JPH0613628A - Semiconductor device and its manufacture - Google Patents

Abstract

PURPOSE:To reduce an occupying space and realize a high integration by a method wherein a channel forming region between a source and a drain is made to stand vertically. CONSTITUTION:A protrusion 302 is formed on a semiconductor substrate 301. An impurity region is formed in the crest part of the protrusion and a gate oxide film 306 is formed by a thermal oxidation method and a semiconductor film 305 is formed on it. A semiconductor 307 which is formed along a trench continuously is divided into respective chips. After a gate oxide film 308 is formed by a thermal oxidation method, a film 309 of which gate wirings 310 are to be composed is formed. Then the parts of the film 309 other than the parts on the side surfaces of the protrusion are removed to form the gate wirings 310. After impurity regions 311 are formed by an ion implantation method, a layer insulating film 312 and a contact hole 313 are formed and a metal wiring 313 are formed. The side surfaces of the protrusion are used as channel regions which are made to stand vertically.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a technology for highly integrating a semiconductor integrated circuit. In the present invention, a semiconductor device suitable for high integration is proposed and a manufacturing method thereof is described. The semiconductor device according to the present invention is used for a nonvolatile memory device having a so-called MNOS structure.

[0002]

2. Description of the Related Art Many researches and developments have been made on miniaturization and high integration of semiconductor devices. Especially MOSFE
The progress of miniaturization technology of an insulating gate field effect type semiconductor device called T is remarkable. What is MOS?
It is an acronym for Oxide-Semi-condeuctor. Even if the metal is not pure metal,
It is used in a broad sense, including semiconductor materials with sufficiently high conductivity and alloys of semiconductors and metals. Further, not only a pure oxide but also an insulator such as a nitride may be used instead of an oxide between a metal and a semiconductor. In such a case, strictly speaking, the term MOS is correct. However, hereinafter, in the present specification, a field effect element having such a structure including a nitride and other insulators is referred to as a MOSF.
Referred to as ET or MOS transistor.

In a normal MOS transistor, an oxide (insulator) such as silicon oxide is formed as a gate oxide (gate insulator) on a semiconductor substrate, and a metal or semiconductor acting as a gate electrode is formed thereon. Is provided, and the conductivity of the underlying semiconductor is controlled by controlling the potential of the gate electrode.

However, an electrically independent semiconductor film (this is called a floating gate) is formed on the gate oxide, and an insulating film is formed again thereon, and a gate electrode (this is called a control gate) is formed. It is known that the element can be used as an element of a non-volatile memory by providing. A memory having such a structure is usually E
It is commercially available as PROM or EEPROM. The principle is that by applying a strong electric field to the control gate electrode, charges such as electrons and holes are trapped in the intermediate floating gate film and charged to a specific conductivity type, so that the conductivity of the underlying semiconductor is improved. It tries to fix it semipermanently. Of course, for example, when the charges injected into the floating gate are removed by irradiation of ultraviolet rays or an electric effect, the state returns to the original state, that is, the data is erased.

It has been known that a similar effect can be obtained by using a silicon nitride film instead of a semiconductor floating gate. That is, in a structure in which silicon nitride and a gate electrode are laminated on a semiconductor, an irreversible characteristic was observed once a voltage was applied to the gate electrode. It is considered that this is because the charge is trapped in the silicon nitride film itself, or in the semiconductor clusters and other defects formed inside the silicon nitride film, and the same action as the floating gate is obtained. When the silicon nitride film is formed directly on the semiconductor,
Since there are many localized levels on the surface of a semiconductor (silicon or the like) and there is a problem with reliability, an oxide film with excellent interfacial characteristics such as silicon oxide is usually formed on a semiconductor, and a silicon nitride film is formed on the oxide film. And MNOS (Metal-Nitride-Oxide-S)
emiconductor) structure. However, if the silicon nitride film is in direct contact with the gate electrode, the charge accumulated in the silicon nitride film leaks and the memory disappears. Therefore, in order to make a stable memory element, the silicon nitride film and the gate electrode are formed. MONOS ((M
et al-Oxide-Nitride-Oxide-Semi-conductor) has been developed.

Such a MOS transistor having a floating gate, MNOS, and MONOS transistor are used as a memory cell transistor.

A memory using such an element is D
Unlike a RAM such as a RAM or an SRAM, a power supply is not required to retain the memory, and a capacitor is not required, especially when compared with a DRAM, so that a memory of 16 Mbits or more will not be manufactured in the future. In this case, the cell area per bit can be made small, which is suitable for high integration, so that research has become particularly active in recent years. In particular, EEPROMs that can be electrically erased have attracted particular attention.

