Patent Application
20180166495
This application claims the benefit of priority of Singapore application No. 10201505778X filed on Jul. 24, 2015, the contents of it being hereby incorporated by reference in its entirety for all purposes.
Various aspects of this disclosure relate to sensor elements for sensing optical light, methods of forming the sensor elements, as well as methods of operating the sensor elements. Various aspects of this disclosure relate to image sensors for sensing optical light, methods of forming the image sensors, as well as methods of operating the image sensors.
The widespread of smart consumer electronic devices of which we are in incessant contact with in our daily lives has boosted huge market opportunities in recent times. One of the common features of these smart devices lies in their ability to instantly capture digital images of objects of interest. At present, this imaging capability is achieved via an integrated complementary metal oxide semiconductor (CMOS) image sensor, which is essentially made up of an array of minute optically sensitive electronic elements (p-n semiconductor junctions or photodiodes). The captured image is then transferred and stored in an on-board memory chip for further manipulation. Here, the NAND flash memory is also made up of an array of electronic elements called floating gate transistor. For both the sensor and memory components, information is represented in terms of either the absence or presence of electronic charges. The primary success of the CMOS image sensor and NAND flash memory technologies hinges on the use of well-established CMOS fabrication processes, which allow for integration of the image sensor and memory circuits, along with other control circuits on a given chip at a relatively low cost. Moving forward, the current technologies are unlikely to fulfill the future need for higher resolution image sensing devices. This is because the quantity of charges that individual sensor and memory components can store is fast reducing as their dimensions are being scaled down to achieve higher sensing resolution and storage capacity. With minimal charge storage, the loss of charges can induce substantial parameter variability issues and shrink the program/erase memory window beyond the acceptable range. In addition, due to the large difference in aspect ratios of the sensing and memory elements, direct on-chip integration of the image sensor and NAND flash memory circuits cannot usually be achieved. With integration made at the component level, the overall operating speed is limited by the external data buses.
In various embodiments, a sensor element for sensing optical light may be provided. The sensor element may include a first electrode for electrically coupling to a first supply voltage. The sensor element may further include a second electrode for electrically coupling to a second supply voltage. The sensor element may additionally include an oxide dielectric element between the first electrode and the second electrode. The oxide dielectric element may be configured to form a conductive filament upon a potential difference between the first supply voltage and the second supply voltage exceeding a threshold level, thereby decreasing a resistance of the oxide dielectric element. The sensor element may also include a detector. The first electrode may be configured to allow the optical light to pass through the first electrode to the oxide dielectric element. The detector may be configured to detect an increase in the resistance of the oxide dielectric element upon the oxide dielectric element receiving the optical light.
In various embodiments, a method of forming a sensor element for sensing optical light may be provided. The method may include forming a first electrode for electrically coupling to a first supply voltage. The method may also include forming a second electrode for electrically coupling to a second supply voltage. The method may further include forming an oxide dielectric element between the first electrode and the second electrode. The oxide dielectric element may be configured to form a conductive filament upon a potential difference between the first supply voltage and the second supply voltage exceeding a threshold level, thereby decreasing a resistance of the oxide dielectric element. The method may additionally include providing a detector. The first electrode may be configured to allow the optical light to pass through the first electrode to the oxide dielectric element. The detector may be configured to detect an increase in the resistance of the oxide dielectric element upon the oxide dielectric element receiving the optical light.
In various embodiments, a method of operating a sensor element for sensing optical light may be provided. The method may include electrically coupling a first electrode to a first supply voltage. The method may also include electrically coupling a second electrode to a second supply voltage. The method may further include forming a conductive filament in an oxide dielectric element between the first electrode and the second electrode to decrease a resistance of the oxide dielectric element, by applying a potential difference between the first supply voltage and the second supply voltage that exceeds a threshold voltage across the oxide dielectric element. The method may additionally include detecting, by using a detector, an increase in the resistance of the oxide dielectric element upon the oxide dielectric element receiving the optical light to sense the optical light which passes through the first electrode to the oxide dielectric element.
In various embodiments, an image sensor may be provided. The image sensor may include a plurality of first electrodes for electrically coupling to a first supply voltage. The image sensor may also include a plurality of second electrodes for electrically coupling to a second supply voltage. The image sensor may further include an oxide dielectric layer between the plurality of first electrodes and the plurality of second electrodes. The oxide dielectric layer may include a plurality of oxide dielectric elements, each respective oxide dielectric element configured to form a conductive filament upon a potential difference between the first supply voltage and the second supply voltage exceeding a threshold level, thereby decreasing a resistance of the respective oxide dielectric element. The image sensor may additionally include a detector. The plurality of first electrodes may be configured to allow the optical light to pass through the plurality of first electrodes to the oxide dielectric layer. The detector may be configured to detect an increase in the resistance of the respective oxide dielectric element upon the respective oxide dielectric element receiving the optical light.
In various embodiments, a method of forming an image sensor may be provided. The method may include forming a plurality of first electrodes for electrically coupling to a first supply voltage. The method may also include forming a plurality of second electrode for electrically coupling to a second supply voltage. The method may additionally include forming an oxide dielectric layer between the plurality of first electrodes and the plurality of second electrodes, the oxide dielectric layer including a plurality of oxide dielectric elements, each respective oxide dielectric element configured to form a conductive filament upon a potential difference between the first supply voltage and the second supply voltage exceeding a threshold level, thereby decreasing a resistance of the respective oxide dielectric element. The method may additionally include providing a detector. The plurality of first electrodes may be configured to allow the optical light to pass through the plurality of first electrodes to the oxide dielectric layer. The detector may be configured to detect an increase in the resistance of the respective oxide dielectric element upon the respective oxide dielectric element receiving the optical light.
In various embodiments, a method of operating an image sensor may be provided. The method may include electrically coupling a plurality of first electrodes to a first supply voltage. The method may also include electrically coupling a plurality of second electrodes to a second supply voltage. The method may further include forming a conductive filament in each respective oxide dielectric element of the oxide dielectric layer between the plurality of first electrodes and the plurality of second electrodes to decrease a resistance of the respective oxide dielectric element, by applying a potential difference between the first supply voltage and the second supply voltage that exceeds a threshold voltage across the respective oxide dielectric element. The method may additionally include detecting, by using a detector, an increase in the resistance of the respective oxide dielectric element upon the respective oxide dielectric element receiving the optical light to sense the optical light which passes through the plurality of first electrodes to the oxide dielectric layer.
The invention will be better understood with reference to the detailed description when considered in conjunction with the non-limiting examples and the accompanying drawings, in which:
The following detailed description refers to the accompanying drawings that show, by way of illustration, specific details and embodiments in which the invention may be practiced. These embodiments are described in sufficient detail to enable those skilled in the art to practice the invention. Other embodiments may be utilized and structural, and logical changes may be made without departing from the scope of the invention. The various embodiments are not necessarily mutually exclusive, as some embodiments can be combined with one or more other embodiments to form new embodiments.
Embodiments described in the context of one of the methods or devices (e.g. sensor elements or sensors) are analogously valid for the other methods or devices. Similarly, embodiments described in the context of a method are analogously valid for a device, and vice versa.
Features that are described in the context of an embodiment may correspondingly be applicable to the same or similar features in the other embodiments. Features that are described in the context of an embodiment may correspondingly be applicable to the other embodiments, even if not explicitly described in these other embodiments. Furthermore, additions and/or combinations and/or alternatives as described for a feature in the context of an embodiment may correspondingly be applicable to the same or similar feature in the other embodiments.
In the context of various embodiments, the articles “a”, “an” and “the” as used with regard to a feature or element include a reference to one or more of the features or elements.
In the context of various embodiments, the term “about” or “approximately” as applied to a numeric value encompasses the exact value and a reasonable variance.
As used herein, the term “and/or” includes any and all combinations of one or more of the associated listed items.
Various embodiments may seek to address or mitigate the abovementioned issues. Various embodiments may provide a smaller imaging or sensing device.
