Patent Application
20110121359
Embodiments of the present invention are related to nanoscale electronic devices, and, in particular, to re-configurable diode switches that can be implemented in crossbar arrays.
Significant research and development efforts are currently directed towards designing and manufacturing nanoscale electronic devices, such as nanoscale memories. Nanoscale electronics promise a number of advantages over microscale, photolithography-based electronics, including significantly reduced features sizes and the potential for self-assembly and for other relatively inexpensive, non-photolithography-based fabrication methods. However, the design and manufacture of nanoscale electronic devices present many new problems need to be addressed before large-scale commercial production of nanoscale electronic devices and incorporation of nanoscale electronic devices into microscale and larger-scale systems, devices, and products.
Studies of switching in nanometer-scale crossed-wire devices have previously reported that these devices could be reversibly switched and had an “on-to-off” conductance ratio of ˜103. These devices have been used to construct crossbar circuits and provide a promising route for the creation of ultra-high density nonvolatile memory. A series connection of cross-wire switches that can be used to fabricate a latch has also been demonstrated, such a latch is an important component for logic circuits and for communication between logic and memory. New logic families that can be constructed entirely from crossbar arrays of switches or as hybrid structures composed of switches and transistors have been described. These new logic families have the potential to dramatically increase the computing efficiency of CMOS circuits, thereby enabling performance improvements of orders of magnitude without having to shrink transistors, or to even replace CMOS for some applications if necessary. However, it is desired to improve the performance of the devices that are presently fabricated.
Various embodiments of the present invention are direct to nanoscale, reconfigurable, two-terminal electronic switches. In one embodiment, an electronic switch includes a first electrode, a second electrode, and an active region disposed between the first electrode and the second electrode and including at least one dopant. The switch can be re-configured to operate as a forward rectifier, a reverse rectifier, a shunted rectifier, or a head-to-head rectifier by positioning the dopant within the active region in order to control the flow of charge carriers through the switch.
Various embodiments of the present invention are directed to nanoscale, two-terminal, electronic switches, which are nonvolatile and combine reconfigurable diode rectifying states with memristive switching. A switch configured in accordance with embodiments of the present invention is composed of an active region sandwiched between two electrodes. The two interfaces between the active region and the electrodes are Schottky contacts. The active region is a diode that can be switched into one of four different rectifying states by applying an electrical field of an appropriate magnitude and polarity across the active region. The electric field changes the Schottky contacts at the interfaces to have Ohmic-like barriers and/or Schottky-like barriers, thus enabling the active region to be configured and the switch to operate as one of the four types of rectifiers: a forward rectifier, a reverse rectifier, a shunted rectifier, and a head-to-head rectifier. The active region remains in a particular rectifying state provided operating voltages applied to the switch do not exceed the magnitude of the electric field used to switch the rectifying state of the active region.
The detailed description is organized as follows. A description of two-terminal electronically actuated switches is provided in a first subsection. A description of switching the rectifying state of the switches is provided in a second subsection. Various materials that can be used to fabricate the switches are provided in a third subsection. Implementing the switches in crossbar arrays is provided in a fourth subsection. Finally, a switch composed of platinum electrodes and a TiO2 active region is described in a fifth subsection.
The active region 102 is composed of a primary active region, or layer, and a secondary active region, or layer. The primary active region comprises a thin film of a material that is electronically semiconducting or nominally electronically insulating and can also be a weakly ionic conductor. The primary active material is capable of transporting and hosting ions that act as dopants to control the flow of electrons through the switch 100. The basic mode of operation is to apply an electrical field of an appropriate magnitude and polarity across the active region 102. When the magnitude of the electrical field, also called a “drift field,” exceeds some threshold for enabling the motion of the dopants in the primary material the dopant can drift into or out of the primary material via ionic transport. The ionic species are specifically chosen from those that act as electrical dopants for the primary material, and thereby change the rectifying state of the primary active material. For example, a rectifier can be changed from low conductivity (i.e, an undoped semiconductor or insulator—switch “off” configuration) to high conductivity (doped to provide a higher conductivity—switch “on” configuration) or from high conductivity to low conductivity (switch “on” to switch “off”). In addition, the primary active material and the dopants are chosen such that the drift of the dopants into or out of the primary active material is possible but not too facile in order to ensure that the active region 102 remains in a particular rectifying state for a reasonable period of time, perhaps for many years at room temperature. This ensures that the active region 102 is nonvolatile. In other words, the active region 102 is memristive (i.e., memory resistive) and holds its rectifying state after the drift field has been removed. Applying a drift field with a large enough magnitude causes both electron current and dopant to drift, whereas applying biases with lower relative voltage magnitudes than the drift field causes negligible dopant drift enabling the switch to hold its rectifying state.