However, in order to achieve higher integration, the current technology has various problems. When a memory device is to be manufactured by using such an element, the memory cannot be configured only by this element, and the selection transistor must be formed at the same time as this element. FIG. 2A shows the structure of an EEPROM in which the highest integration has been achieved conventionally. In the figure, 201 is a source region, 202 is a drain region, 203 is a floating gate, and 204 is a control gate. The portion 203 may be a silicon nitride film, in which case it becomes a MONOS type element. Further, although 203 and the control gate 204 are described separately from each other in the figure, when 203 is made of silicon nitride and formed in close contact with each other, the element is an MNOS type element. In any case, it is better to understand that the MONOS type device is a special example of the MNOS type device. Although it is a structurally integrated element, in operation, the portion within the dotted line P in the figure functions as a selection transistor, and Q
The portion within the dotted line of (2) functions as a memory cell transistor.

When writing data in the memory transistor, a high voltage (usually 10 V or more) is applied to the drain 202 and the control gate 204 to inject charges into the floating gate 203, and when erasing the data, the control gate 204 and the source. When a high electric field is applied to the drain with 201 kept at the same potential, the charge of the floating gate is taken out. When such writing / erasing operations are repeated many times, the characteristics of the insulating film are deteriorated. Therefore, it is said that the limit is 100,000 times in the current technology.

To check whether or not there is data, it is sufficient to apply a normal voltage to the control gate 204 while applying a normal operating voltage of the MOS transistor to the drain 202. By this operation, the selection transistor P is turned on. If electric charges (holes (electrons) when the semiconductor substrate is P type (N type) and the source and drain are N type (P type)) are trapped in the floating gate of the memory cell transistor Q, the memory cell transistor is Since it is already on, current flows. However, when the floating gate has no charge or is charged in the opposite direction, the memory cell transistor is in the OFF state, so that no current flows even if the select transistor is in the ON state.

The above is an example of a general EEPROM, and in practice, whether electrons are injected into the floating gate, holes are injected, or charges are injected or removed at the time of writing. Then, there are some variations, but the basic idea is that the state of the memory cell transistor is pulled out to the outside by the selection transistor.

Actually, a large number of such elements are arranged,
It functions as a memory device only after the wiring as shown in FIG. 2B is formed. In this figure, X,
X'is a word line, and Y and Y'are bit lines. If one of the word lines is selected and you are paying attention to a specific bit line, if there is data in the memory cell transistor at the intersection, the signal is detected, otherwise the signal is not detected. .

In the simplest case, one integrated element may be used as a 1-bit storage cell, but in order to improve accuracy, two identical cells are prepared and data is written in one of them. , A method of comparing signals coming from these two cells is taken without writing data to the other. That is, it can be determined that there is data if there is a potential difference between the signals sent from the two cells and that there is no data if there is no potential difference. If such a method is adopted, the memory capacity will be halved, but with high integration,
When noise is superimposed on the bit line, it is desirable to adopt such a method in order to improve the accuracy.

The above is an MO having a floating gate.
Although it is an example of the S-transistor, MNOS and MONOS operate similarly. Now, such a transistor element has some problems in terms of high integration. First, as is clear from FIG. 2A, L ₁ + L ₂ is the minimum width of the element. Here, the minimum value of L ₁ and L ₂ is the processing accuracy itself, and in the present technology, considering the mass productivity, the limit is 0.5 μm.

Therefore, at least 1 μm is required only for the gate portion of this element. In addition, this is a serious problem for MOS type semiconductor devices in general.
It is necessary to provide a contact in the source and drain regions, and the contact needs to occupy a large area because it is located below the gate portion. In particular,
All element source regions were connected to a ground level, contacts having a diameter of several μm were formed in the drain regions, and metal wirings were formed orthogonal to the gate wirings. In this case, the metal wiring needs to be located in the upper layer of the gate wiring and come down to the drain region formed on the substrate surface through the hole formed in the interlayer insulating film. Therefore, the connecting portion from the contact portion to the bit line has a long distance, and disconnection of wiring and contact failure are serious problems. Actually, by the CVD film forming method of tungsten,
The technique of embedding a contact hole was devised to deal with such a problem. If you do not use a special technique of filling contact holes,
Techniques such as increasing the area of the contact, widening the contact hole, or forming the contact hole into a cone shape are required, but these are the opposite techniques for high integration.

The following problem is that the self-alignment method cannot be adopted when manufacturing an EEPROM, and therefore,
The mask process is increased. In fact, EEPRO
In the operation of M, due to the necessity of hot carrier injection from the drain, the floating gate 203
Are required to overlap the drain region 202. However, if planar photolithography technology is adopted to obtain the overlap, the current technology is
The deviation of the photomask must be taken into consideration to be 0.2 μm or more. Therefore, in order for the drain region and the floating gate to surely overlap, the overlapping region is 0.4
μm or more is required. If it is less than this, the variation in the overlap becomes 50% or more, which causes a serious problem in the yield of the device.

