Method of manufacturing an array of MRAM-cells, method of writing the MRAM-cells
阅读说明：本技术 制造mram单元的阵列的方法、写入该mram单元的方法 (Method of manufacturing an array of MRAM-cells, method of writing the MRAM-cells ) 是由 应继锋 王仲盛 牛宝华 于 2019-07-15 设计创作，主要内容包括：一种制造磁性随机存取存储器单元的阵列的方法包括写入磁性随机存取存储器单元。写入存储器单元包括确定所述存储器单元的阵列的最佳写入电流；以及将所述最佳写入电流应用于所述阵列中的第一存储器单元。将第一读取电流应用于第一存储器单元以响应于应用所述最佳写入电流,确定第一存储器单元的磁取向是否已经改变。当第一存储器单元的磁取向没有改变时,将第二写入电流应用于第一存储器单元。第二写入电流不同于最佳写入电流。将第二读取电流应用于第一存储器单元,以响应于应用第二读取电流,确定第一存储器单元的磁取向是否改变。本发明实施例还涉及写入磁性随机存取存储器单元的方法。(A method of fabricating an array of magnetic random access memory cells includes writing magnetic random access memory cells. Writing to a memory cell comprises determining an optimal write current for the array of memory cells; and applying the optimal write current to a first memory cell in the array. A first read current is applied to the first memory cell to determine whether the magnetic orientation of the first memory cell has changed in response to applying the optimal write current. The second write current is applied to the first memory cell when the magnetic orientation of the first memory cell is not changed. The second write current is different from the optimal write current. A second read current is applied to the first memory cell to determine whether the magnetic orientation of the first memory cell has changed in response to applying the second read current. Embodiments of the present invention also relate to methods of writing magnetic random access memory cells.)
1. A method of fabricating an array of magnetic random access memory cells, comprising:
a write magnetic random access memory cell comprising:
determining an optimal write current for the array of magnetic random access memory cells;
applying the optimal write current to a first magnetic random access memory cell in the array;
applying a first read current to the first magnetic random access memory cell to determine whether a magnetic orientation of the first magnetic random access memory cell has changed in response to the applying the optimal write current;
applying a second write current to the first magnetic random access memory cell when the magnetic orientation of the first magnetic random access memory cell is not changed, wherein the second write current is different from the optimal write current; and
applying a second read current to the first magnetic random access memory cell to determine whether a magnetic orientation of the first magnetic random access memory cell has changed in response to the applying the second read current.
2. The method of fabricating an array of magnetic random access memory cells of claim 1, further comprising:
applying a third write current to the first magnetic random access memory cell when the magnetic orientation of the first magnetic random access memory cell is not changed after applying the second write current, wherein the third write current is different from the optimal write current and the second write current; and
applying a third read current to the first magnetic random access memory cell to determine whether a magnetic orientation of the first magnetic random access memory cell has changed in response to the applying the third write current.
3. The method of fabricating an array of magnetic random access memory cells of claim 2, wherein the second write current is greater than the optimal write current and the third write current is less than the optimal write current.
4. The method of fabricating an array of magnetic random access memory cells of claim 2, wherein the second write current is less than the optimal write current and the third write current is greater than the optimal write current.
5. The method of fabricating an array of magnetic random access memory cells of claim 2, further comprising: iteratively repeating applying a write current and a read current when the magnetic orientation of the first magnetic random access memory cell is not changed,
wherein iteratively applying the write current and the read current is stopped when the magnetic orientation of the first magnetic random access memory cell has changed, an
Wherein a write current at each application of the write current is different from any other write current.
6. The method of fabricating an array of magnetic random access memory cells of claim 5, wherein the write current applied at each iteration alternates between being greater than the optimal write current and being less than the optimal write current.
7. The method of fabricating an array of magnetic random access memory cells of claim 5, wherein each successive write current less than the optimal write current is less than a previous write current less than the optimal write current, and each successive write current greater than the optimal write current is greater than a previous write current greater than the optimal write current.
8. The method of fabricating an array of magnetic random access memory cells of claim 5, wherein the write current applied after said applying said optimal write current alternates between a write current less than said optimal write current and a write current greater than said optimal write current.
9. A method of fabricating an array of magnetic random access memory cells, comprising:
determining an optimal write current for the array of magnetic random access memory cells, comprising:
(a) applying a first write current to one of a plurality of magnetic random access memory cells in the array of magnetic random access memory cells;
(b) applying a first read current to one of the plurality of magnetic random access memory cells to determine whether a magnetic orientation of the one of the plurality of magnetic random access memory cells has changed in response to the applying the first write current;
(c) applying a second write current to one of the plurality of magnetic random access memory cells when the magnetic orientation of the one of the plurality of magnetic random access memory cells is not changed, wherein the second write current is different from the optimal write current;
(d) applying a second read current to one of the plurality of magnetic random access memory cells to determine whether the magnetic orientation of the one of the plurality of magnetic random access memory cells has changed in response to the applying the second write current, wherein the second read current has the same value as the first read current;
(e) applying a third write current to one of the plurality of magnetic random access memory cells when the magnetic orientation of the one of the plurality of magnetic random access memory cells has not changed after applying the second write current, wherein the third write current is different from the optimal write current and the second write current;
(f) applying a third read current to one of the plurality of magnetic random access memory cells to determine whether the magnetic orientation of the one of the plurality of magnetic random access memory cells has changed in response to the applying the third write current, wherein the third read current has the same value as the first read current and the second read current;
(g) iteratively repeating applying a write current and a read current when the magnetic orientation of the one of the plurality of magnetic random access memory cells is not changed;
wherein iteratively repeating applying a write current and a read current stops when the magnetic orientation of one of the plurality of magnetic random access memory cells has changed,
wherein the write current at each application of the write current is different from any other write current, an
Wherein the first write current has a first magnitude and the magnitude of the subsequently applied write current is increased in a stepwise manner;
(h) determining a value of a write current that causes a change in magnetic orientation of the magnetic random access memory cell;
(i) repeating (a) through (h) for each magnetic random access memory cell of the plurality of magnetic random access memory cells in the array of magnetic random access memory cells; and
(j) determining the optimal write current based on the determined write current for each of the plurality of magnetic random access memory cells.
