Patent Application
20130161629
1. Field of the Invention
Aspects of the present invention generally relate to methods and devices for stacks in 3D memory vertical gate applications. Further aspects relate to low or zero shrinkage stacks achieved from smooth interfaces between alternating layers of oxide and nitride films or oxide and amorphous silicon films.
2. Description of the Related Art
Computer memory devices are ever in pursuit of smaller geometries with increased capacity at less cost. To this end, components of memory cells are stacked on top of each other to create 3D cells. One such technology is NAND flash memory, which may be found in memory cards, USB flash drives, solid-state drives and similar products, for data storage and transfer. In NAND flash memory, memory cells made from transistors are connected in series, and can be stacked into vertical layers to create densely packed, high capacity devices. With no moving parts, flash drives use less power and are more durable than ordinary hard drives. Accordingly, there is great interest in increasing the capacity of flash drives, while reducing their size and cost.
To create 3D structures for memory cells, charge trapping transistors may be stacked into vertical layers. An electrical diagram of a flash cell string 200 in a 3D structure is illustrated in
However, as flash technology has progressed, limitations exist in how to create high capacity devices on a small scale. For example, different materials that are combined on a microscopic scale have different physical properties that lead to non-uniformities in a flash memory device. Further, high heat process steps can cause the different materials to undergo volume changes at different rates. These problems can cause a deposited stack of different layers to warp. Warping problems limit the number of layers that can be effectively deposited in manufacturing, and it can reduce the number of functioning memory strings available to the overall memory device.
Nom
Next, in poly-channel process step 330, a plurality of channel holes 332 are provided in stack 311 by punching through-holes in the stack. These channel holes 332 may also be referred to as poly-channel holes, because polysilicon (i.e., polycrystalline silicon) is used to fill the channels in a subsequent polysilicon deposition step 340. Thus, in step 340, polysilicon channels 342 (or poly-channels) are created by depositing polysilicon in the channel holes 332 created in step 330. The polysilicon deposition process uses very high temperatures of 700° C. and above. However, the stack 311 of alternating silicon oxide and silicon nitride films is deposited at much lower temperatures to avoid AlFx building up in the reactor during cleaning operations. Thus, the higher temperature polysilicon deposition process acts as an anneal and causes the silicon oxide and silicon nitride films to shrink. Since silicon oxide has different properties than silicon nitride, the films shrink at different rates, which causes the stack 311 to stress and bow, producing a warped shaped as illustrated in step 340 of
Warping limits the number of layers that can be stacked and can reduce the number of working memory strings that are ultimately fabricated in a memory device such as a flash drive. Stack warping can also cause variations in channel length of transistor gates, which negatively impacts memory strings. Therefore, a need exists for improved methods and devices for 3D memory structures.
Devices and methods for 3D memory structures are provided. In one embodiment, a method is provided for depositing a stack of film layers for use in vertical gates for 3D memory devices, the method comprising a sequence of supplying one or more process gases suitable for depositing a nitride film into a processing chamber of a deposition reactor, depositing a sacrificial nitride film layer at a nitride film deposition temperature greater than about 400° C., supplying one or more process gases suitable for depositing an oxide film into a processing chamber of a deposition reactor, and depositing an oxide film layer over the nitride film layer, at an oxide deposition temperature greater than about 400° C., wherein the sequence is repeated to deposit a film stack having alternating layers of the sacrificial nitride films and the oxide films, forming a plurality of holes in the film stack, and depositing polysilicon in the plurality of holes in the film stack at a polysilicon process temperature of about 700° C. or greater, wherein the nitride film layers and the oxide film layers experience near zero shrinkage during the polysilicon deposition.
In a further embodiment, the nitride and oxide deposition temperatures are about 600° C. or greater. In a still further embodiment, a showerhead having a straight hole faceplate is used for supplying the process gases into the processing chamber. In additional embodiments, the method provides for coating at least a portion of the deposition reactor with yttrium oxide to reduce AlFx deposits during subsequent cleaning operations. In other embodiments, the sacrificial nitride film layers are silicon nitride and the oxide film layers are silicon oxide.
In addition, the one or more process gases used to deposit silicon nitride layers comprise silane and ammonia, and the ammonia exceeds the silane on a volumetric basis. Other embodiments provide that the ammonia is at least 100 times as much as the silane, on a volumetric basis. In further embodiments, the one or more process gases used to deposit silicon nitride layers further comprises molecular nitrogen. In additional embodiments, the process gases used to deposit silicon nitride and the process gases used to deposit silicon oxide further comprise one or more dilution gases that are inert at process conditions. In another embodiment, the one or more dilution gases is argon and/or helium. Other embodiments provide that the one or more process gases used to deposit silicon oxide comprise tetraethoxysilane, N2O and a dilution gas that is inert at process conditions.
