps, which is equivalent to 1 / 7 of the 19.1 ps for the two-story type.
[0050] As a result, the single-storey type can theoretically capture images at a speed about one order of magnitude faster than silicon photoelectric conversion means.
[0051] A layer of crystal defects, typically dangling bonds, is formed at the interface between the silicon crystal and the silicon oxide insulating layer, which can cause signal electron traps and dark currents.
[0052] To prevent this, image sensors typically first perform hydrogen annealing to terminate the dangling bonds.
[0053] In addition, by applying a negative charge to the interface, the probability of repulsion of negatively charged electrons can be increased. To achieve this, a technique has been used in which boron is injected near the interface and diffused.
[0054] Recently, a technique has been used in which a thin layer of aluminum oxide or hafnium oxide having a fixed charge is formed parallel to the insulating layer.
[0055] Germanium is often used as a photoelectric conversion means in the near-infrared region. Also, as shown in Figure 8, its absorption coefficient in the visible light region is several tens of times that of silicon. Therefore, when using germanium to capture visible light, a photoelectric conversion means thickness of 100 nm is sufficient.
[0056] On the other hand, as shown in FIG. 7, in germanium, a drift velocity comparable to that of silicon can be achieved with an electric field that is a fraction of that of silicon. [Prior art documents] [Patent documents]
[0057] [Patent Document 1] Patent No. 6188679, Solid-state imaging device, Patent application 2014-502360 [Patent Document 2] US9503663B2, SOLID-STATE IMAGING APPARATUS, Filed on August 25, 2014, Date of Patent, November 22, 2016. [Patent Document 3] Patent application 2016-63530, High-sensitivity, high-speed solid-state imaging element and device [Patent Document 4] Patent application 2020-155754, High-speed imaging means [Non-patent literature]
[0058] [Non-Patent Document 1] TG Etoh, et al., Toward One Giga Frames per Second-Evolution of in Situ Storage Image sensors, Sensors, vol.13, No.4, Sensors, 4640-4658, 2013. [Non-Patent Document 2] HN Nguyan, et al., A Pixel Design of a Branching Ultra-Highspeed Image Sensor, Sensors, vol.21, No.7, 2506, 2021. [Non-Patent Document 3] Hideki Muto, ITE Technical Report, vol.46, No.14, IST2022-14, 17-20, 2022. [Non-Patent Document 4] TG Etoh, et al., The Theoretical Highest Frame Rate of Silicon Image Sensors, Sensors, vol.17, No.3, 483, 2017. Summary of the Invention [Problem to be solved by the invention]
[0059] The biggest problem with branching image sensors is that the number of shots that can be taken is the same as the number of transfer means or storage means, and the number of continuous shots that can be taken is low.
[0060] The greatest advantage of a branching image sensor is that it can capture successive images at very short time intervals.
[0061] By making the transfer means of the branching image sensor a single-storey type using a vertical gate as proposed in Non-Patent Document 3, it is possible, in principle, to reduce the time resolution to a fraction of that of the two-storey branching image sensor using the planar gate used in Patent Documents 1 to 4. The present invention aims to take advantage of this effect while increasing the number of images that can be captured, which is an issue with branching image sensors.
[0062] Another challenge is to introduce new technologies to further increase speed.
[0063] First, factors limiting the number of shots that can be taken will be described.
[0064] In the case of a branching image sensor structure having a transfer means consisting of two electrodes as shown in Figure 4 disclosed in Non-Patent Document 3, in order to sequentially transfer signal electrons in the K (K = 16 in Figure 4) direction from the central photoelectric conversion means 105, 2K vertical transfer electrodes 108 are required on both sides of the transfer path 109 of each transfer means 106.
[0065] Furthermore, when the electrode voltage of the kth (here, k≦K−1) transfer means is VH, the electrode voltages of the (k+1)th and (k−1)th transfer means are VL.
[0066] At this time, a high electric field and capacitance are generated between one of the two electrodes of the kth transfer means and the electrode of the adjacent transfer means, so that, in principle, the insulating layer 115 between these electrodes needs to have a thickness necessary to reduce the electric field and RC delay.
[0067] The actual fabrication of this insulating layer poses a major process challenge. First, it is necessary to create slits for the two vertical electrodes. It is difficult to fabricate deep parallel slits close together, and the slit spacing required for this is much larger than the insulating layer thickness required to mitigate the electric field and RC delay.