There has already been a lot of effort to develop an alternative memory technology that could replace the present NAND Flash memory technology. Resistive Random Access Memory, or RRAM, has been at the forefront of non-volatile memory technology research for the past few years and is touted to be the most suitable candidate. The popularity of RRAM lies in its simple structure and design, use of standard CMOS compatible materials, fast operating speed (on the order of nanoseconds) and low power consumption. Most importantly, RRAM makes use of a very localized change of the insulator resistance to denote binary information storage and thus is able to be scaled down to much smaller dimensions. Numerous materials have been tested so far for RRAM devices, including transition and rare-earth metal oxides, various electrolytes, semiconductor materials, and even some organic compounds. Among the many materials examined, transition metal oxides, such as the hafnium dioxide (HfO2), are particularly promising due to their established compatibility with the CMOS technology. A recent study reported excellent performance metrics for a HfO2 RRAM (10 nm×10 nm cell size, excellent endurance up to 109 cycles, fast switching speed in the nanosecond range, and power consumption per transition of ˜0.1 pJ. Most of the leading institutions in microelectronic manufacturing and research areas have been researching on the RRAM.
Basically, a RRAM device operates by alternating the resistance of a conducting filament embedded in the oxide dielectric between a high (“0”) and a low level (“1”). Typically, such changes are induced by applying voltages of different polarities across the device, following a high-voltage procedure that created the conducting filament in the oxide.
Apart from the electrical means, it is found that visible light may also cause the filament embedded in oxide dielectrics such as HfO2, ZrO2, SiO2, etc. to change from a low to high resistance state, i.e. light may partially or totally eliminate the filament. This observation may pave the way for the inclusion of novel optical functionalities in CMOS-based integrated circuits (e.g. a new oxide-based image sensor with built-in non-volatile memory capability, i.e. a RRAM array performing dual memory and light sensing roles). As will also be shown, the ability for the filament to be eliminated by light may also enable the reliability of oxide dielectrics to be extended significantly. This is extremely crucial for basic circuit elements such as the metal-oxide-semiconductor field-effect transistor (MOSFET), for which such oxide dielectrics play an integral role on its operation.
Most of the present research on the RRAM has thus far focused mainly on the enhancement of electrical performance through optimization of various design metrics (e.g. material, process, architecture and operational). The relatively large bandgap energy of HfO2, ZrO2, and SiO2 (>5 eV) may make them non-responsive to the visible light spectrum (which has photon energy ranging from 1.8-3 eV). Therefore, all studies reported to-date on RRAM based on these oxide dielectrics have only focused on electrical stimulation, and photo-induced effect has not been considered.
Various embodiments relate to a new negative photoconductivity behavior in soft-breakdown large bandgap dielectric materials such as HfO2, ZrO2 and SiO2. Soft-breakdown of the oxide dielectric, via an electrical stressing means during which a nanoscale conducting filament may be created in the oxide network, forms the basis for the operation of the RRAM. Via an electrical method, the resistance of this nanoscale conducting filament may be made to repeatedly toggle between a high and low level, which are used to represent the binary memory states. Our newly discovered negative photoconductivity effect makes it possible to incorporate optical functionality into the RRAM based on large-bandgap oxide dielectrics (e.g. a light-mediated multi-level memory cell, an image sensor with built-in non-volatile memory capability, etc.). As these oxide materials are already widely used in mainstream integrated circuit (IC) manufacturing, it may also be possible to integrate novel optical functionalities directly into such ICs and therefore has a widespread impact.
In the last few years, there have been several studies on the effect of light on the conduction behavior of MIM (metal-insulator-metal) or MOS (metal-oxide-semiconductor) structures. For instance, in Ungureanu et al., (“A light-controlled resistive switching memory,” Adv. Mater., vol. 24, no. 18, pp. 2496-2500, May 2012.), light was used to enhance the injection of optically generated electrons from a p-type Si substrate into a dielectric stack comprising Al2O3 and SiO2 layers. Subsequent trapping of electrons in the dielectric stack induced a current hysteresis (which appeared only in the presence of light) that could be potentially exploited for memory encryption. Numerous studies were focused on the positive photoconductivity behavior of ZnO nanorod and ZnWO4 nanowires and multiferroic dielectrics. The main reason attention has been put on alternative oxide materials is because conventional ICs have so far been fabricated using wide bandgap oxides (such as SiO2, HfO2, and ZrO2) which are non-photo-responsive. The motivation is therefore to identify suitable alternative photo-responsive oxides that could be integrated with the mainstream IC technology to enhance its functionalities. It should be noted that although the approach described in Ungureanu et al. makes use of the conventional MOS structure (which is a basic building block of ICs), its operating principle is the same as that used in current memory technology, in that different memory states are distinguished through varying the amount of charge trapped or stored. As mentioned earlier, this approach is already facing difficult challenges in addressing the needs of future memory technology due to the diminishing amount of charge stored as a result of device dimension scaling.
The most important benefit of our finding lies in resistance modulation, by visible light, of a nanoscale conducting filament created in an oxide dielectric (such as HfO2, ZrO2 and SiO2) by electrical stress induced breakdown. The basic operating principle entirely differs from that which the current image sensing technology is based on. In particular, the resistance of the nanoscale conducting filament can be increased, by several orders of magnitude upon exposure to visible light. In RRAM terminology, the resistance increase is referred to as reset. Since the resistance is increased in the presence of light, we termed this behavior negative photoconductivity, as opposed to the typical positive conductivity behavior of photo-responsive, narrow bandgap materials where increased conductivity (or decreased resistivity) results from light-induced generation of excess charges. Furthermore, the light-induced resistance modulation phenomenon does not interfere with the conventional electrical operation of the RRAM device.
The observed impact of visible light on RRAM resistance reset opens the door to many opportunities where optical functions could be directly incorporated into present mainstream ICs without the need for foreign-material integration. One such potential application is the direct engagement of a RRAM array in digital imaging application. In this case, one has an entirely new oxide-based image sensor with built-in non-volatile memory capability. This should be distinguished from the present CMOS image sensor which relies on a p-n junction to store the charges created by light and then transfer the charges out of the imaging array to be stored in another memory array. In various embodiments, an image sensor may be provided. The image sensor directly leverages the main advantages already identified with the RRAM technology. One very important advantage is scalability, by virtue of the fact that resistance switching is determined by a minute conductive filament (having a cross-sectional area on the order of several tens of nanometer square) embedded in the oxide layer. Hence, the dimensions of individual RRAM device may be reduced down to a level which may not be possible to achieve by the present CMOS imaging technologies, offering a quantum jump in spatial resolution improvement. It should also be highlighted that such scalability may be achieved without a loss of sensitivity, because the resistance of the nanoscale conducting filament can be modulated over several orders of magnitude. In fact, sensitivity may be independently enhanced by adjusting the thickness of the oxide layer, thus eliminating the size-sensitivity tradeoff of the current technology. Moreover, integrity of the image captured may be preserved over a longer period without requiring an intermediate data transfer to a separate memory storage, since RRAM device itself functions as a non-volatile memory device. In comparison, in the current CMOS image sensor, the generated charges must be quickly transferred out before they are lost through recombination.
Various embodiments may relate to multi-level or continuous resistance tuning using light. The ability to achieve different resistance levels may not only allow multiple memory states to be stored in a single filament, it may be very crucial for neuromorphic computing applications which require the basic computing element to exhibit a considerable amount of plasticity. Achieving different resistance values of a nanoscale conducting filament via an electrical means is challenging as it requires very precise control of electrical pulse timings. On the other hand, through adjusting the exposure period and light intensity, a nearly continuous variation of resistance may be achieved.
Various embodiments may relate to gate oxide reliability rejuvenation. The gate oxide is the heart of the MOSFET, the basic building block of nearly all ICs. Being an extremely thin oxide (comprising only several atomic layers) that is constantly subjected to electrical voltage biasing, defect generation over time may cause the gate oxide to lose its insulating property when the generated defects line up in a path that directly shorts the gate to the channel. When this happens, the functionality of the MOSFET may be compromised and this would in turn trigger IC failure. Thus, oxide breakdown is a pressing reliability issue in advanced ICs. Since light can eliminate a nanoscale conducting filament, various embodiments may offer an opportunity to reverse oxide breakdown and extend the gate oxide reliability significantly.
In other words, the sensor element 100 may include an oxide dielectric element 106 sandwiched between a first electrode 102 and a second electrode 104. A potential difference may be applied between the first electrode 102 and the second electrode 104 to form a conductive filament in the oxide dielectric element 106 so that the electrical resistance of the oxide dielectric element 106 is decreased. Optical light may then be provided to increase the electrical resistance of the oxide dielectric element 106. The first electrode 102 may allow the optical light to pass through to reach the oxide dielectric element 106. The increase in the electrical resistance may be detected or determined by the detector 108.
