WO2019115459A1

WO2019115459A1 - Pixel structure, image sensor device and system with pixel structure, and method of operating the pixel structure

Info

Publication number: WO2019115459A1
Application number: PCT/EP2018/084188
Authority: WO
Inventors: Guy Meynants
Original assignee: Ams AG
Current assignee: Ams Osram AG
Priority date: 2017-12-15
Filing date: 2018-12-10
Publication date: 2019-06-20
Anticipated expiration: 2020-06-15
Also published as: EP3499872B1; TWI837107B; US11405602B2; US20200382762A1; CN111684791B; EP3499872A1; TW201931618A; CN111684791A; US20220321862A1

Abstract

A photodetector (PD) in semiconductor material (S) is provided with a first transfer gate (TX1) between the photodetector and a first diffusion region (FD1) in the semiconductor material, a second transfer gate (TX2) between the photodetector (PD) and a second diffusion region (FD2) in the semiconductor material, a capacitor (C) connected between the first diffusion region (FD1) and the second diffusion region (FD2), a first switch (RST1) connected between the first diffusion region (FD1) and a first reference voltage (Vref1), and a second switch (RST1) connected between the second diffusion region (FD2) and a second reference voltage (Vref2).

Description

PIXEL STRUCTURE, IMAGE SENSOR DEVICE AND SYSTEM WITH PIXEL STRUCTURE, AND METHOD OF OPERATING THE PIXEL STRUCTURE

The present disclosure applies to the field of three- dimensional imaging, especially differential imaging.

Active stereovision systems make use of a projector and at least one image sensor. Typically the projector projects a structured light pattern that is composed of dots, lines or other shapes which are projected on the scene by infrared light, which may typically have a wavelength of 940 nm, for instance. The pattern is then recorded by a global-shutter two-dimensional image sensor including a two-dimensional matrix of pixels. The image sensor may be optimized for a good response at the used wavelength.

The light pattern can be projected by a strong light source, such as a VCSEL (vertical-cavity surface-emitting laser) during a short time. The maximum emitted light power is constrained by eye safety limitations and by the maximal power that can be output by the VCSEL at a specified duty cycle. Usually short pulses of few microseconds to few milliseconds are used. A pattern is created, e.g. by a mask, which is projected by an optical system (e.g a lens and/or an optical diffractive element) onto the scene.

A two-dimensional camera records the projected pattern. The position of the pattern as seen in the recorded image depends on the distance of the objects to the camera and projector.

By triangulation the distance can be recorded. Other

computation techniques may involve deep learning, for example in a case where the projected patterns changes appearance depending on object distance, thanks to some optical effect like focus distance or interference.

Some systems use two cameras and employ stereovision to make a depth map. A pattern is projected to create some contrast in the scene. This may be needed for objects without much texture, e.g. a painted wall. In those cases a flashed illumination is used.

In the above described stereovision and structured light cases, the light is pulsed and illuminated only during short time. The image sensor works in 'global shutter' mode. Before the flash, all pixels are reset and start integrating light. Then the pattern is projected. After the light pulse, the exposure time stops. In the pixels, the signal from the photodiode is sampled on an in-pixel memory. That memory is then read out row-by-row after the exposure time. This is a typical way of operation of global shutter image sensors, in so-called 'triggered global shutter mode' .

A rolling shutter pixel can instead be used. Because of the significantly larger contribution of background light to the captured signal, fast readout and off-chip data buffer storage are required.

Background image subtraction can be used to improve the contrast in the image. A second image is captured with the same exposure time but without projected light. The

differential image is then calculated by software. The subtraction operation is preferably executed inside the pixel during the image acquisition phase. It is favourable if multiple image acquisitions can be accumulated in the pixel, because a large amount of acquisition cycles enables to enhance the signal-to-noise ratio. After each image

acquisition, the signal corresponding to the difference between the images with and without structured light

projection remains stored in the pixel. Signals from

successive subtraction operations are thus accumulated in the pixel .

US 8 569 671 B2 discloses a pixel comprising a photo

sensitive element for generating charges in response to incident radiation. A transfer gate is positioned between the photo-sensitive element and a sense node for controlling transfer of charges to the sense node. A reset switch is connected to the sense node for resetting the sense node to a predetermined voltage. A first buffer amplifier has an input connected to the sense node. A sample stage is connected to the output of the first buffer amplifier and is operable to sample a value of the sense node. A second buffer amplifier has an input connected to the sample stage. Control circuitry operates the reset switch and causes the sample stage to sample the sense node while the photo-sensitive element is being exposed to radiation. An array of pixels is

synchronously exposed to radiation. Sampled values for a first exposure period can be read while the photo-sensitive element is exposed for a second exposure period.