[0068] Because the number of electrodes is twice K and K wide insulating layers are required, when the total space required for these is subtracted for the same perimeter of the photoelectric conversion means, the space that can be allocated to the transfer path 109 that actually transfers the signal charges becomes 2 / 3 to 1 / 2, which significantly limits the number of transfer means that can be incorporated for the same size of photoelectric conversion means, i.e., the number of continuous images that can be captured.
[0069] If the shooting speed is not an issue, additional memory can be created on the joined chip. However, a new invention is required to continuously transfer image signals to the joined chip while maintaining a certain shooting speed.
[0070] The number of shots of a branching image sensor is essentially the number of transfer means. In this case, the time resolution is the time it takes for the signal electrons to reach the transfer means. In this case, the drive speed of the storage means and FD after the signal electrons pass through the transfer means is not related to the shooting speed.
[0071] However, when sending an image signal from the sensor chip to an additional image signal storage means on the joint chip, the image signal can be sent from the image signal storage means on the sensor chip or the FD as quickly as possible to minimize the decrease in shooting speed and increase the number of shots.
[0072] Next, we will explain a technology that can further increase the speed in addition to the speed achieved by the vertical transfer means. The travel time of electrons also depends on the shape of the potential between the starting point and the end point. The drift speed of electrons on a semiconductor is proportional to the electric field until it approaches the saturated drift speed. In this case, the arrival time of electrons moving between two points with different potentials can be minimized by maximizing the minimum electric field. To do this, the electric field between the two points is made constant. In other words, the potential is made as linear as possible.
[0073] The diffusion coefficient is larger for smaller electric fields. Whether a linear gradient strictly gives the minimum time even with this effect has not been examined. However, experience based on simulations shows that the principle of shortening the transfer time by linearizing the electric field is essentially correct when the starting and ending potentials are given.
[0074] Furthermore, if the thickness of the photoelectric conversion means and the circuit diffusion layer can be reduced by changing the photoelectric conversion material, etc., it will be easier to form the slits required for the vertical transfer means, and further speedup will be possible by reducing the transfer time. [Means for solving the problem]
[0075] A branching image sensor in which each pixel has one photoelectric conversion means and three or more transfer means arranged around it in a substantially radial direction from the photoelectric conversion means, each of the transfer means having one vertical electrode.
[0076] Also provided is a means for generating a repulsive force against the charges around the vertical electrodes.
[0077] In addition, when the number of vertical transfer electrodes is K (K≧3), when substantially the same voltage is applied to K−1 of the vertical transfer electrodes and a different voltage is applied to one of the vertical transfer electrodes, a means is provided for making the potential along the transfer path of the signal electrons of the photoelectric conversion means closer to a straight line.
[0078] The means for linearizing the potential along the transfer path is a layer that is located substantially at the center of the photoelectric conversion means in a plan view and is formed by injecting type 3 or type 5 atoms into a region that is substantially narrower than the area of the photoelectric conversion means.
[0079] Furthermore, the photoelectric conversion means and the transfer means are substantially in the same plane.
[0080] In addition, in terms of the circuit configuration, the charge storage means, which is located farther away from the photoelectric conversion means than the transfer means, or the means for converting the number of stored charges into a voltage, is substantially on the same plane as the photoelectric conversion means and the transfer means.
[0081] The photoelectric conversion layer is germanium and the storage means is silicon.
[0082] The photoelectric conversion means is InGaAs and the storage means is InP. Effect of the Invention
[0083] Simulation results for a simplified model are used to demonstrate the effectiveness of the measures to solve the above problems.
[0084] A major issue with branching image sensors is the low number of continuous shots they can take, but by using only one vertical transfer electrode, the number of shots can be significantly increased. However, it is necessary to check whether there are any issues associated with transferring signal charges using a single vertical electrode. In the simulation, in addition to the effects, we also consider whether these issues exist.
[0085] Furthermore, if further speed increases are possible under those conditions, the effect will be even greater.
[0086] Figure 9a shows the planar shape of a pixel model for explaining the effect of the present invention. The cross-sectional shape is the inverted two-story type shown in Figure 3c. The feature of this example is that each transfer means has one vertical transfer electrode, and the transfer paths for signal charges are on both sides of the electrode.
[0087] In the pixel model shown in FIG. 9, circuit elements and structures that are not directly related to the effect of the present invention are omitted as much as possible for simplification.
[0088] For example, in the model of Fig. 9, the FD also serves as the storage means. Also, the source follower circuit 132, which is the latter half of the readout means shown in Fig. 1, can be fabricated on the junction chip, and is therefore not shown in Fig. 9.
