Technical Field

[1] This disclosure relates to tuneable spectrum endoscope components, such as photo diodes that sense light. The photo diodes may also be used in other applications than endoscopes.

Background

[2] Endoscopes have improved significantly and are now an important tool for surgeons. For example, keyhole surgery has become possible because the surgeon can visually follow the surgery making a small incision through which an endoscope with a camera at the proximal or distal tip is inserted into the patient's body. Distal tip cameras are generally very small so that they fit into a relatively thin endoscope. In addition to the lens and the image processing electronics, the heart of the camera is the light sensor itself.

[3] A large range of light sensors are based on silicon technology, such as charge couple devices (CCD) or complementary metal oxide semiconductor (CMOS) chips. These devices provide photo sensitive areas also referred to as photo sites. A colour filter can be provided at each photo site in order to limit the sensitivity to a particular band of the light spectrum. For example, a 2x2 square of four photo sites can comprise one red, one blue and two green filters in front of respective photo sites. The measured signal then represents the red, blue and green (RGB) values of one pixel. Many such squares of photo sites together form a multi-pixel image sensor.

[4] There are, however, problems with this technology. The first problem is related to the location of the photo sites. In particular, the red, green and blue sites are off- set from one another but need to be combined into a single pixel value. Therefore, the colour values are interpolated, which can lead to colour artefacts especially on sharp edges. The second problem is that the colour filters absorb most of the light, which is then not available anymore at the photo site. As a result, the low-light performance is limited which often results in noisy images at low-light conditions. A further disadvantage is the complex manufacturing process, which results in an ever decreasing size of photo sites while increasing the number of pixels. This again leads to a reduced area that is available to capture light and reduced low-light performance.

[5] Any discussion of documents, acts, materials, devices, articles or the like which has been included in the present specification is not to be taken as an admission that any or all of these matters form part of the prior art base or were common general knowledge in the field relevant to the present disclosure as it existed before the priority date of each claim of this application.

[6] Throughout this specification the word "comprise", or variations such as "comprises" or "comprising", will be understood to imply the inclusion of a stated element, integer or step, or group of elements, integers or steps, but not the exclusion of any other element, integer or step, or group of elements, integers or steps.

Summary

[7] There is provided a photo diode made from multiple layers of organic semiconductor. The organic semiconductor is less complex to manufacture which means that the area of photo sites can remain larger. The multiple layers form an energy cascade that can be activated by the bias voltage, such that a response to certain bands can be switched on and off. This way, multiple bands can be sensed at the same photo site.

[8] A sensor to sense light comprises:

a first electrode; a first layer of organic semiconductor material connected to the first electrode and having first energy levels corresponding to a first band of the light;

a second layer of organic semiconductor material in contact with the first layer and having second energy levels corresponding to a second band of the light; and

a second electrode connected to the second layer, wherein

the first energy levels are different to the second energy levels

to block charge carriers generated by the second band of light in the second layer under a first voltage between the first electrode and the second electrode, and to conduct charge carriers generated by the second band of light in the second layer under a second voltage between the first electrode and the second electrode such that charge carriers detected at the first electrode and the second electrode are indicative of an intensity of light in the second band of the light when the second voltage is applied to the first electrode and the second electrode.

[9] It is an advantage that the voltage across the electrodes switches the sensitivity to the second band on (at the second voltage). This way, two bands of light can be selectively sensed with a single photosite.

[10] Blocking charge carriers may comprise inhibiting charge carriers generated in the second layer from crossing a junction between the first layer and the second layer into the first layer from the second layer

[11] Conducting charge carriers may comprise allowing charge carriers generated in the second layer to cross a junction between the first layer and the second layer into the first layer from the second layer.

[12] The charge carriers may be excitons and each exciton may comprise a positive charge carrier and a negative charge carrier [13] The first energy levels may comprise a first lower energy level and the second energy levels may comprise a second lower energy level, and the first lower energy level may be lower than the second lower energy level to block positive charge carriers generated in the second layer under the first voltage between the first electrode and the second electrode.