On the other hand, the secondary active material comprises a thin film that is a source of dopants for the primary active material. These dopants may be impurity atoms such as hydrogen or some other cation, such as alkali or transition metals, that act as electron donors for the primary active material. Alternatively, the dopants can be anion vacancies, which in the primary active material are charged and therefore are also electron donors for the lattice. It is also possible to drive the anions into the primary active material, which become electron acceptors or hole donors.
The primary active material can be nanocrystalline, nanoporous, or amorphous. The mobility of the dopants in such nanostructured materials is much higher than in bulk crystalline material, since diffusion can occur through grain boundaries, pores or through local structural imperfections in an amorphous material. Also, because the primary active material film is thin, the amount of time needed for dopants to diffuse into or out of region of the film to substantially change the film's conductivity is relatively rapid. For example, the time needed for a diffusive process varies as the square of the distance covered, so the time to diffuse one nanometer is one-millionth the time to diffuse one micrometer.
The primary active and secondary active materials of the active region 102 are contacted on either side by metal electrodes 104 and 106, or one of the electrodes can be composed of a semiconductor material and the other a metal. When the active region 102 is composed of semiconductor material, the contract between a metal electrode and the active region 102 depletes the active region 102 of free charge carriers. Thus, the active region 102 has a net charge that depends on the identity of the dopant which is positive in the case of electron donors and negative in the case of electron acceptors. The traditional description of electrode/semiconductor Schottky and Ohmic barriers is modified by the fact that the materials are structured at the nanoscale, and so the structural and electronic properties are not averaged over the large distances over which the theory of metal-semiconductor contracts have been developed. Thus, the undoped electrode/active region interfaces electronically resemble Schottky barriers and are called “Schottky-like barriers,” and the doped electrode/semiconductor interfaces electronically resemble Ohmic barriers and are called “Ohmic-like barriers.”
Conduction of electrons through the primary active material is via quantum mechanical tunneling through the Ohmic-like barrier. When the semiconducting material has a low dopant concentration or is essentially intrinsic, the tunneling barrier is a Schottky-like barrier, which is high and wide. Thus, the conductivity through the switching material 102 is low and the device 100 is in the “off” state. When a significant number of dopants have been injected into the semiconductor, the tunneling barrier is an Ohmic-like barrier and the width and perhaps the height of the tunneling barrier are diminished by the potential of the dopants, which results in an increase in the conductivity, and the device 100 is in the “on” state.
Each of the four rectifiers has a different dopant arrangement. When the dopant is located at or near an electrode/active region interface, the interface has an Ohmic-like barrier. Thus, charge carriers can readily tunnel through the Ohmic-like barrier into and out of the active region 102. On the other hand, an undoped portion of the active region 102 at or near an electrode/active region interface has a Schottky-like barrier that is either too high or wide to permit most charge carriers from tunneling through the active region 102.
Switching from one rectifier to another can be accomplished by applying an electric field of an appropriate magnitude and polarity across the active region 102. The electric field forces the dopants to drift into or out of the electrode/active region interface regions thus changing the rectifying state of the device 100. For example, as shown in
The ability of the charged species to diffuse into and out of the primary active material is substantially improved if one of the interfaces connecting the active region 102 to a metal or semiconductor electrode is non-covalently bonded. Such an interface may be caused by a void in the material or it may be the result of an interface that contains a material that does not form covalent bonds with the electrode, the primary active material, or both. This non-covalently bonded interface lowers the activation energy of the atomic rearrangements that are needed for drift of the dopants in the primary active material. This interface is essentially an extremely thin insulator, and adds very little to the total series resistance of the switch.
One potentially useful property of the primary active material is that it can be a weak ionic conductor. The definition of a weak ionic conductor depends on the application for which a switch 100 is designed. The mobility μd and diffusion constant D for a dopant in a lattice are directly proportional to one another as characterized by the Einstein relation:
D=μdkT
where k is Boltzmann's constant, and T is absolute temperature. Thus, if the mobility μd of a dopant in a lattice is high so is the diffusion constant D. In general, it is desired for the active region 102 of the switch 100 to maintain a particular rectifying state for an amount of time that may range from a fraction of a second to years, depending on the application. Thus, it is desired that the diffusion constant D be low enough to ensure a desired level of stability, in order to avoid inadvertently turning the active region 102 from one rectifier to another rectifier via ionized dopant diffusion, rather than by intentionally setting the state of the active region 102 with an appropriate voltage. Therefore, a weakly ionic conductor is one in which the dopant mobility μd and the diffusion constant D are small enough to ensure the stability or non-volatility of the active region 102 for as long as necessary under the desired conditions. On the other hand, strongly ionic conductors would have relatively larger dopant mobilities and be unstable against diffusion.