[0089] However, when using a junction-type image sensor, the sensor chip structure will be the upside-down structure of FIG. 3c.
[0090] Each pixel in FIG. 9a includes a square photoelectric conversion means 116, which is a branching image sensor shape with the minimum number of transfer means, four vertical transfer means 117, 123, four FDs 118 connected to each transfer means, and a p-layer 119 for pixel isolation.
[0091] Each transfer means includes a single vertical transfer electrode 120, 121. Signal electrons reach the FD via transfer paths 122 on both sides of each electrode.
[0092] FIG. 9b is a further simplification of FIG. 9a, leaving only the circuit elements necessary to explain the effect of the present invention.
[0093] In Fig. 9b, the generation of relatively low potentials occurring in three vertical transfer means 123 to which VL is applied among the four vertical transfer means in Fig. 9a is replaced by the generation of potentials by p-type implantation 119 for pixel separation. There is no substantial difference in the potential distributions in the photoelectric conversion means for the structures in Fig. 9a and Fig. 9b.
[0094] Fig. 13 shows an example of a cross-sectional potential (corresponding to the planar potential in Fig. 12) for the simplified plan view of Fig. 9b. The cross-sectional structure for Fig. 9a is Fig. 3c, but in Fig. 9b, the transfer means 123 diagonally opposite the transfer means 117 is replaced with a p-well, so that in the cross-sectional view of Fig. 13, one side of the two vertical transfer gates shown on both sides of Fig. 3c is omitted.
[0095] The pixel size is 1 μm. The depth of the vertical electrodes is up to 0.55 μm. A simulation example is shown in which the ion implantation mask is changed for the structure in FIG. 9b.
[0096] Figures 10a, 11a and 12a show three types of pixel mask shapes. The mask in Figure 10a also shows the pixel element correspondence with the model in Figure 9b.
[0097] 10b, 11b and 12b show the impurity injection shapes of the three types of pixels.
[0098] In Figures 11b and 12b, the boron layer 124 diffused laterally into the transfer gate is formed by injecting boron into the side of the slit created by the etching after vertical etching using thermal diffusion. The slit is then filled with polysilicon to form the electrode 120. Since no mask is used, the side boron does not appear in Figures 11a and 12a.
[0099] The boron layer 126 in the center of Fig. 12b is implanted from the light-receiving surface side using a rectangular mask 125 shown in Fig. 12a. Since boron diffuses at the ends, the concentration distribution 126 in Fig. 12b has an elongated elliptical shape.
[0100] Figures 10c, 11c and 12c show potential diagrams.
[0101] As can be seen from FIG. 13, the center of the distribution of the boron layer 126 in the depth direction is about 0.4 μm from the light-receiving surface side.
[0102] In Fig. 11c, the potential at the outlet drops and the potential contour line 127 spreads around the center due to the boron implanted region 124 around the electrode in Fig. 11b, which causes a delay in the transfer time.
[0103] 12c, the boron implantation 126 in the central portion raises the potential in the central portion, and the potential contour interval 128 in the central portion is made uniform. In other words, the potential is distributed in a linear fashion toward the downstream.
[0104] FIG. 14 shows the change over time in the number of remaining electrons in a PD immediately after a group of electrons placed in the PD with all vertical transfer means set to VL and one of the vertical transfer means is set to VH.
[0105] Curves 181, 182, and 183 show the relationship between the elapsed time and the number of remaining electrons for the three cases of Figures 10, 11, and 12. As expected from the potential contours, the rate of decrease in the number of remaining electrons is in the order 183<181<182. In curve 183, the time required for the number of remaining electrons to decrease to 1 / 10 is 16 ps.
[0106] The FWHM (Full Width at Half Maximum) value is often used to define the time resolution. This represents the time interval from when the current value reaches half of the peak current value until it falls below that value. If the same data processing is applied to the remaining electron number curve, the FWHM equivalent value is 6 ps.
[0107] According to this study example, the present invention has the following effects.
[0108] When signal electrons are transferred from the periphery of the photoelectric conversion means of each pixel of a single-storey branching image sensor using K vertical transfer means, the number of electrodes can be reduced from 2K to K by using one electrode for each transfer means and using its side as a transfer path, and arranging these essentially on the circumference. Also, when using the space between two electrodes as a transfer path, an insulating layer with a thickness necessary to reduce the electric field and capacitance between the adjacent electrodes on the side of the transfer path is required, but this is also unnecessary. This makes it possible to arrange more transfer gates for the same pixel size, and capture more continuous images.