Another embodiment provides for a method for depositing a stack of film layers for use in vertical gates for 3D memory devices, the method comprising supplying one or more process gases suitable for depositing an amorphous silicon film into a processing chamber of a deposition reactor, depositing an amorphous silicon film layer at an amorphous silicon film deposition temperature greater than about 550° C., supplying one or more process gases suitable for depositing a silicon oxide film into a processing chamber of a deposition reactor, depositing an oxide film layer over the nitride film layer, at a silicon oxide deposition temperature greater than about 550° C., repeating the above steps to deposit a film stack having alternating layers of the amorphous silicon films and the silicon oxide films, forming a plurality of holes in the film stack, and depositing polysilicon in the plurality of holes in the film stack at a polysilicon process temperature of about 700° C. or greater, wherein the amorphous silicon film layers and the oxide film layers experience near zero shrinkage during the polysilicon deposition.
In a further embodiment, the amorphous silicon and the silicon oxide deposition temperatures are about 600° C. or greater. In a still further embodiment, a showerhead having a straight hole faceplate is used for supplying the process gases into the processing chamber. In another embodiment, the method provides for using yttrium oxide as a coating for at least a portion of the deposition reactor to reduce AlFx building up during subsequent cleaning operations. Other embodiments provide that the one or more process gases used to deposit silicon oxide comprise tetraethoxysilane, N2O and a dilution gas that is inert at process conditions.
Embodiments are also disclosed for computer memory devices made by any of the above methods, whether alone or in combination with other embodiments. In one embodiment, a 3D vertical gate computer memory device is formed by a process comprising at least the steps of supplying one or more process gases suitable for depositing a sacrificial film into a processing chamber of a deposition reactor, depositing a sacrificial film layer at a sacrificial film deposition temperature greater than about 550° C., supplying one or more process gases suitable for depositing an oxide film into a processing chamber of a deposition reactor, depositing an oxide film layer over the nitride film layer, at an oxide deposition temperature greater than about 550° C., repeating the above steps to deposit a film stack having alternating layers of the sacrificial films and the oxide films, forming a plurality of holes in the film stack, and depositing polysilicon in the plurality of holes in the film stack at a polysilicon process temperature of about 700° C. or greater, wherein the sacrificial film layers and the oxide film layers experience near zero shrinkage during the polysilicon deposition.
Additional embodiments provide that the sacrificial film layers are silicon nitride, and the oxide film layers are silicon oxide. Further embodiments provide that the one or more process gases suitable for depositing the sacrificial silicon nitride layers comprise silane and ammonia, and the ammonia exceeds the silane on a volumetric basis. Still other embodiments provide that the sacrificial film and the oxide film deposition temperatures are about 600° C. or greater.
So that the manner in which the above recited features of the present invention can be understood in detail, a more particular description of the invention, briefly summarized above, may be had by reference to embodiments, some of which are illustrated in the appended drawings. It is to be noted that the appended drawings illustrate only example embodiments for discussion, and are therefore not drawn to scale and are not limiting of claim scope.
It is contemplated that features of one embodiment may be beneficially incorporated in other embodiments without further recitation.
Embodiments discussed herein provide for improved stacks for 3D memory devices, methods for producing 3D memory devices and apparatuses for producing 3D memory devices. Further embodiments are described for approaches to reduce and/or eliminate shrinkage in alternating film layers in a stack when those layers are exposed to a high temperature process such as an anneal.
In order to micronize memory cells in vertical 3D arrangements, film layers are deposited into a stack, which is then further processed to create arrays of string cells. Some examples discussed herein relate to a Terabit Cell Array Transistor (TCAT) flash memory structure, in which a cell string has six-NAND cell transistors. But it is to be understood that the ideas disclosed herein may be applied to other 3D or vertical gate memory structures as well. For example, other configurations may use three-dimensional Bit-Cost Scalable (BiCS) flash memory. Other flash memory devices may use pipe-shaped Bit Cost Scalable (P-BiCS) structures, in which NAND strings are folded like a U-shape so that the select-gate transistors are at the top of each end section in the “U.” Thus, the bit line is positioned at one terminal end of the “U,” and the source line is positioned at the second terminal end of the “U.” P-BiCS flash memory may use two adjacent NAND strings connected at the bottoms by a “pipe-connection”, which is gated by the bottom electrode. Additionally, orientations may be changed, such as by vertical-gate (VG) NAND, in which cell strings are positioned horizontally so that the control gate is embedded in the perpendicular direction and connected directly to the word line on the bottom layer. Other types of flash memory are also possible, such as NOR flash, in which each cell has one end connected directly to ground, and the other end connected directly to a bit line.