The active region 102 is non-volatile and re-configurable and exhibits diode rectifying states with memristive switching.
The plots 210-213 of the I-V characteristic curves reveal the response of the switch 100 to different voltage polarities and magnitudes. In particular, plot 210 reveals that when the switch 100 is configured as the forward rectifier 201, current flows from the first electrode 104 to the second electrode for positive polarity voltages exceeding a voltage 214 and resistance is large for negative polarity voltages. Plot 211 reveals that when the switch 100 is configured as the reverse rectifier 202, current flows from the second electrode 106 to the first electrode 104 for negative polarity voltages exceeding a voltage 215 and resistance is large for positive polarity voltages. Plot 212 reveals that when the switch 100 is configured as the shunted rectifier 203, current substantially flows undisturbed through the switch 100 for positive and negative polarity voltages with magnitudes exceeding voltages 216 and 217. Finally, plot 213 reveals that when the switch 100 is configured as a head-to-head rectifier 204, the resistance of the switch 100 is high for positive and negative polarity voltages between voltages 218 and 219. Note that plots 210-213 show only operating voltage ranges. In other words, the magnitudes of voltages applied to the rectifiers 201-204 represented in plots 210-213 are not large enough to change the rectifier to a different rectifier or destroy the switch 100.
The dopants are mobile under an appropriate drift field because the active region 102 may only be a few nanometers thick. The reconfiguration of the dopant profiles due to the drift of dopants under a drift field leads to electrical switching between the four rectifiers. As shown in
Opening is switching between the reverse rectifier 202 and the head-to-head rectifier 204. In this case, the undoped interface 220 remains unchanged and only the doped interface 222 is switched. The undoped interface contains few dopants and remains rectifying instead of Ohmic-like. A bias of an appropriate polarity and magnitude on the first electrode 104 forces dopants away from the interface 222 and switches the reverse rectifier 202 into the head-to-head rectifier 204, and vice versa. The switching between the forward rectifier 201 and the back-to-back rectifier 204 is also opening.
Inverting between the forward rectifier 201 and the reverse rectifier 202 involves simultaneously applying oppositely polarized biases to the electrodes 104 and 106. For example, switching from the forward rectifier 201 to the reverse rectifier 202 is accomplished by applying oppositely polarized biases to the electrodes 104 and 106 to forces dopants away from the interface 220 and at the same time attracts dopants to the interface 222. Switching from the reverse rectifier 202 to the forward rectifier 201 is accomplished by applying oppositely polarized biases to the electrodes 104 and 106 to force dopants away from the interface 222 and at the same time attract dopants to the interface 220. Therefore, the dopant profile across the active region 102 is essentially inverted and so is the rectifying orientation, resulting in a switching between a reverse rectifier and a forward rectifier.
The electrodes 104 and 106 can be composed of platinum, gold, silver, copper, or any other suitable metal, metallic compound (e.g. some perovskites such as BaTiO3 and Ba1-xLaxTiO3) or semiconductor. The primary and secondary active materials of the active region 102 can be oxides, sulfides, selenides, nitrides, phosphides, arsenides, chlorides, and bromides of the transition and rare earth metals, with or without the alkaline earth metals being present. In addition, there are various alloys of these compounds with each other, which can have a wide range of compositions if they are mutually soluble in each other. In addition, the active region 102 can be composed of mixed compounds, in which there are two or more metal atoms combined with some number of electronegative elements. The dopants can be anion vacancies or different valence elements doped in the active region 102. One combination of materials is a primary active material that is undoped and stoichiometric, and thus a good insulator, combined with a secondary source/sink of the same or related parent material that either contains a large concentration of anion vacancies or other dopants that can drift into the primary material under the application of an appropriate bias.