[0109] By adding an electron repellent layer to the vertical electrode, the probability of signal electrons colliding with the gate oxide film can be reduced, and trapping of signal electrons and generation of dark current can be suppressed.
[0110] By providing a means for making the potential distribution in the photoelectric conversion means closer to a straight line at the center of the photoelectric conversion means, the imaging time resolution can be further reduced. [Brief description of the drawings]
[0111] [Figure 1] Circuit diffusion layer of a typical pixel of a branching image sensor [Diagram 2] Conceptual diagram of a branching image sensor [Diagram 3] Cross-sectional view of a two-tiered branching image sensor [Figure 4] Circuit diffusion layer of one pixel of a single-storey branching image sensor [Diagram 5] Conceptual diagram of a cross section of one pixel of a single-storey branching image sensor [Figure 6] Example of potential simulation for a single-story building [Figure 7] Dependence of drift velocity and diffusion coefficient on electric field in silicon and germanium. [Figure 8] Absorption coefficients of silicon and germanium versus wavelength of incident light [Figure 9]Branching-gate image sensor with square four single vertical electrode transfer means and FD [Figure 10a] Mask drawing of a branching gate image sensor with one square single vertical electrode transfer means and FD [Figure 10b] Impurity distribution of a branching-gate image sensor with a square single vertical electrode transfer means and FD [Figure 10c] Maximum potential distribution in a branching-gate image sensor with a square single vertical electrode transfer means and FD [Figure 11a] Mask diagram of a square pixel with a single vertical electrode with peripheral boron ion implantation. [Figure 11b] Impurity profile of a square pixel with a single vertical electrode with boron ion implantation around the periphery [Figure 11c] Maximum potential distribution in a square pixel with a single vertical electrode with peripheral boron ion implantation. [Figure 12a] A mask diagram of a single vertical electrode with boron ion implantation on the periphery and a square pixel with boron ion implantation in the center of the photoelectric conversion means. [Figure 12b] Impurity distribution of a single vertical electrode with boron ion implantation on the periphery and a square pixel with boron ion implantation in the center of the photoelectric conversion means [Figure 12c] Maximum potential distribution of a square pixel with a single vertical electrode with boron ion implanted around the periphery and a photoelectric conversion means with boron ion implanted in the center [Figure 13] Impurity distribution in oblique cross section for the case of Figure 12 [Figure 14] Time waveform of the number of remaining electrons in a PD in response to pulsed incident light [Figure 15] A mask diagram of a plan view of one pixel in the first embodiment. [Figure 16] Impurity distribution in a plan view of one pixel in the first embodiment [Figure 17] Maximum potential distribution in a plan view of one pixel in the first embodiment [Figure 18] 1 shows an example of the potential and signal electron paths in a cross section of one pixel according to the first embodiment. [Figure 19] Time waveform of the number of remaining electrons in a PD in response to pulsed light incident on one pixel in the first embodiment [Figure 20] 2 is a cross-sectional view of the second embodiment; [Figure 21] 11 is a cross-sectional view of one pixel according to a third embodiment. [Figure 22] I-type and T-type electrodes [Figure 23] Square pixel array and hexagonal pixel array [Figure 24] FIG. 1 is a block diagram of a photographing device according to a first embodiment of the present invention; DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS First embodiment
[0112] The first embodiment is made of a silicon semiconductor. Figures 15 to 17 show the planar structure of one pixel of the first embodiment, and Figure 18 shows an example of the cross-sectional potential structure. Figure 15 shows a mask diagram 129.
[0113] Sixteen transfer means 131 and an FD 133 connected to each transfer means are provided around a circular photoelectric conversion means 130. In this example, the FD is a readout means within the pixel and also serves as a storage means for image signals.
[0114] As shown in Fig. 1, the general readout means 132 is composed of an output gate 180, an FD 133, a reset gate 134, a reset drain 135, and a source follower amplifier 136. In this embodiment, the latter stage of the readout means is mounted on a junction chip. This circuit is a general readout circuit that is widely used, so the details are not shown in Fig. 15.
[0115] The pixel diameter is 1 μm, and the vertical transfer electrodes are arranged from 0 μm to 2.5 μm in the depth direction.
