The present invention relates to a variable resistance nonvolatile storage device having memory cells each of which includes a transistor and a variable resistance element whose resistance value reversibly changes based on electrical signals, and a method of forming such memory cells.
In recent years, research and development of nonvolatile storage devices having memory cells formed with variable resistance elements has moved forward. A variable resistance element is an element with a property that its resistance value reversibly changes based on electrical signals, and can store data corresponding to the resistance value in a nonvolatile manner.
Commonly known as a nonvolatile storage device that includes variable resistance elements is a nonvolatile storage device which includes memory cells known as 1T1R memory cells arranged in a matrix at positions at which bit lines intersect with word lines and source lines that are arranged orthogonally to the bit lines. Here, the 1T1R memory cells refer to memory cells each of which is formed by connecting in series a MOS (metal oxide semiconductor) transistor and a variable resistance element.
Patent Literature 1 discloses a nonvolatile storage device including 1T1R memory cells in which oxides having a perovskite-type crystalline structure are used as variable resistance elements.
A memory cell 1011 is formed by electrically connecting a transistor 1006 and a variable resistance element 1010 in series.
The transistor 1006 includes, on a semiconductor substrate 1001, a source region 1002 serving as a first diffusion layer region, a drain region 1003 serving as a second diffusion layer region, and a gate electrode 1005 formed on a gate oxide film 1004.
The variable resistance element 1010 is formed by having a variable resistance layer 1008 between a lower electrode 1007 and an upper electrode 1009. Here, the variable resistance layer 1008 is a layer whose resistance value changes upon voltage application.
The drain region 1003 and the lower electrode 1007 are electrically connected.
The upper electrode 1009 is connected to a metal line serving as a bit line 1012, the gate electrode 1005 is connected to a word line, and the source region 1002 is connected to a metal line serving as a source line 1013.
Here, although Patent Literature 1 discloses Pr1-xCaxMnO3 (PCMO), La1-xCaxMnO3 (LCMO) and so on as the materials of the variable resistance layer 1008, there is no particular mention of electrode materials.
Furthermore, as to a method for writing into the memory cell 1011, PATENT LITERATURE 1 discloses that: application of a pulse voltage Vpp, a pulse voltage Vss, and a pulse voltage having a predetermined voltage amplitude Vwp to the upper electrode 1009, the source region 1002, and the gate electrode, respectively, changes a state of the memory cell 1011 from a low resistance state to a high resistance state; and inversely, application of the pulse voltage Vss, the pulse voltage Vpp, and the pulse voltage having a predetermined voltage amplitude Vwe to the upper electrode 1009, the source region 1002, and the gate electrode, respectively, changes the state from the high resistance state to the low resistance state.
Patent Literature 2 discloses a nonvolatile storage device which includes 1T1R memory cells having variable resistance elements that exhibit changes in the resistance based on a principle that is different from that of the above-mentioned variable resistance elements that exhibit changes in the resistance according to electrical signals. This storage device is called a phase-change memory.
The phase-change memory stores data by making use of the fact that a phase-change material called a chalcogenide material has different resistance between the crystalline state and the amorphous state. Data is rewritten by changing the state of the phase-change material through passage of a current therein so as to generate heat near its melting point. A resistance change to high resistance state (amorphization) called a reset operation is performed by control of maintaining the phase-change material at a relatively high temperature, and a resistance change to low resistance state (crystallization) called a set operation is performed by control of maintaining the phase-change material at a relatively low temperature for a sufficient period.
Moreover, Patent Literature 2 discloses that with the phase-change memory, a current necessary for data rewriting is different between the reset operation and the set operation, and the reset operation requires a relatively large current.
A memory cell 1021 has a 1T1R structure in which a storage unit 1022 and an NMOS transistor 1027 are used. The NMOS transistor 1027 includes N-type diffusion layer regions 1029 and 1030 respectively corresponding to a source and a drain, and a gate electrode 1031 located between the N-type diffusion layer regions 1029 and 1030.
The storage unit 1022 includes a phase change element 1024, a second metal line layer 1023 above the phase change element 1024, and a contact via 1025 and a first metal line layer 1026 below the phase change element 1024, and is connected to the N-type diffusion layer region 1029 of the NMOS transistor 1027.
The other N-type diffusion layer region 1030 of the NMOS transistor 1027 is connected to a third metal line layer 1028 via each of line layers.
Here, the second metal line layer 1023, the third metal line layer 1028, and the gate electrode 1031 of the NMOS transistor 1027 correspond to a source line, a bit line, and a word line, respectively.
Patent Literature 2 discloses adopting into a phase-change memory device a mechanism for controlling the source line and switching the current passage direction in a set operation and a reset operation.
In the reset operation requiring passage of a relatively large current, the source line is set to a predetermined high level and the bit line to a low level, and in the set operation requiring only a relatively small current, the bit line is set to a predetermined high level and the source line to a low level.
With such settings, the current direction in the reset operation is a direction in which the electric potential at the source of the NMOS transistor 1027 of the memory cell (in this case, corresponding to the electric potential at the N-type diffusion layer region 1030) is maintained at a low level almost equal to the electric potential at the semiconductor substrate. This reduces the influence of a so-called substrate bias effect in the MOS transistor, and thus the reset operation is performed with a driving capability of the transistor being high (a large current can be obtained).
On the other hand, the current direction in the set operation is a direction in which the electric potential at the source of the NMOS transistor 1027 of the memory cell (in this case, corresponding to the electric potential at the N-type diffusion layer region 1029) rises to a voltage value that is determined according to a divided voltage relationship between an on-resistance value of the NMOS transistor 1027 and a resistance value of the phase-change element 1024. This increases the influence of the substrate bias effect in the MOS transistor, and thus the set operation is performed with the current flowing in the transistor kept relatively small.
The above structure makes it easier to separately supply a current of different magnitudes suitable for each of the set operation and the reset operation, thereby allowing the outcome of each operation to be stably obtained.
Generally, to form a high density memory cell array, memory cells need to be formed in the smallest possible areas, and thus it is important to form a transistor, as well as a variable resistance element which is another structural element of each memory cell, in the smallest possible areas.
An effective way of forming a transistor in the smallest possible area is to make a gate length L of the transistor as short as possible and a gate width W of the transistor as small as possible in an efficient manner.
The following considers application of the above structure to the nonvolatile storage device disclosed in PATENT LITERATURE 1.
According to PATENT LITERATURE 1, the nonvolatile storage device shown in
Here, the electric potential at the source region 1002 serving as the first diffusion layer region of the transistor 1006 (in this case, the source region 1002 functions as the source of the transistor 1006) is almost equal to the electric potential at the semiconductor substrate 1001, that is, 0 V, and the substrate bias effect in the transistor 1006 small.
On the other hand, the nonvolatile storage device changes the state of the memory cell 1011 from the high resistance state to the low resistance state (resistance change to low resistance state) by setting the bit line 1012 to 0 V and the source line 1013 to Vpp.
Here, the electric potential at the drain region 1003 that is the second diffusion layer region (in this case, the drain region 1003 functions as the source of the transistor 1006) rises to a voltage determined according to a divided voltage relationship between the resistance value of the variable resistance element 1010 and the on-resistance of the transistor 1006, and the substrate bias effect in the transistor 1006 becomes greater than in the case of the resistance change to high resistance state.
As described above, it is rational, in terms of forming the transistor of the memory cell in an optimum size, to perform the resistance change to high resistance state, which requires a current larger than for the resistance change to low resistance state, by using a current flowing in the direction in which the substrate bias effect in the transistor is smaller, because that way the transistor does not need to have a driving capability with an unnecessary margin.
It is to be noted that the semiconductor device disclosed in Patent Literature 2 also adopts the same concept that the reset operation requiring a larger current is performed using a current flowing in a direction in which the substrate bias effect in the transistor is smaller.
The inventors of the present application have examined, as one of variable resistance nonvolatile storage devices, a variable resistance nonvolatile storage device which includes 1T1R memory cells each having a variable resistance layer comprising an oxygen-deficient transition metal oxide.
Here, the oxygen-deficient oxide is an oxide whose composition is deficient in oxygen compared to its stoichiometric composition. Taking tantalum, one of transition metals, as an example, Ta2O5 is an oxide having a stoichiometric composition. Ta2O5 contains oxygen 2.5 times more than tantalum, and has an oxygen content percentage of 71.4%. An oxide having an oxygen content percentage lower than 71.4%, that is, a tantalum oxide having a non-stoichiometric composition which satisfies 0<x<2.5 when expressed as TaOx, is called an oxygen-deficient tantalum oxide.
To facilitate description of problems to be solved, the following describes some properties which have been measured of a variable resistance element having an oxygen-deficient tantalum oxide as a variable resistance layer.
As
On the other hand, in gradual application of a negative voltage to the upper electrode 3303 with reference to the lower electrode 3301 (equivalent to application of a positive voltage to the lower electrode 3301 with reference to the upper electrode 3303) in the high resistance state, the current rapidly increases when the negative voltage exceeds the voltage at the point B. That is, it shows a change from the high resistance state to a low resistance state (resistance change to low resistance state).
The variable resistance element exhibiting the characteristics shown in
Further, the characteristics shown in
As described above, it is rational, in terms of forming the transistor of the memory cell in an optimum size, to perform the resistance change to high resistance state, which requires a larger current, by applying a voltage of a polarity that reduces the substrate bias effect in the transistor, and to perform the resistance change to low resistance state, which requires a smaller current, by applying a voltage of the reversed polarity.
However, the inventors of the present application have discovered during the examination that the voltage application direction (driving polarity) stably causing a resistance change in one direction (resistance change to low resistance state or resistance change to high resistance state) is not always the same, and that the driving polarity is different in some of the variable resistance elements which are made of the same materials, that is, Pt for the upper and lower electrodes and an oxygen-deficient tantalum oxide for the variable resistance layer.
For instance, it has been verified that a variable resistance element undergoes the resistance change to low resistance state upon application of a pulse voltage of +2.0 V in amplitude and 100 ns in width between the upper electrode 3303 and the lower electrode 3301, and the resistance change to high resistance state upon application of a pulse voltage of −2.6 V in amplitude and 100 ns in width between the upper electrode 3303 and the lower electrode 3301, given that a voltage applied between the upper electrode 3303 and the lower electrode 3301 is positive when the voltage at the higher electrode 3304 is higher than the voltage at the lower electrode 3301.
It has also been verified that another variable resistance element undergoes the resistance change to low resistance state upon application of a pulse voltage of −2.0 V in amplitude and 100 ns in width between the upper electrode 3303 and the lower electrode 3301, and the resistance change to high resistance state upon application of a pulse voltage of +2.7 V in amplitude and 100 ns in width between the upper electrode 3303 and the lower electrode 3301, given that a voltage applied between the upper electrode 3303 and the lower electrode 3304 is positive when the voltage at the upper electrode 3303 is higher than the voltage at the lower electrode 3301.
As shown in
This relationship between the direction of the resistance change and the polarity of the applied voltage is called A mode for descriptive purposes.
As shown in
This relationship between the direction of the resistance change and the polarity of the applied voltage is called B mode for descriptive purposes. The hysteretic current-voltage characteristics shown in
It is to be noted that the above pulse voltage values denote output voltage values set for a pulse generator, and that an effective voltage value applied between the end terminals of the variable resistance element is considered to be a voltage value smaller than a corresponding one of the pulse voltage values due to a voltage drop across a measurement system.
In the variable resistance element showing such a result, the upper electrode 3303 and the lower electrode 3301 both comprise Pt, and the variable resistance layer 3302 which comprises the oxygen-deficient tantalum oxide and is located between the upper and lower electrodes has a top-bottom symmetrical relationship with the electrodes electrically.
Accordingly, it is not always self-evident as to which of the A mode and the B mode appears as the resistance change characteristics, and the mode is determined based on empirical rules or experimental measurement results. It is expected that these phenomena are determined by some sort of anisotropy factors unexplained in the resistance change mechanism.
In the case where the A mode and the B mode appear indeterminately, the following problems are expected in forming a 1T1R memory device.
The first problem is that the size of the transistor cannot be optimized.
If the resistance change characteristics can be determined as either the A mode or the B mode, the transistor can be formed in a minimum size that allows driving of a current necessary for the resistance change to high resistance state assuming that the transistor operates under a condition that causes a small substrate bias effect according to a conventionally-known concept.
However, if the mode is indeterminate, it is necessary to form the transistor in a size sufficient for driving the current necessary for the resistance change to high resistance state, considering also that the transistor operates under a condition that causes a large substrate bias effect. For this reason, unlike in the case where the mode is determined, it is necessary to make the gate width W of the transistor wide. This is not desirable as the wide gate width is a significant impediment to miniaturization of memory cells.
The second problem is that it is necessary to manage information identifying the mode of the resistance change characteristics.
When the mode is indeterminate, the correspondence is indeterminate between the polarity of a voltage to be applied for changing a resistance state, and a resistance state (high resistance state or low resistance state) to be read after the voltage application. This creates a need of information which identifies the mode in order to actually use the variable resistance element as a storage element.