The production of the memory area of the conventional EEPROM requires the following steps even if it is the main one. Circled numbers are mask process numbers. (1) LOCOS is formed on a semiconductor substrate (2) Drain region 202 is formed (3) Floating gate is formed (4) Control gate (word line) is formed (5) Source region 201 is formed (6) Interlayer insulator Formation and contact hole formation in the drain (7) Bit line formation Most of these steps require a mask process (a mask is unnecessary only for self-aligned formation (5)). The accuracy is required to be 0.2 μm or less. Therefore, as a result, DRA
Compared with M (five mask processes) or the like, the yield is reduced.

In the future, the EEPROM will be used as a special non-volatile memory called a flash memory in the DRA.
It is regarded as an element that replaces M, but if the yield is high, the unit price per bit will be relatively high, leading to a decrease in competitiveness.

[0019]

SUMMARY OF THE INVENTION The present invention has been made to solve some or all of the problems raised above. That is, the present invention proposes a device structure having a higher degree of integration and a process therefor.
Further, the present invention proposes an element structure and its process in which the number of steps of the mask process is reduced or the accuracy required for the mask process is relaxed to improve the yield.

[0020]

The present invention solves these problems by arranging three-dimensionally arranged MOS transistors, which are conventionally arranged in a plane. That is,
By vertically standing the channel formation region between the source and the drain, which has been conventionally arranged in a plane, the area of that portion is reduced. The basis of the present invention is to form a convex portion on a semiconductor substrate, and make the side surface a channel forming region,
The top is one of the impurity regions (source or drain), the bottom is another impurity region, and the gate electrode is formed on the side surface of the channel formation region.
As a result, the gate is also required to stand vertically. However, for example, in an EEPROM, it is necessary to form two types of transistors, a storage transistor having a floating gate and a selection transistor having a normal structure. When the present invention is used, the selection transistor portion in the EEPROM is arranged in a plane and the memory cell transistor portion is verticalized for simplification of manufacturing.

Further, it can be easily understood that the area can be saved (high integration) by applying the present invention to a general MOS transistor even if it is not a special transistor such as an EEPROM. However, as with the present invention, one of the impurity regions is provided with a gate electrode.
Forming at a position higher than the wiring is extremely convenient when forming a contact in the impurity region. Therefore, for example, even when manufacturing an EEPROM device, it is desirable to apply the present invention to both the memory region and the peripheral region.

FIG. 1 is a schematic diagram for expressing the technical idea of the present invention. The details are not necessarily exactly described, but they are sufficient to understand the invention.
FIG. 1A shows a schematic sectional view of an EEPROM element according to the present invention. In this figure, four elements are depicted. The second element from the right will be described below, but the other elements are exactly the same. As is apparent from the figure, a convex portion is provided on the semiconductor substrate, and the top thereof is the drain region 102.

The present invention does not particularly define the method of manufacturing such a convex portion. For example, such a convex portion may be formed by etching a single crystal semiconductor substrate, or such a portion may be formed by forming a semiconductor region over the substrate and etching it. You may. At this time, a single crystal semiconductor substrate may be used as the substrate and homoepitaxial growth may be performed thereon. Further, the area of the top of the convex portion may be smaller or larger than the area of the bottom. These items can be changed according to the design items required by those who implement the present invention.

In the case of forming an EEPROM, the drain region is precisely formed so that the drain region has an optimum overlap with the floating gate.
When used as a normal MOS transistor,
It is formed to have almost no thickness. Or LD
When a structure similar to that of D is to be formed, impurity layers having different impurity concentrations may be formed in two or more steps.

Further, in the EEPROM, in order to effectively inject or remove the tunnel current into the floating gate by field emission, the conventional EEPROM is used.
In the OM, the thickness of the gate oxide film at a specific portion was made extremely thin, but for that purpose, a special patterning process was required.

For example, if a similar portion is to be formed according to the present invention, this can be done by changing the structure of the impurity region. That is, a layer having a high impurity concentration may be sandwiched between layers having a low impurity concentration. By doing so, the electric field is effectively concentrated in the central region having a high impurity concentration, and the tunnel current can flow.

The impurity region can be formed by controlling the vertical diffusion of the impurity, so that the impurity region can be controlled substantially as fine as 50 nm. Therefore, it is possible to form an extremely fine impurity region as compared with the formation of the impurity region in the conventional planar type MOS transistor.

On the other hand, a source region 101 is provided on the bottom of the semiconductor substrate. The floating gate 103 and the control gate 104 are formed so as to stick to the side surfaces of the convex portion. It is a normal MOS transistor if the floating gate is not formed.
If the portion corresponding to the floating gate is made of silicon nitride, it becomes MNOS or MONOS.

Further, the wiring 106 functioning as a bit line connects the impurity region 102 through the contact hole 107.

Here, it should be noted that a memory transistor as shown by a dotted line portion Q in the drawing is vertically formed in the element having such a shape, while a selection shown by a dotted line portion P is made. That is, the transistor is formed horizontally.