The phenomenon of increasing electrical resistance of element 106 upon exposure to optical light may be referred to as negative photoconductivity.
A sensor element 100 may be referred to as a sensor cell or an optical sensor. A sensor element 100 may have a metal-insulator-metal (MIM) structure, wherein the oxide dielectric element 106 is the insulator and the electrodes 102, 104 are the metal structures sandwiching the insulator.
The first electrode 102 may be in direct contact with the oxide dielectric element 106 or may be separated from the oxide dielectric element 106 by one or more electrically conductive layers or elements. The second electrode 104 may be in direct contact with the oxide dielectric element 106 or may be separated from the oxide dielectric element 106 by one or more electrically conductive layers or elements. In various embodiments, the oxide dielectric element 106 may be between a portion of the first electrode 102 and a portion of the second electrode 104. The oxide dielectric element 106 may be in contact with a portion of the first electrode 102 and may be in contact with a portion of the second electrode 104.
In various embodiments, the first electrode 102 may be translucent or transparent (to the optical light). The first electrode 102 may include indium-doped tin oxide (ITO) glass or any other suitable transparent conducting oxide (e.g. IGZO which denotes indium gallium zinc oxide) .
In various embodiments, the second electrode 104 may be opaque, while in other embodiments, the second electrode 104 may be transparent or translucent. In various embodiments, the second electrode may include any suitable electrically conductive material. In various embodiments, the second electrode 104 may include any suitable material selected from a group consisting of titanium nitride, copper, titanium, nickel, aluminum, silicon or polysilicon.
The first electrode 102, the second electrode 104, and the oxide dielectric element 106 may form a vertical stacked arrangement. The oxide dielectric element 106 may be on or over the second electrode 104, and the first electrode 102 may be on or over the oxide dielectric element 106.
The oxide dielectric element 106 may include any one oxide selected from a group consisting of zirconium oxide, silicon oxide, hafnium oxide, aluminum oxide, yttrium oxide, scandium oxide, zinc oxide, a bismuth based oxide, and any combination thereof. The oxide dielectric element 106 may be or may include a wide bandgap oxide such as of zirconium oxide, silicon oxide, hafnium oxide, aluminum oxide, yttrium oxide or scandium oxide. It may also be envisioned that the oxide dielectric element 106 may be or may include a photoconductive, narrow bandgap oxide such as zinc oxide or a perovskite such as BiMnO3 or BiFeO3.
In various embodiments, the oxide dielectric element 106 may have a thickness selected from a range of about 5 nm to about 10 nm for wide bandgap oxides, or a thickness selected from a range of more than about 10 nm for narrow bandgap oxides.
In various embodiments, the optical light may be or may include visible light and/or ultraviolet light and/or infrared light. In other words, the optical light may include visible light, ultraviolet light, infrared light, or any combination thereof. In various embodiments, the optical light may have a single wavelength or range of wavelengths (e.g. between about 400 nm and about 780 nm for visible light). The optical light may additionally or alternatively have a single wavelength or range of wavelengths corresponding to the regime of infrared light or ultraviolet light.
The threshold level may be any suitable non-zero value. The threshold value may be a positive voltage or a negative voltage, which may be determined by the thickness and/or processing conditions of the oxide dielectric element. In various embodiments, the threshold level may be any value selected from a range of between about 0.5 V to about 5 V. In various embodiments, the conductive filament may be formed in the oxide dielectric element 106 when the oxide dielectric element 106 is subjected to a potential difference above the threshold level.
In various embodiments, a current flowing through the oxide dielectric element to form the conductive filament may be below a predetermined compliance limit, for instance below 200 nA, or below 150 nA. In various embodiments, the predetermined compliance limit may be any suitable value selected from a range between about 10 nA to about 100 nA.
The detector 108 may be configured to determine or detect the increase in the resistance by applying a sensing voltage across the oxide dielectric element and determining the current flowing through the oxide dielectric element.
The sensing voltage may be smaller than the potential difference applied across the oxide dielectric element to form the conductive filament.
In various embodiments, the first supply voltage may be either positive or negative, depending on the structure of the sensor element. For example, for the silicon dioxide/copper/titanium/silicon (SiO2/Cu/Ti/Si) structure, a negative first supply voltage may be needed to effect the photoconductivity modulation described herein, whereas a positive or negative first supply voltage may be suitable for the hafnium oxide/titanium nitride (HfO2/TiN) structure). The second supply voltage may be at ground or 0 V. The first supply voltage may be chosen such that the potential difference between the first supply voltage and the second supply voltage exceeds the threshold level to form or reform the conductive filament.
In various embodiments, an image sensor may be provided. The image sensor may include a plurality of sensor elements as described herein.
In various embodiments, the image sensor may be a cross-bar array image sensor.
The image sensor may include a plurality of first electrodes for electrically coupling to a first supply voltage. The image sensor may also include a plurality of second electrode for electrically coupling to a second supply voltage. The image sensor may further include an oxide dielectric layer between the plurality of first electrodes and the plurality of second electrodes. The oxide dielectric layer may include a plurality of oxide dielectric elements, each respective oxide dielectric element configured to form a conductive filament upon a potential difference between the first supply voltage and the second supply voltage exceeding a threshold level, thereby decreasing a resistance of the respective oxide dielectric element. The image sensor may additionally include a detector. The plurality of first electrodes may be configured to allow the optical light to pass through the plurality of first electrodes to the oxide dielectric layer. The detector may be configured to detect an increase in the resistance of the respective oxide dielectric element upon the respective oxide dielectric element receiving the optical light.
In other words, the image sensor may include a number of first electrodes electrically coupled or connected to a first surface of an oxide dielectric layer, and a number of second electrodes electrically coupled or connected to a second surface of the oxide dielectric layer. During operation, a conductive filament may be formed in a portion of the oxide dielectric layer between a particular first electrode, or a portion of the particular first electrode, and a particular second electrode, or a portion of the particular second electrode, when a potential difference between the first electrode and the second electrode exceeds a particular threshold value. The potential difference between corresponding first electrodes and second electrodes may be increased to beyond the threshold value, so that a conductive filament is formed in the corresponding portions of the oxide dielectric layer between the corresponding first electrodes and second electrodes. The resistance of the different portions of the dielectric layer may be of an initial common or substantially equal resistance value. Optical light from an object may be incident on the different portions of the oxide dielectric layer. Different amounts of optical light may be on different portions, leading to different increase in resistance between the individual portions of the oxide dielectric layer. By sensing the resistance of the individual portions of oxide dielectric layer, the amount of light incident on the different portions of the oxide dielectric layer may be determined. Consequently, an image of the object may be determined.
The plurality of first electrodes may extend in parallel to one another along a first direction in a first plane. The plurality of second electrodes may extend in parallel to one another along a second direction in a second plane parallel to the first plane. The second direction (in the second plane) may be substantially perpendicular to the first direction (in the first plane).
Each oxide dielectric element may be a portion of the oxide dielectric layer between a portion of the second electrode and a portion of the first electrode over the portion of the second electrode.
The image sensor may further include a lens system for directing the optical light to the plurality of sensor elements.
The image sensor may also include a plurality of selectors, such as diodes. Each of the plurality of selectors may be electrically coupled to a sensor element. A particular selector may be activated to determine a resistance value of a corresponding oxide dielectric element at a time. The remaining selectors may be deactivated so that effects of parasitic leakage current from nearby sensor elements or corresponding oxide dielectric elements may be reduced or prevented.
The image sensor may additionally include a colour filter over the plurality of sensor elements for achieving colour imaging. The image sensor may also include an antireflection coating layer over the plurality of sensor elements for enhancing light transmission to the oxide dielectric layer.
In other words, the method may include forming an arrangement of two electrodes and an oxide dielectric element between the two electrodes. One of the two electrodes may be configured to allow optical light to pass through. A detector may be provided for determining an increase in the resistance of the oxide dielectric element caused by the optical light.
In various embodiments, the first electrode, the second electrode and the oxide dielectric element may form a vertical stacked arrangement. The method may include forming the second electrode, forming the oxide dielectric element on or over the second electrode, and forming the first electrode on or over the oxide dielectric element.
In various embodiments, the first electrode may be transparent or translucent. The first electrode may include indium-doped tin oxide (ITO) glass or any other suitable transparent conducting oxide (e.g. IGZO which denotes indium gallium zinc oxide). The method may include depositing the transparent conductive oxide (such as ITO or IGZO) on the oxide dielectric element. The second electrode may include any suitable electrically conductive material. The method may also include depositing the suitable electrically conductive material on the oxide dielectric element.