[0116] FIG. 16 shows the concentration distribution of implanted ions for FIG. 15, FIG. 17 shows a plan view of the potential, and FIG. 18 shows a cross-sectional view of the potential along the center line passing through the electrodes to which VH is applied.
[0117] Each transfer means is composed of a vertical electrode 137 made of polysilicon and an insulating layer 138 made of silicon dioxide surrounding it. Also, boron is implanted into a circular region 139 at the center of the photoelectric conversion means.
[0118] A voltage VL of 0 V is applied to the fifteen electrodes 140. A voltage VH of 3.3 V is applied to the remaining electrode 141.
[0119] Figure 18 shows the average path 143 of electrons generated on the incident light side, the average path 144 of electrons generated at an intermediate depth, and the average path 145 of electrons generated near the bottom surface at the most upstream position 142 of the center line viewed on the plane of this cross section.
[0120] FIG. 19 shows the time change 146 of the number of remaining electrons generated in the PD by the instantaneous incident light.
[0121] The FWHM equivalent value of the current change is about 4 ps, and the time required for it to decay to 10% is about 13 ps.
[0122] As can be seen from the above, the present invention provides an image sensor that can capture a large number of consecutive images at an ultra-high speed of 16 consecutive frames with a very short time resolution of approximately 13 ps, which is something that could not be achieved with existing multi-framing cameras.
[0123] As shown in FIG. 18, in this embodiment, the transfer electrodes are vertical and extend deep, and a deep rod-shaped FD is provided on the outside thereof, so that the signal electrons move nearly horizontally within the photoelectric conversion means.
[0124] In this model, there is no need to transfer the signal charge vertically within the FD. The moment the signal charge is transferred horizontally to the vertically long FD, a voltage drop occurs in the FD, so the signal can be immediately transferred as a voltage signal by the source follower amplifier on the junction chip.
[0125] This contributes to increasing the shooting speed when the number of continuous shots is increased by adding additional memory on the bonded chip.
[0126] A single-storey sensor with a germanium photoelectric conversion means and a silicon post-stage circuit
[0127] The planar shape of the second embodiment is the same as that of the first embodiment. However, there is no circular boron layer 139 in the center of FIG. 16. A cross-sectional view 147 is shown in FIG. 20. A junction chip 149 is joined to a sensor chip 148. The feature is that the photoelectric conversion means 150 is made of germanium, and the transfer means 151 and FD 152 are made of silicon layers. Readout means 153 and 154 span the sensor chip and junction chip, and are electrically joined by Cu-Cu junctions (not shown). The junction chip is provided with readout means 155 to outside the pixel. The effect is that it enables ultra-high speed photography in NIR (near infrared) and SWIR (shortwave infrared).
[0128] Although germanium generates a large dark current, the branching image sensor is designed for ultra-high speed continuous shooting, with shooting time intervals on the order of picoseconds to nanoseconds, which is difficult to achieve with existing sensors.
[0129] The image signal obtained during this time is transferred to the storage means of the silicon layer at the subsequent stage at the above-mentioned time intervals, and is stored until it is read out after the shooting is completed. The dark current generated in the germanium photoelectric conversion means during the readout after the shooting is completed is discharged outside the pixel, so it does not actually get lost in the storage means (FD).
[0130] The simplest method to drain the dark current out of the pixel after shooting is to use one of the 16 transfer means for the drain. If the output gate 180 and reset gate 134 in the readout circuit shown in Figure 1 are set to the VH side, they will become the drain. However, the number of shots will be 15.
[0131] The dark current generated in the downstream silicon storage means during the long time before it is read out is much larger than the dark current generated during imaging, but defect control in silicon semiconductors is highly advanced, and even if it is relatively large, it can be reduced to an acceptable level by cooling.
[0132] As described above, even when capturing NIR or SWIR images using germanium photoelectric conversion means, dark current can be effectively suppressed by using a branching image sensor structure to transfer and store image signal packets in silicon storage means at very short time intervals.
[0133] In imaging devices for scientific measurement technology, the additional cost required for cooling technology of image sensors is small compared to the development cost of the sensor. The synergistic effect of sub-nanosecond integration of signal charges generated by germanium photoelectric conversion, suppression of dark current during readout by silicon post-stage circuits, and sufficient cooling of the image sensor makes it possible to develop NIR or SWIR branching ultra-high-speed image sensors with sufficient suppression of dark current. Third embodiment
[0134] Two-stage visible light sensor with germanium photoelectric conversion means
[0135] In the case of a branching image sensor in which both the photoelectric conversion means and the transfer means are silicon, the single-story type is superior to the two-story type in terms of shooting speed and simplicity of structure. Most image sensors are made entirely of silicon semiconductors, but when made from other materials, the single-storey type is not necessarily superior.