Further noteworthy is the above-mentioned conventional EE.
Although it is a limit of the fine processing which is a problem in the PROM, in the example of FIG. 1, it is substantially limited to the width of this convex portion. This is apparent from the process described later, but there is substantially no mask process for forming the gate portion. The shape of the gate part is
It is determined by forming the convex portion of the base, and the thickness thereof is determined by the thickness of the film used for forming the gate portion and the degree of anisotropic etching thereof. Therefore, if the present invention is adopted, even if the minimum processing accuracy is 0.5 μm, the width required for each element can be 1 μm or less. The conventional method (FIG. 2) required at least 5 μm.

FIG. 1B is a top view of an EEPROM formed by assembling such elements in a matrix. In this figure there is an 8-bit memory. In the figure, 101 is a source region, which are all integrated and arranged in parallel with the gate wiring, and function as a power supply line. 102 is a drain region formed on the top of the convex portion. Reference numeral 103 is a floating gate formed vertically on the side surface of the convex portion, and 104 is a control gate. This control gate 104
In the figure, they are connected vertically to form a word line. Reference numeral 105 denotes a thick insulator provided to separate each element of the convex portion, and has a function similar to that of the conventional LOCOS. Reference numeral 106 denotes a bit line, which is connected to each drain region via a contact hole 107.

The present invention is also advantageous in forming wiring such as bit lines. That is, in the present invention,
Since one of the impurity regions is located above the gate electrode, a deep contact hole as in the conventional case is not required. Therefore, the area of the portion required for the contact may be remarkably smaller than that in the conventional method, and there are few problems such as disconnection and contact failure, leading to an improvement in yield.

In the conventional method, when trying to make the depth of the contact hole shallow, the wiring was moved up and down by the step difference of the gate wiring. The existence of such a step directly leads to a problem of disconnection of wiring. However, in order to suppress such vertical movement, when the wiring is formed after the interlayer insulating material is flattened, the contact hole becomes deep.

In the present invention, it is possible to make the contact hole shallow and reduce the vertical and stepped portions of the wiring, and the yield improvement brought about only by this is remarkable.

FIG. 6 is a circuit diagram of FIG. 1 (B). Now, assume that no electrons are injected into the floating gates of all the elements. Then, suppose that electrons are injected into only the floating gate of the element of T ₁₃ among them. Therefore, the power supply line (source line) S ₁ ,
The potentials of S ₂ and S ₃ are kept at 0, and the word line (gate wiring)
It is assumed that the potentials of G ₁ , G ₂ , and G ₄ are V ₀ , the potential of G ₃ is V ₁ , the potential of the bit line (drain wiring) D ₁ is V ₂ , and the potential of D ₂ is V ₃ . At this time, the elements T ₁₁ , T ₁₂ , T
The voltage between ₁₄ the gate and the drain is (V ₀ -V _2), a the element _{T 13 (V 1 -V 2)} , the element T _21, T
₂₂ and T ₂₄ are (V ₀ −V ₃ ), and element T ₂₃ is (V ₀ −V ₃ ).
₁ is a -V _3). Here, if a tunnel current occurs and electrons are injected when a potential difference of V _th or more occurs, the following inequalities occur. -V _th <V ₀ -V ₂ <V _th V ₁ -V ₂ > V _th -V _th <V ₀ -V ₃ <V _th -V _th <V ₁ -V ₃ <V _th

Here, the lower limit is set for each potential difference because the data already stored is erased when a voltage lower than that is applied. this is,
Although it is a quaternary simultaneous inequality and it is troublesome to find a general solution, for example, when V _th = 2 V, V ₀ =
_{0, V 1 = 2V, V} 2 = -V, V 3 = V is one of the solution, this _{time, V 0 -V 2 = V,} V 1 -V 2 = 3
Since V, V ₀ −V ₃ = −V, and V ₁ −V ₃ = V, the above conditions are satisfied. That is, there are two types of voltage applied to the word line, 0 and 2V, and the voltage applied to the bit line is V
Information can be input to an arbitrary element by using two types, ie, -V.

This is an example of the operation, and other various operation modes can be assumed, but they are not taken up here individually.

Several methods can be considered for producing the one having the structure as shown in FIG. 1, but the typical process is as follows. (1) Formation of impurity region (drain) on semiconductor surface (2) Formation of convex portion (3) Formation of film to become floating gate (deposition and anisotropic etching) (4) Formation of element isolation region Etching of unnecessary parts of floating gate (5) Formation of control gate (deposition and anisotropic etching) (6) Formation of interlayer insulator and formation of contact hole (7) Formation of source region (8) Drain wiring Formation Although the number of steps seems to have increased in this way, the actual mask process could be significantly reduced. Further, since the mask process does not particularly relate to the production of the portion related to the EEPROM structure, the present invention can be applied to a general M process.
The number of mask processes does not change when manufacturing an OS transistor.