For example, in the case where the same mode appears in a lot or a slice, a storage element dedicated for management is provided in a chip. Then, in the manufacturing stage, identification information indicating in which one of the A mode and the B mode the variable resistance element undergoes a resistance change is recorded in the storage element dedicated for management. Further, in the use stage, according to the identification information, the polarity of the applied voltage is reversed in a write operation or the polarity of output data is reversed in a read operation.
It appears that this allows the actual use of the variable resistance element as the storage element. However, it is not desirable because this complicates the circuit configuration and the control method. Further, in the case where different modes appear across smaller units in a lot or a slice, such as memory cells, it is practically impossible to provide, for each memory cell, a storage element dedicated for management and record the identification information of the mode in the storage element.
The present invention has been conceived in consideration of the above problems, and it is an object of the present invention to provide a technique that enables a 1T1R nonvolatile storage device including a variable resistance element to control the appearance of the A mode and the B mode of the resistance change characteristics of the variable resistance element and that enables designing of a memory cell with an optimum transistor size.
In order to solve the above problems, the variable resistance nonvolatile storage device according to an aspect of the present invention is a variable resistance nonvolatile storage device including: a semiconductor substrate; a variable resistance element including: a first electrode; a second electrode; and a variable resistance layer which is provided between the first electrode and the second electrode and in contact with the first electrode and the second electrode and which has a resistance value that reversibly varies based on each of voltage signals which have different polarities and are applied between the first electrode and the second electrode; and a MOS transistor formed on a main surface of the semiconductor substrate, wherein the first electrode comprises one of tantalum nitride and tungsten, the second electrode comprises at least one metal selected from platinum, iridium, and palladium, the variable resistance layer includes: a first region which is in contact with the first electrode and contains a first oxygen-deficient transition metal oxide having a composition MOx; and a second region which is in contact with the second electrode and contains a second oxygen-deficient transition metal oxide having a composition MOy (where x<y), and a drain of the MOS transistor is connected with either the first electrode or the second electrode included in the variable resistance element to form a memory cell so that a substrate bias effect in the MOS transistor is smaller when one of the voltage signals having a polarity for changing the variable resistance layer into a high resistance state is applied to the MOS transistor and the variable resistance element, than when an other one of the voltage signals having a polarity for changing the variable resistance layer into a low resistance state is applied to the MOS transistor and the variable resistance element.
Further, the second electrode may comprise a material having a standard electrode potential which is higher than a standard electrode potential of a transition metal of the first and second oxygen-deficient transition metal oxides, and the first electrode may comprise a material having a standard electrode potential which is lower than the standard electrode potential of the second electrode.
Furthermore, the MOS transistor may be an N-type MOS transistor including, on the main surface of the semiconductor substrate: a first N-type diffusion layer region; a gate; and a second N-type diffusion layer region formed on a side of the gate opposite the first N-type diffusion layer region, and the first electrode may be connected with the first N-type diffusion layer region of the N-type MOS transistor to form the memory cell.
Moreover, the MOS transistor may be a P-type MOS transistor including: an N well formed on the main surface of the semiconductor substrate; a first P-type diffusion layer region; a gate; a second P-type diffusion layer region formed on a side of the gate opposite the first P-type diffusion layer region, the first P-type diffusion layer region, the gate, and the second P-type diffusion layer region being formed in the N well, and the second electrode may be connected with the first P-type diffusion layer region of the P-type MOS transistor to form the memory cell.
According to the nonvolatile storage device according to an aspect of the present invention, a variable resistance nonvolatile storage element is used in which: a first region containing an oxygen-deficient transition metal oxide which does not easily exhibit a resistance change due to a lower oxygen content percentage is provided in contact with a first electrode; and a second region containing an oxygen-deficient transition metal oxide which easily exhibits a resistance change due to a higher oxygen content percentage is provided in contact with a second electrode. Thus, the voltage application direction (driving polarity) which causes resistance changes can be uniquely determined for each of memory cells such that: the resistance change to high resistance state is performed upon application of a positive voltage to the second electrode with reference to the first electrode; and the resistance change to low resistance state is performed upon application of a positive voltage to the first electrode with reference to the second electrode.
Generally, the resistance change to high resistance state of the variable resistance element requires a larger driving current compared to the resistance change to low resistance state, because a voltage needs to be generated to cause a resistance change in the variable resistance element in a low resistance state.
Thus, in the case of forming each memory cell with the variable resistance element and an N-type MOS transistor, the first electrode of the variable resistance element is connected with a first N-type diffusion layer region of the N-type MOS transistor. With this connection, when the variable resistance element is to perform the resistance change to high resistance state, a second N-type diffusion layer region of the N-type MOS transistor is grounded and thus it is possible to supply the driving current to the variable resistance element using a ground bias which causes a smaller substrate bias effect in the N-type MOS transistor.
Moreover, when forming each memory cell with the variable resistance element and a P-type MOS transistor, the second electrode of the variable resistance element is connected with a first P-type diffusion layer region of the P-type MOS transistor. With this connection, when the variable resistance element is to perform the resistance change to high resistance state, a second N-type diffusion layer region of the P-type MOS transistor is connected to a power source and thus it is possible to supply the driving current to the variable resistance element using a power bias which causes a smaller substrate bias effect in the P-type MOS transistor.
As a result, it is no longer necessary to form the transistor in a size sufficient for driving the current necessary for the resistance change to high resistance state considering that the transistor operates under a condition that causes a large substrate bias effect. As a result, it is possible to design the memory cells with an optimum transistor size.
In other words, it is possible to provide, in a smaller layout area, a variable resistance nonvolatile storage device in which 1T1R memory cells are used, and to achieve increase in the integration degree and cost reduction.
The following describes in detail an embodiment of the present invention with reference to the drawings.
The variable resistance nonvolatile storage device according to an embodiment of the present invention is a 1T1R nonvolatile storage device in which a variable resistance element and a MOS transistor are connected in series, and which fixes the mode of the resistance change characteristics of the variable resistance element and optimizes the structure of the MOS transistor according to the mode fixed.
The following describes, as preparation, basic data regarding three types of variable resistance elements used for the variable resistance nonvolatile storage device according to the present invention.
These variable resistance elements are formed by providing, between two electrodes, a variable resistance layer comprising one of the following oxides: an oxygen-deficient tantalum oxide; an oxygen-deficient hafnium oxide; and an oxygen-deficient zirconium oxide.
These variable resistance elements were invented by the inventors of the present application for the purpose of obtaining a nonvolatile storage element which has reversible and stable rewriting characteristics and which makes use of a resistance change phenomenon. Such variable resistance elements are described in detail in the following related patent applications: Japanese Patent Application No. 2007-149032 (Patent Literature); Japanese Patent Application No. 2008-180946 (Patent Literature); and Japanese Patent Application No. 2008-180944 (Patent Literature).
A feature of these variable resistance elements that the resistance change characteristics can be fixed to either the aforementioned A mode or B mode is applied to the variable resistance nonvolatile storage device according to the present invention. Hereinafter, the above-mentioned related patent applications are partially cited for purposes of illustration.
It is to be noted that the term “variable resistance element” and the term “variable resistance nonvolatile storage element” (or briefly, “nonvolatile storage element”) are used synonymously in the present DESCRIPTION.
First, the following describes the first experiment regarding a variable resistance nonvolatile storage element which uses an oxygen-deficient tantalum oxide as the variable resistance layer and which performs a bipolar operation.
As shown in
Here, the variable resistance layer 104 includes: a first tantalum containing layer (hereinafter referred to as “first tantalum oxide layer”) 104a having a lower oxygen content percentage; and a second tantalum containing layer 104b (hereinafter referred to as “second tantalum oxide layer”) formed on the first tantalum oxide layer 104a and having a higher oxygen content percentage.
To drive the variable resistance element 100, a voltage satisfying a predetermined condition is applied between the lower electrode 103 and the upper electrode 105 by an external power source. According to the voltage application direction, the resistance value of the variable resistance layer 104 of the variable resistance element 100 increases or decreases reversibly. For example, when a pulse voltage larger than a predetermined threshold voltage is applied, the resistance value of the variable resistance layer 104 increases or decreases, whereas when a pulse voltage smaller than the predetermined threshold voltage is applied, there is no change in the resistance value of the variable resistance layer 104.
The materials of the lower electrode 103 and the upper electrode 105 are, for example, Pt (platinum), Ir (iridium), Pd (palladium), Ag (silver), and Cu (copper).
Although the substrate 101 may be a single-crystal silicon substrate or a semiconductor substrate, it is not limited to these substrates. The variable resistance layer 104 can be formed at a relatively low substrate temperature and thus can be formed on a resin material and the like.
Next, a manufacturing method of the variable resistance element 100 used in the present experiment is described with reference to
First, as shown in
Next, as shown in
Subsequently, by sputtering, a Pt thin film having a thickness of 150 nm is formed on the second tantalum oxide layer 104b as the upper electrode 105.
Lastly, photolithography processing creates a pattern 106 using a photoresist, and dry etching produces an element region 107 as shown in
With the above-described manufacturing method, elements A to C were manufactured. The following describes the details.
First, the layered structure of the substrate 101, the oxide layer 102, and the lower electrode 103 comprising Pt was formed in the above-described manner. After that, the first tantalum oxide layer 104a was formed on the lower electrode 103 by reactive sputtering, that is, a process of sputtering a tantalum target in an argon gas and an oxygen gas.
The film forming condition here was as follows: vacuum (back pressure) in the sputtering apparatus was approximately 7×10−4 Pa before sputtering; power was 250 W when sputtering; the total gas pressure of the argon gas and the oxygen gas was 3.3 Pa; the flow ratio of the oxygen gas was 3.4%; the preset temperature of the substrate was 30° C.; and the time period for film formation was seven minutes. With this, the first tantalum oxide layer 104a having an oxygen content percentage of about 58 at %, i.e., TaO1.4, was deposited with a thickness of 30 nm.
The manufacturing of the elements A to C involved the formation of the first tantalum oxide layer 104a and the second tantalum oxide layer 104b, followed by the formation of the upper electrode 105, both performed in the sputtering apparatus. More specifically, after depositing the first tantalum oxide layer 104a, a shutter was inserted between the tantalum target and the oppositely placed substrate 101 under the same sputtering condition, including the condition of the gas pressure and the power, for example, and it was kept in this state for a predetermined time period.
This caused oxidation of the top surface of the first tantalum oxide layer 104a due to oxygen plasma. As a result, the second tantalum oxide layer 104b having an oxygen content percentage higher than that of the first tantalum oxide layer 104a was formed on the surface of the first tantalum oxide layer 104a.
Thereafter, the upper electrode 105 comprising Pt was formed on the second tantalum oxide layer 104b in the above-described manner.
Then, the element region 107 was formed through the photolithography processing. It is to be noted that the element region 107 of the elements A to C has a circular pattern with a diameter of 3 μm.
In this experiment, the elements A to C were manufactured with different time periods for the oxidation using the oxygen plasma (oxygen plasma exposure time). Table 1 summarizes the initial resistance value of each element manufactured, and the thickness and the oxygen content percentage x of the first tantalum oxide layer (TaOx layer) as well as the thickness and the oxygen content percentage y of the second tantalum oxide layer (TaOy layer) which were measured through X-ray reflectivity measurement (described later).
It is to be noted that the oxygen plasma exposure time of the element A being zero minute means that the deposition of the first tantalum oxide layer 104a was immediately followed by the deposition of Pt as the upper electrode 105, without exposure to the oxygen plasma.
The following describes the characteristics and so on of the variable resistance elements manufactured in the above manner.
First, the initial resistance of the variable resistance layer 104 of each of the elements A to C is measured and the result is examined. Here, the initial resistivity of the variable resistance layer 104 of each element was determined by measuring a current flowing upon application of a weak voltage of 50 mV that is lower than a threshold voltage (1 V, for example) between the lower electrode 103 and the upper electrode 105 of each element.
Table 1 shows that the element A (oxygen plasma exposure time: zero minute) has resistance of 1.7Ω, the element B (oxygen plasma exposure time: 0.5 minute) has resistance of 650Ω, and the element C (oxygen plasma exposure time: 1 minute) has resistance of 1850Ω. This indicates that the resistance value of the variable resistance layer 104 rises with increase in the oxygen plasma exposure time.
Probably, the resistance of 1.72Ω seen in the element A manufactured by zero-minute oxygen plasma exposure is equivalent to contact resistance, and the large resistance values above several hundred ohms seen in the elements B and C are mostly due to the second tantalum oxide layer 104b formed by the oxygen plasma processing.
Next, the following describes the characteristics observed when a resistance change occurs upon application of electrical pulses to the elements A to C.
First, referring to
Further, as is clear from
Further, it was found that the elements B and C do not exhibit reversible resistance changes in A mode which is similar to the characteristics shown in
Referring to
These results appear to indicate a need for the second tantalum oxide layer 104b to cause resistance changes. As indicated, with the elements B and C having the second tantalum oxide layer 104b, the reversible resistance changes in B mode can be observed.
The following describes a closer examination of the variable resistance layer 104 in these experiments.