The stack 411 may comprise alternating layers of oxide and nitride films. The oxide layer may be a silicon oxide. The nitride layer may be a silicon nitride. Silicon oxide may comprise SiO, although SiO2 and mixtures of SiO and SiO2 are not excluded. Silicon nitride may comprise SiN, although Si3N4, other molecular formulations and mixtures of the same are not excluded. Additionally, embodiments discussed herein also allow for use of alternating layers of silicon oxide and amorphous silicon (a-Si) to form a stack. In this illustrative example, a first or bottom layer of silicon nitride 412 is deposited, followed by a first layer of silicon oxide 413, then a second layer of silicon nitride 414, followed by a second layer of silicon oxide 415, followed by a third layer of silicon nitride 416, and then a third layer of silicon oxide 417. It should be understood that additional layers will also be provided in practice. Further, the top layer may also comprise a silicon nitride film (not shown). Alternatively, the bottom layer may be a silicon oxide. As will be discussed below for
Various techniques have been discovered that reduce and/or eliminate the warping effect, which are discussed in the embodiments presented herein. These embodiments may further be combined to provide even greater process uniformity and to provide increased margins for error in manufacturing. Further, eliminating stack warping allows for increasing the number of gates and/or decreasing the size of channel length for such gates.
In one embodiment, high temperature deposition of oxide and nitride layers achieve near zero shrinkage when exposed to high temperature process steps such as the polysilicon anneal. Near zero shrinkage of a film layer means that the change in thickness before and after high temperature exposure (e.g., 700° C. and above) is less than about 0.3%. Near zero shrinkage rates were seen in oxide layers, nitride layers and amorphous silicon layers. This is a significant improvement over previous methods, in which a silicon nitride layer may shrink in thickness about 2.4%, whereas a silicon oxygen layer would shrink in thickness about 1%. Moreover, near zero or zero stress changes may be achieved in addition to near zero or zero shrinkage rates. Further, obtaining near zero shrinkage rates in different layers reduces interface stresses because even if there is some small amount of shrinkage, it is much less and more uniform across the stack.
As discussed above, the polysilicon deposition process may involve temperatures of about 700° C. and above. To deposit the alternating film layers in stack 411, a PECVD chamber may be used to deposit the films at temperatures over 500° C. In further embodiments, film layers may be deposited at temperatures over about 550° C. Preferably, temperatures of about 600° C. and greater are used to deposit the film layers in the stack 411. In a further embodiment, temperatures of about 650° C. are used to deposit the film layers. In still further embodiments, the film layers in the stack 411 may be deposited at temperatures near the temperature of the polysilicon process, such as temperatures that vary less than 15% or 10% or even 5% from the temperature of the polysilicon process. In other embodiments, temperatures of about 400° C. and above may be used. Temperatures in ranges of 400° C. and above may be combined with other embodiments discussed herein, as exemplified in Tables 1, 2 and 3 below.
It has been discovered that the high-temperature deposited films in the stack 411 have higher density and reduced hydrogen content than in the past. This is especially critical to reducing warping in the stack 411, since higher density films with less hydrogen have less opportunity to shrink when exposed to a high temperature anneal. This further reduces interface stresses between the alternating layers. When combined into a stack, increased interface stresses between multiple layers lead to warping, such as shown in
As discussed above, the reason previous deposition methods use low temperatures is to avoid AlFx building up in the reactor during cleaning operations. AlF3 is used to provide cleaning, and is known to leave deposits behind when higher temperatures are used. In order to avoid these problems with higher temperatures, a coated heater may be used in the PECVD process to run at temperatures over 600° C. with minimum AlFx build-up during cleaning. A coating may be selected that is resistant to floride. In one embodiment, an oxide compound is used for the coating. In a further embodiment, yttrium oxide is used to coat the reactor. In another embodiment, yttrium oxide is used to coat parts of a PECVD reactor used for heating, such as the plasma electrodes.
Additional embodiments reduce the surface roughness of the film layers in the stack 411. Faceplate engineering can improve roughness. Embodiments have been discovered to alter ion bombardment in ways that make the film surfaces smoother or that reduce interface roughness. These embodiments are applicable to oxide and nitride layers as well as to a-Si layers. In one example, a showerhead having a conical hole faceplate in a PECVD reactor was replaced with a straight hole faceplate. The linear holes resulted in a smoother surface on the top layer deposited, which resulted in less stress at the interface between the next deposited layer.
Providing a smooth surface was found to be critical, because the top surface of the stack becomes rougher as more layers are added. When successive layers are deposited on a rough surface, the roughness measurements are compounded as the stack grows. By reducing surface roughness of each film layer, more layers can be uniformly deposited.