[14] The first lower energy level may be lower than the second lower energy level to conduct positive charge carriers generated in the first layer through the second layer to the second electrode under the first voltage between the first electrode and the second electrode

[15] The second voltage may shift the first lower energy layer towards the second energy level to allow conduction of positive charge carriers generated in the second layer into the first layer.

[16] The first energy levels may comprise a first upper energy level and the second energy levels may comprise a second upper energy level, and the first upper energy level is lower than the second upper energy level to block negative charge carriers generated in the first layer under the first voltage between the first electrode and the second electrode.

[17] The first upper energy level may be lower than the upper lower energy level to conduct negative charge carriers generated in the second layer through the first layer to the first electrode under the first voltage between the first electrode and the second electrode.

[18] The second voltage may shift the first energy levels towards the second energy levels to allow conduction of negative charge carriers generated in the first layer into the second layer. [19] The first layer and the second layer may comprise a respective semiconductor material that allows triplet state exciton generation at the corresponding band of the light.

[20] The first layer may comprise rubrene and the second layer may comprise pentacene.

[21] The first layer may have a thickness of between 5 nm and 100 nm and the second layer may have a thickness between 5 nm and 100 nm.

[22] The sensor may further comprise a third layer of organic semiconductor material between the second layer and the electrode.

[23] The first layer may have a thickness of between 5 nm and 100 nm and the second layer may have a thickness between 5 nm and 100 nm and the third layer may have a thickness between 2 nm and 20 nm.

[24] The first layer, second layer and third layer may define a well for the charge carriers.

[25] The well may be a well for positive charge carriers.

[26] Each layer may be one of:

donor;

acceptor; and

ambipolar.

[27] A multi-pixel image sensor comprises a matrix of multiple sensors as defined above.

[28] A camera comprises the multi-pixel image sensor above. [29] An endoscope comprises the camera above.

[30] A method for computer vision comprises:

exposing a photodiode having two layers of organic semiconductor material to light while applying a first bias voltage across the photodiode to extract first charge carriers generated by the light out of one of the two layers of organic semiconductor material;

exposing the photodiode to light while applying a second bias voltage across the photodiode to extract second charge carries generated by light out of another of the two layers of organic semiconductor material;

performing computer vision based on the extracted first charge carriers and the extracted second charge carriers.

[31] Performing computer vision may comprise classifying an object from which the light is reflected.

[32] It is noted that optional features of one aspect of photodiode, image sensor, endoscope and method equally apply to the other aspects.

Brief Description of Drawings

[33] An example will now be described with reference to the following drawings:

Fig. 1 illustrates a photo diode.

Fig. 2 illustrates the photo diode of Fig. 1 in reverse bias while illuminated. Fig. 3 illustrates a structure of energy levels of an organic semiconductor junction.

Fig. 4 illustrates the photo diode of Fig. 3 under a different bias voltage.

Fig. 5 illustrates a further example of a three-layer photo diode.

Fig. 6 illustrates experimental data for the EQE of the two layer configuration of

Figs. 3 and 4. Fig. 7 shows experimental data for the EQE of the three layer configuration of Fig. 5.

Fig. 8 illustrates the current density across the bias voltage for the three layer configuration of Fig. 5 for dark conditions and light conditions.

Fig. 9 illustrates the EQE of a pentacene/C ₆ o configuration for various different bias voltages.

Fig. 10 illustrates the EQE of a pentacene/DCV3T/C6o configuration.

Fig. 11 illustrates a method for computer vision.

Fig. 12 illustrates a multi-pixel image sensor.

Fig. 13 illustrates a camera.

Fig. 14 illustrates an endoscope.

Fig. 15 illustrates the difference in sensitivity between 0 V and - I V bias voltage.

Description of Embodiments

[34] Many industrial applications rely on machine vision. However, in many cases, common RGB systems are not sufficiently reliable. For example, it is difficult to distinguish from endoscope imaging data between healthy tissue and diseased tissue. Both types may appear red but due to different levels of oxygen saturation, could be distinguished based on their reflectance spectrum. The captured spectrum may also be indicative of a target tissue under a drug. For example, a particular drug may cause the spectrum of cancer cells be different to the spectrum of healthy cells. This may aid a surgeon in isolating and removing the cancer tissue.