It is an object of the present invention to provide improved differential imaging.

This object is achieved with the pixel structure for a semiconductor imaging device according to claim 1, the image sensor device with pixel array according to claim 9 the imaging system according to claim 10, and the method of operating the pixel structure according to claim 13.

Embodiments and variants derive from the dependent claims.

The definitions as described above also apply to the

following description unless stated otherwise.

The pixel structure for a semiconductor imaging device comprises a photodetector in a semiconductor material, a first transfer gate between the photodetector and a first diffusion region in the semiconductor material, a second transfer gate between the photodetector and a second

diffusion region in the semiconductor material, a capacitor connected between the first diffusion region and the second diffusion region, a first switch connected between the first diffusion region and a first reference voltage, and a second switch connected between the second diffusion region and a second reference voltage.

In an embodiment of the pixel structure, the photodetector is a pinned photodiode in a substrate comprising the

semiconductor material.

A buffer amplifier is connected to one of the diffusion regions to read out the pixel. In a further embodiment, buffer amplifiers are connected to the first diffusion region and the second diffusion region. In such an embodiment there may also be an additional anti-blooming gate connected to the photodetector to avoid carriers flowing into the diffusion region after the exposures.

In a further embodiment, the first reference voltage and the second reference voltage are equal, in particular equal to a supply voltage. The first reference voltage or the second reference voltage may be connected to a supply voltage.

In a further embodiment, a select transistor is connected to a column bus. A source follower formed by a field-effect transistor can be connected to the select transistor, a gate of the source follower being connected to the first diffusion region. This is a possible way of implementing the buffer amp1ifier .

The image sensor device comprises a two-dimensional array of such pixel structures.

The imaging system including such an image sensor device further comprises a light source synchronized to the image sensor device, the light source being configured to be active when the signal is acquired on the first diffusion region and not active when the signal is acquired on the second

diffusion region. The light source may instead be configured to be active when the signal is acquired on the second diffusion region and not active when the signal is acquired on the first diffusion region. In particular, the light source can be configured to project a pattern or sequence of patterns on a scene to be imaged.

The method of operating the pixel structure comprises discharging the capacitor by closing both the first switch and the second switch into a connecting state, acquiring a first signal on the photodetector and transferring the first signal to the first diffusion region by a charge transfer pulse on the first transfer gate, while the second switch is closed in a connecting state, acquiring a second signal on the photodetector and transferring the second signal to the second diffusion region by a charge transfer pulse on the second transfer gate, while the first switch is closed in a connecting state, and after acquisition of the signals, reading the pixel signal out from the first diffusion region while the second diffusion region is connected to the second reference voltage via the second switch.

In a variant of the method, the second transfer gate is at a higher electric potential than the first transfer gate during readout of the pixel signal.

Using the method, first and second signals can repeatedly be acquired and accumulated on the capacitor.

In a further variant of the method, the charge transfer pulse on the first transfer gate, the charge transfer pulse on the second transfer gate, or each of the charge transfer pulses is a multiple pulse, so that the corresponding transfer gate is pulsed several times during the same charge transfer operation .

The following is a detailed description of examples of the pixel structure in conjunction with the appended figures.

Figure 1 shows a pixel circuit including a schematical cross section of a pinned photodiode.

Figure 2 shows a circuit diagram of the pixel circuit

according to Figure 1.

Figure 3 is a timing diagram for the pixel structure. Figure 4 is a timing diagram according to Figure 3 for control of the maximal amount of stored charges.

Figure 5 is an alternative timing diagram according to Figure

3 for voltage control.

Figure 6 is an alternative timing diagram according to Figure

5 for repeated pulsing of the transfer gate(s) .

Figure 7 is a layout example with a capacitor comprising

interdigitated conductor tracks.

Figure is a layout example with a plate capacitor.

Figure 9 is a layout example with a plate capacitor and both transfer gates on the same side.