[0136] Example 156 shown in FIG. 21 is a case where germanium is used as the photoelectric conversion means, and is an example in which it is easier to implement in a two-story type than in a single-story type.
[0137] In Figure 21, a bonding chip 158 is also shown together with a sensor chip 157. The sensor chip looks like a single-story type because the photoelectric conversion means is thin, but it is structurally a two-story type. The photoelectric conversion means 159 in the upper layer is made of germanium. The circuit diffusion layer in the lower layer of the sensor chip is made of silicon. The bonding chip is also made of a silicon layer.
[0138] In visible light photography, the thickness of the germanium photoelectric conversion means can be 100 nm, and the thickness of the circuit conversion layer can be 500 nm because photoelectric conversion is not required.
[0139] In a typical silicon visible light image sensor, photoelectric conversion is performed in the silicon layer, so a thickness of at least 2.5 μm is required. In this embodiment, the thickness is one-fifth of that. Moreover, because the germanium photoelectric conversion means is thin, there are almost no limiting factors in forming a vertical electrode on the silicon layer from the light receiving surface side.
[0140] The reason for setting the layer thickness of the photoelectric conversion means to 100 nm will be explained. The average penetration depth of 550 nm green light into germanium is 20 nm (Figure 5). If red light is also included, a thickness of 100 nm is desirable.
[0141] The effect of this structure is that the manufacturing method of the branching image sensor of the vertical transfer means is simplified and the imaging speed can be further increased. The reason will be explained below.
[0142] The first reason is that the thickness of the photoelectric conversion means is sufficiently thin, so that even if the photoelectric conversion means is mounted thereon, a slit for forming a vertical gate can be formed from the light incident surface side.
[0143] Secondly, the vertical gate has a low height, which makes the slit process easier.
[0144] The third reason is that because the height of the vertical gate is short, both R and C in the RC (resistance x capacitance) that governs the time delay of the drive voltage to the vertical gate are small, allowing for faster imaging speeds.
[0145] The fourth reason is that, with this thickness of the photoelectric conversion means, even if the FD is formed on the light incident side or the opposite side of the circuit diffusion layer, the signal charge can be easily transferred obliquely from the transfer means, and the time resolution can be further shortened.
[0146] Adding a Gate
[0147] In order to introduce a CDS (Correlated Double Sampling) function to reduce readout noise, a storage means 102 and an output gate 180 between the storage means and the FD may be added as shown in FIG. 1. Also, to increase the number of images that can be captured, a second transfer means may be provided in the rear stage of each transfer means. These means may be of either a vertical or planar type. An InGaAs photoelectric conversion means and an InP circuit diffusion layer or a rear circuit
[0148] In SWIR imaging, a combination of InGaAs as the photoelectric conversion means and InP as the transfer and storage means is often used. This combination may be used instead of the combination of germanium and silicon. In this combination, the crystal constants of InGaAs and InP can be made equal, so there are no traps or dark currents due to defects caused by the difference in the crystal constants between germanium and silicon.
[0149] The boron layer around the transfer electrode may be a fixed charge layer such as aluminum oxide or hafnium oxide.
[0150] Instead of the single vertical electrode 160 shown in Fig. 22a, a T-shaped or inverted T-shaped electrode 161 shown in Fig. 22b may be used. This contributes to shortening the transfer time by oblique transfer within the transfer means when the FD or the first-stage readout means is on the light-receiving surface side or the bottom side.
[0151] The vertical transfer electrodes may be circular or elliptical in shape. If the electrodes are rectangular, the tips of the electrodes may be angled like the tip of a spear, rather than being straight and parallel to the circumference.
[0152] This allows signal electrons that are generated in the vicinity of the straight line connecting the transfer electrode to which VH is applied and the photoelectric conversion means and that reach the tip of the transfer electrode to be distributed smoothly to the left and right.
[0153] Also, a means for increasing or decreasing the peripheral potential of the transfer electrode in the radial direction may be used.
[0154] In addition, if the distance from the pixel center to the center position of each transfer electrode is equal, the central axis of the transfer electrode does not need to face the pixel center.
[0155] The shape of the photoelectric conversion means does not have to be a circle or a hexagon or more, but may be a square or a rectangle.