Further, the straight hole faceplate embodiment can be combined with embodiments to modify the ratio of reactive species in the PECVD process. It has been discovered that a high ammonia (NH3) to silane (SiH4) ratio reduced the surface roughness of silicon nitride films, and reduced the interface roughness between silicon nitride and silicon oxide. Moreover, both near zero shrinkage and near zero stress changes may be achieved. Previously, a 1:2 silane to ammonia ratio was used to deposit the silicon nitride layers. In one embodiment, the silane to ammonia ratio may be altered so that there is more ammonia than silane on a volumetric basis. In a further embodiment, 100 to 200 times as much ammonia is used as silane. Molecular nitrogen (N2) may also be added to the gas mixture used in the reactor. In some embodiments, there is less N2 than NH3. In other embodiments, 10 to 20 times as much N2 as NH3 may be used.
Moreover, when films are deposited so as to experience near zero or zero shrinkage, non-reactive dilution gases can be added to the deposition process to make the films more stable. Dilution gases may be used for depositing layers of oxides, nitrides or amorphous silicon. In some embodiments, Argon or Helium may be added as a diluting agent to the reactive species in a plasma. Further, it has been found that using inert plasmas such as Argon or Helium lowers the stress close to neutral. For example, a dense nitride layer may be very tensile. The inert diluting agents reduce the internal stresses in the nitride layer. This features allows for zero or near zero stress changes to be achieved when the stack 411 is exposed to a high temperature anneal. Accordingly, both zero shrinkage and zero stress changes may be achieved. Additionally, refractive index may be measured to examine stresses in one or more layers. Thus, stresses may be measured as a function of the refractive index.
Additionally, in some embodiments, the various techniques discussed herein may be combined without using temperatures above 500° C. for the deposition. Thus, temperatures may be used in a range of about 400° C. to about 650° C. Example ranges of process parameters for the films are provided in the three tables below. Table 1 provides process conditions for deposition of oxide films to be used in an oxide/nitride or oxide/silicon stack. Table 2 provides process conditions for deposition of nitride films to be used in oxide/nitride stacks. Table 3 provides process conditions for deposition of silicon films to be used in oxide/silicon stacks with tunable doping. In these tables, HF means high frequency, LF means low frequency, TEOS means tetraethoxysilane, TMB means trimethylboron. In Table 3, TMB and B2H6 can be used for boron doped a-Si, and PH3 can be used for phosphorous doped a-Si films.
The embodiments discussed herein to reduce surface roughness (whether used alone or combined with one or more of the other techniques discussed below) allow targeting specific surface roughness metrics for quality. For example, previous surface roughness was found to be about 4-5 nm RMS on a top surface. To maintain low interface stress, a top surface roughness may be targeted of about 3 nm RMS or less, and preferably 1-2 nm RMS. Further, the techniques discussed herein to reduce surface roughness also may be used for other alternating layers besides nitrides and oxides. For example, the stack 411 could also use alternating layers of silicon oxide and amorphous silicon.
Moreover, by substantially reducing or eliminating stack warping, variability in channel length of later formed gates can also be reduced or eliminated. So that the importance of eliminating warping can be better understood, four additional process steps to form metal gates in a stack are shown in
In process step “A,” a plurality of word line cuts are made in the stack 511, such as illustrated by word line cut 550. The word line cuts may be made by a dry etching process. The word line cut 550 exposes vertical edges of the alternating stack layers 520 and 530 for further processing. The word line cuts 550 open an area for later steps that make word line connections to subsequently deposited metal gates. Next, in process step “B,” the exposed portion of the silicon nitride film is removed, such as by a wet etching step. This leaves gaps 552, having an exposed surface 554, between the oxide layers 530, and bordered on one side by the polysilicon channel 540.
Afterwards, in process step “C,” a gate dielectric film 560 is deposited on the exposed surface 554. The gate dielectric may comprise a tunnel oxide film, which is deposited on the exposed surface 554 of the silicon oxide layers 530 and the polysilicon channel 540. The gate dielectric 560 does not fill the entirety of gaps 552. Next, in step “D,” a gate material 565 is deposited in the gaps 552. The gate material may be a metal or metal alloy. A preferred material for TCAT structures is tungsten. Tungsten alloys may be used as well. Step D illustrates that excess metal may be deposited so that the metal overlaps all exposed surfaces of the dielectric. Excess metal may be removed in step “E,” forming metal gates 570, which are isolated from each other, and separated from the polysilicon electrode 540 by the dielectric film 560. The metal gates 570 serve as control gates for the word lines in the memory strings. Additionally, one or more gates at the top and also at the bottom of the stack 411 may be used as selection gates for the memory string.
It should be appreciated from
While the foregoing is directed to embodiments of the present invention, other and further embodiments of the invention may be devised without departing from the basic scope thereof, and the scope thereof is determined by the claims that follow.