The oxide dielectric element may include any one oxide selected from a group consisting of zirconium oxide, silicon oxide, hafnium oxide, aluminum oxide, yttrium oxide, scandium oxide, zinc oxide, a bismuth based oxide, and any combination thereof. The oxide may be a mixed oxide. The method may include depositing the oxide to form the oxide dielectric element.
In various embodiments, the optical light may be visible light and/or ultraviolet light and/or infrared light. In other words, the optical light may include visible light, ultraviolet light, infrared light, or any combination thereof. In various embodiments, the optical light may have a single wavelength or range of wavelengths (e.g. between about 400 nm and about 780 nm for visible light). The optical light may additionally or alternatively have a single wavelength or range of wavelengths corresponding to the regime of infrared light or ultraviolet light.
The threshold level may be a non-zero voltage. The threshold level may be a positive voltage or negative voltage determined by the thickness and processing conditions of the oxide dielectric element. The threshold level may have a value ranging from 0.5 V to 5V.
In various embodiments, a current flowing through the oxide dielectric element to form the conductive filament may be below a predetermined compliance limit.
In various embodiments, a method of forming an image sensor may be provided. The image sensor may include a plurality of sensor elements. Each sensor element of the plurality of sensor elements may be formed by any method as described herein.
The method may include forming a plurality of first electrodes for electrically coupling to a first supply voltage. The method may also include forming a plurality of second electrode for electrically coupling to a second supply voltage. The method may additionally include forming an oxide dielectric layer between the plurality of first electrodes and the plurality of second electrodes, the oxide dielectric layer including a plurality of oxide dielectric elements, each respective oxide dielectric element configured to form a conductive filament upon a potential difference between the first supply voltage and the second supply voltage exceeding a threshold level, thereby decreasing a resistance of the respective oxide dielectric element.
The method may additionally include providing a detector. The plurality of first electrodes may be configured to allow the optical light to pass through the plurality of first electrodes to the oxide dielectric layer. The detector may be configured to detect an increase in the resistance of the respective oxide dielectric element upon the respective oxide dielectric element receiving the optical light.
In various embodiments, the method of forming the image sensor may include forming the plurality of first electrodes extending in parallel to one another along a first direction in a first plane. The method may also include forming the plurality of second electrodes extends in parallel to one another along a second direction in a second plane parallel to the first plane. The second direction may be substantially perpendicular to the first direction. A plurality of oxide dielectric elements may be included in an oxide dielectric layer between the plurality of first electrodes and the plurality of second electrodes. The method may include forming the oxide dielectric layer.
The method may include forming the plurality of second electrodes, forming the oxide dielectric layer on the plurality of second electrodes, and forming the plurality of first electrodes on the oxide dielectric layer.
Each oxide dielectric element may be a portion of the oxide dielectric layer between a portion of the second electrode and a portion of the first electrode over the portion of the second electrode.
The method may further include forming a lens system for directing the optical light to the plurality of sensor elements. The method may also include forming a plurality of selectors, each of the plurality of selectors electrically coupled to a sensor element. The method may also include forming a colour filter over the plurality of sensor elements. The method may additionally include forming an antireflection coating layer over the plurality of sensor elements.
In other words, the method may include applying a potential difference which is above a threshold between the first electrode and the second electrode across the oxide dielectric element to form a conductive filament, which decreases the resistance of the oxide dielectric element. The method may also include applying optical light to the oxide dielectric element, and using a detector to sense an increase in the resistance.
In various embodiments, applying the optical light to the oxide dielectric element may partially or completely remove the conductive filament, thereby increasing the resistance of the oxide dielectric element.
In various embodiments, the optical light may be applied or provided by an optical light source.
In various embodiments, a method of operating an image sensor may be provided. The image sensor may include a plurality of sensor elements. Each sensor element may be operated by a method as described herein.
The method may include electrically coupling a plurality of first electrodes to a first supply voltage. The method may also include electrically coupling a plurality of second electrodes to a second supply voltage. The method may further include forming a conductive filament in each respective oxide dielectric element of the oxide dielectric layer between the plurality of first electrodes and the plurality of second electrodes to decrease a resistance of the respective oxide dielectric element, by applying a potential difference between the first supply voltage and the second supply voltage that exceeds a threshold voltage across the respective oxide dielectric element. The method may additionally include detecting, by using a detector, an increase in the resistance of the respective oxide dielectric element upon the respective oxide dielectric element receiving the optical light to sense the optical light which passes through the plurality of first electrodes to the oxide dielectric layer.
By applying a potential difference that exceeds a threshold voltage across each oxide dielectric element of the plurality of oxide dielectric elements, the resistance of each of the plurality of oxide dielectric elements may be decreased to a common level. In other words, the resistance of one oxide dielectric element may be substantially equal to the resistance of another oxide dielectric element. The plurality of oxide dielectric elements may then be exposed to optical light emitted or directed from an object. The plurality of oxide dielectric elements may be exposed to different amounts of optical light, and the resistance of one oxide dielectric element may increase by a different amount compared to the resistance of another oxide dielectric element. By determining the increase in resistance for each of the plurality of oxide dielectric elements, an image of the object may be sensed or determined.
The resistance or increase in resistance of an oxide electric element may be determined or detected by applying a sensing voltage and determining or detecting the current flowing through the oxide dielectric element. The detector may be configured to apply the sensing voltage, and further configured to determine or detect the current flowing through the oxide dielectric element.
The resistance of one or a few oxide electric elements may be detected or determined at a time. An oxide dielectric element may be selected by activating a corresponding selector or diode in electrical connection with the oxide dielectric element
In various embodiments, a method of increasing a resistance of an oxide dielectric element (which may be referred to as a gate oxide) may be provided.
In various embodiments, the second resistance value may be substantially equal to an initial resistance value of the oxide dielectric element before electrical breakdown.
It has been found that exposure of an oxide dielectric material to optical light may repair the oxide dielectric material.
In various embodiments, the method may further include providing the optical light. The optical light may be provided by an optical light source, such as a light emitting diode (LED). The method may further include determining that the resistance of the oxide dielectric element has been increased. The resistance of the oxide dielectric element may indicate the reliability of the oxide dielectric element. The increase in the resistance of the oxide dielectric element may indicate that the reliability of the oxide dielectric element has been improved.
In various embodiments, a setup for increasing a resistance of an oxide dielectric element may be provided. The setup may include a holder for holding the oxide dielectric element or a sample including the oxide dielectric element, such as a field effect transistor. The method may further include an optical light source. The setup may also include a testing device, such as a device for obtaining a voltage-current characteristic of the oxide dielectric element, for determining the resistance of the oxide dielectric element.
In various embodiments, the first electrode 502a may be configured to allow the optical light to pass through the first electrode 502a to the oxide dielectric element 502c.
In various embodiments, the resistance of the oxide dielectric element 502c may be changed, i.e. increased or decreased to further resistance values for representing further logic states. In various embodiments, the memory element 502 may be capable of representing more than two logic states.
In various embodiments, the memory element 502 may be a conductive bridge resistive random access memory (CBRAM) element or cell.
Experiment 1
The negative photoconductivity phenomenon was first discovered on the hafnium oxide (HfOx) and was found to apply to the zirconium oxide (ZrOx) and silicon oxide (SiOx).
The sample structures used may be classified as: 1) oxide/metal (RRAM structures 600a, 600b, 600c); 2) oxide/Si (gate stack structures 600d, 600e). Fabrication of the former involves the sputter deposition of a titanium (Ti) adhesion layer 604a, 604b, 604c on a pre-cleaned p-type doped (100)-oriented silicon (Si) wafer 602a, 602b, 602c. This is followed by the deposition of either a titanium nitride (TiN) as the bottom metal electrode 606a, 606b in 600a, 600b, or a copper (Cu) layer as the bottom metal electrode 606c in 600c. The active oxide layer 608a, 608b, 608c may be then grown by atomic-layer deposition (ALD). Other oxide growth methods, such as sputter deposition, chemical vapor deposition (CVD), etc. also give qualitatively similar results although the exact performance (e.g. the chemistry of the nanoscale filament formed and thus its responsivity to illumination) may vary with the oxide growth method. For our study, the HfOx and ZrOx were grown by ALD using commercially available metal precursors (e.g. tetrakis (dimethylamino) hafnium for HfOx). The oxidizer was H2O vapor. The SiO, was grown by plasma-enhanced CVD.