To analyze the structure of the variable resistance layer 104 in this experiment, samples were prepared by, under the same conditions as that of the elements A to C, depositing a tantalum oxide on a substrate in which an oxide layer having a thickness of 200 nm is formed on a single-crystal silicon substrate, followed by oxygen plasma irradiation.
These samples are referred to as samples A, B, and C. Table 1 summarizes the oxygen plasma exposure time and a later-described analysis of each sample. It is to be noted that Pt serving as the upper electrode 105 is not deposited on the samples A to C, and thus the variable resistance layer is exposed.
The samples A and C exhibited the same spectra as that of the sample B, revealing that all of the samples have the variable resistance layer primarily comprising an amorphous tantalum oxide. However, the X-ray diffraction measurement showed very similar spectra among the samples and did not show any dependency on the oxygen plasma exposure time.
As described above, the sensitivity of the X-ray diffraction measurement is not so high for the samples used in this experiment. That is to say, the variable resistance layer in the samples A to C is very thin (30 nm thick) and is amorphous as described above, making it difficult to conduct a detail analysis of the tantalum oxides using an ordinary X-ray diffraction spectrum.
In view of the above, a more detailed analysis was conducted using a method known as X-ray reflectometry. This is a method of emitting an X-ray entering the sample surface at a shallow angle, and measuring the intensity of the X-ray reflected.
Then, fitting is performed presuming an appropriate structural model for the spectrum, so as to evaluate the thickness and the refractive index of the variable resistance layer in the samples A to C. Here, the parameters for the fitting are the layered structure of the variable resistance layer and the thickness and δ (=1−refractive index) of each layer.
Referring to
The result of an analysis of the sample B when fitting was performed presuming this double-layered structure showed the following values as shown in Table 1: the thickness of the first tantalum oxide layer was 28.6 nm and δ was 29.3×10−6; and the thickness of the second tantalum oxide layer was about 1.43 nm and δ was 22.3×10−6. Generally, δ of metal tantalum is 39×10−6, and δ of Ta2O5 is 22×10−6. Comparison between these generally known values and the above obtained values appears to indicate that the first tantalum oxide layer is an oxide deficient in oxygen such as TaO1.43, which is clearly different from the stoichiometric composition of tantalum.
Further, as for the second tantalum oxide layer, calculating its compositional ratio using the δ value gives TaO2.45, indicating an oxide similar to Ta2O5 (TaO2.5). However, the second tantalum oxide layer seems to be an oxygen-deficient oxide which is slightly different from the stoichiometric composition.
Referring to Table 1, the sample C showed almost the same result. More specifically, when the first tantalum oxide layer is expressed as TaOx, its thickness is approximately 29 nm and x is about 1.4, and when the second tantalum oxide layer is expressed as TaOy, its thickness is approximately 1.2 nm and y is about 2.3.
In addition, Table 1 shows that even the sample A manufactured by zero-minute oxygen plasma exposure has the second tantalum oxide layer of about 1 nm in thickness. The sputtering apparatus which deposited the tantalum oxide is kept at high vacuum with a back pressure of 7×10−4 Pa, and thus it is unlikely that this oxide layer was formed in the sputtering apparatus.
Thus, the main part of this layer was probably formed during a time from when the sample A was taken out of the sputtering apparatus after the sputtering and until when the X-ray reflectivity was measured (In fact, the measurement was taken a few days after taking out the sample A from the sputtering apparatus). That is to say, if the second electrode is formed without taking the sample A out of the sputtering apparatus, the second tantalum oxide layer 104b probably does not exist, or is as thin as 1 nm or less even if it does exist.
The same inference indicates a possibility that even with the samples B and C, the thickness of the second tantalum oxide layer slightly increased due to exposure to the air after being taken out from the sputtering apparatus which deposited the tantalum oxide (before the X-ray reflectivity measurement). However, it is known that the oxidation generally proceeds fast at the beginning and gradually slows down.
Therefore, in the case of forming the second tantalum oxide having a higher oxygen content percentage by exposure to oxygen plasma in the sputtering apparatus, it is inferred that the proportion of the second tantalum oxide layer which increased outside the sputtering apparatus is small.
The existence of the second tantalum oxide layer is consistent with the fact described above with reference to Table 1 that the initial resistance of the variable resistance layer 104 of the elements B and C is much higher than in the case of providing a single layer of the first tantalum oxide layer.
That is to say, the resistance values of the elements B and C are higher by two to three digits than the resistance value of the element A which appears to have no second tantalum oxide layer. This is probably because the elements B and C have, between the first tantalum oxide layer 104a and the upper electrode 105, a highly resistant second tantalum oxide layer 104b having a higher oxygen content percentage.
Generally, Ta2O5 having a stoichiometric composition is considered to be an insulator. However, as described above, the second tantalum oxide layer is not an insulator because it is deficient in oxygen compared to Ta2O5. It is to be noted that the present invention follows the general definition of an insulator. To be more specific, a material whose resistivity is 108Ω cm or greater is defined as an insulator (Non Patent Literature: “Semiconductor Engineering for Integrated Circuits” (Kougyou Chousakai, 1992) Akira Usami, Shinji Kanefusa, Takao Maekawa, Hajime Tomokage, and Morio Inoue), and a material whose resistivity is smaller than 108Ω cm is defined as a conductor.
If the second tantalum oxide layer of the present experiment is an insulator and has a resistivity of 108Ω cm, its resistance value should be approximately 1.4×108Ω cm (resistance value=resistivity×thickness/area) given that it has a circular shape with a diameter of 3 μm (diameter of the element region 107 in this experiment) and has a thickness of 1 nm (roughly estimated thickness of the second tantalum oxide layer). Even when the second tantalum oxide layer has a thickness of 0.1 nm, its resistance value is given by 1.4×107Ω cm. On the other hand, in the elements B and C, the resistance value is approximately 103 to 104Ω at most with reference to Table 1, which is lower than in the case of assuming an insulator, by at least three to four digits approximately.
This calculation result also indicates that the second tantalum oxide layer formed in the present experiment is not an insulator but a conductive oxide layer.
Although this experiment used the X-ray reflectivity measurement for analyzing the second tantalum oxide layer, other instrumental analysis techniques can also be used, including Auger electron spectroscopy analysis (AES), fluorescent X-ray photoelectron spectroscopy analysis (XPS), and electron probe microanalysis (EPMA: also known as WDS, EDS, or EDX depending on the detection method).
The element B and the sample B underwent the sputtering and the oxygen plasma irradiation under the same condition. The element C and the sample C also underwent the sputtering and the oxygen plasma irradiation under the same condition. Thus, in the elements B and C as in the samples B and C, the second tantalum oxide layer 104b probably exists between the first tantalum oxide layer 104a and the upper electrode 105.
Therefore, it can be said that the second tantalum oxide layer 104b having a thickness of 1.1 nm is formed in the element B as in the sample B, and the second tantalum oxide layer 104b having a thickness of 1.2 nm is formed in the element C as in the sample C.
As previously described, the elements B and C exhibit a stable resistance change phenomenon in B mode. However, the element A which does not have the second tantalum oxide layer having a higher oxygen content percentage does not exhibit the resistance change phenomenon. This appears to indicate that the second tantalum oxide layer is essential for a resistance change. In the present experiment, the composition of this second tantalum oxide layer, when expressed as TaOy, is sufficient if it has a thickness of about 1.1 nm and y of about 2.1.
Next, a variable resistance element (referred to as an element B′) was manufactured which includes a first tantalum oxide layer having a thickness different from that in the above element B, and the resistance change characteristics of this variable resistance element were examined.
The element B′ is different from the element B only in thickness of the first tantalum oxide layer 104a. While the thickness of the first tantalum oxide layer 104a of the element B was 30 nm, the thickness of the first tantalum oxide layer 104a of the element B′ was 90 nm. In manufacturing the element B′, the oxygen plasma exposure time was 0.5 minutes as in the case of the element B. Therefore, with the element B′ too, the thickness of the second tantalum oxide layer 104b is probably 1 to 2 nm approximately.
When electrical pulses of a negative voltage −2.0 V and a positive voltage 3.0 V with a pulse width of 100 nsec were alternately and repeatedly applied between the upper electrode 105 and the lower electrode 103 of the element B, the element B exhibited such resistance change characteristics that the resistance value changed from about 500Ω to 20Ω when −2.0 V was applied, and thereafter, reversible resistance changes stably occurred in B mode between about 20Ω and about 200 Ω.
The above results show that the thickness of the first tantalum oxide layer does not have a significant impact on the resistance change phenomenon of the nonvolatile storage element according to the present experiment.
As described above, the first experiment involved deposition of the first tantalum oxide layer in the sputtering apparatus, followed by oxidation using oxygen plasma to form the second tantalum oxide layer. However, this method could not thickly form the second tantalum oxide layer due to the apparatus used. Thus, as the second experiment, the following describes the operation of the variable resistance element in the case where a second tantalum oxide layer is thickly formed.
The structure and the manufacturing method of the variable resistance element used in the second experiment are basically the same as in the first experiment. However, for convenience of the oxidation process, the second experiment is different from the first experiment in condition under which tantalum oxide is deposited and size of the nonvolatile storage element formed. Hereinafter, a process for manufacturing a nonvolatile element is described with reference to
First, as shown in
Here, the first tantalum oxide layer 104a was deposited under the condition described below. That is, the substrate was placed in the sputtering apparatus and then the sputtering apparatus was vacuumed to about 8×10−6 Pa. Then, the sputtering was performed for 20 seconds with tantalum used as the target, power set to 1.6 kW, a flow of an argon gas at 34 sccm, a flow of an oxygen gas at 21 sccm, and the pressure in the sputtering apparatus kept at 0.17 Pa. With this, the first tantalum oxide layer was deposited with a thickness of 30 nm, a resistivity of 6 mΩcm, and an oxygen content percentage of about 61 at % (TaO1.6).
Next, as shown in
Subsequently, by sputtering, a Pt thin film having a thickness of 150 nm was formed on the second tantalum oxide layer 104b as the upper electrode 105. To prevent oxidization of the second tantalum oxide layer 104b in the air, the upper electrode 105 was formed immediately after the deposition of the second tantalum oxide layer 104b. Lastly, photolithography processing created the pattern 106 using a photoresist, and dry etching produced the element region 107. The element region 107 here is assumed to be a rectangle with each side being 0.5 μm in length.
Next, the following describes the characteristics observed when a resistance change occurs upon application of electrical pulses to the elements D and E which were actually manufactured in the second experiment.
First, the following describes the result of the element D in which the second tantalum oxide layer was formed through the oxidation using oxygen plasma. As is clear from the result shown in
Next, referring to
As
On the other hand, in gradual application of a negative voltage to the upper electrode 105 side with reference to the lower electrode 103 (equivalent to application of a positive voltage to the lower electrode 103 side with reference to the upper electrode 105) in the high resistance state, the current rapidly increases when the negative voltage exceeds the voltage at the point B. This shows a change in the resistance from the high resistance state to the low resistance state.
Thus, it is clear that the element D exhibits the resistance changes in B mode and requires a larger driving current for the resistance change from the low resistance state to the high resistance state than for the resistance change from the high resistance state to the low resistance state.
To analyze the structure of the variable resistance layer 104 used in this experiment, samples were prepared by, under the same conditions as that of the elements D and E, depositing a tantalum oxide on a substrate in which an oxide layer having a thickness of 200 nm is formed on a single-crystal silicon substrate, followed by oxidation. These samples are referred to as samples D and E. The results of the X-ray reflectivity measurement on these samples are shown in Table 2. It is to be noted that in the samples D and E, the second tantalum oxide layer is exposed as in the case of the samples A to C.
Referring to Table 2, the sample D which was oxidized using oxygen plasma has the second tantalum oxide layer TaOy whose thickness is 8.1 nm, which is, as originally intended, thicker than that of the samples A to C. Further, y is 2.47, indicating that oxygen is more deficient in TaOy than in Ta2O5 having a stoichiometric composition. As for the sample E which was oxidized using the lamp annealing apparatus, the thickness of the second tantalum oxide layer TaOy was 7.3 nm and y was 2.38.
As described above, the second tantalum oxide layer formed in the variable resistance element in this experiment has a thickness of 7 to 8 nm approximately. Such a thickness is sufficient for observing the existence of the second tantalum oxide layer through cross-section observation of the nonvolatile element using a transmission electron microscope. Thus, cross-section observation was actually performed on the variable resistance element in which the second tantalum oxide layer was formed through the oxidation using oxygen plasma.
These values almost match the result of the X-ray reflectivity measurement on the sample D which was manufactured under the same oxidation condition (as Table 2 shows, the thickness of the first tantalum oxide layer is 26.6 nm and the thickness of the second tantalum oxide layer is 8.1 nm).
The above result made it clear that the second tantalum oxide layer does actually exist in the nonvolatile storage element used in the second experiment. In addition, it proves the validity of the analysis using the X-ray reflectivity measurement.
The third experiment is now described.
In the case of the elements A to C used in the above-described first experiment, the oxygen content percentage of the first tantalum oxide layer 104a was 58 at % (TaO1.4). The first tantalum oxide layer 104a of the elements D and E used in the second experiment had a similar oxygen content percentage of 61 at % (TaO1.6).