Figure 1 shows an example of the pixel structure. The circuit of the pixel structure includes a schematical cross section of a semiconductor photodetector device, which can be used for the pixel structure. A photodetector PD, which may be a pinned photodiode, for example, is integrated in

semiconductor material S, in particular in a semiconductor substrate, which may comprise silicon, for instance. The photodetector PD can be connected with a first diffusion region FD1 in the semiconductor material S via a first transfer gate TX1 and with a second diffusion region FD2 in the semiconductor material S via a second transfer gate TX2.

The diffusion regions FD1, FD2 may be arranged on opposite sides of the photodetector PD and may be provided with a first contact region CR1 and a second contact region CR2, respectively. The contact regions CR1, CR2 may have a higher doping concentration to enable the formation of an ohmic contact for external electric connections.

Typically the substrate may be composed of a bulk silicon wafer (e. g. p++ doped) with an epitaxial layer (e. g. p-, 5xl0¹⁴ cm^-3) . The pinned photodiode may comprise a p++ shallow surface implant and a deeper n-type implant. The first contact region CR1 and the second contact region CR2 may be highly doped (e. g. 5xl0¹⁹ cm^-3) n-type implants. They may be surrounded by lower doped n-type implants, and may or may not be located in a p-well area (which may be typically doped with a doping concentration of about lxlO¹⁷ cm^-3) .

The diffusion regions FD1, FD2 are electrically connected with each other via a capacitor C, which may typically have a capacitance of about 15 fF, for instance. The first diffusion region FD1 is also connected to a first reference voltage V_refl via a first switch RST1, and the second diffusion region FD2 is also connected to a second reference voltage V_ref2 via a second switch RST2. The first and second switches RST1,

RST2 can be field-effect transistors, for instance. The first reference voltage V_refl and the second reference voltage V_ref2 may be equal or different.

The first diffusion region FD1 is further connected to a buffer amplifier. This could be a source follower SF as shown in Figure 1. The first diffusion region FD1 may in this case especially be connected to the gate of a further field-effect transistor, for instance. The source follower SF connects a supply voltage V_DD with a source/drain terminal of a select transistor SEL. The other source/drain terminal of the select transistor SEL is connected with a column bus for read-out. The supply voltage V_DD may in particular be used as the first reference voltage V_refl and the second reference voltage V_ref2. The source follower SF and the select transistor SEL may also be integrated in the semiconductor material S. Alternatively to the source follower, a junction FET could be used, or a differential amplifier configuration could be used.

Figure 2 shows a complete circuit diagram of the pixel circuit according to Figure 1. The node of the first

diffusion region FD1 has a first parasitic capacitance Cfdl, and the node of the second diffusion region FD2 has a second parasitic capacitance Cfd2. It is advantageous if the

parasitic capacitances Cfdl, Cfd2 are equal and as small as possible. The capacitance of the capacitor C is considerably larger than the first and second parasitic capacitances Cfdl, Cfd2. Typical values of the first and second parasitic capacitances Cfdl and Cfd2 could be in the range from 1 fF to 4 fF. Techniques could be used to reduce the parasitic capacitances. Each of these parasitic capacitances Cfdl and Cfd2 is composed of the junction capacitance of the diffusion region, the gate-drain overlap capacitance of the transfer gates, the input capacitance of the buffer amplifier and the gate-drain overlap capacitance of the reset transistor. The overlap capacitances could be reduced by dedicated implant schemes, e. g. with modifications of dopant concentrations at lightly doped drain areas (LDDs) located adjacent to the gate under the gate spacer region. The junction capacitance and junction sidewall capacitance in particular can be changed by reducing the doping of the p-well under the n+ diffusion area, or by using a moderately doped n-type diffusion

surrounding the higher doped n++ ohmic contact area. The latter is shown in Figure 1. The capacitance of the buffer amplifier can be reduced by the input transistor topology.

E. g. a buried channel source follower or JFET could be used with lower input capacitance. Both devices have the

additional benefit that they have lower 1/f and random- telegraph-signal noise, which is also a benefit to reduce the read noise of the pixel.

Charges are dumped for successive frames consecutively on the first diffusion region FD1, while the second diffusion region FD2 is kept at the second reference voltage V_ref2 via the second switch RST2, and on the second diffusion region FD2, while the first diffusion region FD1 is kept at the first reference voltage V_refl via the first switch RST1. The

reference voltages V_refl and V_ref2 are choosen such that the signal swing on nodes FD1 and FD2 can be maximized.