The choice of TiN or Si as the bottom electrode is guided by its present widespread use in mainstream IC manufacturing. The former is used as the metal gate in the high-k/metal gate stack technology and as a barrier layer in backend interconnect technology. The latter is used as the channel material for the metal-oxide-semiconductor field-effect transistor, the basic building block of almost all ICs. Cu was used in the case of the SiO2 as the SiO2/Cu structure has been investigated for conducting-bridge RRAM application. The choices of the bottom electrode considered in our study, though not exhaustive, are in-line with current options for large-scale manufacturing. In principle, any metal or semiconductor which does not substantially react with the oxide layer of interest over the range of temperatures involved in the fabrication process would be suitable to be used as the bottom electrode.
The oxide stoichiometry was determined using x-ray photoelectron spectroscopy and the metal to oxygen ratio was typically 1:1.8. Other ratios yield qualitatively similar results.
Although the demonstration of the negative photoconductivity phenomenon was made on the HfOx, ZrOx and SiOx dielectrics, other oxide materials may also exhibit a similar phenomenon. The negative photoconductivity effect may be a consequence of the recombination between vacancy defects which made up the nanoscale conducting filament and photo-stimulated migrating oxygen species in the proximity of the filament. Thus, anyoxides which obey the breakdown mechanism where the oxygen species dislodged from the breakdown site (filament) by the electrical stress are stored in the vicinity of the vacancy-filled filament may exhibit the negative photoconductivity behavior and may be suitable to be considered for commercial applications based on the concept proposed in this document. These may include other wide bandgap oxides (−5 eV or larger bandgap) such as aluminum oxide (Al2O3), yttrium oxide (Y2O3), scandium oxide (Sc2O3), etc. It should be mentioned that photoconductive, narrow bandgap oxides (such as ZnO) or perovskites (e.g. BiMnO3, BiFeO3) may in principle also be suitable provided the decrease in current is not masked out by the increase contributed by the non-breakdown region surrounding the conducting filament. On the other hand, the surrounding non-breakdown region in a wide bandgap oxide is non-photo-responsive, i.e. conduction through this region is not affected by light and remains at a very low level by virtue of the wide bandgap. This enables the negative photoconductivity effect to be prominently manifested.
In the conventional metal/insulator/metal (MIM) structure used for RRAM application, the top metal electrode is relatively thick and has low optical transmittance. Hence, white light is unable to penetrate the top electrode and interact with the conducting filament in this structure.
Thus, the experiment was performed on an uncapped HfOx/TiN sample in an ultra-high vacuum (UHV) conductive atomic force microscope (CAFM)/scanning tunneling microscope (STM) system.
The system 650 may also include a white-light source 662. The white-light source 662, of wavelengths ranging from about 400 nm to about 780 nm, was provided by an ordinary LED (light-emitting diode) lamp, positioned at the quartz window of the UHV chamber 658, at about 25 cm from the sample 656. Interaction between white-light and the conducting filament occurs through the exposed oxide region around the probe 660. The light intensity at the surface of the sample 656 was determined by a Daystar's DS-05A solar meter placed at the same separation from the light source 662 as the sample 656. The inset 666 shows the magnified view of the probe-to-sample contact region, showing the probe 660 contacting a nanoscale conducting filament (cf) formed in the oxide.
It should be mentioned that a UHV environment is not a pre-requisite for the observation of the negative photoconductivity effect. For actual product implementation, an optically transparent top electrode (such as the conductive indium-doped tin oxide (ITO) glass) may be deployed. An alternative may involve high optical transmittance (˜80%) TiN sputter-deposited at a relatively high pressure (>10 Pa). In principle, any optically transparent and highly conductive material that remains thermally stable with the oxide of interest could serve as the top electrode. Lithography patterning followed by etching of the top electrode may have to be performed to achieve MIM structures of the desired dimension.
The nanoscale conducting filament was formed by a voltage-ramp applied to the probe by the parameter analyzer 664, e.g. a Keithley SCS4200 parameter analyzer. The polarity of the voltage-ramp is non-essential. The negative photoconductivity effect is observed on filaments formed using both positive and negative polarities.
The line 670a denotes the current-voltage (I-V) characteristic during the electrical process leading to filament formation. A progressive increase in current with voltage can be seen and filament formation occurs at the instant when an abrupt current increase is detected. The filament formation process may be well controlled in order for the filament formed to exhibit the negative photoconductivity behavior. Allowing the current to rise uncontrollably without any limit may give rise to severe Joule heating which in turn may change the material composition/structure at the breakdown location, leading to a state called hard breakdown. In this circumstance, the negative photoconductivity may not be observed. Thus, filament formation is carried out under a current-limited condition to achieve soft breakdown, with the maximum amount of current allowed to pass through the filament at the instant of its formation capped by the parameter analyzer at a preset level (termed current compliance), e.g. at 100 nA. In other words, when the current reaches this preset level, the parameter analyzer automatically reduces the voltage bias to restrict further evolution towards hard breakdown. In the soft breakdown state, oxygen ions released during the filament formation may populate interstitial sites around the filament—which may be a necessary condition for the negative photoconductivity effect to happen according to theory.
After filament formation at the depicted current compliance of 100 nA, a smaller arbitrary voltage of 1 V was applied to the probe and the current monitored continuously in the dark. The value of the monitoring voltage is large enough to enable the measured current to be distinguished from the noise floor but small enough so as not to induce further breakdown of the oxide region.
To further confirm the photosensitivity of the breakdown oxide region, current maps of the filament and its surrounding area before and after white-light illumination were obtained.
In the current map, a bright shade denotes a region of high conductivity. After soft breakdown, the filament is highly conductive and the breakdown oxide region is therefore much brighter than its surrounding area, as shown in
The resistance of the filament may be precisely modulated by varying the light exposure period (for a given light intensity) or the light intensity (for a given exposure period). This relationship between filament resistance and light energy may make the negative photoconductivity phenomenon suitable for imaging application.
The LIR effect may also be consistently observed on samples made up of other oxides (ZrO2, SiO2) as well as those having the gate stack (i.e. MOS) structure.
As shown in
A similar LIR effect is seen in the SiO2/Cu sample 600c (
It is also of interest to also examine the impact of light intensity and duration on the resistance reset. Results are shown in
Initially, white light of a lower intensity was applied to induce resistance reset after forming (triangle). Then, the conducting filament was reformed by ramping up a positive voltage followed by the exposure to white light of a five times higher intensity. As can be seen in
As the data in
In various embodiments, a light-enabled resistive random access memory (RRAM) resistance tuning approach may be provided as an alternative to the electrical counterpart for neuromorphic computing applications. The electrical tuning approach may be challenging because it requires the control of a set current compliance limit (for DC switching) or set/reset pulse timing (for AC switching).
A possible explanation for the light-induced resistance reset phenomenon is illustrated in
The proposed model may be supported by: 1) studies based on atomic-scale electron-energy loss spectroscopy which show that the breakdown oxide region is depleted of oxygen, i.e. the percolation path or conducting filament comprises of oxygen-vacancy defects; 2) a recent physics based simulation study showing some of the dislodged oxygen ions populating interstitial sites around the vacancy-filled conducting filament when the breakdown process is abruptly terminated.
Various embodiments may provide for different applications of the negative photoconductivity phenomenon of soft-breakdown oxides. Various embodiments may relate to an oxide image sensor which relies on the demonstrated relationship between filament resistance and light energy. Various embodiments may provide a method for rejuvenating oxide reliability, which depends on the restoration of the breakdown oxide region by light via elimination of the conducting filament.
Various embodiments may relate to other structures and/or methods employing the negative photoconductivity phenomenon. Since the negative photoconductivity of soft breakdown oxide is a physical phenomenon which stems from the interaction between photons and the chemical state of the filament, thereby modifying chemically the oxide region concerned, the phenomenon may have a broad range of optical applications that involve light sensing, or information processing applications that use light to change the logic state of a circuit element.
The basic element of the CMOS sensor array employed in nearly all mobile-phone cameras is a p-n semiconductor junction or a photodiode. Imaging is achieved by light generating excess electrical charge in the photodiode, the amount of which depends on the light energy entering the photodiode. The excess charges are then read by a periphery circuitry to determine the amount of light energy falling on a particular photodiode for imaging reconstruction purpose. The performance of a CMOS sensor (e.g. dynamic range, signal-to-noise ratio, etc.) is directly dependent on the size of the photodiode. More charges can be generated in a large photodiode, which translate into a richer tone of colors and enabling the sensor to fully reproduce the light details. A large photodiode, however, has a more dark current and is less sensitive to low-light conditions.