An example process for practicing the present invention is shown in FIG.
And it shows below using FIG. According to the present invention, EE
If a device such as a PROM is to be formed on a large scale,
It has to be manufactured paying attention to the separation between the transistors. For that purpose, for example, as shown in FIG. 4, it is advisable to adopt the conventional LOCOS method and various other element isolation techniques.

An example of the element isolation technique will be described with reference to FIG. First, the impurity region 40 is formed on the semiconductor substrate 401.
Form 2. Although various semiconductors can be used as the semiconductor substrate, the (100) plane is preferably used when silicon is used. The thickness of the impurity region is 10 to 500 nm
It is good to The optimum thickness is designed according to the purpose of the device. For example, when it is used as an EEPROM, its thickness must be 100 to 500 nm to promote injection of charges into the floating gate. When used as a normal MOS transistor,
When the thickness of this layer is large, the overlap with the gate electrode becomes large, and therefore it is preferably thin. When the impurity concentration is formed in two or more stages, the impurity region is close to the LDD structure used in the planar MOS transistor.

As a method of forming the impurity region, an impurity diffusion method using a known ion implantation method or the like may be used, or a semiconductor containing impurities may be epitaxially grown on the semiconductor substrate. Alternatively, a polycrystalline semiconductor containing impurities may be simply formed. Any of these methods is a known technique, and in adopting it, it is necessary to determine the advantages and disadvantages of each method and determine them. The method of impurity diffusion may be the most common method. In the case of silicon, the epitaxial growth method has a high growth temperature, and therefore has a problem of unintended diffusion of impurities.
However, the crystal interface of the semiconductor is clean, and it seems to be suitable for compound semiconductors such as gallium arsenide semiconductor. Deposition of a polycrystalline semiconductor is the simplest method, but defects are likely to occur at the interface between the single crystal semiconductor and the polycrystalline semiconductor.

Next, as shown in FIG. 4B, the semiconductor substrate is etched into a stripe shape to form a groove 403. A known photolithography method may be used for this.

Then, as shown in FIG. 4 (C), a coating 4 having an oxidation resistance such as silicon nitride is selectively formed on the surface of the substrate.
To form 04. In this formation, such a silicon nitride film needs to be formed even on the details of the unevenness. For that purpose, a conventional low pressure CVD method may be used, but an optical CVD method that causes less damage to the substrate and has good step coverage is preferably used. A film of silicon oxide or the like may be formed under the silicon nitride for stress relaxation.

In this way, a part of the semiconductor substrate is exposed and a part is covered with the silicon nitride film or the like. Finally, when the substrate is oxidized by a method such as steam oxidation, the portion not covered with the film 404 such as silicon nitride is oxidized to form a thick oxide 405. By removing the coating such as silicon nitride, FIG. 4D is obtained. In this way, the element isolation region is formed. In the figure, an oxide is formed so as to divide the groove portion. However, if such a pattern is adopted, when it is desired to use the bottom portion (concave portion) of the substrate as a power supply line, the power supply line cannot be formed because it is later divided by the oxide 404. Therefore, for that purpose, it is necessary to cover a part or all of the bottom of the groove with a film of silicon nitride or the like in advance so that the part of the power supply line is not oxidized. The above method is the conventional LOCO
Although the S method is directly applied to the present invention, there are some disadvantages.

For example, when manufacturing an EEPROM or the like, it is necessary to form a semiconductor film for forming the floating gate along the side surface of the convex portion.
Since this is formed continuously along the semiconductor side surface,
After that, it must be divided into each element by the photolithography method. However, since the dividing work is the same as the isolation between the elements, the oxide formation and the floating gate patterning can be performed at the same time. That is, the mask process can be saved by performing the element isolation step later.

The details will be described in the following description based on FIG. Hereinafter, a case of manufacturing an EEPROM will be described. The following description describes a rough process, and some design changes may be necessary to obtain the characteristics required by a person who carries out the invention.

First, the convex portion 302 is formed on the semiconductor substrate 301 by using all the manufacturing methods shown in FIG. The width of the convex portion is determined by the adopted design rule. For example, let's say 500 nm.
Further, it is desirable that the height of the convex portion is approximately the same as the width of the convex portion. However, since this is a factor that determines the channel length of the memory transistor, it cannot be generally stated. 200-800n
m is suitable. The impurity region 30 is formed on the top of the convex portion 302.
3 is formed. Then, an oxide film 306 that functions as a gate oxide film is formed by a method such as a thermal oxidation method. The thickness is preferably 10 to 100 nm. When silicon is used as a semiconductor, silicon oxide formed by a known thermal oxidation method is suitable. In particular, the thermal oxidation method is suitable because an oxide film can be evenly formed on the side surface of the convex portion.