In contrast, the variable resistance element used in the third experiment includes the first tantalum oxide layer whose oxygen content percentage is changed to be a little greater. The structure of the variable resistance element used in the third experiment is the same as in the first and the second experiments, and is thus not shown.
With reference to
First, the following describes the condition under which the tantalum oxide layer of the third experiment was manufactured and an analysis result of the oxygen content percentage in the third experiment. The tantalum oxide was manufactured using the same method as in the first experiment. However, the oxygen content percentage of the tantalum oxide is controlled by adjusting the oxygen flow ratio employed in sputtering.
Describing in accordance with the specific processes of sputtering, first, a substrate is placed in a sputtering apparatus, and the sputtering apparatus is vacuumed to about 7×10−4 Pa. Then, the sputtering is performed with tantalum used as the target, power set to 250 W, the total gas pressure of the argon gas and the oxygen gas set to 3.3 Pa, and the preset temperature of the substrate set to 30° C. Here, the flow ratio of the oxygen gas is varied from 0.8% to 6.7%.
Since the object of the third experiment is to examine the composition, SiO2 deposited 200 nm in thickness on Si (Silicon) was used as the substrate, and the sputtering time was adjusted to make the thickness of the tantalum oxide layer about 100 nm. Unlike in the first experiment, the oxygen plasma exposure was not performed in the third experiment.
It appears that the sample prepared for the composition measurement was oxidized with oxygen in the air during a time between the deposition on the substrate and the measurement, thereby forming a high oxygen content percentage layer on the sample surface. However, because the surface was etched before the measurement using the RBS and the AES, it is possible to ignore the impact that the high oxygen content percentage layer formed on the surface has on the measurement of the oxygen content percentage.
Although the present experiment used the Rutherford backscattering spectrometry (RBS) and the Auger electron spectroscopy (AES) for analyzing the tantalum oxide layer, other instrumental analysis techniques can also be used, including fluorescent X-ray photoelectron spectroscopy analysis (XPS) and electron probe microanalysis (EPMA).
Next, the following describes the resistance change characteristics observed when the variable resistance element 100 is manufactured by forming the variable resistance layer 104 using, as the first tantalum oxide layer 104a and the second tantalum oxide layer 104b, each of tantalum oxide layers having different oxygen content percentages.
The variable resistance element 100 was manufactured using the same method as in the first experiment. To be more specific, thermal oxidation produced the oxide layer 102 having a thickness of 200 nm on the substrate 101 which is a single-crystal silicon, and then, by sputtering, a Pt thin film having a thickness of 100 nm was deposited on the oxide layer 102 as the lower electrode 103. Then, the sputtering was performed to form the first tantalum oxide layer 104a, with tantalum used as the target, power set to 250 W, the total gas pressure of the argon gas and the oxygen gas set to 3.3 Pa, and the preset temperature of the substrate set to 30° C.
In this experiment, each element was manufactured with the flow ratio of the oxygen gas varied from 0.8% to 6.7%.
The sputtering time was adjusted such that the thickness of the first tantalum oxide layer 104a becomes 30 nm. Thereafter, the top surface of the first tantalum oxide layer 104a was exposed to oxygen plasma for 30 seconds, so as to form the second tantalum oxide layer 104b. Lastly, by sputtering, a Pt thin film having a thickness of 150 nm was formed on the second tantalum oxide layer 104b as the upper electrode 105, so as to form the variable resistance element 100.
The resistance change phenomenon of the variable resistance element formed in the above manner was measured. The result of the measurement showed a satisfactory resistance change phenomenon. More specifically, with the nonvolatile storage element using a tantalum oxide film represented by a range in
As described so far in the first to third experiments, it was found that with the variable resistance element shown in
The variable resistance element having such a structure did not exhibit the resistance changes in A mode which are resistance changes of the reversed polarity. Further, with this structure, the thickness of the second oxygen-deficient tantalum oxide layer between 1 nm and 8 nm inclusive was favorable in exhibiting the stable resistance changes in B mode.
Next, the following describes the fourth experiment related to a variable resistance element which exhibits stable resistance changes in A mode, which are opposite to the resistance changes described so far.
Here, the variable resistance layer 104 includes: a first tantalum containing layer (hereinafter referred to as “first tantalum oxide layer”) 104a having a lower oxygen content percentage; and a second tantalum containing layer (hereinafter referred to as “second tantalum oxide layer”) 104b formed on the first tantalum oxide layer 104a and having a higher oxygen content percentage.
The variable resistance element used in the fourth experiment is different from that used in the first to third experiments in that the second tantalum oxide layer 104b is provided in contact with the lower electrode 103, and the first tantalum oxide layer 104a is provided in contact with the upper electrode 105.
Next, a manufacturing method of the variable resistance element 100 used in the present experiment is described with reference to
First, as shown in
Next, as shown in
This produces, on the surface of the second tantalum oxide layer 104b, the first tantalum oxide layer 104a having an oxygen content percentage lower than that of the second tantalum oxide layer 104b. Such a layered structure of the second tantalum oxide layer 104b and the first tantalum oxide layer 104a is the structure of the variable resistance layer 104.
Subsequently, by sputtering, a Pt thin film having a thickness of 150 nm is formed on the first tantalum oxide layer 104a as the upper electrode 105. Lastly, photolithography processing creates a pattern 106 using a photoresist, and dry etching produces an element region 107 as shown in
With the above-described manufacturing method, an element F was manufactured. The element region 107 here is assumed to be a rectangle with each side being 0.5 μm in length.
Next, the following describes the resistance change characteristics observed when a resistance change occurs upon application of electrical pulses to the element F which was actually manufactured in the fourth experiment.
First, upon application of a positive voltage of 1.5 V to the upper electrode, the resistance value decreases to about 200Ω, and upon application of a negative voltage of −1.8 V, the resistance value increases to about 20000Ω. After that, upon applying the electrical pulses of the positive voltage 1.5 V and the negative voltage −1.8 V alternately, stable resistance changes occur in A mode, with the resistance value varying between 100 Ω to 8000Ω approximately.
As
On the other hand, in gradual application of a positive voltage to the upper electrode 105 side with reference to the lower electrode 103 (equivalent to application of a negative voltage to the lower electrode 103 side with reference to the upper electrode 105) in the high resistance state, the current rapidly increases when the negative voltage exceeds the voltage at the point B. In other words, it shows a change in the resistance from the high resistance state to the low resistance state.
Thus, it indicates that the element F exhibits the resistance changes in A mode and requires a larger driving current for the resistance change from the low resistance state to the high resistance state than for the resistance change from the high resistance state to the low resistance state.
The following examines the structure of the variable resistance layer 104 included in the variable resistance element used in the fourth experiment, and especially examines the composition of the second tantalum oxide layer formed in the present experiment by sputtering, with Ta2O5 used as the target.
Due to the plasma in the sputtering process, the composition of the second tantalum oxide layer is not exactly Ta2O5 but is a composition which slightly lacks oxygen. Thus, it is inferred that a tantalum oxide TaOy (y=2.3 to 2.4) was formed which contains less oxygen than in the composition of the target.
Thus, with the variable resistance element shown in
As described so far regarding the fourth experiment, it was found that with the variable resistance element shown in
The variable resistance element having such a structure did not exhibit the resistance changes in B mode which are resistance changes of the reversed polarity. Further, with this structure, the thickness of the second oxygen-deficient tantalum oxide layer was 3 nm.
By combining the structure of the variable resistance element in the fourth experiment and the compositions of the variable resistance layers in the first to third experiments, it is possible to adequately infer that with the structure of the variable resistance element shown in
In addition, it can be inferred that the variable resistance element having such a structure does not exhibit the resistance changes in B mode which are the resistance changes of the reversed polarity. Furthermore, it is possible to infer that, with this structure too, the thickness of the second oxygen-deficient tantalum oxide layer between 1 nm and 8 nm inclusive is favorable in exhibiting the stable resistance changes in A mode.
Next, the following describes, as an embodiment of the present invention, a 1T1R nonvolatile storage device for which the above-mentioned variable resistance element is used.
(1T1R Nonvolatile Storage Device having NMOS Structure)
As shown in
The nonvolatile storage device 200 further includes, as a power source for writing 211, a power source for low resistance (LR) writing 212 and a power source for high resistance (HR) writing 213. An output V2 of the power source for low resistance (LR) writing 212 is supplied to the row driver 207, whereas an output V1 of the power source for high resistance (HR) writing 213 is supplied to the write circuit 206.
Moreover, the nonvolatile storage device 200 includes: an address input circuit 209 which receives address signals inputted from the outside; and a control circuit 210 which controls operations of the memory main portion 201 based on a control signal inputted from the outside.
The memory array 202 includes: word lines WL0, WL1, WL2 . . . ; bit lines BL0, BL1, BL2 . . . ; NMOS transistors N11, N12, N13, N21, N22, N23, N31, N32, N33 . . . (hereinafter, referred to as “transistors N11, N12 . . . ”); and variable resistance elements R11, R12, R13, R21, R22, R23, R31, R32, R33 . . . (hereinafter, referred to as “variable resistance elements R11, R12 . . . ”). These NMOS transistors, and variable resistance elements form memory cells M11, M12, M13, M21, M22, M23, M31, M32, M33 . . . (hereinafter, referred to as “memory cells M11, M12 . . . ”). Here, the word lines and the bit lines are formed above the semiconductor substrate and arranged to intersect with each other. The NMOS transistors are provided at corresponding points where the word lines WL0, WL1, WL2 . . . and the bit lines BL0, BL1, BL2 . . . intersect with each other. The variable resistance elements are connected in series with the transistors N11, N12 . . . on a one-on-one basis. It is to be noted that each of the variable resistance elements R11, R12 . . . is the variable resistance element described above as the basic data of the present invention.
As shown in
Furthermore, the transistors N11, N21, N31 . . . and the transistors N12, N22, N32 . . . are commonly connected to the source line SL0, and the transistors N13, N23, N33 . . . and the transistors N14, N24, N34 . . . are commonly connected to the source line SL2.
Moreover, the variable resistance elements R11, R12, R13, R14 . . . are connected to the bit line BL0, the variable resistance elements R21, R22, R23, R24 . . . are connected to the bit line BL1, and the variable resistance elements R31, R32, R33, R34 . . . are connected to the bit line BL2.
The address input circuit 209 receives address signals from an external circuit (not shown), and outputs row address signals to the row selection circuit 208 and column address signals to the column selection circuit 203 based on the address signals. Here, the address signals are signals indicating an address of a specific memory cell selected from among the memory cells M11, M12 . . . .
In a data write cycle, the control circuit 210 outputs to the write circuit 206 a write command signal instructing application of a voltage for writing, according to input data Din inputted into the data input and output circuit 205. On the other hand, in a data read cycle, the control circuit 210 outputs to the sense amplifier 204 a read command signal instructing a read operation.
The row selection circuit 208 receives the row address signals outputted from the address input circuit 209, and according to the row address signals, the row driver 207 causes a word line driver WLD corresponding to one of the word lines WL0, WL1, WL2 . . . to apply a predetermined voltage to the selected word line.
Similarly, the row selection circuit 208 receives the row address signals outputted from the address input circuit 209, and according to the row address signals, the row driver 207 causes a source line driver SLD corresponding to one of the source lines SL0, SL2 . . . to apply a predetermined voltage to the selected source line.
In addition, the column selection circuit 203 receives the column address signals outputted from the address input circuit 209, selects one of the bit lines BL0, BL1, BL2 . . . according to the column address signals, and applies the voltage for writing or the voltage for reading to the selected bit line.
In the case of receiving the write command signal from the control circuit 210, the write circuit 206 outputs to the column selection circuit 203 a signal instructing an application of the voltage for writing to the selected bit line.
Moreover, in the data read cycle, the sense amplifier 204 detects an amount of a current flowing in the selected bit line to be read, and determines whether stored data indicates “1” or “0”. The resulting output data DO is outputted to an external circuit via the data input and output circuit 205.
The power source for writing 211 includes the power source for low resistance (LR) writing 212 and the power source for high resistance (HR) writing 213, and the output of the power source for low resistance (LR) writing 212 and the output of the power source for high resistance (HR) writing 213 are inputted to the row driver 207 and the write circuit 206, respectively.
In the case of using the element D, which has been described as the basic data of the present invention, as the variable resistance elements R11, R12 . . . , for example, and given that a voltage at the point A of
Transistors 317 and the variable resistance elements 309 correspond to the transistors N11 and N12 and the variable resistance elements R11 and R12 shown in
The memory cells 300 are formed by sequentially forming, above a semiconductor substrate 301, a second N-type diffusion layer region 302a, first N-type diffusion layer regions 302b, gate insulator films 303a, gate electrodes 303b, first vias 304, first wiring layers 305, second vias 306, second wiring layers 307, third vias 308, the variable resistance elements 309, fourth vias 310, and a third wiring layer 311.
The third wiring layer 311 connected to the fourth vias 310 corresponds to the bit line BL0, and the first wiring layers 305 and the second wiring layers 307 which are connected to the second N-type diffusion layer region 302a of the transistors 317 correspond to the source line SL0 extending perpendicularly to the diagram.