In the oxide image sensor according to various embodiments, the nanoscale conducting filament may replace the photodiode as the light-sensing element. Instead of generating excess electrical charge, the light may modulate the resistance of the filament. The extent of resistance modulation may depend on the amount of light energy that impinges on the filament (as shown in
Experiment 2
The test sample including a 4-nm thick HfO2 layer, formed via the process of atomic-layer deposition on the titanium nitride/titanium/p-type silicon substrate (TiN/Ti/p-Si) is fabricated. Tetrakis (dimethylamino) hafnium was used as the metal precursor and H2O vapor as the oxidizer. The growth temperature and pressure were 250° C. and 0.1 Torr, respectively. No post-deposition annealing was carried out. A control sample of titanium nitride/titanium/p-type silicon substrate (TiN/Ti/p-Si) is also fabricated.
The conducting filament was formed in the HfO2 layer with a setup similar to that shown in
The angle of incidence was about 45°. Measurement under light exposure was also performed on a bare TiN surface (i.e., a control sample without an active oxide layer).
The effect of the conducting filament dimension, determined by the current compliance limit (CCL) used during the DC forming process, on light-induced reset is shown in
When the conducting filament is formed at a low CCL (of 10 nA), the light-induced reset state is relatively stable (i.e., after the light is turned off, the current remains constant at a low level throughout the entire duration of the observation time). On the other hand, when the conducting filament is reformed at a higher CCL of 100 nA, the light induced reset is less complete and unstable (i.e., the current after illumination is much higher than the case of low
CCL and can be seen to gradually increase back to the prereset level). Similar results are obtained on the ZrO2/TiN/Ti/p-Si sample.
A possible explanation for the light-induced resistance reset phenomenon is illustrated in
These observations shown in
It should be noted that in
The resistance of a nanoscale conducting filament (formed using voltage sweep under a certain current compliance limit) in the HfO2 dielectric may be increased by exposing the dielectric film to white light. Resistance values may be modulated by varying the light intensity and exposure duration. The light-induced resistance increase in the reset process may be ascribed to a conducting filament disruption caused by a recombination of the oxygen vacancies there with oxygen ions, whose transport from the surrounding interstitial locations is activated by the white light. These findings point to a possibility of extending the application range of HfO2-based devices to the optical domain, e.g., imaging sensors and photodetectors with built-in non-volatile memory capability.
Experiment 3
Localized breakdown of SiO2 or HfO2 film formed on a p-Si substrate was achieved using a diamond-coated Si probe in an RHK 3500 ultra-high vacuum (˜5×10−10 Torr) conductive atomic force microscope (C-AFM). The contact area between the C-AFM probe and oxide is estimated to be 25.5 nm2, based on an applied force of 28 nN and the properties of the diamond-coated Si probe. The SiO2 film is 5-nm thick, formed by a standard plasma-enhanced chemical vapor deposition process with SiH4 and N2O. The HfO2 film was achieved by atomic layer deposition involving Tetrakis (dimethlyamino) hafnium and water vapor and has a 4-nm thickness. The C-AFM probe was connected to a Keithley SCS4200 parameter analyzer and the substrate was grounded.
Breakdown stressing, with a constant voltage at the probe, was applied and interrupted when the current exceeded a preset compliance limit. A LED lamp, positioned at a quartz window of the C-AFM system, functioned as a white-light source. The light intensity at the sample surface was 1 mW/cm2, estimated based on the separation between the sample and light source, using a Daystar DS-05A solar meter. The entire experiment was carried out at 297 K. Negligible light-induced heating was confirmed by periodic measurement of the sample temperature using an infrared thermometer. As SBD exhibits initial relaxation after stress interruption, light exposure was carried out only after a 45-minute wait period. Probe stability was verified by monitoring the breakdown current for extended period (>8 hours).
As can be seen from
Since electrical induced recovery (via an opposite polarity voltage sweep) is commonly observed in the MIM resistance random access memory (RRAM) structure, it may also be pertinent to examine the effect of an opposite-voltage sweep on the post-SBD recovery of gate stacks.
The opposite polarity sweep is shown to have a limited effect on the recovery of the SiO2/Si gate stack after suffering soft breakdown during negative constant-voltage stressing (CVS). On the other hand, full recovery via white light illumination is achieved, even after the gate stack had suffered further breakdown during the positive-voltage sweep. Unlike the RRAM structure, the gate structure is generally more resistant to electrical-induced recovery. As can be seen from
The lack of breakdown-oxide restoration under restricted positive-voltage sweep (
This may imply the concurrent role of other mechanisms and/or defect-passivating agents (e.g. hydrogen), which require further investigation
Physical analysis of the breakdown location in SiO2 by high-resolution electron-energy-loss spectroscopy has revealed an increasing proportion of Si oxidation states having values lesser than +4 when the current compliance limit is raised. In addition, a lateral growth of the oxide region having lower Si oxidation state values is correspondingly observed. These observations imply that more O is depleted from the stressing location when the stress is interrupted at a higher compliance limit, due to greater O migration by increased Joule heating. Based on this oxide breakdown framework, a possible explanation for the light-induced breakdown restoration is proposed.
Although a directed motion of O ions 1610, released from oxide-field-induced bond dissociation, towards the anode occurs during breakdown transient, the stochastic nature of ion transport, exacerbated by Joule heating (estimated to produce a substantial local temperature increase), results in lateral propagation and some O ions populating interstitial sites in the vicinity of the percolation path 1608 when the breakdown is arrested. Migration of the interstitial O ions 1610 back to the vacancy-rich percolation path 1608 may entail the overcoming of an energy barrier (˜0.3-0.6 eV) and thus does not readily occur in the absence of an external excitation. Under illumination by white light (with energies ranging from 1.8-3 eV), photon absorption electronically excites the interstitial O 1610, thus decreasing their migration barriers.
Aided by the vacancy-interstitial dipole field, the excited interstitial O 1610 may then be able to migrate towards the vacancy sites in the percolation path 1608. Subsequent recombination with the vacancies there eliminates the percolation path 1608 and restores the insulating property of the oxide 1606. Compared to the larger bandgap energy of SiO2 or HfO2, the relatively low photon energies of white light may be insufficient to “dislodge” lattice O and thus would not cause harder breakdown.
At a higher current compliance limit, increased Joule heating enhances the lateral propagation of O ions 1610, thus decreasing the portion of ions populating interstitial sites in the proximity of the percolation path 1608 when stress is terminated. As a consequence, only a limited number of the photon-excited interstitial O ions 1610 can migrate back to the percolation path 1608, resulting in only a partial restoration (
Experiment 4
The study was carried out on uncapped samples including SiO2/p-Si, HfO2/p-Si, HfO2/TiN/Ti/p-Si, ZrO2/TiN/Ti/p-Si and SiO2/Cu/Ti/p-Si stacks in an ultra-high vacuum (UHV) conductive atomic force microscope (C-AFM) system (RHK 3500 HT AFM/STM). The C-AFM probe was made up of diamond-coated Si, which performed the role of a top electrode mimicking a metal-oxide-Si (MOS gate stack) or metal/insulator/metal (RRAM) structure of a very small dimension. From the force-distance relationship and the mechanical properties of the cantilever, the probe-to-oxide contact was estimated to be 26 nm2. The UHV environment helped avoid surface contamination. The SiO2 was 5-nm thick for the SiO2/p-Si sample and 10-nm thick for the SiO2/Cu/Ti/p-Si sample, formed by a conventional plasma-enhanced chemical vapour deposition process. The HfO2 and ZrO2 were each of thickness 4 nm and were grown by an atomic-layer deposition process. Before the oxide deposition step, sputter-deposition of a 10-nm or 30-nm Ti layer, followed by a 70-nm TiN or 30-nm Cu layer, respectively, on a precleaned p-Si substrate were carried out to form the bottom metal electrode in some of the test samples. For simplicity, only the type of oxide and the metal electrode in contact with the oxide is referenced in subsequent discussion.
Experimental setup is similar to the one shown in
The time-dependent dielectric breakdown (TBBD) reliability of the pristine and restored oxide is compared for the SiO2/Si sample.