[0156] In the case of a square, one or two transfer means are provided on each side, for a total of four or eight transfer means, while in the case of a rectangle, two transfer electrodes are provided on each of the two opposing sides.
[0157] In the case of a square or rectangular pixel, two vertical electrodes that form a substantially 45° angle with the side may be placed near the center of each side, and one side of each electrode may be used as a transfer path.
[0158] When the light receiving surface is substantially circular, a square pixel array 162 has large marginal spaces 163 at the corners, as shown in FIG. 23a.
[0159] In the hexagonal pixel array 164 shown in FIG. 23b, the spare space 165 at the corners is narrower than in the case of a square, and the pixel density can be increased for a given light receiving area. Imaging equipment
[0160] The pixel arrangement on the light receiving surface, the configuration of the imaging device, and the method of operation in this patent technology are described in detail in Patent Documents 1 and 2, which are basic patents for branching image sensors.
[0161] The following is a brief description of the configuration of the image capturing device according to the first embodiment of the present invention shown in Fig. 24. The image capturing device 166 is made up of an optical system 167 and a camera body 168.
[0162] The camera body comprises an image sensor 169 according to the first embodiment of the present invention, an AFE (analog front-end circuit) 170, a drive circuit 171, a pixel configuration circuit 172, a DSP (digital signal processing circuit) 173, and an image display circuit 174.
[0163] The image sensor includes a pixel array 175 consisting of 1200×700 pixels, a drive signal receiving circuit 176 and an image capturing signal sending circuit 177 .
[0164] The AFE is equipped with a TG (timing generator) and a first-stage image signal recording means (not shown in the diagram). A drive circuit creates a drive signal for the image sensor according to the signal generated by the TG. It also reads out the captured image signal from the image sensor.
[0165] The image signals are rearranged in the pixel configuration circuit and converted into digital image signals by the DSP, after which the image display circuit adjusts the display speed, etc., and outputs the resulting image signal to an external monitor or recording device (not shown). Application Areas
[0166] High-speed video cameras are often used in mechanical engineering research and technology development for the motion of objects, high-speed flows, shock waves, deformations caused by collisions, etc. The high-speed multi-framing camera of the present invention is several orders of magnitude faster than conventional high-speed video cameras, creating a wide range of applications that were previously unimaginable for high-speed photography.
[0167] Here's one example. If we create an image sensor of the present invention using existing technology, we can achieve a time resolution of 100 ps. In principle, we can achieve 10 ps in the very near future. Furthermore, even if the image is taken with a time resolution of 100 ps, if it is repeated 100 times or more, the error will be reduced to 1 / 10 or less, making it possible to analyze phenomena with a time resolution equivalent to 10 ps.
[0168] In 10 ps, light (including X-rays, gamma rays, etc.) travels approximately 2.3 mm through the human body. When applied to TOF PET (Time-of-Flight Positron Emission Tomography), in addition to the conventional method of identifying the location from the intersection of multiple gamma rays emitted from the affected area, it is possible to know the distance from the sensor to the affected area, thereby significantly improving the probability and accuracy of cancer detection. [Explanation of symbols]
[0169] 100 Photoelectric conversion means of two-stage branching image sensor 101 Means of transfer 102 Preservation means 103 Reading means 104 Circuit diffusion layer of single-storey branching image sensor 105 Photoelectric conversion means 106 Vertical Transfer Means 107 Preservation means 108 2 vertical electrode pairs 109 Horizontal Transfer Path 110 Reading means 111 Relationship between electric field and drift velocity in silicon 112 Relationship between electric field and drift velocity in germanium 113 Absorption coefficient of silicon for each wavelength of incident light 114 Absorption coefficient of germanium for each wavelength of incident light 115 Insulating Layer 116 Photoelectric conversion means of a square flat-type branching image sensor for explanation 117 Vertical transfer means adding VH 118FD 119 P-type implantation for pixel isolation 120,121 Single electrode transfer means 122 Transfer paths on both sides of a single electrode 123 Add VL Vertical Transfer Means 124 Laterally diffused boron layer around the transfer electrode 125 Rectangular mask for implanting boron layer in the center of photoelectric conversion means 126 Boron layer in the center of the photoelectric conversion means 127 Potential contour lines near the center of the photoelectric conversion means (when there is no boron implantation in the center) 128 Potential contour lines near the center of the photoelectric conversion means (when boron is