[35] In a different example, the endoscope does not comprise a camera to generate images but a more general light sensor. This can be used in neural imaging for brain examination, for example, where a calcium indicator dye is injected and the dye is fluorescent under a voltage difference between two synapses associated with calcium transmission. The brain endoscope may comprise a monochromatic light source to stimulate the dye. The spectrum of light emitted from the dye can be used to identify neurons that are activated using a single photo diode in a device referred to as optrode. It is difficult to detect this spectrum using common RGB sensors.

[36] In yet another example that relates to agriculture, it is difficult for an

autonomous picker of green capsicum amongst green foliage to rely on the green channel of a common RGB camera because the green channel would have the same value for the capsicum and the leaves of the plant. As a result, a large number of leaves are picked and a large number of capsicums are left on the plants.

[37] This disclosure provides a tuneable spectrum endoscope component in the form of a photodiode that can be tuned "on the fly" to adapt to the problem at hand. In particular, the photodiode can be used at different configurations within a short period of time and a classification or other computer vision task can be performed based on the corresponding two values or images. The disclosed photodiode has a sensitivity spectrum to light that changes with the bias voltage. As a result, the bias voltage can be controlled to achieve an optimal outcome of the machine vision process. Referring back to the example of picking capsicums, two images can be taken with different bias voltages. The difference between the two images can then be used to distinguish the capsicums from the leaves because despite both being green, their spectra differ in other bands.

[38] Fig. 1 illustrates a photo diode 100 comprising a donor 101 joined to an acceptor 102 at a junction 103. A first electrode 104 and a second electrode 105 are connected at the outside to apply a bias voltage 106. The donor 101 provides charge carriers, such as holes, and the acceptor 102 accepts the photoexcited charge carriers, i.e. electrons.. The working mechanism is as such that when acceptor 102 is photoexcited it can transfer its hole to donor 101. In a silicon scenario, at zero bias voltage, the holes from the donor 101 recombine with the electrons from the acceptor leaving a depletion region 107 depleted of charge carries. The depletion region 107 acts as an insulator such that no current can flow. In organic donor- acceptor junction this is where the charge transfer states are formed. The energy level alignment of this junction is as such that when a positive potential is applied to the donor 101, the holes are attracted by that potential and flow to the left in Fig. 1. Equally, a negative potential applied to second electrode 105 causes the electrons to flow to the right. The charge selective nature of contacts 104 and 105 makes sure that high current flow in forward bias whereas no current can flow in the diode 100 in the reverse bias voltage. Most photo diodes, however, are operated in reverse bias. In standard structure photodiodes the contact 304 is transparent or semitransparent to incident light spectrum. Without loss of generality, reversing the charge selective role of contacts 104 and 105 is used to make inverted structure photodiodes. In inverted structure photodiode the contact 105 has to be transparent or semitransparent for the incident light spectrum.

[39] Fig. 2 illustrates the photo diode in reverse bias when light 200 hits the junction 103. The light 200 has sufficient energy to generate a positive hole 201 and a negative electron 202. The application of reverse bias voltage 106 efficiently pulls the hole 201 to the first electrode 104 and the electron 202 to the second electrode 105. This continues to occur while the light 200 hits the junction resulting in a current 203 that can be measured despite the photo diode 100 being in reverse bias. The value of this photo current 203 depends on the intensity of the light 200. As a result, the value of the current can be used as an indication of the light that has been received by the sensor. Converting the current into a digital value results in a colour value for that pixel (filters are omitted here for clarity).

[40] It is noted that Fig. 1 and Fig. 2 are simplified examples for illustration purposes only. The corresponding explanation applies to silicon and other non-organic semiconductors. Organic semiconductors, on the other hand have complex

photophysical requirements and behave differently in the sense that electrons and holes are not readily separated as they are in silicon. In other words, photons of the light 200 generate electron-hole pairs, also referred to as excitons, but there are certain energy requirements at the donor-acceptor junction to break these pairs into separate electrons and holes. This is referred to as the binding energy.