Soft breakdown of the SiO2 was induced by constant-voltage stressing (CVS) applied to CAFM probe. The experimental sequence used in the study is summarized in
For the two random locations depicted in
Experiment 5
In this study, the effect of light on the nanoscale property of a conductive filament in the silicon oxide/copper (SiO2/Cu) stack 2002 (Cu sample) is investigated.
By applying a negative-voltage ramp to the C-AFM probe, an abrupt current increase is observed at −10 V for the no-Cu sample (
Subsequently, a positive-voltage ramp was applied (2nd sweep). No resistance reset is observed in the former; note that there is no evident drop in the low-resistance state (LRS) current. This is supported by a subsequent negative-voltage ramp (3rd sweep), in which the current-voltage curve is similar to that of the 1st reverse-sweep. In contrast, a significant current drop is observed in the Cu sample (
Breakdown was induced at 30 random locations on the control sample and all of them did not display breakdown recovery. This result draws similarity to the general lack of breakdown recovery in the polysilicon/HfO2/Si gate stack. On the other hand, breakdown recovery can be commonly observed in the TiN/HfO2/Si gate stack. The relative ease of recovery in this gate stack is ascribed to the migration of O ions from the TiN interface region to the breakdown path during negative gate-voltage sweep. A possible reason for the absence of breakdown recovery in the control sample is the lack of O supply from the Si interface region due to the strong bonding between O and Si. Moreover, O ions driven towards the Si substrate during the breakdown may be “locked-in” by the formation of strong Si—O bonds, thus limiting their drift back to the breakdown path during the positive sweep.
On the other hand, more than 80% of the 30 conducting filaments (CFs) formed at random locations on the Cu sample exhibit a near-complete resistance reset after the positive sweep. This significantly enhanced breakdown recovery may be attributed to 1) the relative ease for Cu ions in the filament region to migrate back to the Cu electrode; 2) the availability of O ions from the Cu interface region for reducing the Vo defects in the filament region. As Cu has a lower oxygen affinity (the Gibbs free energy of the formation of the metal oxide) than Si, the oxidation of Cu is less likely as compared to Si. Thus, it is believed that 0 ions driven towards the Cu electrode during breakdown would accumulate in the Cu interface region and then drift back to the filament during the positive sweep.
Light may help reduce the oxygen vacancies (Vo's) in SiO2 and rupture the CF for both samples, even for the no-Cu sample in which reset is not observed under voltage ramp (
It is worth mentioning that both samples can be repeatedly “cycled” between the LRS (by voltage-induced set or breakdown) and the HRS (by light-induced reset or recovery (LIR)), similar to that shown in the study on the HfO2/TiN stack. The LIR effect may be attributed to the photo-induced migration of O ions from the vicinity of the CF or breakdown path and their recombination with the VO defects in the CF or breakdown path. These O ions were released and driven outward from the filament or breakdown path during the forming or breakdown process by the thermal effect and some reside interstitially in the surrounding oxide after the process was aborted. Ab-initio simulation study has shown the activation energy for the migration of interstitial O ions in the HfO2 to be ˜0.3-0.6 eV, corresponding to 4.8-9.6×10−20 J. Considering that the intensity of the visible light at the sample surface is 1 mW/cm2, equivalent to 1×10−19 J/s·Å2, it is thus plausible that visible-light illumination may reduce part of the VO defects in the CF or breakdown path by stimulating the back-migration of interstitial O ions located in the vicinity. On the other hand, such back-migration of interstitial O ions to the CF or breakdown path from the vicinity would not occur during voltage sweeping, as the Joule-heating effect tends to favor the further outward propagation of these O ions. This may explain the lack of breakdown recovery in the SiO2/p-Si control samples during post-forming voltage sweeps.
It has been shown that the conducting filament in the Cu sample may be rendered non-photo-responsive to the white light illumination after it is subjected to a negative-voltage ramp (0 V to about 8 V) several times before illumination. However, such an effect is not observed on the breakdown path in the no-Cu sample, which continues to exhibit a negative photo-response despite being subjected to the same number of negative-voltage ramps.
As shown in
By repeatedly subjecting the CF in the Cu sample to a 0→−8 V sweep (at the same current compliance used for forming), the Cu sample may eventually lose the photo-response seen in
To further probe the mechanism behind the lack of LIR after the repeated negative sweeps, a comparison is made between the averaged pre-illumination I-V curve of the 20 CFs which lost their LIR behavior after the repeated negative sweeps and the averaged pre-illumination I-V curve of the same 20 locations that were photo-responsive in the initial stage (cf.
These findings indicate that the photosensitivity of the CF in the Cu sample can be modulated by the repeated voltage sweeps.
A possible explanation for the loss of photo-response of a CF in the Cu sample after repeated negative sweeps is depicted in
In summary, it is shown that visible light is able to disrupt the CF in the SiO2/Cu stack and the breakdown path in the SiO2/Si stack. The observations may be attributed to the photo-stimulated recombination between Vo defects within the filament or breakdown region and O ions in the vicinity. However, unlike the latter where a consistent photo-response may still be observed, the CF in the SiO2/Cu stack would become non-photo-responsive after repeated negative-voltage sweeps were applied to the C-AFM probe (top electrode). The photosensitivity may, however, be restored by electrically resetting the CF and reforming it again. These results show the existence of an electrically controllable photosensitivity effect in the SiO2/Cu stack. The loss of photosensitivity may be ascribed to additional Cu ions drawn into the filament by the applied negative voltage filling the Vo defects there, thus suppressing their photo-stimulated recombination with O ions in the vicinity.
Experiment 6
The structure of this test sample is zirconium oxide/titanium nitride/silicon (ZrO2/TiN/Si). Interestingly, a larger number of repeated negative voltage sweeps (˜70) is needed to cause the breakdown path to lose its negative photo-response, as compared to the case of the copper (Cu) sample in the previous experiment.
Very interestingly, a single positive voltage sweep is sufficient to restore the negative photo-response. These findings suggest the possibility of achieving photo-response modulation via a combination of electrode type and number of voltage sweeps.
Among the various electrodes examined, Ni has the lowest percentage. This result agrees with other reports that it is relatively easy to form a Ni filament (i.e. a breakdown path made up of Ni) after forming. The highest percentages are obtained for the control sample (SiO2/Si) and Al sample (SiO2/Al/Ti/Si). Intermediate values are obtained for the Cu sample (SiO2/Cu/Ti/Si) and Ti sample (SiO2/Ti/Si).
Experiment 7
The infra-red light-induced RESET is observed on ZrO2/TiN sample in ultra-high-vacuum (5×10−10 Ton) conductive atomic force microscope (CAFM) chamber.
The effect of the infra-red light-induced-RESET is evidenced by the CAFM current map as well, as shown in
The RESET time (defined as the time for a decade decrease in current) follows log-normal distribution, with a mean value of 2.03 s and a median value of 1.90 s among 26 measured spots.
The sample is at fresh state before forming, showing a low current in the I-V curve. At a certain voltage, the forming occurs. The oxygen ions move away from the current path, driven by electrical field and Joule heating, and finally distribute around the CF to be interstitial oxygen ions. Upon illumination, these interstitial oxygen ions may be re-activated to be movable by absorbing the photon energy provided by the infra-red light source. By virtue of the oxygen concentration difference, the activated oxygen ions would naturally diffuse towards the low concentration area, namely the conductive filament, resulting in the disruption of the current path.
Experiment 7
The impact of red light and blue light on the speed of the negative photoresponse is examined. The red and blue coloured lights were achieved by passing white light (from an LED (light-emitting diode) illuminator) through the relevant colour filter. The coloured light was focused using a collimator lens. The intensity was measured by a solar meter placed at 25 cm (similar to the separation between the test sample and light source) from the light source.
The data presented here are preliminary. Further measurements are in progress after the following issues were resolved recently.
“Shadow” effect: Because the intensity of the light after the colour filter is significantly reduced, a collimator lens has to be used to focus the light onto the test sample so that sufficient light intensity is achieved. Due to the now narrower beam of light, a shadow effect occurs due to light blockage by one of the legs of the “beetle” scan head of the conductive atomic force microscope. Hence, the actual intensity of the light arriving at the sample surface is lower than that measured separately using a solar meter. This problem was successfully resolved by optimizing the sample height so that the leg would not be in the path of the light source after the probe has made contact to the sample surface.