The active region 102 can be composed of oxides that contain at least one oxygen atom (O) and at least one other element. In particular, the active region 102 can be composed of titania (TiO2), zirconia (ZrO2), and hafnia (HfO2). These materials are compatible with silicon (Si) integrated circuit technology because they do not create doping in the Si. Other embodiments for the active region 102 include alloys of these oxides in pairs or with all three of the elements Ti, Zr, and Hf present. For example, the active region 102 can be composed of TixZryHfzO2, where x+y+z=1. Related compounds include titanates, zirconates, and hafnates. For example, titanates includes ATiO3, where A represents one of the divalent elements strontium (Sr), barium (Ba) calcium (Ca), magnesium (Mg), zinc (Zn), and cadmium (Cd). In general, the active region 102 can be composed of ABO3, where A represents a divalent element and B represents Ti, Zr, and Hf. The active region 102 can also be composed of alloys of these various compounds, such as CaaSrbBacTixZryHfzO3, where a+b+c=1 and x+y+z=1. There are also a wide variety of other oxides of the transition and rare earth metals with different valences that may be used, both individually and as more complex compounds. In each case, the mobile dopant can be an oxygen vacancy or an aliovalent element doped into the active region 102. The oxygen vacancies effectively act as dopants with one shallow and one deep energy level. Because even a relatively minor nonstoichiometry of about 0.1% oxygen vacancies in TiO2-x is approximately equivalent to 5×1019 dopants/cm3, modulating oxygen vacancy profiles have strong effect on electron transport.
In other embodiments, the active region 102 can be a sulfide or a selenide of the transition metals with some ionic bonding character, essentially the sulfide and selenide analogues of the oxides described above.
In other embodiments, the active region 102 can be a semiconducting nitride or a semiconducting halide. For example, semiconducting nitrides include AlN, GaN, ScN, YN, LaN, rare earth nitrides, and alloys of these compounds and more complex mixed metal nitrides, and semiconducting halides include CuCl, CuBr, and AgCl. The active region 102 can be a phosphide or an arsenide of various transition and rare earth metals. In all of these compounds, the mobile dopant can be an anion vacancy or an aliovalent element.
A variety of dopants can be used and are selected from a group consisting of hydrogen, alkali, and alkaline earth cations, transition metal cations, rare earth cations, oxygen anions or vacancies, chalcogenide anions or vacancies, nitrogen anions or vacancies, pnictide anions or vacancies, or halide anions or vacancies.
In other embodiments, the active region 102 can also be composed of a wide variety of semiconductor materials including various combinations of direct and indirect semiconductors. A direct semiconductor is characterized by the valence band maximum and the conduction band minimum occurring at the same wavenumber. In contrast, indirect semiconductors are characterized by the valence band maximum and the conduction band minimum occurring at different wavenumbers. The indirect and direct semiconductors can be elemental and compound semiconductors. Indirect elemental semiconductors include Si and germanium (Ge), and compound semiconductors include III-V materials, where Roman numerals III and V represent elements in the IIIa and Va columns of the Periodic Table of the Elements. Compound semiconductors can be composed of column IIIa elements, such as aluminum (Al), gallium (Ga), and indium (In), in combination with column Va elements, such as nitrogen (N), phosphorus (P), arsenic (As), and antimony (Sb). Compound semiconductors can also be further classified according to the relative quantities of III and V elements. For example, binary semiconductor compounds include semiconductors with empirical formulas GaAs, InP, InAs, and GaP; ternary compound semiconductors include semiconductors with empirical formula GaAsyP1-y, where y ranges from greater than 0 to less than 1; and quaternary compound semiconductors include semiconductors with empirical formula InxGa1-xAsyP1-y, where both x and y independently range from greater than 0 to less than 1. Other types of suitable compound semiconductors include II-VI materials, where II and VI represent elements in the IIb and VIa columns of the periodic table. For example, CdSe, ZnSe, ZnS, and ZnO are empirical formulas of exemplary binary II-VI compound semiconductors.
The dopants can be p-type impurities, which are atoms that introduce vacant electronic energy levels called “holes” to the electronic band gaps of the active region 102. These impurities are also called “electron acceptors.” The dopants can be n-type impurities, which are atoms that introduce filled electronic energy levels to the electronic band gap of the active region 102. These impurities are called “electron donors.” For example, boron (B), Al, and Ga are p-type impurities that introduce vacant electronic energy levels near the valence band of Si; and P, As, and Sb are n-type impurities that introduce filled electronic energy levels near the conduction band of Si. In III-V compound semiconductors, column VI impurities substitute for column V sites in the III-V lattice and serve as n-type impurities, and column II impurities substitute for column III atoms in the III-V lattice to form p-type impurities. Moderate doping of the active region 102 can have impurity concentrations in excess of about 1015 impurities/cm3 while more heavy doping of the active region 102 can have impurity concentrations in excess of about 1019 impurities/cm3.