An image with active illumination (e. g. when a structured light pattern is projected) may be integrated using the first diffusion region FD1, while a background reference image may be integrated using the second diffusion region FD2. In this case the signal at the first diffusion region FD1 will drop faster than the signal at the second diffusion region FD2. After several acquisitions of illuminated and background reference images, as per the later described timing diagrams, where the illuminated image is always transferred to FD1 and the background reference voltage is always transferred to FD2, the voltage on FD1 will drop below V_refl and the voltage on FD2 will rise above V_ref2. It is then beneficial to use a low voltage for V_ref2 and a high voltage for V_refl . It should be ensured that the voltage levels on FD1 and FD2 are always higher than the depletion voltage of the pinned photodiode at the moment that the respective transfer gate TX1 or TX2 is opened. This is needed to guarantee a charge transfer of all carriers from the photodiode to the nodes FD1 and FD2. Figure 3 shows a timing diagram for this pixel structure. The operation of the pixel structure is synchronously performed with the operation of a structured-light projector for multiple image acquisitions in global shutter mode. A first image is taken under structured-light conditions. The

photocarriers from this image exposure are integrated on the photodiodes of the pixel. The photodiode is then read out by a charge transfer through the first transfer gate TX1 with the first switch RST1 in its disconnecting state and the second switch RST2 in its connecting state, so that the signal is transferred to the first diffusion region FD1. The first diffusion region FD1 is thus floating when the first switch RST1 is in its disconnecting state. The diffusion region must be floating when the corresponding transfer gate is operated. A further image is taken without structured light. The photocarriers are again integrated on the

photodiode. The photodiode is then read out by a charge transfer though the second transfer gate TX2 with the first switch RST1 in its connecting state and the second switch RST2 in its disconnecting state, so that the signal is transferred to the second diffusion region FD2.

The first and second switches RST1, RST2 are alternatingly operated to remove the common mode signal corresponding to the background light. The first switch RST1 is in its connecting state only when the second switch RST2 is in its disconnecting state and vice versa, in order to prevent the capacitor C from being discharged. The driving circuits should guarantee that the RST1 and RST2 signals are non overlapping, in order to avoid this discharge. Only at the start of the image acquisition process, both switches are switched on together, in order to reset the capacitor C. After the time period of exposure, the differential image is read out from the first diffusion region FD1 using the source follower SF and the select transistor SEL. For this purpose, the first switch RST1 is in its disconnecting state, and the second switch RST2 is in its connecting state, so that the second diffusion region FD2 is at least approximately on the second reference voltage V_ref2, and the first diffusion region FD1 provides the stored signal. The voltage at the second transfer gate TX2 is higher than the voltage at the first transfer gate TX1, and the second transfer gate TX2 is acting as anti-blooming drain during readout. When the photodetector PD saturates, excess charge is removed by means of the second transfer gate TX2 and the second switch RST2. Alternatively, the second transfer gate TX2 could be in a high state during readout .

The pixel can also be operated as a standard rolling shutter pixel, by not clocking the transfer gate TX2 and the second switch RST2.

Figure 4 shows a further timing diagram. The timing according to this diagram allows to control the maximal amount of charges to be stored on the photodiode. This avoids overflow from the photodetector to the respective diffusion region. This can especially be useful if the illumination is bright. Three distinct voltage levels will be needed for the transfer gate control line if this timing diagram is used: a high voltage to enable charge transfer; a low voltage to prohibit overflow of carriers in case when the photodiode is saturated (under strong light conditions) and a middle voltage level to ensure that overflow of carriers at saturation flows thorugh the transfer gate. The overflow should happen to the

diffusion region where the respective RST switch is in its connecting stage, so that this diffusion region is connected to its respective V_ref voltage. Figure 4 shows a particular instance of such timing, but slight modifications can be considered that still are according to the described

operation .

Figure 5 shows a further timing diagram. This timing avoids excessive voltage at the second diffusion region FD2 by delaying the rising edge of the first switch RST1. The excessive voltage could otherwise cause leakage through the second switch RST2. The timing depends on the different dimensions of the components, including the full-well

capacity of the photodiode PD, the capacitance of the

capacitor C, and the parasitic capacitances Cfdl and Cfd2.