Further, a semiconductor film (silicon, germanium, etc.) 305 is formed thereon. The thickness is preferably 10 to 500 nm. Especially for the purpose of high integration, the thinner one is preferable. When a silicon nitride film having a thickness of 3 to 500 nm is formed instead of the semiconductor film, a MNOS or MONOS type element can be obtained. It is necessary that the formation of this coating also be performed with good step coverage. In particular, care should be taken because the coating is difficult to form on the side surface of the convex portion. For example, if a silicon nitride film is formed, a method called thermal nitriding can also be used. As described above, FIG. 3A is obtained.

Next, the semiconductor film is etched by a known anisotropic (also called directionality) etching method such as bias reactive ion etching. Although the process may be completed only by etching the semiconductor film, in order to completely remove the semiconductor film except the side surface by etching,
The underlying silicon oxide film and the substrate may be slightly etched.
In this way, FIG. 3B is obtained. During this etching process, the semiconductor film other than the side surface of the convex portion is completely removed. Although the semiconductor film 307 remains on the side surface, the semiconductor film is peeled off if the adhesion between the semiconductor film and the base oxide film is not good, so that care must be taken in manufacturing the semiconductor film.

Although FIG. 3B shows that the semiconductor substrate is also etched, such over-etching has the effect of lengthening the channel formation region when a select transistor is formed later. Have. If the channel length is extremely short, the problem of short channel effect arises, so it is necessary to keep it to an appropriate value, and such a part effectively acts in determining the length of the channel length of the present invention. Is a feature of.

Although not shown in the figure, after the step of FIG. 3B is completed, a step of dividing the semiconductor film 307 formed continuously along the groove into each element is performed. If the element isolation process is not performed first, the element isolation region can be formed simultaneously at this time. This is shown in FIG.

FIG. 5 is a cross section of two convex portions. A semiconductor film is formed on each side surface of the convex portion by the process up to FIG. Then, a silicon oxide film (thickness: 10 to 100 nm) and a silicon nitride film 5 are formed thereon.
01 (thickness 20 to 400 nm) is formed. The formation of this film is required to have good step coverage. Then, a photoresist 502 is applied to the entire surface and exposed to remove the resist in a portion to be an element isolation region (a portion where an oxide is formed). FIG. 5 (A) shows such a state.

Next, isotropic etching is performed in this state. Anisotropic etching is not preferable because the side surface of the convex portion may not be etched. The silicon nitride film is removed by this etching process. The state is shown in FIG.

Finally, only the region where the silicon nitride is removed is selectively oxidized by the thermal oxidation method to form the oxide 503. A suitable oxide thickness is 0.1 to 1.0 μm. This state is shown in FIG. In this oxidation step, the oxidation condition of the LOCOS method used for manufacturing the conventional planar type semiconductor device may be adopted.

By this oxidation step, the semiconductor film on the side surface (corresponding to 307 in FIG. 3) is also oxidized at the same time.

If this coating is made of a material such as silicon nitride, the side surface will not be oxidized, so that another means must be taken for element isolation. In addition, if the film is made of an insulating material such as silicon nitride, it is not necessary to divide it into individual elements.
However, since the impurity regions at the top are connected along the convex portion, it is necessary to divide each element.

For that purpose, as in the case of FIG. 5A, the etching may be carried out in a state where the resist is removed only in a necessary portion. This etching may be isotropic etching or anisotropic etching. After that, for example, by photo-CVD method, an oxide is deposited with the resist attached, and by lift-off method, the oxide deposited on the resist is removed, and only the oxide deposited on the region without the resist remains. Good. When adopting this method, it is required that the oxide can be deposited at a low temperature and that the step coverage be good.

No matter which method is adopted, the mask process required to form the element isolation region is only once. When the element isolation region is formed in advance by the method as shown in FIG. 4, it is sufficient to divide the semiconductor film 307. In that case, the element isolation region is formed and the semiconductor film 307 is formed. The mask process needs to be performed once for each of the divisions, that is, twice in total, and there is a concern that the yield may be reduced.

When a silicon nitride film is used for the film 307, it is not necessary to divide the silicon nitride film if the element isolation region is formed in advance by the method shown in FIG. The process up to this point is summarized as follows.

(A) When the film 307 is a semiconductor film A-1 (1) Formation of impurity layer (402 in FIG. 4) (2) Formation of groove (403 in FIG. 4) (3) Formation of element isolation region ( (405 in FIG. 4) (4) Formation of semiconductor film (307 in FIG. 3) (5) Separation of semiconductor film (307 in FIG. 3)

A-2 (1) Formation of impurity layer (402 in FIG. 4) (2) Formation of groove (403 in FIG. 4) (4) Formation of semiconductor film (307 in FIG. 3) (5) Semiconductor film ( Division of 307) of FIG. 3 and formation of element isolation regions

(B) When the coating 307 is silicon nitride or the like B-1 (1) Formation of impurity layer (402 in FIG. 4) (2) Formation of groove (403 in FIG. 4) (3) Formation of element isolation region (405 in FIG. 4) (4) Formation of silicon nitride film (307 in FIG. 3)

B-2 (1) Formation of Impurity Layer (402 in FIG. 4) (2) Formation of Groove (403 in FIG. 4) (4) Formation of Silicon Nitride Film (307 in FIG. 3) (5) Convex Section On top of silicon, oxide lift-off method

As is clear from the above, the maximum number of masks required in the processes up to here is three,
The minimum is 2.