The voltage at the semiconductor substrate 301 is 0 V, and is supplied by a commonly-known structure via a 0-V power source line (not shown).
As shown in the enlarged view in
Here, the variable resistance layer 309b comprises an oxygen-deficient transition metal oxide and, in the present example, has a layered structure in which the following layers are stacked: a conductive, first oxygen-deficient tantalum oxide layer 309b-1 having a composition TaOx (where x=1.54); and a conductive, second oxygen-deficient tantalum oxide layer 309b-2 having a composition TaOy (where y=2.47). Here, the thickness of TaOx (where x=1.54) is 26.6 nm, and the thickness of TaOy (where y=2.47) is 2.47 nm. Further, the lower electrode 309a and the upper electrode 309c comprise the same material, which is Pt (platinum) in this example. The lower electrode 309a is connected to the second N-type diffusion layer region 302b of the transistor through the vias, whereas the upper electrode 309c is connected, through the via, to the bit line BL0 formed by the third wiring layer 311.
With reference to timing charts shown in
In
In
The method for determining V1 and V2 is described later.
In
Furthermore, in
In the write cycle in which the data “0” is written into the memory cell M11 as shown in
Then, the selected bit line BL0 is set to the voltage 0 V for a predetermined period, and after the predetermined period, a pulse waveform corresponding to the voltage V2 is applied again to the bit line BL0. At this stage, in the variable resistance element 309, a negative voltage having an absolute value larger than the threshold voltage for low resistance writing VAR is applied to the upper electrode 309c with reference to the lower electrode 309a, so that a writing is performed to change from a high resistance value to a low resistance value. Then, the word line WL0 is set to the voltage 0 V to turn off the transistor 317, thereby completing the writing of the data “0”.
In the write cycle in which the data “1” is written into the memory cell M11 as shown in
Then, the selected bit line BL0 is set to the voltage V1 for a predetermined period, and after the predetermined period, a pulse waveform corresponding to the voltage 0 V is applied again to the bit line BL0. At this stage, in the variable resistance element 309, a positive voltage higher than the threshold voltage for high resistance writing VHR is applied to the upper electrode 309c with reference to the lower electrode 309a, so that a writing is performed to change from a low resistance value to a high resistance value. Then, the word line WL0 is set to the voltage 0 V, thereby completing the writing of the data “1”.
In the read cycle in which data is read from the memory cell M11 as shown in
Then, the selected bit line BL0 is set to the read voltage Vread for a predetermined period, and the sense amplifier 204 detects the value of the current flowing in the selected memory cell M11 to determine whether the stored data is the data “0” or the data “1”. Then, the word line WL0 is set to the voltage 0 V, thereby completing the data reading operation.
The following describes the 1T1R memory cells M11, M12 . . . according to an embodiment, and especially describes the structures of the NMOS transistors N11, N12 . . . .
According to the present embodiment, as shown in the enlarged view in
A voltage application direction 1 shows the characteristics observed when a predetermined positive voltage is applied to the bit lines BL0, BL1 . . . and 0 V is applied to the source lines SL0, SL1 . . . in
In addition, a voltage application direction 2 shows the characteristics observed when 0 V is applied to the bit lines BL0, BL1 . . . and a predetermined positive voltage is applied to the source lines SL0, SL1 . . . in
For example, when the element resistance value is 1000Ω, the voltage application direction 1 indicates that approximately 2.1 V can be applied to the variable resistance element and that a current of 2.1 mA can be driven (2.1 V/1000Ω). In the case of the voltage application direction 2, it indicates that approximately 1.25 V can be applied to the variable resistance element and that a current of 1.25 mA can be driven (1.25 V/1000Ω).
This shows that in the voltage application direction 1 in which the influence of the substrate bias effect in the NMOS transistor is small, a current 1.7 times larger than in the case of the voltage application direction 2 can be driven.
Further, the value of the voltage V2 generated by the power source for LR writing 212 shown in
For instance, when the resistance value of the variable resistance element 309 is 10 kΩ in the high resistance state, it is clear that with application of 2.2 V to the both ends of the memory cell, a voltage up to approximately 1.5 V can be applied to the variable resistance element 309 (point C in
When the threshold voltage for low resistance writing VLR, which is applied for the resistance change to low resistance state of the variable resistance element 309, is −1.1 V (point B in
Likewise, the value of the voltage V1 generated by the power source for HR writing 213 shown in
For instance, when the resistance value of the variable resistance element 309 is 1000Ω in the low resistance state, it is clear that with application of 2.2 V to the both ends of the memory cell, a voltage up to approximately 2.1 V can be applied to the variable resistance element 309 (point D in
When the threshold voltage for high resistance writing VHR, which is applied for the resistance change to high resistance state of the variable resistance element 309, is 1.2 V (point A in
It is also possible, at the designing stage, to determine an approximate voltage using the above-mentioned method, and at the product inspection stage, to also use a commonly-known conventional method of making, while checking the operation, fine adjustments so as to determine optimum voltages of the voltages V1 and V2 at which changes in the resistance occur stably.
As described above, the variable resistance nonvolatile storage device of the present embodiment includes the variable resistance element in which: the second oxygen-deficient tantalum oxide layer having a higher oxygen content percentage is provided on the upper electrode side; and the first oxygen-deficient tantalum oxide layer having a lower oxygen content percentage is provided on the lower electrode side. Thus, the voltage application direction (driving polarity) which stably causes a resistance change in one direction (resistance change to low resistance state or resistance change to high resistance state) is uniquely determined for each of the memory cells.
Further, the memory cell is formed by connecting the lower electrode and one of the N-type diffusion layer regions of the NMOS transistor. Therefore, voltage application for the resistance change from the low resistance to the high resistance, which requires a larger current, can be reliably performed in the voltage application direction 1, thereby eliminating the need of taking into account the case of the voltage application direction 2. As a result, the memory cell can be designed with an optimum transistor size.
This means that especially the voltage V1 of the power source for HR writing can be further reduced when there is a margin in the current drive capability, thereby leading to reduction in voltage and reduction in power consumption.
In addition, because the driving polarity is uniquely determined, there is no need to manage information identifying the modes of the resistance change characteristics, thereby allowing for a simple and inexpensive circuit configuration.
Here, a variable resistance layer 309e comprises an oxygen-deficient tantalum oxide as is the variable resistance layer 309b, and has a structure in which: a second oxygen-deficient tantalum oxide layer 309e-2 having a higher oxygen content percentage is provided in contact with the lower electrode; and a first oxygen-deficient tantalum oxide layer 309e-1 having a lower oxygen content percentage is provided in contact with the upper electrode.
The description of
In this case, as in the case of
In the configuration shown in
Unlike
In this case, as in
The memory cells 400 are formed by sequentially forming, above a semiconductor substrate 301, an N well 418, a second P-type diffusion layer region 402a, first P-type diffusion layer regions 402b, gate insulator films 303a, gate electrodes 303b, first vias 304, first wiring layers 305, second vias 306, second wiring layers 307, third vias 308, the variable resistance elements 409, fourth vias 310, and a third wiring layer 311.
The third wiring layer 311 connected to the fourth vias 310 corresponds to the bit line BL0, and the first wiring layers 305 and the second wiring layers 307 which are connected to the second P-type diffusion layer region 402a of transistors 417 correspond to the source line SL0 extending perpendicularly to the diagram. The N well is supplied with a power supply voltage VDD of a nonvolatile storage device 200 through a commonly-known structure via a VDD power source line (not shown).
As shown in the enlarged view in
Here, unlike in the case of the 1T1R memory cell (
The driving direction of the transistors 417 in which the influence of the substrate bias effect is small and the current drive capability is large is the direction in which the second P-type diffusion layer region 402a serves as a source and a voltage at the second P-type diffusion layer region 402a approaches the voltage (VDD) at the N well 418 serving as the substrate voltage for the PMOS transistor; that is, it is the direction in which the lower electrode 309d is set to a high level and the upper electrode 309f is set to a low level.
In order to match the resistance change direction from the low resistance state to the high resistance state, which requires a larger current, with the above-mentioned voltage application direction, the second oxygen-deficient tantalum oxide layer 309e-2 having a higher oxygen content percentage is provided on the lower electrode 309d side, and the first oxygen-deficient tantalum oxide layer 309e-1 having a lower oxygen content percentage is provided on the upper electrode 309f side. When a positive voltage is applied to the lower electrode 309d with reference to the upper electrode 309f, the oxidation phenomenon progresses near the interface of the lower electrode 309d, thereby causing a change into the high resistance state.
Although the NMOS transistor is often used in the 1T1R memory cell in general, there could be cases where the PMOS transistor is used in the memory cell, including the following cases:
For instance, there is a case where in a selected memory cell, only a threshold voltage of a transistor of the memory cell is set low in order to obtain a larger current for driving the transistor. In this case, there is an increase in leak current to non-selected memory cells connected to a bit line to which the selected memory cell belongs. This will probably result in deterioration in the read characteristics.
As a method for avoiding the increase in the leak current while maintaining the driving current of the selected memory cell, there is a method of electrically dividing a region of the semiconductor substrate 301 into some blocks and adjusting substrate voltage of the blocks so as to increase the threshold voltage of transistors in blocks other than a block to which the selected memory cell belongs, thereby reducing the leak current.
Generally, many CMOS semiconductor devices use a P-type silicon semiconductor for the semiconductor substrate 301. Thus, to implement such a structure, in the case of forming the transistor of the memory cell with an NMOS transistor, it is necessary to adopt a well structure known as a triple-well structure, for instance, and to electrically divide the substrate region into some blocks. In such a case, a new manufacturing step needs to be added, which leads to an increase in cost.
On the other hand, in the case of forming the transistor of the memory cell with a PMOS transistor, it is only necessary to design the layout of the N well 418 in a desired unit and electrically divide the N well 418 into some blocks. Thus, there seems to be an advantage that the N well 418 can be divided into the blocks without an additional manufacturing step.
The cross-sectional views shown in
The cross-sectional views corresponding to
Further, the cross-sectional views corresponding to
Table 3 shows, for each of the memory cell structures corresponding to
indicates data missing or illegible when filed
For each memory cell, the voltage application direction (driving polarity) which stably causes a resistance change in one direction (resistance change to low resistance state or resistance change to high resistance state) is uniquely determined based on Table 3. Thus, there is no need to manage the information identifying the modes of the resistance change characteristics, thereby allowing for a simple circuit configuration.
It is to be noted that although Pt is used as the electrode material in the present embodiment, Ir, Pd, Ag or Cu may be used instead.
Although the present embodiment has shown the example of the variable resistance element in which the tantalum oxide is used as the variable resistance layer, the material of the variable resistance layer is not limited to the tantalum oxide. To manufacture an element having the same structure as in the present embodiment, it is possible to use any material as long as it stably shows the resistance change characteristics in A mode or B mode.
The inventors of the present application have acquired from an experiment basic data which allows inference that the resistance change characteristics in A mode or B mode stably occur in, for example, a variable resistance element in which a hafnium oxide is used as the variable resistance layer and in a variable resistance element in which a zirconium oxide is used as the variable resistance layer.
The following describes that experiment.
The following describes the fifth experiment regarding a variable resistance nonvolatile storage element which uses an oxygen-deficient hafnium oxide as the variable resistance layer.
As shown in
Here, the variable resistance layer 1107 includes: a second hafnium containing layer (hereinafter referred to as “second hafnium oxide layer”) 1105 having a low oxygen content percentage; a first hafnium containing layer (hereinafter referred to as “first hafnium oxide layer”) 1104 formed under the second hafnium oxide layer 1105 and having an oxygen content percentage higher that that of the second hafnium oxide layer 1105; and a third hafnium containing layer (hereinafter referred to as “third hafnium oxide layer”) 1106 formed on the second hafnium oxide layer 1105 and having an oxygen content percentage higher that that of the second hafnium oxide layer 1105.
The materials of the lower electrode 1103 and the upper electrode 1108 are, for example, Pt (platinum), Ir (iridium), Pd (palladium), Ag (silver), Ni (nickel), W (tungsten), Cu (copper), Al (aluminum), Ta (tantalum), Ti (titanium), TiN (titanium nitride), TaN (tantalum nitride), and TiAIN (titanium aluminum nitride).
Although the substrate 1101 may be a single-crystal silicon substrate or a semiconductor substrate, it is not limited to these substrates. The variable resistance layer 1107 can be formed at a relatively low substrate temperature and thus can be formed on a resin material and the like.
Next, a manufacturing method of the variable resistance element 1100 used in the fifth experiment is described with reference to
First, as shown in
The first hafnium oxide layer 1104 whose oxygen content percentage is higher than that of the second hafnium oxide layer 1105 is formed by exposure of the surface of the lower electrode 1103 to the air at the time of forming the second hafnium oxide layer 1105. Further, after forming the second hafnium oxide layer 1105, the third hafnium oxide layer 1106 whose oxygen content percentage is higher than that of the second hafnium oxide layer 1105 is formed by exposure to plasma of the Ar gas and the O2 gas used in the sputtering.