[41] Fig. 3 illustrates a structure of energy levels of an organic semiconductor junction 300 comprising a donor 301 and an acceptor 302 where higher levels indicate higher energy and lower levels indicate lower energy corresponding to energy axis 303 (units in eV with respect to vacuum being at 0 eV). The optical gap of donor could be as such that when light enters from the left hand side and the light that is not absorbed in donor 301 reaches acceptor 302 through donor. In inverted structure photodiodes, same principle can be applied to acceptor 302. In other words, the donor 301 is transparent to light at part of the spectrum where no excitons are created in donor 301. There is also a first electrode 304 and a second electrode 305. The donor 301 comprises a highest occupied molecular orbital (HOMO) 310 and a lowest unoccupied molecular orbital (LUMO) 311. Equally, acceptor 302 comprises HOMO 313 and LUMO 312. Photons excite electrons from HOMOs 310/313 into the LUMOs 311/312 leaving holes in the HOMO 310/313 bound to the excited electrons.

[42] The energy scheme of the donor-acceptor junction is as such that at zero bias voltage photoexcited electrons in the LUMO will only "fall down" an energy step while holes will only "rise up" an energy step. In particular, electrons in LUMO 311 of donor 301 can fall down into LUMO 312 of acceptor 302 but holes in HOMO 313 of acceptor 302 will not fall down into HOMO 310 of donor 301. The introduction of energy barrier for flow of holes from HOMO 313 to HOMO 310 provides the bias activated spectrum function to claimed photodiode. It is further noted that the difference between LUMO 311 of donor 301 and LUMO 312 of acceptor 302 is greater than the binding energy to dissociate excitons to create free electrons and holes.

[43] In one example, the photodiode 300 is reverse biased that is when a relatively negative potential is applied to the first electrode 304 and a positive potential is applied to the second electrode 305. Under this bias voltage, electrons in the LUMO 311 of the donor 301 are conducted through acceptor 302 to second electrode 305 while holes are conducted from the HOMO 310 of the donor 301 to the first electrode 304. As a result, a current can be measured that is indicative of the light incident on the donor 301. However, while electrons from LUMO 312 of acceptor 302 are conducted to second electrode 305, holes from HOMO 313 of acceptor are blocked from passing through donor 301 and therefore do not reach first electrode 304. Fig. 3 shows example hole 320 that is blocked from moving towards the first electrode 304. As a result, the measured photo current is indicative of only the light incident onto donor 301 and not indicative of light incident on acceptor 302.

[44] Fig. 4 illustrates the photo diode 300 under a different bias voltage that exceeds the energy barrier at HOMO 313 and HOMO 310 where now the HOMO 313 of acceptor lies below the HOMO 310 of donor 301. This happens due to the electric field dependent lowering of energy-level alignment at organic donor-acceptor junctions. As a result, holes are conducted from acceptor 302 through donor 301 to first electrode 304. In particular, example hole 320 is now conducted to first electrode 304 through donor 301. As a result, the photo current is indicative of light incident onto both donor 301 and acceptor 302.

[45] As can be seen in Fig. 4, the optical band gap, that is usually taken as the difference between HOMO and LUMO energy levels, is different between the donor

301 and the acceptor 302. In one example, the donor 301 is a rubrene layer with a HOMO of 3.2 eV and a LUMO of 5.4 eV, that is, a band gap of 2.2 eV. The acceptor

302 may be a pentacene layer with a HOMO of 4.9 eV and a LUMO of 3.0 eV, that is, a band gap of 1.9 eV. There is also a triplet state in rubrene with energy of 1.12 eV and in pentacene with energy of 0.86 eV. Both rubrene and pentacene undergo singlet fission that allows additional routes for electrons in the LUMO to transition through their respective triplet states. The multiple electron-hole pair generation per photon absorbed in singlet fission process results is an increased photo current which means a better low-light performance of the sensor. [46] The energy levels defining the band gap also directly correspond to the band of light at which the respective layer absorbs light and thereby generates electrons/holes. For example, the lowest singlet state S 1 state of rubrene relates to about 500 nm, the higher singlet states S2-Sn state of rubrene to about 330 nm and the S I state of pentacene to about 680 nm. In general, the wavelength correlates to the size of the band gap shown in Figs. 3 and 4. In this sense, a smaller band gap, that is, a shorter bar in the figures, indicates that less energy is required to generate an exciton, which relates to a longer wavelength. On the other hand, a large band gap, that is a longer bar in the figures, indicates that more energy is required to generate an exciton, which relates to a shorter wavelength. The above mentioned wavelength values are understood to be the centre values of the respective bands of light. The actual band of light is typically about 50 nm on either side of the centre value. Again, this highlights how the layers are sensitive to different bands of light, which can be activated by the appropriate bias voltage.