The first transfer gate TX1 may be pulsed several times during the time period when the first switch RST1 is in its disconnecting state and the second switch RST2 in its

connecting state, the signal corresponding to the first switch RST1 thus being low. And similarly the second transfer gate TX2 may be pulsed several times during the time period when the first switch RST1 is in its connecting state and the second switch RST2 is in its disconnecting state, the signal corresponding to RST2 thus being low. This multiple pulsing allows to transfer more carriers from the photodiode to the integration capacitor C. Using multiple transfers can reduce the attenuation caused by parasitic capacitances at the nodes formed by the first and second diffusion regions FD1, FD2. A corresponding time diagram is shown in figur 6.

Figures 7 and 8 show layout examples of the proposed pixel structure, in particular for backside-illuminated image sensor technologies. Figure 7 shows an example with a capacitor formed by

interdigitated conductor tracks, which can be produced in CMOS technology. For backside illumination, this capacitor can be arranged on top of the photodetector PD, so that the incident light is not shielded by the conductor tracks. An optional reflector R can be provided at the bottom of the layers that are produced in the back end of line (BEOL) .

Figure 8 shows an example with a plate capacitor comprising a sequence of a metal plate, a dielectric layer, and a further metal plate, in particular a metal-isolator-metal (MiM) capacitor, which is known per se. In the example shown in Figure 8, the capacitor C is formed by two individual

capacitors connected in anti-parallel fashion, in order to keep the parasitic capacitances of the bottom plate identical for the nodes of the first diffusion region FD1 and the second diffusion region FD2.

Other types of capacitors may be appropriate as well, depending on the individual application of the pixel

structure. Examples are trench capacitors and polysilicon- insulator-polysilicon capacitors, which are known per se in semiconductor technology.

Figure 9 shows a structure according to Figure 8 with the difference that the transfer gates TX1, TX2 are located on the same side of the photodetector PD. This arrangement may be advantageous for a more symmetrical pixel layout.

Additionally, a separate anti-blooming gate or photodiode reset gate may be connected. It may also be possible to evacuate and overflow charges from the pinned photodiode via a vertical anti-blooming structure. In that case the

photodiode can be reset by a voltage pulse on the substrate under the photodiode.

The pixel can also include a sample stage behind the pixel source follower, especially as described in US 8 569 671 B2, for example. This enables various new readout modes. It allows to expose the next image during readout, for example. It can also be used to sample the initial reset level and reduce the kTC noise of the pixel.

A two-dimensional array of pixel structures is especially suitable for an image sensor device. An imaging system including such an image sensor device may further comprise a light source synchronized to the image sensor device. A light source L is schematically indicated in Figures 7 and 8. The light source can be configured to be active when the signal is acquired on the first floating diffusion FD1 and not active when the signal is acquired on the second floating diffusion FD2, or vice versa. The light source can be used to project a pattern or a sequence of patterns.

In typical applications of the described pixel structure, difference signals are repeatedly obtained by detecting and subtracting two frames, which may be an illuminated frame and a dark frame, for example. The difference signals are

additively stored. Thus the signal-to-noise ratio is increase by charge accumulation. This enables to realize structured- light systems or stereovision with improved tolerance of background light. The amount of accumulated photocharges is

are the charges

accumulated in the k-th time interval of exposure of frame number 1, which may be an illuminated frame, for example, and frame number 2, which may be a dark frame. The duration of the corresponding time intervals can be different.

In embodiments of the pixel structure, more than two

different frames can be subtracted (e. g. for m different image acquisitions) , by use of more than one capacitor in the pixel. Each of the frames qi,_k, i = 1, 2, 3,..., m, can have an exposure time that differs from the exposure times of the other frames. Such pixels allow to do charge arithmetic and frame-to-frame calculations in the pixel.

It is advantageous that the subtraction operation can be executed inside the pixel during the image acquisition phase. It is also favourable that, after subtraction, successive acquisitions can be accumulated in the pixel. This improves the saturation level that can be accumulated in the pixel.

The described pixel structure allows a compact

implementation, because it requires only one photodetector, one capacitor, two transfer gates, two reset switches, and a source follower and select transistor in the pixel (6

transistors in total) . It is compatible with backside

illumination. Different timing allows to store a simple non- subtracted image with large full well charge.

The described image sensor can also be used in a camera system where an object is tracked. This object is illuminated by a flashed illumination or contains a flashed light source, which is synchronized to the images collected on FD1 or on FD2.

The described image sensor can also be used in a system where two structured light patterns are projected, and the first structured light pattern is integrated on FD1 while the second structured light pattern is integraded on FD2. This can further enhance the information that can be extracted from the projected patterns. Such information could include depth information. This depth information could be calculated from triangulation principles.