This layered structure in which the first hafnium oxide layer 1104, the second hafnium oxide layer 1105, and the third hafnium oxide layer 1106 are stacked is the structure of the variable resistance layer 1107.
Subsequently, by sputtering, a Pt thin film having a thickness of 150 nm is deposited on the third hafnium oxide layer 1106 as the upper electrode 1108.
Lastly, photolithography processing and dry etching create an element region 1109. The element region 1109 has a circular shape with a diameter of 3 μm.
With the above-described manufacturing method, plural variable resistance elements were manufactured under different manufacturing conditions. The following describes the details.
First, the following describes the conditions under which the oxygen-deficient hafnium oxide layer of the fifth experiment was manufactured and an analysis of the oxygen content percentage in the fifth experiment. The oxygen-deficient hafnium oxide layer was manufactured by reactive sputtering, that is, a process of sputtering the hafnium target in an atmosphere of an Ar (argon) gas and an O2 gas.
First, a substrate was placed in a sputtering apparatus and then the sputtering apparatus was vacuumed to about 3×10−5 Pa. Then, sputtering was performed with hafnium used as the target, power set to 300 W, the total gas pressure of the argon gas and the oxygen gas set to 0.9 Pa, and the preset temperature of the substrate set to 30° C. In this experiment, five types of hafnium oxide layers were formed with the flow ratio of the O2 gas to the Ar gas varied from 2% to 4%.
First, with an object of this experiment being an examination of the composition, SiO2 deposited 200 nm in thickness on Si was used as the substrate, and the sputtering time was adjusted to make the thickness of the five types of the hafnium oxide layers about 50 nm.
Although this experiment used the Rutherford backscattering spectrometry (RBS) for analyzing the hafnium oxide layers, other instrumental analysis techniques can also be used, including Auger electron spectroscopy (AES), fluorescent X-ray photoelectron spectroscopy analysis (XPS), and electron probe microanalysis (EPMA).
The above result made it clear that: the oxygen content percentage of the hafnium oxide layer can be controlled using the oxygen flow ratio; and a hafnium oxide layer (sample K) which seems to excessively contain oxygen was formed as well as oxygen-deficient hafnium oxide layers (samples G to 3) which have an oxygen content percentage lower than 66.7 at %, which is the oxygen content percentage of HfO2 that is a stoichiometric oxide of hafnium.
Fives types of variable resistance elements were manufactured by depositing a hafnium oxide as the variable resistance layer 1107 under the same conditions as for the samples G to K, and then forming the upper electrode 1108 on the variable resistance layer 1107. The thickness of all the variable resistance layers 1107 was set to 30 nm. These elements are referred to as elements G, H, I, J, and K.
The resistance change characteristics of the nonvolatile storage elements manufactured in the above manner were measured.
First, the resistance change characteristics observed upon application of electrical pulses to the element I (the oxygen flow ratio of about 3.0% and the oxygen content percentage of the hafnium oxide layer of about 56.8 at %) were closely examined.
First,
Although the details are omitted, an additional experiment showed a result indicating that the resistance changes in B mode occur near the upper electrode 1108.
Next,
Although the details are omitted, an additional experiment showed a result indicating that the resistance changes in A mode occur near the lower electrode 1103.
Further, an experiment was conducted with the other elements in the same manner as the element I, so as to find out with what percentage of oxygen content in an oxygen-deficient hafnium oxide used for the variable resistance layer, the elements exhibit resistance changes.
The result of the experiment showed that aside from the element I, the element H (the oxygen flow ratio of about 2.7% and the oxygen content percentage of the hafnium oxide layer of about 46.6 at %) and the element 3 (the oxygen flow ratio of about 3.3% and the oxygen content percentage of the hafnium oxide layer of about 62 at %) exhibit favorable resistance changes in which the high resistance value is equal to or greater than four times the low resistance value.
From the above findings, it can be said that the bipolar variable resistance element exhibits high-speed resistance changes when its variable resistance layer is manufactured at an oxygen flow ratio of 2.6% to 3.3% and has an oxygen content percentage of 46.6 to 62 at %. In other words, when the variable resistance layer is expressed as HfOx, a more appropriate compositional range of the variable resistance layer is 0.9≦x≦1.6 (the oxygen content percentage of 46.6 at % and the oxygen content percentage of 62 at % correspond to x=0.9 and x=1.6, respectively).
The element G, that is, the variable resistance element which was manufactured at the oxygen flow ratio of 2.0% and includes the hafnium oxide layer having an oxygen content percentage of about 37.7 at % (HfO0.6), has a small initial resistance. Thus, the forming was not possible, and the resistance did not change. Also, the element K, that is, the non-volatile element which was manufactured at the oxygen flow ratio of 4.2% and includes the hafnium oxide layer having an oxygen content percentage of about 69.4 at % (HfO2.3), has a great initial resistance. Thus, soft breakdown was not possible even upon application of a direct current voltage of 5 V, and the resistance did not change.
Since the resistance changes in A mode and B mode appear to occur near the upper electrode or near the lower electrode, a closer analysis of the structure of the oxygen-deficient hafnium oxide layer near the upper electrode and the lower electrode was further conducted.
To analyze the structure of the variable resistance layer 1107 included in the variable resistance element 1100, a hafnium oxide layer of the sample I was formed to have an oxygen content percentage of 56.8% and a thickness of 50 nm. Since the analysis requires a relatively large region in which an oxygen-deficient hafnium oxide is deposited, a sample specifically for the analysis was separately prepared by depositing an oxygen-deficient hafnium oxide on a substrate with no element pattern, in which an oxide layer having a thickness of 200 nm is formed on a single-crystal silicon substrate.
In this experiment, too, using the above-described X-ray reflectometry, an X-ray was emitted to enter the sample surface at a shallow angle, and the intensity of the X-ray reflected was measured. Then, fitting was performed presuming an appropriate structural model for an obtained spectrum, so as to evaluate the thickness and the refractive index of the variable resistance layer in the analysis sample which was specifically prepared for the analysis. Here, the parameters for the fitting are the layered structure of the oxygen-deficient hafnium oxide and the thickness and δ (=1−refractive index) of each layer.
Referring to
The above results appear to indicate that the oxygen-deficient hafnium oxide layer includes: the first oxygen-deficient hafnium oxide layer 1104 on the lower electrode side; the second oxygen-deficient hafnium oxide layer 1105 in the middle; and the third oxygen-deficient hafnium oxide layer 1106 on the upper electrode side.
The analysis result obtained when fitting was performed presuming this triple-layered structure showed the following values: the thickness of the first oxygen-deficient hafnium oxide layer was 3.9 nm and δ was 24.2×10−6; the thickness of the second oxygen-deficient hafnium oxide layer was 45.5 nm and δ was 26.0×10−6; and the thickness of the third oxygen-deficient hafnium oxide layer was 3.3 nm and δ was 24.2×10−6.
Although it is difficult to derive a correct composition from these δ values, a rough estimation can be made from the facts that δ of hafnium metals is 31.2×10−6 and δ of HfO2 having a stoichiometric composition is 24.2×10−6, and so on. That is to say, since δ of the second oxygen-deficient hafnium oxide layer is intermediate between δ of hafnium metals and δ of HfO2, it appears to be a hafnium oxide having a non-stoichiometric composition as originally intended.
In addition, based on the δ values, the first and the third oxygen-deficient hafnium oxide layers are predicted to be HfO1.94 approximately, and are thus estimated to be hafnium oxides very close to HfO2 (oxygen content percentage of 66.7%) having a stoichiometric composition.
The following describes the analysis using the X-ray reflectometry.
First, for the purposes of simplifying later calculations, fitting is performed presuming that the variable resistance layer has a double-layered structure. More specifically, the fitting is performed to calculate δ and the thickness of a high oxygen containing layer and a low oxygen containing layer, assuming that the high oxygen containing layer exists near the upper electrode and that the layer other than the high oxygen containing layer is the low oxygen containing layer (calculation 1). It is to be noted that the fitting is performed using a least squares method.
The calculation 1 roughly determines the values of the thickness and δ of the high oxygen containing layer which exists near the upper electrode.
The following calculation is performed assuming that the variable resistance layer has a triple-layered structure.
Fitting is performed to calculate a new δ and thicknesses of the first, second, and third variable resistance films under such conditions that δ and the thickness of the high oxygen containing layer are the initial values of δ and the thickness of the first variable resistance film and that δ of the first variable resistance film is the same as δ of the third variable resistance film (calculation 2). With this calculation, δ and the thicknesses of the first, second, and third variable resistance films were determined through the fitting.
Generally, in the case of using the X-ray reflectometry, the layer farthest from the surface in the triple-layered structure cannot be easily measured. The reason why this calculation of the layered structure of the variable resistance films was performed using highly precise data on the third variable resistance layer as the data on the first layer is because the element I exhibited a resistance change phenomenon both in B mode and A mode, which led to an inference that a high oxygen containing layer similar to the third layer was formed also near the lower electrode.
Table 4 shows the result of measuring the samples G to K using the X-ray reflectometry in the same manner as above.
From the result of measuring the samples H, I and J which were made under the same conditions as the well-operated elements H, I and J, respectively, it can be said that a preferable thickness of the m first or third layer is between 3 nm and 4 nm inclusive. It can also be said that a preferable y value of the first or third layer is 1.8<y<2.
It is to be noted that the first and third hafnium oxide layers can also be formed through deposition using sputtering or chemical vapor deposition. In the case of sputtering, for example, first, sputtering is performed under a condition of a high oxygen gas flow ratio at the time of deposition, so that a hafnium oxide having a high oxygen content percentage and high resistance can be formed. For example, in the case of the sputtering condition described above in relation to the manufacturing method of the variable resistance element 1100, the hafnium oxide layers can be formed by setting the oxygen gas flow ratio to 4% approximately.
The following considers the mechanism of the resistance change phenomenon observed in the fifth experiment, and examines based on the mechanism, the structure of the variable resistance element which can uniquely exhibit the resistance change phenomenon either in A mode or B mode.
First, it is to be noted that the roles of the first hafnium oxide layer 1104 and the third hafnium oxide layer 1106 are not clearly known because the mechanism of the resistance change phenomenon is not understood well at present.
In this experiment, however, considering that the resistance changes of the variable resistance nonvolatile element occur due to movements of oxygen ions in the interface between an electrode and a hafnium oxide layer, the first hafnium oxide layer 1104 and the third hafnium oxide layer 1106 may have a role of effectively applying a voltage to a region near the interface. That is, the resistance change phenomenon seems to occur as oxygen ions gather and diffuse near the interface between the lower electrode 1103 and the first hafnium oxide layer 1104 and near the interface between the upper electrode 1108 and the third hafnium oxide layer 1106 due to an electric field.
To be more specific, applying a positive voltage to the upper electrode 1108 causes negatively-charged oxygen ions to gather on the upper electrode 1108 side, forming a high resistance layer, which means that the element performs the resistance change to thigh resistance state. On the other hand, applying a negative voltage causes diffusion of the oxygen ions in the hafnium oxide layer, resulting in the resistance change to low resistance state of the element.
Here, since the third hafnium oxide layer 1106, which is a high resistance layer, exists in contact with the upper electrode 1108, a large voltage is applied to the third hafnium oxide layer 1106, causing injection of oxygen into the hafnium oxide layer 1106. As a result, the oxygen content percentage increases and the third hafnium oxide layer 1106 approaches HfO2 having a stoichiometric composition known as an insulator. That is to say, the third hafnium oxide layer 1106 plays a part in the resistance changes.
The gathering and diffusion of oxygen ions can also take place on the lower electrode 1103 side. In that case, the first hafnium oxide layer 1104, which is a high resistance layer provided in contact with the lower electrode 1103, plays a part in the resistance changes.
With such a mechanism as described above, in the case where the third hafnium oxide layer 1106 plays a larger part in the resistance changes, it appears that the resistance changes in B mode occur; that is, the resistance change to high resistance state is performed upon application, to the upper electrode 1108 provided in contact with the third hafnium oxide layer 1106, of electrical pulses having a voltage higher than a voltage applied to the lower electrode 1103, whereas the resistance change to low resistance state is performed upon application of a negative voltage.
On the other hand, in the case where the first hafnium oxide layer 1104 plays a larger part in the resistance changes, it appears that the resistance changes in A mode occur; that is, the resistance change to high resistance state is performed upon application, to the lower electrode 1103 provided in contact with the first hafnium oxide layer 1104, of electrical pulses having a voltage higher than a voltage applied to the upper electrode 1108, whereas the resistance change to low resistance state is performed upon application of a negative voltage.
The experimental result that the element G, in which the first hafnium oxide layer 1104 and the third hafnium oxide layer 1106 have a low oxygen content percentage, did not exhibit the resistance changes indicates that the voltage is evenly applied to the hafnium oxide layer 1105 if the first hafnium oxide layer 1104 and the third hafnium oxide layer 1106, which are high resistance layers, do not exist, and thus high resistance layers which are almost insulators are not easily formed near the electrodes. As a result, it appears that the resistance change phenomenon does not occur easily.