[47] In one example, the first electrode is a transparent, conducting layer, such as ITO/PEDOT:PSS, while the second electrode 305 is a metallic back plate, such as Sm/Al. The electrodes may also be swapped in the sense that the transparent electrode may be on either side, the donor 301 or the acceptor 302. It is further noted that there is an ohmic contact between the electrodes 304/305 and the connected layer of organic semiconductor and any other material with ability to modify work function for ohmic contact formation will deliver similar results.

[48] Fig. 5 illustrates a further example of a three-layer photo diode. In this example, there is a rubrene layer 501, a pentacene layer 502 and a C ₆ o layer 503. Again, there is a first electrode 504 with an ohmic contact to the rubrene layer and a second electrode 505 with an ohmic contact to the C ₆ o layer. The C ₆ o layer 503 significantly increases the light sensitivity, that is, the external quantum efficiency (EQE) of the photodiode. In addition, the C ₆ o layer adds another band at about 350 nm. [49] It is noted that in one example, in the two layer system of Figs. 3 and 4, the thickness range of the layers may be 5-100 nm. In three layer system of Fig. 5 the middle thickness range will be from 2-20 nm. And the outer two can range from 5-100 nm. In general, there is a trade-off in the sense that a thinner middle layer provides more efficient extraction of charge carriers before recombination especially at lower bias voltages while a thicker layer provides increased light absorption providing more charge carriers. The values provided above represent a good compromise between the two competing objectives.

[50] It is noted that more than three layers may also be used. It would be an advantage to have further layers with absorption bands in different regions of the spectrum. In particular, bands in the near or far infrared or the ultra violet (UV) range can be used.

[51] Examples

[52] Fig. 6 illustrates experimental data for the EQE of the two layer configuration of Figs. 3 and 4. At a bias voltage of -2 V, there is a clear response from the S2 level of rubrene 601, the S I level of rubrene 602 and pentacene 603. At a bias voltage of -1 V and 0 V the pentacene sensitivity 603 is almost not visible anymore. This illustrates again that the sensitivity to light in the band around 680 nm can be selectively switched on and off. It is noted that the high energy peak 601 is due to the direct dissociation of singlet excitons from rubrene's S2 level (hot excitons). Further, it is possible to extract triplet excitons generated in pentacene through rubrene layer.

[53] Fig. 7 shows experimental data for the EQE of the three layer configuration of Fig. 5. There is now a clearly improved EQE across the spectrum compared to Fig. 6. There is also a significant peak from C ₆ o indicated at 701. Again, at about 500 nm there is a rubrene peak 702 and at about 680 nm there is a pentacene peak 703. As in Example 1, the pentacene peak 703 vanishes as the bias voltage is changed from - I V to O V.

[54] Fig. 8 illustrates the current density across the bias voltage for the three layer configuration of Fig. 5 for dark conditions 801 and light conditions 802. It can be seen that the photo current generation is a function of the voltage bias and has different slopes. This indicates that with a larger voltage, charge carriers originating from different layers can be selectively extracted before they recombine. A value of open- circuit voltage; Voc of about 0.44 V at 803 can be observed that is defined by the pentacene/C ₆ o interface. It is noted that due to voltage bias controlled operation of such photodiode in this example, the following condition holds:

[55] Fig. 9 illustrates the EQE of a pentacene/C ₆ o configuration for various different bias voltages ranging from -1.5 V to voltages above Voc where the band of light around 350 nm 901 can be selectively activated. Fig. 10 illustrates the EQE of a pentacene/DCV3T/C6o configuration. Here, the band from 550 nm to 700 nm can be selectively activated.