Measurement speed limitation: Current measurement was previously made using a semiconductor parameter analyser. The average measurement time of the analyser is ˜2 s or longer for sub-picoampere current level. This limitation had severe impact on the accuracy of the response time measurement. This problem was successfully resolved via real-time waveform capture using a high-speed oscilloscope. Very recent measurements have revealed response time in the range of tens to hundreds of milliseconds, giving much potential for further reduction in the case of a prototype device.
Image Sensor
The instrument that enables mobile devices such as smartphones, tablet computers, etc. to instantly capture an image of interest and share it on social media is the digital camera. At the heart of the digital camera is the CMOS image sensor, which is basically made up of a two-dimensional array of photodiodes, where each photodiode makes up one pixel in the digital image. The imaging process occurs when light, from an external object, falls on the array of photodiodes, generating within individual photodiodes varying amount of excess electrical charges which are in turn detected by a peripheral circuitry to form the digital image. Due to the constant drive towards higher image resolution, the size of individual photodiodes has continuously decreased over the years. The reduction in size of a photodiode has resulted in several critical problems/challenges for future CMOS image sensor development. These challenges include: 1) increased “cross-talk” between pixels due to their reduced spacing; 2) decreased signal-to-dark-current-noise ratio due to lesser electrical charge generation within the reduced pixel area.
Recently, it has been observed that a wide bandgap oxide such as SiO2, HfO2, etc. exhibits a negative photoconductivity behavior after suffering an electrical induced soft-breakdown. It should be mentioned that because of the relatively large bandgap of these oxide materials, they are non-photo-responsive (i.e. visible light is unable to induce any electrical response from them). Hence, although these oxide materials have been widely deployed in current integrated circuit (IC) manufacturing technology, stimulating them using visible light for realizing novel functionalities has never been a consideration; the operation of current CMOS ICs are limited to only electrical stimulation. It has been found that after suffering an electrical soft-breakdown, these oxide materials may become photo-responsive. In particular, the insulating property of the breakdown oxide may be restored upon illumination by visible light. As will be shown later, the restoration may be manifested via a significant decrease in electrical conduction through the breakdown oxide region. This response, which may be referred to as negative photoconductivity, suggests possible exploitation for digital imaging. Since these oxides are already part of the existing IC manufacturing technology, full manufacturing compatibility may immediately be realized after successful prototype development and testing.
Further reducing the size of the photodiodes in the current CMOS image sensor to improve image resolution may be challenging due to the issues on pixel cross-talk and dark current noise, which basically limit how close and how small each photodiode can be. The oxide image sensor according to various embodiments may be based on a different operating principle—the detection of the change in the electrical resistance, due to visible-light induced restoration of the insulating property of a nanoscale breakdown oxide region. As a consequence, the individual sensing elements in the oxide image sensor may be very small and spaced relatively far apart. This may help eliminate the crosstalk problem (leakage of electrical charge from one to another adjacent photodiode) that the current CMOS image sensor is facing. Moreover, the oxide sensing element may work based on an inverse or negative photoconductivity principle, i.e. the current flowing through the sensing element in the dark (i.e. the dark current) is higher than when it is exposed to visible light, with the rate of current decrease increasing with the dark current. In this way, the sensitivity of the oxide sensing element may actually improve with the level of the dark current. In contrast, the photodiodes in the CMOS image sensor relies on the additional electrical charges generated by light to achieve a further current increase as a means for light sensing. In this manner, the sensitivity of the photodiodes is compromised by the dark current already present before exposure to visible light, as a sufficient number of electrical charges may have to be generated during light exposure so as to create a current increase that can be distinguished from the inherent dark current noise due to thermal generation.
These inherent advantages of the oxide image sensor according to various embodiments mentioned above may be crucial to the realization of digital cameras with much improved image resolution (in the several tens of mega pixels) and that may be free from crosstalk and dark current noise issues. With the widespread use of digital cameras in mobile devices, which now form an integral part in the daily life of a common man on the street, various embodiments may thus have very a significant societal and market impact.
As shown in
The modified resistance of each filament 2708 may then be read by applying appropriate voltages at the metal lines 2702, 2704 for digital image reconstruction. A potentially advantageous feature of the oxide sensor 2700 may lie in the broad dynamic range of resistance of ˜3 to 4 orders of magnitude as shown above. As the resistance of the filament 2708 may be electrically varied at the time of its formation, it may be possible to use the same oxide sensor 2700 under a wide range of lighting conditions. For instance, for imaging under low-light condition, the current compliance applied during filament 2708 formation may be set to a lower value so that fewer oxygen ions may be displaced into the interstitial positions. This may in turn require lesser light energy to reset the resistance, making the oxide sensor 2700 more sensitive under low-light conditions.
In other words, before the image capture step, each sensing element 2704′, 2704″ may be set into the same default state by electrically inducing a soft breakdown to the oxide 2706′, 2706″ sandwiched between the top electrodes 2702′, 2704′ and bottom electrodes 2702″, 2704″ through appropriate voltage pulses applied. The time needed to complete this step for each sensing element 2704′, 2704″ may be in the range of several tens of nanoseconds, as is already shown in other studies aimed at engaging a similar structure for non-volatile memory applications. As such a step may be simultaneously carried out on all the sensing elements 2704′, 2704″, the overall time needed may be extremely short and may not be noticeable to the user. Following this, exposure to the light coming from a lens system may take place. As a result of the spatial variation in the light intensity, the breakdown oxide 2706′, 2706″ in each sensing element 2701′, 2701″ may be restored by a different degree. As illustrated in
In contrast, the basic configuration block for the oxide image sensor 2700 may be much simpler, with each sensing element's operation controlled by one selector or diode (meant for eliminating parasitic leakage current from nearby elements).
For various embodiments of the image sensor 2700, as light sensing takes place within a nanoscale breakdown region in the oxide, individual sensing elements (e.g. 2701′, 2701″) may be made small and spaced far apart (relative to the size of the sensing element). This may help overcome the crosstalk issue that the CMOS image sensor 2750 currently faces. In the CMOS image sensor 2750, the minimum spacing between individual photodiodes may be limited by the wavelength of red light (˜700-800 nm). Given that the total depth of the photodiode must be sufficiently deep (˜1.5 μm) in order for a substantial part of the red light to be absorbed, a spacing below the minimum may result in diffraction and distortion of the image. At present, the spacing between the photodiodes in a CMOS image sensor 2750 is about 1 μm. In contrast, the oxide layer 2706 in the proposed oxide image sensor 2700 may be only several nanometers in thickness. Thus, the minimum spacing between sensing elements may be reduced below the wavelength of red length without suffering from any significant diffraction effect. This could help to improve the resolution of the image sensor 2700.
To-date, the inventors have found SiO2, HfO2 and ZrO2 dielectrics to exhibit light-sensing capability. These dielectric materials are already deployed in IC manufacturing and may be fully compatible with the mainstream fabrication technology.
Gate Oxide Reliability Rejuvenation
The gate oxide is a critical component of the metal oxide field effect transistor (MOSFET), the basic building block of nearly all ICs. Under continuous electrical stimulation, breakdown of the gate oxide after some time of IC operation. Gate oxide breakdown causes parametric drifts to occur in the MOSFET, which in turn impact circuit operation. Depending on the extent of such drifts (which in turn depends on the degree of gate oxide breakdown), circuit functionality failure may occur. The negative photoconductivity phenomenon may be exploited for restoring soft-breakdown gate oxide to its pre-breakdown state, since exposure to light may permanently eradicate the conducting filament formed during the breakdown.
Actual implementation of the gate oxide reliability rejuvenation method may include a light-emitting diode (LED) integrated on-chip next to the circuit of interest. During normal operation, the LED is turned off. Upon the detection of a gate oxide breakdown event, the LED is turned on for the required period of time for the purpose of restoring the breakdown oxide region.
While the invention has been particularly shown and described with reference to specific embodiments, it should be understood by those skilled in the art that various changes in form and detail may be made therein without departing from the spirit and scope of the invention as defined by the appended claims. The scope of the invention is thus indicated by the appended claims and all changes which come within the meaning and range of equivalency of the claims are therefore intended to be embraced.
| Number | Date | Country | Kind |
|---|---|---|---|
| 10201505778X | Jul 2015 | SG | national |
| Filing Document | Filing Date | Country | Kind |
|---|---|---|---|
| PCT/SG2016/050343 | 7/19/2016 | WO | 00 |