The switch 100 can be implemented at nanowire intersections of nanowire crossbar arrays.
Although individual nanowires in
The layers can be fabricated by mechanical nanoimprinting techniques. Alternatively, nanowires can be chemically synthesized and can be deposited as layers of approximately parallel nanowires in one or more processing steps, including Langmuir-Blodgett processes. Other alternative techniques for fabricating nanowires may also be employed. Thus, a two-layer nanowire crossbar comprising first and second layers, as shown in
Oxygen vacancies in TiO2 operate as n-type dopants transforming a wide band-gap oxide into a material that operates as an electrically conductive doped semiconductor. As described above with reference to
Any pair of electrodes 505-512 form a switch, from which an I-V curve can be obtained, as shown in
The initial resistance state of the switches, i.e. the oxygen vacancy profile, in large degree determines the rectifying state of the switch. In practice, the oxygen vacancy profile can be controlled by engineering the structure and/or the fabrication condition of the active region, such as deposition gas species, annealing environment, inserting pure metal (e.g., Ti) at the interface. The following description provides results representing the realization of all three types of switching in real switches.
Circuit models accompanying the following experimental results include memristors. The term “memristor” is short for “memory resistor.” Memristors are a class of passive two-terminal circuit elements that maintain a functional relationship between the time integrals of current and voltage. This results in resistance varying according to the device's memristance function. Specifically engineered memristors provide controllable resistance useful for switching current. The memristor is a special case in so-called “memristive systems,” a class of mathematical models useful for certain empirically observed phenomena, such as the firing of neurons. The definition of the memristor is based solely on fundamental circuit variables, similar to the resistor, capacitor, and inductor. Unlike those more familiar elements, the necessarily nonlinear memristors may be described by any of a variety of time-varying functions. As a result, memristors do not belong to Linear Time-Independent circuit models. A linear time-invariant memristor is simply a conventional resistor.
The memristor is formally defined as a two-terminal element in which the magnetic flux Φm between the terminals is a function of the amount of electric charge q that has passed through the device. Each memristor is characterized by its memristance function describing the charge-dependent rate of change of flux with charge as follows:
Based on Faraday's law of induction that magnetic flux Φm is the time integral of voltage, and charge q is the time integral of current, the memristance can be written as
Thus, the memristance is simply charge-dependent resistance. When M(q) is constant, the memristance reduces to Ohm's Law R=VII. When M(q) is not constant, the equation is not equivalent because q and M(q) vary with time. Solving for voltage as a function of time gives:
V(t)=M(q(t))I(t)
This equation reveals that memristance defines a linear relationship between current and voltage, as long as charge does not vary. However, nonzero current implies instantaneously varying charge. Alternating current may reveal the linear dependence in circuit operation by inducing a measurable voltage without net charge movement, as long as the maximum change in q does not cause change in M. Furthermore, the memristor is static when no current is applied. When I(t) and V(t) are 0, M(t) is constant. This is the essence of the memory effect.
An oxygen vacancy is the only dopant used for the concept demonstration of the three the switching types described above. However, in principle, other dopants (e.g., C and N) with different properties, such as mobility, charge, and diffusivity, can be introduced to the system to intestinally build an asymmetric device. Only one dopant like oxygen vacancies can be sufficient for the inverting switch since the device is symmetric and equal but opposite changes at the two interfaces are needed for this type of switching. As for opening and shunting switching, one interface is heavily reduced for shunting or oxidized for opening in order to minimize the change at that interface during switching. A different dopant that is much less mobile than oxygen vacancies at the unchanged interface would serve that purpose even better.
The foregoing description, for purposes of explanation, used specific nomenclature to provide a thorough understanding of the invention. However, it will be apparent to one skilled in the art that the specific details are not required in order to practice the invention. The foregoing descriptions of specific embodiments of the present invention are presented for purposes of illustration and description. They are not intended to be exhaustive of or to limit the invention to the precise forms disclosed. Obviously, many modifications and variations are possible in view of the above teachings. The embodiments are shown and described in order to best explain the principles of the invention and its practical applications, to thereby enable others skilled in the art to best utilize the invention and various embodiments with various modifications as are suited to the particular use contemplated. It is intended that the scope of the invention be defined by the following claims and their equivalents:
| Filing Document | Filing Date | Country | Kind | 371c Date |
|---|---|---|---|---|
| PCT/US08/09246 | 7/31/2008 | WO | 00 | 1/26/2011 |