Further, taking into account that the hafnium oxide layers near the electrodes have a role of locally applying a large voltage to a region near the electrodes in the variable resistance layer, it seems that the resistance of the hafnium oxide layers near the electrodes need to be larger than the resistance of the second hafnium oxide layer, which is an oxygen supplying layer. For this reason, it is sufficient if the first or third hafnium oxide layer expressed as HfOy satisfies x<y<2. As also for the thickness of the first or third hafnium oxide layer, it is sufficient as along as the thickness falls within a range which allows the first or third hafnium oxide layer to locally apply a large voltage.
From the viewpoint of simple manufacturing, the first and third hafnium oxide layers having a thickness of 1 nm or greater are suitable for implementation. Moreover, from the viewpoint of increase in the element resistance which is to be brought about by future miniaturization, a thickness of 5 nm or smaller is probably suitable for implementation.
(Preferable Structure of Variable Resistance Element using Hafnium Oxide as Variable Resistance Layer)
Such a mechanism as described above seems to indicate that the variable resistance element which can uniquely exhibit the resistance changes either in A mode or B mode can be obtained by providing only one of the first hafnium oxide layer 1104 and the third hafnium oxide layer 1106 of the variable resistance layer.
This idea is also supported by the results of the first to fourth experiments related to the variable resistance element in which a variable resistance layer containing a tantalum oxide is used; that is, the results that the variable resistance element in which a tantalum oxide layer having a higher oxygen content percentage is provided in contact with only one of the upper electrode and the lower electrode uniquely exhibited the resistance changes in A mode or B mode.
With a variable resistance element 1100A shown in
With the variable resistance element 1100A having such a structure, the hafnium oxide layer 1104 provided in contact with the lower electrode 1103 appears to play a part in the resistance changes and to cause the resistance changes in A mode.
As for a variable resistance element 1100B shown in
With the variable resistance element 1100B having such a structure, the hafnium oxide layer 1106 provided in contact with the upper electrode 1108 appears to play a part in the resistance changes and to cause the resistance changes in B mode.
The present invention also includes a variable resistance nonvolatile storage device having the variable resistance element 1100A and a variable resistance nonvolatile storage device having the variable resistance element 1100B. Such variable resistance nonvolatile storage devices also achieve the same advantageous effect as that of the variable resistance nonvolatile storage device described in the embodiment.
Next, the following describes the sixth experiment regarding a variable resistance nonvolatile storage element which uses an oxygen-deficient zirconium oxide as the variable resistance layer.
As shown in
The materials of the lower electrode 2103 and the upper electrode 2107 are, for example, Pt (platinum), Ir (iridium), Pd (palladium), Ag (silver), Ni (nickel), W (tungsten), Cu (copper), Al (aluminum), Ta (tantalum), Ti (titanium), TiN (titanium nitride), TaN (tantalum nitride), and TiAIN (titanium aluminum nitride).
Although the substrate 2101 may be a single-crystal silicon substrate or a semiconductor substrate, it is not limited to these substrates. The variable resistance layer 2106 can be formed at a relatively low substrate temperature and thus can be formed on a resin material and the like.
Next, a manufacturing method of the variable resistance element 2100 used in the sixth experiment is described with reference to
First, as shown in
Next, as shown in
Subsequently, by sputtering, a Pt thin film having a thickness of 150 nm is formed on the second zirconium oxide layer 2105 as the upper electrode 2107. Lastly, photolithography processing creates a pattern 2108 using a photoresist, and dry etching produces an element region 2109.
With the above-described manufacturing method, three types of variable resistance elements were manufactured under different manufacturing conditions. The following describes the details.
First, the layered structure of the substrate 2101, the oxide layer 2102, and the lower electrode 2103 comprising Pt was formed in the above-described manner. After that, the first zirconium oxide layer 2104 was formed on the lower electrode 2103 by reactive sputtering, that is, a process of sputtering a zirconium target in an argon gas and an oxygen gas (
The film forming condition here was as follows: vacuum (back pressure) in the sputtering apparatus was approximately 2×10−5 Pa before sputtering; power was 300 W when sputtering; the total gas pressure of the argon gas and the oxygen gas was 0.93 Pa; the flow ratios of the oxygen gas were 2.0%, 2.7% and 3.3%; the preset temperature of the substrate was 25° C.; and the time period for film formation was approximately four minutes.
With this, the first zirconium oxide layer 2104 was deposited with a thickness of about 30 nm to 40 nm and an oxygen content percentage of about 48%, about 59%, or about 67% corresponding to one of the above mentioned oxygen gas flow ratios; in other words, the first zirconium oxide layer 2104 was deposited which can be expressed as ZrOx, where x is 0.9, 1.4 or 2.0.
In manufacturing the variable resistance elements, the formation of the first zirconium oxide layer 2104 and the second zirconium oxide layer 2105 was followed by the formation of the upper electrode 2107, both performed in the sputtering apparatus. More specifically, after depositing the first zirconium oxide layer 2104, a shutter was inserted between the zirconium target and the oppositely placed substrate 2101 with no change to the sputtering condition, including the condition of the gas pressure and the power, and it was kept in this state for about 10 to 30 seconds.
This caused oxidation of the top surface of the first zirconium oxide layer 2104 due to oxygen plasma. As a result, the second zirconium oxide layer 2105 whose oxygen content percentage is higher than that of the first zirconium oxide layer 2104 was formed on the surface of the first zirconium oxide layer 2104.
Thereafter, the upper electrode 2107 comprising Pt was formed on the second zirconium oxide layer 2105 in the above-described manner (
Then, the element region 2109 was formed through the photolithography processing (
With the above processes, three types of variable resistance elements were manufactured under different manufacturing conditions. These variable resistance elements are referred to as elements L, M, and N. It is to be noted that the element region 2109 of the elements L, M, and N has a circular pattern with a diameter of 3 μm.
To analyze the structure of the variable resistance layer 2106 in the sixth experiment, samples were prepared by, under the same conditions as that of the elements L, M, and N, depositing a zirconium oxide on a substrate in which an oxide layer having a thickness of 200 nm is formed on a single-crystal silicon substrate, followed by oxygen plasma irradiation. These samples are referred to as samples L, M, and N.
Table 5 summarizes the oxygen gas flow ratio and a later-described analysis of each sample. It is to be noted that Pt equivalent to the upper electrode 2107 is not deposited on the samples L to N, and thus the variable resistance layer 2106 is exposed.
In the sixth experiment, too, using the above-described X-ray reflectometry, an X-ray was emitted to enter the surface of each sample at a shallow angle, and the intensity of the X-ray reflected was measured. Then, fitting was performed presuming an appropriate structural model for an obtained spectrum, so as to evaluate the thickness and the refractive index of the variable resistance layer in the samples L to N. Here, the parameters for the fitting are the layered structure of the variable resistance layer and the thickness and δ (=1−refractive index) of each layer.
Referring to
The result of an analysis of the sample M when fitting was performed presuming this double-layered structure showed the following values as shown in Table 5: the thickness of the first zirconium oxide layer was 38.5 nm and δ was 17.2×10−6; and the thickness of the second zirconium oxide layer was about 3.9 nm and δ was 16.5×10−6.
Generally, δ can be theoretically calculated based on the density of the film; δ of metal zirconium having a density of 6.798 g/cm3 is 19.0×10−6, and δ of ZrO2 having a density of 5.817 g/cm3 is 16.9×10−6. Comparison between these values and the above obtained values appears to indicate that the first zirconium oxide layer is a zirconium oxide such as ZrO1.42 which is an oxide deficient in oxygen and is clearly different from the stoichiometric composition of zirconium oxide. Further, as for the second zirconium oxide layer, calculating its compositional ratio using the δ value gives ZrO1.97. This indicates that the second zirconium oxide layer is an oxide similar to ZrO2 yet seems to be an oxygen-deficient oxide which is different from the stoichiometric composition.
As shown in Table 5, the sample L exhibited almost the same results. More specifically, when the first zirconium oxide layer is expressed as ZrOx, its thickness is approximately 33.5 nm and x is about 0.93, and when the second zirconium oxide layer is expressed as ZrOy, its thickness is approximately 5.0 nm and y is about 1.79.
In comparison, referring to the analysis result of the sample N shown in
Here, as shown in Table 5, when the single zirconium oxide layer is expressed as ZrOx, its thickness is approximately 41.9 nm and the x is about 2.0 which indicates a stoichiometric compositional ratio.
The element L and the sample L underwent the sputtering and the oxygen plasma irradiation under the same condition. The element M and the sample M also underwent the sputtering and the oxygen plasma irradiation under the same condition. Thus, in the elements L and M as in the samples L and M, the second zirconium oxide layer 2105 probably exists between the first zirconium oxide layer 2104 and the upper electrode 2107.
Thus, it can be said that the second zirconium oxide layer 2105 having a thickness of 3.9 nm and a compositional ratio in which y is 1.97 is formed in the element M as in the sample M.
Although the sixth experiment used the X-ray reflectivity measurement for analyzing the second zirconium oxide layer, other instrumental analysis techniques can also be used, including Auger electron spectroscopy analysis (AES), fluorescent X-ray photoelectron spectroscopy analysis (XPS), and electron probe microanalysis (EPMA: also known as WDS, EDS, or EDX depending on the detection method).
Next, the following describes: the characteristics observed when electrical pulses are applied to the element M to cause the resistance change in the element M, which is a representative example of the double-layered structure including the first zirconium oxide layer 2104 and the second zirconium oxide layer 2105 which were actually manufactured in the sixth experiment; and the characteristics observed when electrical pulses are applied to the element N in which the second zirconium oxide layer 2105 does not seem to exist.
First, referring to
After that, the element M exhibited the above-described resistance changes in B mode in which application of electrical pulses having a negative voltage of −1.0 V causes a decrease in the resistance value to about 110Ω and application of electrical pulses having a positive voltage of +1.7 V causes an increase in the resistance value to about 3 kit. Then, reversible resistance changes occurred stably between 110Ω and 3 kΩ.
Referring to
These results clarified that reversible resistance changes can be observed only when the second zirconium oxide layer 2105 exists. However, when the second zirconium oxide layer 2105 having a high oxygen content percentage does not exist, a pulse having a width of 100 ns, for example, cannot bring about the resistance change phenomenon. This indicates that the second zirconium oxide layer is essential to bring about high-speed resistance changes. The composition of this second zirconium oxide layer, when expressed as ZrOy, is sufficient as long as it satisfies x<y<2. A further preferable value of y is 1.9<y<2.
From the viewpoint of simple manufacturing, the second zirconium oxide layer having a thickness of 1 nm or greater is suitable for implementation. Moreover, from the viewpoint of increase in the element resistance which is to be brought about by future miniaturization, a thickness of 5 nm or smaller is probably suitable for implementation.
The above results show a tendency identical to that shown in the result of the fifth experiment in relation to the variable resistance element in which the variable resistance layer comprising the hafnium oxide is used. The variable resistance element in which the variable resistance layer comprising the zirconium oxide is used also appears to exhibit resistance changes with the same mechanism as that examined in the fifth experiment.
This idea is supported also by the fact that the element M exhibited the stable resistance changes in B mode after a few and low voltage forming process. That means, the element M exhibited the resistance changes in B mode probably because the zirconium oxide layer 2105 provided in contact with the upper electrode 2107 and having a high oxygen content percentage played a part in the resistance changes.
This suggests that by providing the zirconium oxide layer 2105, which has a high oxygen content percentage, in contact with the lower electrode 2103, a variable resistance element can be obtained which exhibits the resistance changes in A mode.
With a variable resistance element 2100A shown in
With the variable resistance element 2100A having such a structure, the zirconium oxide layer 2104 provided in contact with the lower electrode 2103 appears to play a part in the resistance changes and to cause the resistance changes in A mode.
The present invention also includes a variable resistance nonvolatile storage device having the above-described variable resistance element 2100 and a variable resistance nonvolatile storage device having the above-described variable resistance element 2100A. Such variable resistance nonvolatile storage devices also achieve the same advantageous effect as that of the variable resistance nonvolatile storage device described in the embodiment.
Although only an exemplary embodiment of the variable resistance nonvolatile storage device according to the present invention has been described above, the present invention is not limited to this embodiment. Various modifications which can be conceived by those skilled in the art without materially departing from the novel teachings and advantages of the present invention are intended to be included within the scope of the present invention.
For example, although the embodiment has described, as an example, the nonvolatile storage device that uses a variable resistance element in which the upper electrode and the lower electrode both comprise Pt, it is also possible to use a variable resistance element in which the upper electrode and the lower electrode comprise different materials.
From another experiment conducted by the inventors of the present application, it was found that the mode of the resistance change characteristics is also unique in a variable resistance element formed by providing a variable resistance layer comprising a transition metal oxide having a uniform oxygen content percentage, between a second electrode comprising a material having a standard electrode potential which is higher than that of the transition metal of the transition metal oxide and a first electrode comprising a material having a standard electrode potential which is lower than that of the second electrode.
Although the details are omitted, the experiment showed a result that the above-described resistance changes in B mode occurred in, for example, a variable resistance element in which the upper electrode comprises Pt that is an example of the second electrode material and the lower electrode comprises W (tungsten) that is an example of the first electrode material. Here, as compared to W, Pt has a high standard electrode potential and is not easily oxidized.