implanted in the center) 129 Mask diagram of one pixel in the first embodiment 130 Photoelectric conversion means 131 16 transports 132 Reading means 133FD 134 Reset Gate 135 Reset Drain 136 Source Follower Readout Amplifier 137 Vertical Electrode 138 Silicon dioxide insulating layer 139 Circular boron implanted region in the center of the photoelectric conversion means 15 electrodes with 140 VL applied 141 Electrode with VH applied 142 Position of signal electron generation on a plane 143 Average path of electrons generated on the incident light side 144 Average path of electrons generated at intermediate depth 145 Average path of electrons generated near the bottom 146 Time waveform of the number of residual electrons in a PD generated by pulsed light 147 Cross-sectional view of the second embodiment 148 Sensor Chip 149 Bonding Chip 150 Germanium photoelectric conversion means 151 Silicon Transfer Means 152 Silicon FD 153, 154 Reading means Off-pixel readout means on 155 bonded chips 156 Cross-sectional view of one pixel of the third embodiment 157 Sensor Chip 158 Bonding Chip 159 Thin germanium photoelectric conversion means 160 I-type vertical electrode 161 Inverted T-shaped electrode 162 square pixel array 163 Spatial space of square pixel array 164 hexagonal pixel array 165 Hexagonal pixel array spare space 166 First embodiment of the imaging device 167 Optical system 168 Camera body 169 Image Sensor 170 AFE (Analog Front-End Circuit) 171 Drive circuit 172 Pixel configuration circuit 173 DSP (Digital Signal Processing Circuit) 174 Image display circuit 175 pixel array 176 Drive signal receiving circuit 177 Shooting signal transmission circuit 178 Circuit Diffusion Layer 179 Charge collection means 180 Output Gate 181 Time waveform of the number of PD remaining electrons in Figure 10 182 Time waveform of the number of PD remaining electrons in Figure 11 183 Time waveform of the number of PD remaining electrons in Figure 12 184 Figure 1 shows a transfer means with charge collection and transfer functions.
Claims
1. A semiconductor device; means for focusing incident electromagnetic waves or incident charged particles onto said semiconductor device; means for controlling said semiconductor device; and a means for processing a signal output from the semiconductor device, Here, the semiconductor device is M rows and N columns of photosensitive units, where M and N are positive integers; Each of the photosensitive units is (1) a photoelectric conversion device that converts the incident electromagnetic wave or the incident charged particle into J electric charges (where J is an integer); (2) having m subunits (where m is 3 or more); The subunits each include at least one transfer device for selectively collecting and transferring the charges in turn; a storage device for temporarily storing the collected electric charges; and a voltage conversion device for converting the number of electric charges stored in the storage device into a voltage; The average direction of incidence of the incident electromagnetic wave or the incident charged particle is called the normal direction, and the direction substantially perpendicular to the normal direction and moving away from the photoelectric conversion device is called the radial direction. the collecting devices of the m subunits are positioned substantially equidistant from the photoelectric conversion devices; Each of said collection devices is characterized by comprising a single electrode having a substantially vertically extending portion.
2. 2. An imaging device according to claim 1, further comprising means for moving said charge away from said electrode.
3. 3. The imaging apparatus according to claim 2, wherein the means for moving said charges away from said electrodes is type 3 or type 5 ions implanted into a semiconductor layer around an oxide film around said electrodes.
4. 3. The imaging device of claim 2, wherein said means for directing said charges away from said electrodes is a layer of aluminum oxide or hafnium oxide adjacent to an oxide film around the periphery of said electrodes.
5. 2. The imaging device according to claim 1, further comprising an ion implantation region for type 3 or type 5 ions, the region having a planar shape substantially similar to that of the photoelectric conversion device and smaller than that of the photoelectric conversion device, in a central portion of the photoelectric conversion device.
6. The vertical length of the single electrode according to claim 1 is substantially equal to the vertical length of the photovoltaic conversion device according to claim 1.
7. The vertical length of the single electrode of claim 1 is substantially equal to the vertical length of the voltage-equivalent device of claim 1.
8. The photoelectric conversion device according to claim 1 is made of a semiconductor substantially containing germanium as a main component, and the storage device is made of a semiconductor substantially containing silicon as a main component.
9. The photoelectric conversion device of claim 1 consists essentially of InGaAs and the storage device consists essentially of InP.
Citation Information
Patent Citations
Data identification circuit
JP1986088679A
High-sensitivity / high-speed solid-state imaging device and apparatus
JP2016063530A