Next, returning to FIG. 3 again, the gate oxide layer 308 is formed by, eg, thermal oxidation. If the film 307 is silicon nitride, no oxide grows on it by the thermal oxidation method, but an oxide film is obtained in the select transistor portion. Therefore, in that case, the memory transistor is of the MNOS type. Dare, MONOS
To form the mold, an oxide film may be formed by a film forming method such as CVD. However, it should be noted that in that case, the interface state becomes larger than that of an oxide film obtained by thermal oxidation. In order to obtain a good quality silicon oxide film by a CVD method, an organic silicon material such as tetraethoxyoxysilane (TEOS) is decomposed and deposited by heat or high frequency power, and further, 600 ° C. or higher, preferably Obtained by applying a temperature of 700 ° C. or higher.

After that, a film 309 for forming a gate wiring (control gate) is formed. this is,
A semiconductor film such as polycrystalline silicon, a metal film such as tungsten and chromium, a silicide thereof, or silicon and a multilayer structure thereof are preferable. In this way, FIG. 3C is obtained. After that, the coating 309 other than the side surface portion of the convex portion is removed again by anisotropic etching,
Gate wiring 310 is formed. It should be noted that this gate wiring runs along the side surface of the convex portion. Further, it is a feature of the present invention that the formation of the gate wiring does not depend on the mask process.

Then, the impurity region 311 is formed by a known impurity diffusion method such as an ion implantation method. This impurity region is formed in a self-aligned manner using the gate wiring 310 as a mask. Further, in order to form the LDD region used in the conventional planar type MOSFET,
Another insulating film may be formed on the gate wiring, anisotropic etching may be performed similarly to form a spacer, and the spacer may be used as a mask for further impurity diffusion. The details will not be given here. Further, the techniques of Japanese Patent Application Nos. 3-238709 to 3-238712, which are the inventions of the present inventors, can be used for forming the LDD in the present invention.

Finally, an interlayer insulator 312 is formed. A known planarization technique such as an etch-back method can be used for forming the interlayer insulating material. And contact hole 3
13 is formed, and the metal wiring 313 is formed. Masks are required for forming the contact holes and patterning the metal film, respectively. As described above, in the present invention, this contact hole may be shallow and suitable for fine processing.

Through the above steps, the EEPROM element is formed. Since the required number of masks is 4 to 5, the number of masks required in the conventional process can be significantly reduced. Needless to say, the processes described above are only basic ones, and it is necessary to add some processes to form a device with higher added value. Further, it may be necessary to change the above process due to the difference in the forming process of the peripheral circuit and the memory part. However, the invention does not refer to the individual processes in any further detail. Hereinafter, several examples using the present invention will be taken up and described.

[0071]

[Embodiment] "Embodiment 1" FIG. 7A shows a first embodiment. The example shown in FIG. 7A shows a MOS transistor (two transistors are depicted in the figure) formed on a semiconductor substrate having a convex region of the present invention.

In the figure, reference numerals 701 and 702 denote impurity regions having an impurity concentration of 0.1 × 10 ^{20 to} 2.0 × 10 ^20.
cm ^-3 . Particularly, the depth of the impurity region 702 is 10 to 10.
It is 20 nm. 703 is a gate electrode. As is clear from the figure, the overlap between the gate electrode and the impurity region 2 is small, and the parasitic capacitance in that part is small.

[Embodiment 2] FIG. 7B shows a second embodiment. The example shown in FIG. 7B is a MOS transistor formed on a semiconductor substrate having a convex region according to the present invention and having a low-concentration impurity region (LDD region) (two transistors are illustrated in the drawing). Is shown). In the figure, 704 and 708 are high-concentration impurity regions, and the impurity concentration thereof is 0.1 to 2.0 × 10 ²⁰ cm ⁻³ . The regions 705 and 708 provided adjacent to these impurity regions are LDD regions, and the impurity concentration thereof is 0.2 to 5.0 × 10 ¹⁸ cm ⁻³ . The region 706 is a gate electrode. Of these, the regions 707 and 708 are formed before the convex portion is formed on the semiconductor substrate. Areas 704 and 705 are normal M
It is manufactured by using the LDD manufacturing technique with the OSFET and the technique according to the invention of the present inventors.

In the figure, L is provided on both the source side and the drain side.
Although the DD region is provided, it is possible to form the LDD region on only one of them.