In this case, when voltages having different polarities are alternately applied between the upper electrode and the lower electrode, the oxygen ions are not absorbed into the upper electrode comprising Pt that is not easily oxidized, but are absorbed into the lower electrode comprising W that is easily oxidized. Thus, the gathering and diffusion of the oxygen ions occur more actively near the interface between the variable resistance layer and the upper electrode, thereby causing the resistance changes in B mode.
In view of the above, the electrode provided in contact with a transition metal oxide layer having a higher oxygen content percentage may comprise Pt or the like having a higher standard electrode potential (not easily oxidized), whereas the electrode provided in contact with a transition metal oxide layer having a lower oxygen content percentage may comprise W or the like having a lower standard electrode potential (easily oxidized).
This way, the oxygen ions are not absorbed into the Pt electrode from the transition metal oxide layer having a higher oxygen content percentage, but are absorbed into the W electrode from the transition metal oxide layer having a lower oxygen content percentage, and thus there is a greater tendency in which the oxygen ions gather and diffuse in the transition metal oxide layer having a higher oxygen content percentage. As a result, the mode of the resistance change characteristics is strongly fixed.
This is more preferable in implementing a characteristic structure of the present invention in which a memory cell is formed by connecting a variable resistance element and a transistor in a direction in which a substrate bias effect in the transistor is smaller, depending on the mode of the variable resistance element.
The inventors of the present application have verified the aforementioned concept holds true by conducting the seventh experiment using the actual variable resistance element having a layered structure in which variable resistance layers comprising a transition metal oxide are stacked.
In the seventh experiment, an element O and an element P were manufactured according to the method of manufacturing a variable resistance element described in Embodiments. The element O includes (i) a variable resistance layer comprising a tantalum oxide as an example of one of oxygen-deficient oxides, (ii) an upper electrode comprising Pt as an example of a material having a higher standard electrode potential than Ta, and (iii) a lower electrode comprising TaN as an example of a material having a lower standard electrode potential than that in the upper electrode. The element P includes (i) a variable resistance layer comprising the tantalum oxide as an example of one of the oxygen-deficient oxides, and (ii) the upper electrode and the lower electrode both comprising Pt as an example of a material having a higher standard electrode potential than Ta.
According to the measurement by the inventors, the standard electrode potential of TaN is +0.48 eV. Furthermore, according to Non Patent Literature 2 (referred to as NPL 2 hereinafter), “CRC HANDBOOK of CHEMISTRY and PHYSICS”, DAVID R. LIDE, Editor-in-chief, 84th Edition 2003-2004, CRC PRESS, the standard electrode potential of Pt is +1.18 eV, and the standard electrode potential of Ta is −0.6 eV.
The element O includes the lower electrode comprising TaN and having a thickness of 50 nm, the first tantalum oxide layer comprising TaOx (x=1.57) and having a thickness of 43 nm, the second tantalum oxide layer comprising TaOy (y=2.47) and having a thickness of 7 nm, and the upper electrode comprising Pt and having a thickness of 50 nm.
The element P includes the lower electrode comprising Pt and having a thickness of 50 nm, the first tantalum oxide layer comprising TaOx (x=1.59) and having a thickness of 44 nm, the second tantalum oxide layer comprising TaOy (y=2.38) and having a thickness of 7 nm, and the upper electrode comprising Pt and having a thickness of 50 nm.
Each of the elements O and P was a square having a side of 0.5 μm. The second tantalum oxide layer of each of the elements O and P was formed by oxidizing the top surface of the first tantalum oxide layer through the plasma oxidation process. Each of the values of the thicknesses and compositions is an actual measured value obtained by measuring an X-ray reflectivity.
Since in each of the elements O and P, the second tantalum oxide layer that is in contact with the upper electrode was set to an oxygen content percentage higher than that of the first tantalum oxide layer that is in contact with the lower electrode, the element underwent a resistance change in the B mode.
Furthermore, intending that a mode of the resistance change in the element O was stably fixed to the B mode, the upper electrode comprised Pt having the standard electrode potential higher than that of Ta (i.e. Pt is hardly oxidized), and the lower electrode comprised TaN having the standard electrode potential lower than that of Pt (i.e. TaN is easily oxidized).
The hysteretic resistance-voltage characteristics in the resistance change of these elements O and P were measured. In this measurement, the electrical pulses were applied to each of the elements without passing though a load resistor or a transistor, and the resistance value of the element was measured.
As apparent from the comparison between
As a result, a mode of the resistance change in the element O was more stably fixed to the B mode than that of the element P as intended. Consequently, it has been verified that the element O stably exhibited the clear resistance change with the larger variation range of resistance values.
Next, the eighth experiment has been conducted to measure the resistance change characteristics of a variable resistance element with application of electrical pulses.
In the eighth experiment, elements Q and R each comprising different materials in the upper and lower electrodes were further manufactured.
The element Q includes the lower electrode comprising TaN and having a thickness of 50 nm, the first tantalum oxide layer comprising TaOx (x=1.57) and having a thickness of 44 nm, the second tantalum oxide layer comprising TaOy (y=2.47) and having a thickness of 6 nm, and the upper electrode comprising Ir and having a thickness of 50 nm.
The element R includes the lower electrode comprising TaN and having a thickness of 50 nm, the first tantalum oxide layer comprising TaOx (x=1.57) and having a thickness of 44 nm, the second tantalum oxide layer comprising TaOy (y=2.47) and having a thickness of 6 nm, and the upper electrode comprising Pd and having a thickness of 50 nm.
According to NPL 2, the standard electrode potential of Ir is +1.156 eV, and the standard electrode potential of Pd is +0.951 eV.
Each of the elements Q and R was a square having a side of 0.5 p.m. The second tantalum oxide layer of each of the elements Q and R was formed by oxidizing the top surface of the first tantalum oxide layer through the plasma oxidation process. Each of the values of the thicknesses and compositions is an actual measured value obtained by measuring an X-ray reflectivity.
Furthermore, intending that a mode of the resistance change in each of the elements Q and R was stably fixed to the B mode as the element O, the upper electrode comprised Ir and Pd having the standard electrode potential higher than that of Ta (i.e. Ir and Pd are hardly oxidized), and the lower electrode comprised TaN having the standard electrode potential lower than those of Ir and Pd (i.e. TaN is easily oxidized).
The resistance values of each of the elements Q, R, O, and P were measured with each application of alternate electrical pulses of a negative voltage of −1.5 V and a positive voltage of +2.0 V, through a load resistor of 1 kΩ. The resistance change characteristics of each of the elements O, P, Q, and R underwent the resistance change in the B mode, where the element is changed to the low resistance state with application of the negative voltage, and the element is changed to the high resistance state with application of the positive voltage.
As apparent from the comparison between
As a result, it has been verified that the variable resistance element comprising different materials in the upper and lower electrodes such that the resistance change characteristics were stably fixed to one of modes is superior to the variable resistance element comprising the same material in the upper and lower electrodes, because the variable resistance element comprising the different materials stably exhibited the clear resistance change with the larger variation range of resistance values.
Next, the ninth experiment has been conducted to study the characteristics of a memory cell with a transistor for practical use, using a 1T1R circuit in which a variable resistance element is connected in series with a transistor. Such a 1T1R circuit is an example of an equivalent circuit of a memory cell for practical use.
In the ninth experiment, 1T1R circuits were manufactured which include: 1T1R circuits in each of which a transistor is connected in series with a variable resistance element including a lower electrode comprising TaN and an upper electrode comprising Pt as the electrode materials comprised in the element O; and 1T1R circuits in each of which a transistor is connected in series with a variable resistance element including a lower electrode and an upper electrode both comprising Pt as the electrode material comprised in the element P.
Each of the transistors in the 1T1R circuits was turned on with application of a gate voltage of +2.4 V, and a resistance change occurred in each of the variable resistance elements with application of alternate electrical pulses of −1.8 V and +1.8 V between the upper electrode of the variable resistance element and a source of the transistor that are both ends of the 1T1R circuit, with reference to the source. In each resistance change, a current value was measured with application of a reading voltage of 0.4 V to each of the 1T1R circuits.
As apparent from
As a result, it has been verified that the variable resistance element including the upper and lower electrodes comprising different materials such that the resistance change characteristics were stably fixed to one of modes is superior to the variable resistance element including the upper and lower electrodes comprising the same material, in that even the 1T1R circuit including the variable resistance element and the transistor stably exhibited the clear resistance change with the larger difference in the average current values between the high resistance state and the low resistance state.
As described above, the variable resistance element includes (i) a variable resistance layer having a layered structure in which transition metal oxide layers with different oxygen content percentages are stacked, and (ii) the upper and lower electrodes comprising different materials such that the resistance change characteristics are stably fixed to one of modes. Such a variable resistance element has an advantage of the clear resistance change with the larger variation range of resistance values than that of the variable resistance element including the upper and lower electrodes comprising the same material.
The advantage is probably based on the next mechanism.
Since in each of the elements O, P, Q, and R, the second tantalum oxide layer that is in contact with the upper electrode was set to an oxygen content percentage higher than that of the first tantalum oxide layer that is in contact with the lower electrode, the gathering and diffusion of oxygen ions occurred more actively near the interface with the upper electrode of the second tantalum oxide layer, thereby causing the resistance change in the B mode.
Furthermore, intending that the resistance change characteristics of the elements O, Ω, and R were stably fixed to one of modes, the upper electrode comprised Pt, Ir, and Pd having the standard electrode potential higher than that of Ta (hardly oxidized), and the lower electrode comprised TaN having the standard electrode potential lower than those of Pt, Ir, and Pd (easily oxidized).
In this case, when voltages having different polarities were alternately applied between the upper electrode and the lower electrode, the mode of the resistance change was stably fixed to the B mode. This is because oxygen ions were not absorbed into the upper electrode comprising Pt, Ir, and Pd that are not easily oxidized, but were absorbed into the lower electrode comprising TaN that is easily oxidized. Thus, gathering and diffusion of the oxygen ions probably occurred more actively near the interface with the upper electrode of the variable resistance layer.
Probably, such a mechanism generally works when the aforementioned magnitude relationship in standard electrode potential between a transition metal comprised in a variable resistance layer, an upper electrode, and a lower electrode is satisfied.
Thus, it can be expected that a variable resistance element including a variable resistance layer having a layered structure in which hafnium oxide layers or zirconium oxide layers are stacked stably exhibits the clear resistance change with the larger variation range of resistance values, based on the mechanism using different and suitable materials in the upper and lower electrodes, as the variable resistance element including a variable resistance layer having a layered structure in which tantalum oxide layers are stacked.
Further, one may consider forming the first transition metal oxide layer and the second transition metal oxide layer included in the variable resistance layer of the variable resistance element, using oxide layers comprising different types of transition metals. In this case, according to the types of the transition metals used, the oxygen content percentage of each oxide layer is determined such that the resistance value of the first transition metal oxide layer is greater than that of the second transition metal oxide layer.
As described earlier, in the variable resistance layer, the second transition metal oxide layer whose resistance value is higher than that of the first transition metal oxide layer has a role of locally applying a large voltage. As described above, it appears that with this voltage, the gathering and diffusion of the oxygen ions playing a part in the resistance changes occur more actively in the second transition metal oxide layer, and as a result, the mode of the resistance change characteristics is fixed.
With this idea, even when different types of transition metals are used, it is probably possible, as in the embodiment, to obtain a variable resistance element which can fix the mode of the resistance change characteristics by making the resistance value of the second transition metal oxide layer larger than the resistance value of the first transition metal oxide layer.
That is to say, the present invention also includes the variable resistance element in which the first transition metal oxide layer and the second transition metal oxide layer of the variable resistance layer are oxide layers comprising different types of transition metals.
Although not described above, it is a known technique to implant predetermined impurities, such as an additive for adjusting the resistance value, into the variable resistance layer of the variable resistance element. This technique may be applied to the variable resistance element used in the variable resistance nonvolatile storage device of the present invention. For example, adding nitrogen to the variable resistance layer leads to an increase in the resistance value of the variable resistance layer, thereby improving the reactivity of the resistance changes.
More specifically, regarding the variable resistance element in which an oxygen-deficient transition metal oxide is used for the variable resistance layer, the limitation in the claims that the variable resistance layer includes: a first region containing a first oxygen-deficient transition metal oxide having a composition MOx; and a second region containing a second oxygen-deficient transition metal oxide having a composition MOy (where x<y) is not intended to exclude the first region and the second region containing predetermined impurities (e.g. an additive for adjusting the resistance value) other than the corresponding compositional transition metal oxides.
As described above, the present invention makes it possible to provide, in a small layout area, a variable resistance nonvolatile storage device formed with 1T1R memory cells each using a variable resistance element. Thus, the present invention is useful in providing a memory which is high in integration and small in area, for example.
Number | Date | Country | Kind |
---|---|---|---|
2009-252407 | Nov 2009 | JP | national |
Filing Document | Filing Date | Country | Kind | 371c Date |
---|---|---|---|---|
PCT/JP2010/006453 | 11/2/2010 | WO | 00 | 10/28/2011 |