Similarly, the system could be operated with two light sources at two different wavelengths. E. g. for eye tracking applications, it is known that the eye reflects certain wavelengths and absorbs other wavelengths, leading to a high contrast in a differential image which is the difference between these two images. The proposed image sensor pixel allows to generate such profile.

List of reference logograms

C capacitor

Cfdl first parasitic capacitance

Cfd2 second parasitic capacitance

CR1 first contact region

CR2 second contact region

FD1 first diffusion region

FD2 second diffusion region

L light source

PD photodetector

R reflector

RST1 first switch

RST2 second switch

S substrate

SEL select transistor

SF source follower

TX1 first transfer gate

TX2 second transfer gate

V_DD supply voltage

V_refl first reference voltage

V_ref2 second reference voltage

Claims

1. A pixel structure for a semiconductor imaging device, comprising :

a photodetector (PD) in a semiconductor material (S) , a first transfer gate (TX1) between the photodetector (PD) and a first diffusion region (FD1) in the semiconductor material ,

a second transfer gate (TX2) between the photodetector (PD) and a second diffusion region (FD2) in the semiconductor material ,

a capacitor (C) connected between the first diffusion region (FD1) and the second diffusion region (FD2),

a first switch (RST1) connected between the first diffusion region (FD1) and a first reference voltage (V_refl), and a second switch (RST1) connected between the second diffusion region (FD2) and a second reference voltage (V_ref2) .

2. The pixel structure according to claim 1, wherein

the photodetector (PD) is a pinned photodiode in a substrate comprising the semiconductor material (S) .

3. The pixel structure according to claim 1 or 2, further comprising :

buffer amplifiers connected to the first diffusion region (FD1) and the second diffusion region (FD2) .

4. The pixel structure according to one of claims 1 to 3, wherein the first reference voltage (V_refl) and the second reference voltage (V_ref2) are equal.

5. The pixel structure according to claim 4, wherein the first reference voltage (V_refl) and the second reference voltage (V_ref2) are equal to a supply voltage (V_DD) .

6. The pixel structure according to one of claims 1 to3, wherein the first reference voltage (V_refl) or the second reference voltage (V_ref2) is connected to a supply voltage (V_DD) .

7. The pixel structure according to one of claims 1 to 6, further comprising:

a select transistor (SEL) connected to a column bus.

8. The pixel structure according to claim 7, further

comprising :

a source follower (SF) formed by a field-effect transistor, which is connected to the select transistor (SEL) , a gate of the source follower (SF) being connected to the first diffusion region (FD1) .

9. An image sensor device comprising a two-dimensional array of pixel structures according to one of claims 1 to 8.

10. An imaging system including an image sensor device according to claim 9, further comprising:

a light source (L) synchronized to the image sensor device, the light source (L) being configured to be active when the signal is acquired on the first diffusion region (FD1) and not active when the signal is acquired on the second

diffusion region (FD2), or vice versa.

11. The imaging system according to claim 10, wherein the light source (L) is configured to project a pattern or sequence of patterns on a scene to be imaged.

12. A method of operating the pixel structure according to one of claims 1 to 8, comprising:

discharging the capacitor (C) by closing both the first switch (RST1) and the second switch (RST2) into a connecting state,

acquiring a first signal on the photodetector (PD) and transferring the first signal to the first diffusion region (FD1) by a charge transfer pulse on the first transfer gate (TX1), while the second switch (RST2) is closed in a

connecting state,

acquiring a second signal on the photodetector (PD) and transferring the second signal to the second diffusion region (FD2) by a charge transfer pulse on the second transfer gate (TX2), while the first switch (RST1) is closed in a

connecting state, and

after acquisition of the signals, reading the pixel signal out from the first diffusion region (FD1) while the second diffusion region (FD2) is connected to the second reference voltage (V_ref2) via the second switch (RST2) .

13. The method according to claim 12, wherein

the second transfer gate (TX2) is at a higher electric potential than the first transfer gate (TX2) during readout of the pixel signal.

14. The method according to claim 12 or 13, wherein

first and second signals are repeatedly acquired and

accumulated on the capacitor (C) .

15. The method according to one of claims 12 to 14, wherein at least one of the charge transfer pulse on the first transfer gate (TX1) and the charge transfer pulse on the second transfer gate (TX2) is a multiple pulse.