10. A method of writing to a magnetic random access memory cell, comprising:
applying a first write current to the first magnetic random access memory cell;
determining whether a magnetic orientation of the first magnetic random access memory cell changes in response to the applying the first write current;
applying a second write current to the first magnetic random access memory cell when the magnetic orientation of the first magnetic random access memory cell is not changed, wherein the second write current is different from the first write current;
determining whether a magnetic orientation of the first magnetic random access memory cell has changed in response to the applying the second write current;
applying a third write current to the first magnetic random access memory cell when the magnetic orientation of the first magnetic random access memory cell has not changed after applying the second write current, wherein the third write current is different from the first write current and the second write current; and
determining whether a magnetic orientation of the first magnetic random access memory cell has changed in response to the applying the third write current.
Technical Field
Embodiments of the present invention relate to methods of fabricating an array of MRAM cells, methods of writing the MRAM cells.
Background
Magnetic Random Access Memory (MRAM) offers comparable performance to volatile solid State Random Access Memory (SRAM) and provides comparable density and lower power consumption to volatile Dynamic Random Access Memory (DRAM). MRAM provides faster access times and suffers minimal degradation over time compared to non-volatile memory (NVM) flash memory, however flash memory can only be rewritten a limited number of times. The MRAM cell is formed by a Magnetic Tunnel Junction (MTJ) comprising two ferromagnetic layers separated by a thin insulating barrier, and operates by electron tunneling between the two ferromagnetic layers through the insulating barrier.
Disclosure of Invention
An embodiment of the present invention provides a method of fabricating an array of magnetic random access memory cells, comprising writing a magnetic random access memory cell, the writing magnetic random access memory cell comprising: determining an optimal write current for the array of magnetic random access memory cells; applying the optimal write current to a first magnetic random access memory cell in the array; applying a first read current to the first magnetic random access memory cell to determine whether a magnetic orientation of the first magnetic random access memory cell has changed in response to the applying the optimal write current; applying a second write current to the first magnetic random access memory cell when the magnetic orientation of the first magnetic random access memory cell is not changed, wherein the second write current is different from the optimal write current; and applying a second read current to the first magnetic random access memory cell to determine whether a magnetic orientation of the first magnetic random access memory cell has changed in response to the applying the second read current.
Another embodiment of the present invention provides a method of manufacturing an array of magnetic random access memory cells, comprising determining an optimal write current for the array of magnetic random access memory cells, the determining an optimal write current for the array of magnetic random access memory cells comprising: (a) applying a first write current to one of a plurality of magnetic random access memory cells in the array of magnetic random access memory cells; (b) applying a first read current to one of the plurality of magnetic random access memory cells to determine whether a magnetic orientation of the one of the plurality of magnetic random access memory cells has changed in response to the applying the first write current; (c) applying a second write current to one of the plurality of magnetic random access memory cells when the magnetic orientation of the one of the plurality of magnetic random access memory cells is not changed, wherein the second write current is different from the optimal write current; (d) applying a second read current to one of the plurality of magnetic random access memory cells to determine whether the magnetic orientation of the one of the plurality of magnetic random access memory cells has changed in response to the applying the second write current, wherein the second read current has the same value as the first read current; (e) applying a third write current to one of the plurality of magnetic random access memory cells when the magnetic orientation of the one of the plurality of magnetic random access memory cells has not changed after applying the second write current, wherein the third write current is different from the optimal write current and the second write current; (f) applying a third read current to one of the plurality of magnetic random access memory cells to determine whether the magnetic orientation of the one of the plurality of magnetic random access memory cells has changed in response to the applying the third write current, wherein the third read current has the same value as the first read current and the second read current; (g) iteratively repeating applying a write current and a read current when the magnetic orientation of the one of the plurality of magnetic random access memory cells is not changed; wherein iteratively repeating applying a write current and a read current stops when the magnetic orientation of one of the plurality of magnetic random access memory cells has changed, wherein the write current at each application of the write current is different from any other write current, and wherein the first write current has a first magnitude and the magnitude of a subsequently applied write current increases in a stepwise manner; (h) determining a value of a write current that causes a change in magnetic orientation of the magnetic random access memory cell; (i) repeating (a) through (h) for each magnetic random access memory cell of the plurality of magnetic random access memory cells in the array of magnetic random access memory cells; and (j) determining the optimal write current based on the determined write current for each of the plurality of magnetic random access memory cells.
According to yet another embodiment of the present invention, there is provided a method of writing to a magnetic random access memory cell, the method comprising: applying a first write current to the first magnetic random access memory cell; determining whether a magnetic orientation of the first magnetic random access memory cell changes in response to the applying the first write current; applying a second write current to the first magnetic random access memory cell when the magnetic orientation of the first magnetic random access memory cell is not changed, wherein the second write current is different from the first write current; determining whether a magnetic orientation of the first magnetic random access memory cell has changed in response to the applying the second write current; applying a third write current to the first magnetic random access memory cell when the magnetic orientation of the first magnetic random access memory cell has not changed after applying the second write current, wherein the third write current is different from the first write current and the second write current; and determining whether a magnetic orientation of the first magnetic random access memory cell has changed in response to the applying the third write current.
Drawings
FIG. 1A is a schematic diagram of an MTJ MRAM cell according to an embodiment of the disclosure.
Fig. 1B is a schematic cross-sectional view of an MTJ film stack according to an embodiment of the disclosure.
Fig. 2A, 2B, and 2C show schematic cross-sectional views of magnetic layers of an MTJ film stack according to embodiments of the disclosure.
Fig. 3A and 3B illustrate memory operations of the MTJ cell.
Fig. 3C and 3D illustrate memory operations of the MTJ cell.
Fig. 4 shows an MRAM array.
FIG. 5A shows an optimal write current distribution for an array of magnetic random access memory cells.
FIG. 5B shows the error rate of an array of magnetic random access memory cells at different write currents.
FIG. 5C shows the error rate of an array of magnetic random access memory cells at a jump write current.
FIG. 5D compares the error rates of multiple applications of write current at constant write current and at jump write current for an array of magnetic random access memory cells.
FIG. 6 illustrates an iterative process of ramping write current according to an embodiment of the present disclosure.
Fig. 7 is a flow chart illustrating a method of writing an MRAM cell according to an embodiment of the present disclosure.
Fig. 8 shows a circuit for a hopping write scheme, where LV denotes a low voltage, HV denotes a high voltage, and Iref denotes a reference current, in fig. 8, according to an embodiment of the present disclosure.
Fig. 9 is a flow chart of a method of determining an optimal write current for an array of MRAM cells in accordance with an embodiment of the present disclosure.
FIG. 10 shows a programmable circuit for setting the step size and write current range of an iterative write current change, R in FIG. 10, in accordance with embodiments of the present disclosure1Rn represents resistors with different numbers, Is represents current source current, Rs represents current source resistor, Iout represents output current, and Vin tableThe input voltage is shown.
FIG. 11 shows a functional test circuit according to an embodiment of the present disclosure.
Detailed Description
It is to be understood that the following disclosure provides many different embodiments, or examples, for implementing different features of the invention. Specific embodiments or examples of components and arrangements are described below to simplify the present disclosure. These are, of course, merely examples and are not intended to be limiting. For example, the dimensions of the elements are not limited to the disclosed ranges or values, but may depend on the process conditions and/or desired performance of the device. Also, in the description that follows, forming a first feature over or on a second feature may include embodiments in which the first and second features are formed in direct contact, and may also include embodiments in which additional features may be formed between the first and second features such that the first and second portions are not in direct contact. Various components may be arbitrarily drawn in different scales for simplicity and clarity. In the drawings, some layers/components may be omitted for the sake of simplicity.
Furthermore, spatially relative terms (such as "below …," "below …," "lower," "above," "upper," etc.) may be used for ease of description to describe one element or component's relationship to another element(s) or component as illustrated in the figures. Spatially relative terms are intended to encompass different orientations of the device in use or operation in addition to the orientation depicted in the figures. The device may be otherwise oriented (rotated 90 degrees or at other orientations) and the spatially relative descriptors used herein interpreted accordingly. In addition, the term "made of …" may mean "including" or "consisting of …". Further, in the following manufacturing processes, there are one or more additional operations in or between the described operations, and the order of the operations may be changed. In this disclosure, the term "one of A, B and C" means "A, B and/or C" (a, B, C, a and B, a and C, B and C, or A, B and C), and does not mean one element from a, one element from B, and one element from C unless explicitly described otherwise.
The MRAM cell includes a multi-layer film stack that includes a magnetic layer. In some MRAM devices, one or more non-magnetic spacer layers may be inserted between multiple magnetic layers to optimize magnetic interaction, depending on the magnetic design. In some embodiments, one or more diffusion barrier layers may be inserted into the film stack to minimize adverse diffusion phenomena. Furthermore, in MTJ MRAM cells, each layer in the film stack, except for the tunneling barrier, needs to be conductive to maximize the read/write window.
In some embodiments, the materials for the seed layer, spacer layer, and/or diffusion barrier layer are appropriately selected to provide the particular crystal structure and orientation desired, without disrupting the magnetic interaction of the functional layer. In addition, the seed layer, spacer layer and diffusion barrier layer should be smooth, non-specifically oriented (amorphous), conductive and non-magnetic.
The magnetic tunneling function of an MTJ MRAM cell depends on the particular crystal structure and orientation of the MTJ film. In order to have the desired crystal structure and orientation of the MTJ film, the entire film stack needs to be grown on a smooth, amorphous, conductive, non-magnetic seed layer. Among various materials, tantalum (Ta) is most widely used as a seed layer, and can be easily grown as a smooth and amorphous layer. In addition, a non-magnetic spacer layer such as molybdenum (Mo) is commonly used in the MTJ film. In addition to tantalum (Ta) and molybdenum (Mo), cobalt (Co), platinum (Pt), iridium (Ir), and/or nickel (Ni) may also be used as a seed layer or spacer layer.
A plurality of crystalline magnetic layers are grown from the crystal lattice of the MgO layer, or the plurality of crystalline magnetic layers use the crystal lattice as a growth template.
FIG. 1A is a schematic diagram of an MTJ MRAM cell according to an embodiment of the disclosure. The
The
The
In some embodiments, the MTJ
As shown in fig. 1B, the MTJ
The second fixed
Layer 1303 is a spacer layer. In some embodiments, the spacer layer comprises Ta, Mo, Co, Pt, Ir, and/or Ni. In some embodiments, the spacer layer 1303 has a thickness in a range from about 0.2nm to about 0.5 nm. Layer 1304 is a cobalt iron boron (CoFeB) layer, a cobalt/palladium (CoPd) layer, and/or a cobalt iron (CoFe) layer. In some embodiments, the thickness of layer 1304 is in the range of about 0.8nm to about 1.5 nm.
The first fixed
In some embodiments, the free
In some embodiments, as shown in fig. 1B, the MTJ
In some embodiments, the
In some embodiments, the
In some embodiments, spacer layer 1303 and/or spacer layer 1402 include a layer of iridium and/or a layer of a double alloy of iridium and tantalum. Spacer layers for MTJ film stacks are typically required to have very smooth surface morphology and high conductivity, and to be substantially free of diffusion problems. In addition, the spacer layer should also tolerate a low level of oxidation without significant degradation of its conductivity. The thickness of spacer layers 1303 and/or 1402 is, in some embodiments, in a range from about 0.1nm to about 10nm and, in other embodiments, in a range from about 0.5nm to about 5.0 nm.
In some embodiments, the
The fixed magnetic layer, free magnetic layer, antiferromagnetic layer, and spacer/barrier layer may be formed by CVD, PVD or ALD or any other suitable film deposition method. The tunneling barrier layer may be formed by CVD, PVD or ALD or any other suitable film deposition method. The first and second electrode layers may also be formed by CVD, PVD, ALD or electroplating or any other suitable film deposition method.
In some embodiments, the
In some embodiments, the MRAM cell is formed over a dielectric material disposed over a substrate. In some embodiments, the substrate comprises silicon (Si) and/or silicon oxide or other suitable semiconductor material. Transistors, driver circuits, logic circuits, or any other electronic devices are formed from semiconductor materials and integrated with the MRAM cells.
Fig. 3A to 3D illustrate memory operations of the MTJ cell. As shown in fig. 3A-3D, the MTJ cell includes a fixed
In FIG. 3A, the fixed
If the same current I is forced by the
Fig. 4 shows an
A memory cell is read by asserting the word line of the cell, forcing a read current to flow through the bit line BL of the cell, and then measuring the voltage on the bit line BL. For example, to read the state of the target MTJ cell, the word line WL is asserted to turn on the transistor Tr. The free magnetic layer of the target MTJ cell is thus coupled to a fixed potential SL, e.g., a ground potential, through the transistor Tr. Next, a read current is forced to occur at the bit line BL. Since only the given read transistor Tr is turned on, the read current flows through the target MTJ cell to ground. Then, the voltage of the bit line BL is measured to determine the state ("0" or "1") of the target MTJ cell. In some embodiments, as shown in fig. 4, each MTJ cell has one read transistor Tr. Accordingly, this type of MRAM architecture is referred to as 1T 1R. In other embodiments, two transistors are assigned to one MTJ cell, forming a 2T1R system. Other cell array configurations may be employed.
Current MRAM testing and their field applications (eMRAM, cache, DRAM, and flash replacement) use constant bias (voltage or current) conditions during write/read testing and field applications. Because of the high sensitivity of the MRAM stack to intrinsic process variations, large die-to-die variations across the wafer, block-to-block variations across the die, and bit-to-bit variations across the block are observed. Different dies (dies located on different wafer locations, e.g., center die versus edge die), different blocks and individual bits of MRAM cells typically have significantly different read/write windows, and if a constant read/write bias is used (note that for large arrays of MRAM that can be a practical use of several hundred Megabytes (MB) to Gigabytes (GB), the die-chip can be as large as a 22mm x 32mm full mask field), the read/write window/margin (margin) cannot be tested at very high rates. Without correction, the overall wafer ensemble average constant write/degree window is too narrow to be practical and close to unusable for manufacturing, testing, and field applications. Because significant "optimum write current" differences are typically observed on bits within a small array, there is a need to improve write operations.
In embodiments of the present disclosure, a jump algorithm is employed in executing each write command (as compared to a constant write current) to further reduce the write failure rate in operation and improve yield.
In some embodiments, in the first step of the functional test, the "jump" mode of the writing algorithm will be turned off. In this step, based on the test results for a particular cell population, a regulated and optimized write/read current will be found and set.
Each MRAM cell has an optimal write current (Iopt). As shown in fig. 5A, the optimal write current for the MRAM cells in the array varies according to a gaussian distribution. In this embodiment, the unit of Iopt is μ a. When the write current deviates from the optimum write current, the error rate may increase exponentially. Because the optimal write current for a given magnetic random access memory cell may deviate significantly from the optimal write current for the array, applying the optimal write current for the array may not result in a change in the magnetic orientation of the given magnetic random access memory cell. Thus, the application of the write current may be performed multiple times in order to change the magnetic orientation. As shown in fig. 5B, applying the write current five times in a repeat provides an error rate that a given memory cell does not change orientation. The Y-axis in the graph shown in fig. 5B is logarithmic, and the X-axis is in units of μ a. Thus, it is readily understood that slight changes in Iopt can result in large differences in error rates.
In an embodiment of the present disclosure, multiple different write currents are applied to a given MRAM cell in a hopping scheme. As shown in fig. 5C, where the Y-axis is a logarithmic scale and the X-axis is in units of μ a, the use of the hopping scheme of the write current reduces the error rate by two orders of magnitude when the magnetic orientation of the MRAM is not changed compared to the repeated application of the same write current alone. FIG. 5D compares the error rates of a MRAM cell that applies a write current multiple times at a constant write current and a jump write current, where the current (I) is in units of μ A. The hopping scheme will be explained in more detail with reference to fig. 6.
The write current set by the first step of the functional test is the ensemble average Iopt. Within the cell population, there are MRAM cells whose Iopt differs significantly from the population average. When writing a current set in this way, the cells will have a high error rate.
In an embodiment of the present disclosure, a write process is used that applies a write current using a set of varying currents multiple times. This set of write currents will be centered on the global average set in the first step and cover the range of variation. In a second step of the functional test, such a variation range can be determined and set on the basis of an overall error rate analysis. Multiple attempts (multiple shots) will generate jumps by search patterns within this range. In field operation of an array of MRAM cells, the multiple-attempt "jump" scheme thus determined is used for the write process.
In an embodiment of the present disclosure, a write operation that applies a write current multiple times uses a set of varying write currents. The multiple write current application jumps through the search pattern as shown in fig. 6. With the jump write scheme, fewer devices fail. In a skip write scheme, the set of currents will be centered around the global average set in the first step and cover the range of variation. This range/step of variation can be determined and set based on an overall "Iopt" analysis in the second step of the functional test.
FIG. 6 illustrates a transition write current iteration process in accordance with an embodiment of the disclosure. The step size Δ I represents the difference in write current in μ a for a given write operation, while Δ I represents the total range in μ a between the highest and lowest write current values. In this embodiment, the first write current Wrt1 is the optimal write current for the array. The optimum write current for the array is predetermined during functional testing of the MRAM array. After application of the optimal write current Wrt1, a read current is applied to determine whether the magnetic orientation of the MRAM cell has changed. In some embodiments, the read current is fixed. The read current is determined during functional testing. In some embodiments, a fixed read current for a die, block, or array is set according to different design schemes.
An MRAM cell is acceptable and good if the magnetic orientation has been changed due to the application of an optimal write current. If the magnetic orientation of the MRAM cell has not changed, a second write current Wrt2 is applied. The value of the second write current Wrt2 is different from the value of the optimal write
After applying third write
After applying fourth write
After applying fifth write current Wrt5, a read current is applied to determine whether the magnetic orientation of the MRAM cell has changed. An MRAM cell is good if the magnetic orientation has been changed due to the application of fifth write
As shown in FIG. 6, the hopping scheme alternates between write currents having magnitudes less than and greater than the optimal write current, centered around the optimal write current Wrt1 for the MRAM array. The method according to the present disclosure is not limited to 5 iterations to determine if the MRAM is acceptable. In other embodiments, 3 or 4 iterations of applying the write current are performed. In other embodiments, more than 5 iterations of applying the write current are performed.
FIG. 7 is a flow chart illustrating a
In some embodiments, the
In some embodiments, each successive write current that is less than the optimal write current is less than a previous write current that is less than the optimal write current, and each successive write current that is greater than the optimal write current is greater than a previous write current that is greater than the optimal write current. For example, as shown in FIG. 6, in an embodiment, Iopt for the array of MRAM cells is determined to be 45 μ A. If the magnetic orientation of the MRAM cell is not changed after applying the write current of 45 μ A, a write current of 35 μ A is applied. If the magnetic orientation of the MRAM cell has not changed after applying the write current of 35 μ A, a third write current of 55 μ A is applied. If the magnetic orientation of the MRAM cell has not changed after applying the third write current of 55 μ A, a fourth write current of 25 μ A is applied. If the magnetic orientation of the MRAM-cell is not changed after applying the fourth write current of 25 μ A, a fifth write current of 65 μ A is applied. In this embodiment, the MRAM cell is rejected if 5 iterations are not sufficient to change the magnetic orientation. In this embodiment, the step size Δ I from each iteration of Iopt is 10 μ A, and the total range Δ I of write currents applied to the MRAM cells from the lowest write current value to the highest write current value is 40 μ A. As shown in fig. 6, in some embodiments, the write current applied after the application of the optimal write current alternates between a write current that is less than the optimal write current and a write current that is greater than the optimal write current. In an embodiment, the optimal write current is the midpoint of the values of all the alternating write currents.
FIG. 8 illustrates a portion of a
Fig. 9 is a flow chart illustrating a method 400 of determining an optimal write current for an array of MRAM cells in accordance with an embodiment of the present disclosure. In operation S410, a write current is applied to one of the plurality of MRAM cells. Next, a read current is applied to the MRAM cell in operation S420. In operation S430, it is determined whether the magnetic orientation of the MRAM cell has changed. As shown in fig. 9, if the magnetic orientation is not changed, a different write current is applied to the MRAM cell and a read current is repeatedly applied in operation S440. If the magnetic orientation is not changed, the application of different values of the write current is repeated until the magnetic orientation is changed. When the magnetic orientation has changed, in operation S450, a value of a write current that causes the magnetic orientation of the MRAM cell to change is determined. Then, the following steps are repeated in operation S460: applying a write current S410, applying a read current S420, determining whether the magnetic orientation of the MRAM cell has changed S430, and applying a different write current to the MRAM cell if the magnetic orientation of the MRAM cell has not changed S440. Next, in operation S470, it is determined whether the magnetic orientation of each MRAM cell of the plurality of MRAM cells is changed. If not all of the plurality of MRAM cells have changed their magnetic orientation, the step of applying the different write currents is repeated. The initial write current applied is the write current at the low end of the predicted range of suitable write currents. If the initial write current does not result in a change in magnetic orientation, the magnitude of the subsequently applied write current is increased in a stepwise manner until the magnetic orientation of the MRAM cell changes. If the magnetic orientation of each of the MRAM cells has changed, an optimal write current is determined based on the write current to change the magnetic orientation of each of the plurality of MRAM cells in operation S480. In some embodiments, the optimal write current is an average of the write currents to change the magnetic orientation of each of the plurality of MRAM cells.
In some embodiments, a plurality of parameters determined during functional testing are programmed into the MRAM array circuit by blowing fuses or antifuses in the circuit, wherein the plurality of parameters includes Iopt, a magnitude of a current step Δ I for each iteration of changing the write current, a total range of current changes Δ I for all iterations, and a number of iterations. Burn-in sets the parameters permanently. In some embodiments, the parameters for each block in the array of MRAM cells are different. Thus, each block in the array of MRAM cells can be optimized. For example, fig. 10 shows an antifuse block incorporating programmable circuitry for setting a step size for iterative write current changes and a range of write currents, in accordance with an embodiment of the present disclosure. As shown in FIG. 10, the programmable multiplexer inputs blow the appropriate antifuses in the antifuse block to set the parameters of the MRAM array circuit. For example, after the functional test determines the Iopt for a chip, array, module, or block of MRAM cells, the Iopt for the chip, array, module, or block of MRAM cells is burned in. Functional testing and burn-in may be performed on any size grouping of MRAM cells.
Fig. 11 illustrates a circuit such as
In some embodiments, the
Using the zig-zag jump writing pattern of the present disclosure, the time for successful writing will beShorter. In addition, fewer devices fail with this new "jump" write scheme. In some embodiments, the number of failed MRAM cells obtained by using the jump writing scheme of the present disclosure is reduced by two orders of magnitude. In some embodiments, the failure rate of MRAM cells is reduced to 1 × 10 using the jump writing scheme of the present disclosure-6Or smaller. Accordingly, the present disclosure improves the yield of the semiconductor device.
An embodiment of the present disclosure is a method of fabricating an array of magnetic random access memory cells, including writing to the magnetic random access memory cells. Writing to a magnetic random access memory cell includes: determining an optimal write current for the array of magnetic random access memory cells; the optimal write current is applied to a first magnetic random access memory cell in the array. Applying a first read current to the first magnetic random access memory cell to determine whether a magnetic orientation of the first magnetic random access memory cell has changed in response to applying the optimal write current. Applying a second write current to the first magnetic random access memory cell when the magnetic orientation of the first magnetic random access memory cell is not changed. The second write current is different from the optimal write current. Applying a second read current to the first magnetic random access memory cell to determine whether a magnetic orientation of the first magnetic random access memory cell has changed in response to applying the second read current. In an embodiment, the method further comprises: applying a third write current to the first magnetic random access memory cell when the magnetic orientation of the first magnetic random access memory cell is not changed after applying the second write current, wherein the third write current is different from the optimal write current and the second write current; and applying a third read current to the first magnetic random access memory cell to determine whether the magnetic orientation of the first magnetic random access memory cell has changed in response to applying the third write current. In an embodiment, the second write current is greater than the optimal write current and the third write current is less than the optimal write current. In an embodiment, the second write current is less than the optimal write current and the third write current is greater than the optimal write current. In an embodiment, a method comprises: iteratively repeating applying a write current and a read current when the magnetic orientation of the first magnetic random access memory cell has not changed, wherein iteratively applying the write current and the read current is stopped when the magnetic orientation of the first magnetic random access memory cell has changed, and wherein the write current at each application of the write current is different from any other write current. In an embodiment, the write current applied at each iteration alternates between being greater than the optimal write current and being less than the optimal write current. In an embodiment, each successive write current smaller than the optimal write current is smaller than a previous write current smaller than the optimal write current, and each successive write current larger than the optimal write current is larger than a previous write current larger than the optimal write current. In an embodiment, the write current applied after applying the optimal write current alternates between a write current smaller than the optimal write current and a write current larger than the optimal write current. In an embodiment, the optimal write current is at the midpoint of the values of all alternating write currents.
Another embodiment of the present disclosure is a method of manufacturing an array of magnetic random access memory cells, comprising determining an optimal write current for the array of magnetic random access memory cells. Determining an optimal write current for the array of magnetic random access memory cells comprises: (a) applying a first write current to one of a plurality of magnetic random access memory cells in the array of magnetic random access memory cells; and (b) applying a first read current to one of the plurality of magnetic random access memory cells to determine whether a magnetic orientation of the one of the plurality of magnetic random access memory cells has changed in response to the applying the write current. In operation (c), a second write current is applied to one of the plurality of magnetic random access memory cells when the magnetic orientation of the one of the plurality of magnetic random access memory cells is not changed, wherein the second write current is different from the optimal write current. Then in operation (d), a second read current is applied to one of the plurality of magnetic random access memory cells to determine whether a magnetic orientation of the one of the plurality of magnetic random access memory cells has changed in response to applying the second write current, wherein the second read current has the same value as the first read current. Next, in operation (e), when the magnetic orientation of one of the plurality of magnetic random access memory cells is not changed after the second write current is applied, applying a third write current to the one of the plurality of magnetic random access memory cells, wherein the third write current is different from the optimal write current and the second write current. In operation (f), a third read current is applied to one of the plurality of magnetic random access memory cells to determine whether a magnetic orientation of the one of the plurality of magnetic random access memory cells has changed in response to applying the third write current, wherein the third read current has a value that is the same as the first and second read currents. Then, in operation (g), iteratively repeating applying a write current and a read current when the magnetic orientation of one of the plurality of magnetic random access memory cells has not changed, wherein iteratively repeating applying a write current and a read current stops when the magnetic orientation of one of the plurality of magnetic random access memory cells has changed, wherein the write current at each application of the write current is different from any other write current, and wherein the first write current has a first magnitude and the magnitude of the subsequently applied write current increases in a stepwise manner. Subsequently, in operation (h), a value of a write current that causes a change in a magnetic orientation of the magnetic random access memory cell is determined. In operation (i), operations (a) through (h) are repeated for each of the plurality of magnetic random access memory cells in the array of magnetic random access memory cells. In operation (j), the optimal write current is determined based on the determined write current for each of the plurality of magnetic random access memory cells. In an embodiment, the optimal write current for each of the plurality of magnetic random access memory cells is an average of the write currents that cause the magnetic orientation of the magnetic random access memory cell to change. In an embodiment, a method comprises: writing to a magnetic random access memory cell in the array of magnetic random access memory cells, wherein writing to the magnetic random access memory cell comprises: applying the optimal write current to selected magnetic random access memory cells in the array; applying a read current of a write cell to the selected magnetic random access memory cell to determine whether a magnetic orientation of the selected magnetic random access memory cell has changed in response to applying the optimal write current; applying a write current of a write unit to the selected magnetic random access memory cell when the magnetic orientation of the selected magnetic random access memory cell has not changed in response to applying the optimal write current; and iteratively repeating applying the write current of the write unit and the read current of the write unit when the magnetic orientation of the selected magnetic random access memory cell is not changed. Stopping the iterative application of the write current of the write unit and the read current of the write unit when the magnetic orientation of the selected magnetic random access memory cell has changed, and the write current of the write unit at each application of the write current of the write unit is different from the write current of any other write unit. Then, after iteratively applying the write current of the write unit a set number of times, if the magnetic orientation of the selected magnetic random access memory cell is not changed, the selected magnetic random access memory cell is isolated. In an embodiment, the selected magnetic random access memory cell is isolated by blowing a fuse or an antifuse. In an embodiment, a magnitude of a write current of the write unit of a first application is greater than a magnitude of the optimal write current, and a magnitude of a write current of the write unit of a second application is less than the magnitude of the optimal write current. In an embodiment, a magnitude of a write current of the write unit of a first application is smaller than a magnitude of the optimal write current, and a magnitude of a write current of the write unit of a second application is larger than the magnitude of the optimal write current. In an embodiment, the magnitude of the write current of the write unit applied at each iteration alternates between being greater than the optimal write current and being less than the optimal write current. In an embodiment, the write current of the write cell applied after applying the optimal write current alternates between a write current of a write cell smaller than the optimal write current and a write current of a write cell larger than the optimal write current. In an embodiment, the write currents of the plurality of write cells applied after the application of the optimal write current alternate between a write current of a write cell smaller than the optimal write current and a write current of a write cell larger than the optimal write current, the write current of each successive write cell smaller than the optimal write current is smaller than the write current of a previous write cell smaller than the optimal write current, and the write current of each successive write cell larger than the optimal write current is larger than the write current of a previous write cell larger than the optimal write current.
Another embodiment of the present disclosure is a method of writing a magnetic random access memory cell, comprising: a first write current is applied to the first magnetic random access memory cell. Determining whether a magnetic orientation of the first magnetic random access memory cell has changed in response to the applying the first write current. Applying a second write current to the first magnetic random access memory cell when it is confirmed that the magnetic orientation of the first magnetic random access memory cell has not changed, wherein the second write current is different from the first write current. Determining whether a magnetic orientation of the first magnetic random access memory cell has changed in response to the applying the second write current. Applying a third write current to the first MRAM cell after confirming that the magnetic orientation of the first MRAM cell has not changed after applying the second write current, wherein the third write current is different from the first write current and the second write current. Determining whether a magnetic orientation of the first magnetic random access memory cell has changed in response to the applying the third write current. In an embodiment, the method comprises repeatedly applying said write current until the magnetic orientation of said first magnetic random access memory cell has changed or a certain number of applications of the write current is achieved, wherein the value of the write current applied during each application of the write current is different.
An embodiment of the present disclosure is a method of writing to a magnetic random access memory cell, comprising: the method includes iteratively applying a plurality of write currents to the first magnetic random access memory cell, and determining whether a magnetic orientation of the first magnetic random access memory cell has changed in response to each application of the write currents. The value of the write current applied during each application of the write current is different. The write current applied at each iteration alternates between being greater than and less than a first write current applied to a first magnetic random access memory cell. The iterative application of the write current is stopped when it is determined that the magnetic orientation of the first magnetic random access memory cell has changed.
Another embodiment of the disclosure is a method of writing to a plurality of magnetic random access memory cells in an array of magnetic random access memory cells, the method comprising: iteratively applying a plurality of write currents to a first magnetic random access memory cell in the array, and determining whether a magnetic orientation of the first magnetic random access memory cell has changed in response to each application of a write current. The value of the write current applied during each application of the write current is different. The write current applied at each iteration alternates between being greater than and less than the first write current applied to the first magnetic random access memory cell. Stopping the iterative application of the write current when it is determined that the magnetic orientation of the first magnetic random access memory cell has changed or the application of the write current to the first magnetic random access memory cell is performed a set number of times. A plurality of write currents are iteratively applied to a second magnetic random access memory cell in the array and a determination is made whether a magnetic orientation of the second magnetic random access memory cell has been changed in response to each application of a write current. The value of the write current applied during each application of the write current is different. The application of the write current at each iteration alternates between being greater than and less than the first write current applied to the second magnetic random access memory cell. The iterative application of the write current is stopped when it is determined that the magnetic orientation of the second magnetic random access memory cell has changed or the application of the write current to the second magnetic random access memory cell is performed a set number of times.
Another embodiment of the present disclosure is a method that includes applying a first write current to a first magnetic random access memory cell, wherein the first magnetic random access memory cell is located within a first block of magnetic random access memory cells in an array of magnetic random access memory cells. Determining whether a magnetic orientation of the first magnetic random access memory cell has changed in response to said applying the first write current. Applying a second write current to the first magnetic random access memory cell if the magnetic orientation of the first magnetic random access memory cell is not changed, wherein the second write current is different from the first write current. Determining whether a magnetic orientation of the first magnetic random access memory cell has changed in response to said applying the second write current. Applying a third write current to the first magnetic random access memory cell if the magnetic orientation of the first magnetic random access memory cell has not changed after said applying the second write current, wherein the third write current is different from the first write current and the second write current. Determining whether the magnetic orientation of the first magnetic random access memory cell has changed in response to said applying the third write current. The applying of the write current is repeated until the magnetic orientation of the first magnetic random access memory cell changes or a certain number of times of the applying of the write current is realized, wherein the value of the applied write current during each applying of the write current is different. A value of a write current that causes a change in a magnetic orientation of the first magnetic random access memory cell is determined. The first write current is repeatedly applied to each of a plurality of magnetic random access memory cells within a first block of magnetic random access memory cells in the array of magnetic random access memory cells. In response to repeatedly applying the first write current to each of the plurality of magnetic random access memory cells in the first block of magnetic random access memory cells, determining whether a magnetic orientation of each of the plurality of magnetic random access memory cells has changed. The second write current is repeatedly applied for each of the plurality of magnetic random access memory cells in the first block without a change in magnetic orientation, wherein the second write current is different from the first write current. For each magnetic random access memory cell to which the second write current is applied, determining whether the magnetic orientation of each magnetic random access memory cell has changed in response to repeatedly applying the second write current. After applying the second write current, repeatedly applying a third write current to each of the plurality of magnetic random access memory cells in the first block that do not have a change in magnetic orientation, wherein the third write current is different from the first write current and the second write current. For each of the plurality of magnetic random access memory cells to which the third write current is applied, determining whether a magnetic orientation of each of the plurality of magnetic random access memory cells has changed in response to repeatedly applying the third write current. The application of the write current is repeated until the magnetic orientation of each magnetic random access memory cell has changed or a certain number of write current applications are achieved. The value of the write current applied during each application of the write current to a particular magnetic random access memory cell is different. A value of a write current that causes a change in magnetic orientation of each magnetic random access memory cell in the first block is determined. An average of values of a plurality of write currents that cause a change in magnetic orientation of magnetic random access memory cells in the first block is determined. In an embodiment, the average write current for the first block is set for the first block during a burn-in operation of the array. In an embodiment, the array includes a plurality of blocks of magnetic random access memory cells. In an embodiment, the method is repeated for each block of magnetic random access memory cells in the array. In an embodiment, the magnetic random access memory cells of each of the plurality of blocks of magnetic random access memory cells have a different average write current. In an embodiment, the MRAM cells that do not change magnetic orientation after a set number of applications of write current are isolated from the corresponding block of MRAM cells by blowing a fuse or an antifuse. In some embodiments, a block of random access memory cells including greater than a set number of isolated random access memory cells is isolated from the array by blowing fuses or antifuses.
Another embodiment of the present disclosure is a circuit comprising a current source configured to apply a plurality of different write currents and read currents to a magnetic random access memory cell in an array of magnetic random access memory cells. The circuit includes a controller configured to determine an optimal write current for the array of magnetic random access memory cells, determine whether a magnetic orientation of the magnetic random access memory cells has changed in response to the applying a plurality of different write currents, and control the applying of the write current from the current source to the magnetic random access memory cells.
Another embodiment of the present disclosure is a circuit comprising a current source configured to apply a plurality of different write currents to a magnetic random access memory cell. The circuit includes a controller configured to control iterative application of a plurality of write currents from a current source to the magnetic random access memory cell and to determine whether a magnetic orientation of the magnetic random access memory cell has changed in response to each application of a write current. The controller is further configured to control the current source such that the write current applied each time alternates between being greater than and less than a first write current applied to the magnetic random access memory cell, and to cease applying the write current when it is determined that the magnetic orientation of the first magnetic random access memory cell has changed.
Another embodiment of the present disclosure is a functional test circuit comprising an array comprising a plurality of blocks, each block comprising a plurality of magnetic random access memory cells. The circuit includes a current source configured to provide a plurality of different write currents and read currents to each magnetic random access memory cell. The circuit includes a controller configured to control application of a plurality of different write currents from the current source to each magnetic random access memory cell and to determine whether a magnetic orientation of each magnetic random access memory cell has changed in response to each application of the write current. The controller is further configured to: controlling the current source such that the write current per application alternates between being greater than and less than the first write current applied to each magnetic random access memory cell; stopping applying the write current to each magnetic random access memory cell when it is determined that the magnetic orientation of each magnetic random access memory cell has changed; and determining an average of a plurality of write currents that result in a change in magnetic orientation of the magnetic random access memory cells in each block of the array.
It is to be understood that not all advantages need be discussed herein, that no particular advantage of all embodiments or examples is required, and that other embodiments or examples may provide a plurality of different advantages.
The foregoing has outlined features of several embodiments so that those skilled in the art may better understand the aspects of the present invention. Those skilled in the art should appreciate that they may readily use the present disclosure as a basis for designing or modifying other processes and structures for carrying out the same purposes and/or achieving the same advantages of the embodiments introduced herein. Those skilled in the art should also realize that such equivalent constructions do not depart from the spirit and scope of the present disclosure, and that they may make various changes, substitutions and alterations herein without departing from the spirit and scope of the present disclosure.
- 上一篇：一种医用注射器针头装配设备
- 下一篇：存储器控制器中的刷新方案