[56] Method

[57] Fig. 11 illustrates a method 1100 for computer vision. The method 1100 comprises exposing 1101 a photodiode having two or more layers of organic semiconductor material as described above to light while applying a first bias voltage across the photodiode to extract first charge carriers generated by the light out of one of the two or more layers of organic semiconductor material . The method 1100 continues by exposing the photodiode to light while applying a second bias voltage across the photodiode to extract second charge carries generated by light out of another of the two or more layers. Then computer vision is performed 1103 based on the extracted first charge carriers and the extracted second charge carriers. [58] In one example, the two exposures are immediately after one another such that the scene that is captured is essentially identical for both exposures. For example, the two exposures can be performed within 1 ms. Method 1100 can be expanded to configurations with more than two layers where more exposures at different bias voltages are performed. In one example, the result from the first exposure can be subtracted from the result from the second exposure such that the difference indicates the band that is only detected in one of the two exposures, such as the absorption band of pentacene as shown in Figs. 6 and 7.

[59] Image sensor and camera

[60] Fig. 12 illustrates a multi-pixel image sensor 1200 comprising a first electrode 1201, a first layer of organic semiconductor material 1202, a second layer of organic semiconductor material 1203 and a second electrode 1204. In this example, the first electrode is a transparent conductor, such as ITO/PDOT:PSS and the second electrode is a back plate of Sm/Al. The first layer 1202 and the second layer 2013 form a matrix of individual photo sites such that the photo current generated at each site can be read out individually by read out electronics, such as amplifiers and A/D converters (not shown). However, the electrodes 1201 and 1204 reach across the entire matrix such that all photo sites are subject to the same bias voltage.

[61] As described above, there may be a third matrix or ever further matrices made of different organic semiconductor material to provide further bands at each pixel. It is noted that in the image sensor 1200 each photo site corresponds to one pixel and multiple bands are sensed at each individual photo site. As a result, the interpolation performed in most RGB cameras is not necessary since no adjustment of spatial location of photo sites for different bands is needed.

[62] Fig. 13 illustrates a camera 1300 comprising a lens 1301 and the multi-pixel image sensor 1200 from Fig. 12. When in use, the lens 1301 focusses an image of a scene onto the multi-pixel image sensor 1200. Readout electronics (not shown) then convert the photo current generated by the focussed light into pixel values that can then be processed by an image processor. In particular, the image processor controls the multi-pixel image sensor 1200 and potentially a shutter (not shown) to capture multiple images of the scene at multiple different bias voltages. As a result, each captured image reflects a different spectral band or combination of spectral bands.

[63] In one example, the image processor can adapt to different scenarios depending on the vision task at hand. For example, when objects with a peak at a particular wavelength need to be classified, the processor can use images sensor 1200 to capture two images where the first image captures the particular wavelength and the second image does not capture the particular wavelength. While both images capture a range of other wavelengths, processor can subtract or otherwise process the images to isolate the contribution from the desired wavelength. The image processor may also assess the quality of the classification and automatically adjust the bias voltages over time to gradually optimise the classification outcome. This way, the processor can find the optimal bias voltages without prior knowledge about the objects to be classified.

[64] Fig. 14 illustrates an endoscope 1400 comprising a hand piece 1401 and an insertion tube 1402. At the distal tip of the insertion tube there is a camera 1403 including an image sensor 1404 and lens 1405 as shown in Fig. 13. There may also be a light source and directional controls, which are not shown in the figure.

[65] As described above, a processor may process the data from the photo diodes. Again, this may be a single photo diode, a linear array of multiple photo diodes (strip), or a matrix of photo diodes, such as an imaging chip as shown in Figs 12 and 13. The processor may apply a different bias voltage for two successive measurements or images. Fig. 15 shows a first curve 1501 with square markers indicating the sensitivity at 0 V bias. The figure also shows a second curve 1502 with circle markers indicating the difference between the sensitivity 1501 at 0 V and the sensitivity at -1 V. This shows that the difference in sensitivity is significant at a band between 600 and 700 nm.

[66] It will be appreciated by persons skilled in the art that numerous variations and/or modifications may be made to the above-described embodiments, without departing from the broad general scope of the present disclosure. The present embodiments are, therefore, to be considered in all respects as illustrative and not restrictive.