[Third Embodiment] FIG. 8A shows a third embodiment. The example shown in FIG. 8A is an EEPROM formed on a semiconductor substrate having a convex region of the present invention.
A device (two EEPROM devices are depicted in the figure) is shown. Here, 801 and 802 are impurity regions, and 803 is a silicon nitride film (preferably with a thickness of 10 to 10).
50 nm) and 804 are control gates. The above process may be applied to the production of this element.

Here, the thickness of the gate oxide film below the silicon nitride film 803 is thinner in the region 805 indicated by the dotted circle in the figure than in other portions. By thus thinning the gate oxide film, a tunnel current flows when a high voltage (10 to 20 V) is applied to the control gate, and is accumulated in the silicon nitride film 803.

In order to partially thin such a gate oxide film, the gate oxide film is once formed in the step of forming the gate oxide film before forming the silicon nitride film 803 on the surface of the convex portion. After that, plasma isotropic etching may be performed to etch only the oxide film above the convex portion. Then, a gate oxide film is formed again to obtain the structure shown in the figure.

[Fourth Embodiment] FIG. 8B shows a third embodiment. The example shown in FIG. 8B is an EEPROM formed on a semiconductor substrate having a convex region according to the present invention.
A device (two EEPROM devices are depicted in the figure) is shown. Here, 806, 807, 808,
809 is an impurity region, 810 is a silicon nitride film, 81
1 is a control gate. To make this device,
The above process may be incorporated.

Here, the impurity regions 807, 808, 80
The impurity concentration of 9 is formed so that 808 is the highest. Such a sandwich-like structure may be formed by impurity diffusion before forming the convex portion. By using such a structure, the dotted circle 81
A tunnel current is generated from the central part of 2, that is, the part of the impurity layer 808. This is due to the impurity concentration gradient. With such a structure, charge injection can be performed stably.

[0080]

According to the present invention, a highly integrated semiconductor device can be manufactured. The present invention brings a remarkable technological advance to the integration of the EEPROM, which has been delayed in integration because two transistors or two transistor parts are conventionally required. In addition, the present invention is
Even when applied to the fabrication of semiconductor integrated circuits other than EPROM,
Some or all of the features of the present invention may be benefited, for example, when making an EERPM device (integrated circuit), the memory area naturally uses the present invention, and even the peripheral circuitry. Using the element of
This is desirable from the viewpoint of process integration and integration.

[Brief description of drawings]

FIG. 1 shows a schematic of an EEPROM device according to the invention.

FIG. 2 schematically shows an EEPROM device according to a conventional method.

FIG. 3 shows an example of a manufacturing process of an EEPROM device according to the present invention.

FIG. 4 shows an example of a process of manufacturing a convex portion and an element isolation region according to the present invention.

FIG. 5 shows an example of a process of manufacturing an element isolation region according to the present invention.

FIG. 6 shows a circuit diagram of an EEPROM device according to the present invention.

FIG. 7 shows an example of a MOSFET according to the invention.

FIG. 8 shows an example of an EEPROM device according to the present invention.

[Explanation of symbols]

 101 Source Region 102 Drain Region 103 Floating Gate 104 Control Gate (Word Line) 105 Element Isolation Region 106 Bit Line 107 Contact Hole

Claims (4)

[Claims] 1. A convex portion provided on a semiconductor substrate and having a high concentration of impurities at the top thereof, and a semiconductor film or a silicon nitride film formed along the side surface thereof.
A MOS semiconductor device having a gate electrode formed so as to cover a part or all of the semiconductor film or the silicon nitride film. 2. A semiconductor substrate having a plurality of convex portions, the plurality of semiconductor coatings or silicon nitride coatings provided along the side surfaces thereof, and part or all of the semiconductor coatings or silicon nitride coatings. M, which has a gate wiring formed so as to cover it and a wiring which is orthogonal to the gate wiring and has a contact at the top of the convex portion.
OS type semiconductor device. 3. A bit wiring having a plurality of convex portions on a semiconductor substrate and provided along a side surface of the semiconductor substrate, and a bit wiring orthogonal to the word wiring and having a contact at the top of the convex portion. And an impurity region provided at the bottom of the substrate for supplying a voltage, a MOS memory semiconductor device. 4. A step of forming a first impurity region on the surface of a semiconductor substrate, and etching the surface of the semiconductor substrate,
A step of forming a convex portion, a step of forming a semiconductor film or a silicon nitride film, a step of removing the semiconductor film or the silicon nitride film other than the side surface by anisotropic etching, the semiconductor film or the silicon nitride film. A step of forming a gate electrode on the side surface of the convex portion so as to cover the film, and a step of forming an impurity region of the same conductivity type as the first impurity region at the bottom of the semiconductor substrate using the gate electrode as a mask. A method for manufacturing a semiconductor device, comprising: