Method for Producing a Nonvolatile Semiconductor Memory

1. Field of the Invention

The present invention relates to an electrically writable-erasable nonvolatile semiconductor memory element.

2. Description of the Prior Art

A floating gate avalanche injection-type MOS (hereinafter referred to as FAMOS) has been placed in practical use as an electrically programmable MOS-type nonvolatile memory. Among the FAMOS's, the two-layer gate-type nonvolatile memory which is usually used as an N-channel element, comprises a source region and a drain region formed on one surface of a semiconductor substrate having one conductivity type, the source and drain regions having a conductivity type opposite to that of the semiconductor substrate, a first insulating film formed on a channel region which is located between the source region and the drain region, a floating gate which is formed on at least a portion of the first insulating film and which is electrically floated, a control gate formed on the floating gate via a second insulating film, and a high impurity concentration region formed on a portion of the channel region and having the same conductivity type as that of the substrate. With the above device, the source region and the substrate are maintained at ground potential, and a positive voltage is applied to the drain region and to the control gate. Most of the drain voltage is consumed in a depletion region in the vicinity of the drain region, where a high electric field is maintained, whereby hot electrons and hot holes are generated due to the avalanche in the high impurity concentration region formed in a portion of the channel region. The hot electrons are injected into the floating gate beyond the energy barrier of the first insulating film composed of a gate oxide film. With the above device, however, it is necessary to maintain a margin of mask alignment to form the high impurity concentration region on a portion of the channel region. This makes it difficult to make the elements small in size and, further, the high impurity concentration region reduces the channel width so that the read-out speed is decreased.

In order to preclude the above-mentioned defects, Japanese Patent Application No. 8655/77 proposes a method by which the high impurity concentration region is formed on a portion of a region which is in contact with at least one of the two sides that are in contact with neither the source region nor the drain region in the outer periphery of the channel region. The devices produced by the above method permit easier reading than the earlier similar devices. According to the above-proposed method, however, the device is formed in the order of a high impurity concentration region, a thick oxide film of the field region, gate oxide film, floating gate, source region and drain region. Accordingly, impurities in the high impurity concentration region tend to be diffused by the subsequent thermal steps such as formation of thick oxide films and n.sup.+ diffusion, which will be performed at high temperatures for extended periods of time. With the above proposed method, therefore, it is difficult to control the concentrations of impurities and it is also not possible to sufficiently increase the concentrations. Further, since the high impurity concentration region is formed beneath the floating gate, the channel width tends to narrow and the threshold voltage tends to increase. Moreover, the thickness of the gate oxide film should desirably be small to attain high-speed operation. With the above method, however, the impurities are diffused from the floating gate during such a thermal step as n.sup.+ diffusion, causing the insulating properties of the thin oxide films to be degraded. Moreover, according to the device disclosed in Japanese Patent Application No. 8655/77, the floating gate has portions riding on the thick field oxide film on both sides of the channel. This causes the floating gate to be stretched in a direction at right angles to a direction in which the source region and the drain region are connected. To cover the floating gate riding on the field oxide film by means of a control gate via the insulation layer is desirable from the viewpoint of increasing the operation speed of the element. From the viewpoint of the layout, however, the control gate is better wired in a direction connecting the source and the drain. According to the above-mentioned setup, therefore, the width of the control gate tends to increase, causing the necessary areas to increase.

With the above-mentioned FAMOS, the writing can be electrically effected, but the memory must be erased by means of ultraviolet light. In an attempt to make a memory device which electrically effects the writing as well as the erasing and which also precludes the defects of the FAMOS's, there has been proposed a method of implanting ions to a high concentration into the channel after the source region, drain region and a thin oxide film of gate have been formed, followed by the formation of a floating gate. According to this method, it is easy to control the impurity concentration since there is no thermal step that is performed at high temperatures for extended periods of time after the channel has been doped with impurity. However, there still remains a problem resulting from the addition of high concentration impurities in the channel region. Therefore, there is imposed a limit for increasing the concentrations, and further, the implanted ions tend to diffuse in considerable amounts into the gate insulating film to deteriorate the dielectric strength.

The object of the present invention, therefore, is to provide a method of producing a nonvolatile semiconductor memory which is mentioned earlier having a small area or, in other words, in which the high impurity concentration region is in contact with the channel region and is substantially located outside the channel region, the impurity concentration being allowed to be accurately controlled, and in which the floating gate is extending in a direction to connect the source to the drain.

To achieve the above-mentioned object, the method of the present invention comprises forming a floating gate prior to the high impurity concentration regions, and forming the high impurity concentration regions just outside the channel region by self-alignment with the floating gate using the floating gate as part of a mask.

The features and advantages of the present invention are mentioned below in further detail by way of some examples in conjunction with the accompanying drawing. However, it should be noted that the examples are only to illustrate the invention, and it is possible to effect various other modifications and improvements without departing from the scope of the present invention.

FIG. 1 to FIG. 3 are cross-sectional views showing steps for producing a nonvolatile semiconductor memory according to the present invention;

FIG. 4 is a plan view of the device obtained through the steps of FIGS. 1 to 3;

FIG. 5 and FIG. 6 are cross-sectional views along lines V-V' and VI-VI' of FIG. 4;

FIG. 7 and FIG. 8 are plan views of devices according to other embodiments of the present invention;

FIG. 9 is a cross-sectional view along a line IX-IX' of FIG. 8; and

FIG. 10 is cross-sectional view of a device according to a further embodiment of the present invention.

A film of silicon dioxide is formed on the surface of a p-type silicon substrate 1 by way of a conventional thermal oxidation method. Openings are formed by means of a conventional photo-lithographic method in the film of silicon dioxide in places where a source and a drain are to be formed. Impurities of the n-type are thermally diffused into the p-type silicon substrate through the openings thereby to form a source region 2 and a drain region 3. After the film of silicon dioxide is removed, the surface of the resulting silicon substrate is oxidized by a low-temperature wet oxidation method at 800.degree. C. for 210 minutes. In this case, as shown in FIG. 1, a thick oxide film 4 having a thickness of 550 nm is formed on the source and drain regions 2 and 3 containing impurities at high concentrations, and a thin oxide film having a thickness of 250 nm is formed on other regions containing impurities at low concentrations. Part of the thus formed oxide film is then removed using an etching solution (HF:H.sub.2 O=1:10) in such a manner that the thin oxide film is completely removed and the surface of the silicon substrate 1 appears. The oxide film of a thickness of about 300 nm therefore remains on the source and drain regions 2 and 3. The exposed surface of the silicon substrate 1 is once oxidized in an atmosphere consisting of oxygen diluted with nitrogen under a pressure of 10.sup.-2 to 10.sup.-3 atm. in a furnace heated at such a high temperature as 900.degree. C. or higher, and a polycrystalline silicon layer of a thickness of 200 to 500 nm is formed on the surface of the silicon substrate. This polycrystalline silicon layer is doped with impurity by an ion implantation method. For this purpose, B.sup.+ ions are accelerated at a voltage of 30 KV and implanted at a dose density of 5.times.10.sup.13 ions/cm.sup.2, or P.sup.+ ions are accelerated at a voltage of 30 KV and at a dose density of 2.times.10.sup.14 ions/cm.sup.2. The thus obtained oxide film and the polycrystalline silicon layer are formed to a desired shape as shown in FIG. 2 by means of the photo-lithographic method. The oxide film serves as a gate insulating film 5 and the polycrystalline silicon layer serves as a floating gate 6. The floating gate is extending in a direction connecting the source region to the drain region and has the same width as that of the channel region, and its both ends are riding on the oxide film 4 on the source and drain regions 2 and 3. When the control gate is to be provided thereon via another insulating film, the riding portions act such that a voltage applied to the control gate is effectively divided between the floating gate and substrate owing to the capacitance deviation, such that the electric charge is easily injected into the floating gate. In order to increase the writing, erasing and reading speeds, the gate oxide film 5 should preferably have a thickness of smaller than 20 nm and more preferably a thickness within a range of smaller than 10 nm to about 4 nm. With the thickness being smaller than about 3 nm, however, the stored electric charges escape from the floating gate due to the direct tunnel effect, whereby it becomes difficult to retain the written information. As for the elements which do not require an electrical writing-erasing function, the upper limit of the film thickness may be set to be nearly the same as the film thickness of an ordinary gate oxide film.

Then, a mask 7 for forming the high impurity concentration regions is formed as shown in FIG. 3. The mask 7 may be a silicon oxide film or a photoresist film. FIG. 3 is a cross-sectional view of the device cut in a vertical direction with respect to a line connecting the source region to the drain region passing through the channel region. High impurity concentration regions 8 are formed by implanting B+ ions accelerated at a voltage of 30 kV at a dose density of 7.times.10.sup.12 to 2.times.10.sup.14 ions/cm.sup.2 ; regions having a reverse breakdown voltage of 6 to 15 volts at source and/or drain are thus formed. In this case, the floating gate 6 constitutes a part of the mask. Namely, the high impurity concentration regions 8 are formed just outside the channel region by self-alignment with the floating gate 6 which serves as a portion of the mask. The openings formed in the mask 7 may overhang on the source region 2 or the drain region 3. The source region 2 and the drain region 3 have been covered with a thick oxide film 4. Therefore, B.sup.+ ions are not allowed to reach the source region 2 or the drain region 3. Even if small amounts of ions may reach there, the electrical properties of the source region 2 and the drain region 3 remain unchanged because of their very high impurity concentration.

Referring to FIG. 3, the high impurity concentration regions 8 have been provided on both sides of the channel region. The high impurity concentration regions 8, however, need not necessarily be provided on both sides, but may be provided on one side only.

After the mask 7 has been removed, the surface of the floating gate 6 is thermally oxidized to form a second insulating film 9. At this moment, the surface of the substrate is also oxidized. A film of silicon nitride may also be coated thereon. On the second insulating film 9 is further formed a control gate 10 as shown in a plan view of FIG. 4. FIG. 5 is a cross-sectional view along a line V-V' of FIG. 4, and FIG. 6 is a cross-sectional view along a line VI-VI'.

As will be obvious from the above-mentioned embodiment, according to the present invention, the high impurity concentration layer is formed after the floating gate, and the subsequent steps are all carried out at low temperatures of 900.degree. C. or lower. Therefore, the layer of high impurity concentrations is not excessively diffused, and the concentrations of impurities can be well controlled, thereby overcoming these difficulties encountered in the prior art. Since the high impurity concentration regions have been formed outside the channel region but in contact therewith, neither the width of the channel region nor the threshold voltage need be increased, unlike the prior art. Besides, since the floating gate is so formed as to ride on the source and drain regions, there is no need to increase the width of the control gate which is arrayed in a direction to connect the source region to the drain region and which is so formed as to cover the floating gate. Therefore, the necessary areas can be reduced.

In the foregoing description, the field regions other than the active region are constructed in the same manner as the second insulating film. It is, however, possible to leave on the field regions a thick insulating film which has been formed prior to the formation of source and drain regions, or to apply a thick phosphosilicate glass after the second gate insulating layer has been formed. Most simply, the control gate may be formed with aluminum, but it is of course possible to form the gate using silicon.

Referring to FIG. 4, the control gate has been described to be broader than the floating gate. The operation, however, will not be much affected even if the control gate is designed to be equal to, or slightly narrower than, the floating gate. However, it is safe to design these two gates as described in FIG. 4, so that the writing speed is not affected by the reduced capacitance between the two gates, and so that the reliability is not decreased. Moreover, the high impurity concentration regions 8 need not necessary be brought into contact with either one of the source region 2 and the drain region 3, but may be formed in places where the high impurity concentration regions 8 are in contact with neither the source region 2 nor the drain region 3.

The high impurity concentration region may lie over the whole areas of source and drain regions except the channel region. In this case, the first avalanche takes place in either a boundary between the high impurity concentration region and the source region or a boundary between the high impurity concentration region and the drain region. FIG. 7 is a plan view of this embodiment, in which the same reference numerals as those of FIG. 4 designate the same members as those of FIG. 4, B.sup.+ ions can be implanted in any regions 11 inclusive of the floating gate 6. Like that of Embodiment 1, the electrical properties of the source and drain regions 2 and 3 are not changed even when the region for implanting ions is extended to the source and drain regions. Consequently, the surface of the substrate located between the source region 2 and the drain region 3 in the region 11 and not coated with the floating gate 6 is converted into a high impurity concentration region having the same conductivity type as that of the substrate.

Thus, when the high impurity concentration region is extended through the whole length between the source region and the drain region, it is possible to eliminate the ion implantation step in the polycrystalline silicon layer that was effected in Embodiment 1. Instead, the above purpose can be attained by implanting ions for forming high impurity concentration regions. Besides, the length of the channel region can be shortened. Further, to minimize the length of the high impurity concentration region which is in contact with the source region or the drain region, is effective to minimize the undesirable avalanche current.

The thickness of the gate insulating film need not be reduced throughout the whole channel region, and B.sup.+ ions need not either be implanted over a wide area as shown by the region 11 of FIG. 7. In an embodiment shown in FIG. 8, the silicon substrate is exposed in regions 12 and 13 by etching a gate insulating film 17 of an ordinary thickness using a mask; a thin gate insulating film is formed only on the regions 12 and 13. FIG. 9 is a cross-sectional view along a line of IX--IX of FIG. 8. As will be apparent from FIG. 9, the insulating film 17 on other channel regions has been formed beforehand maintaining an ordinary thickness (50 to 120 nm). The insulating layer 14 on the field regions can be most easily formed simultaneously with the gate insulating film 17 having ordinary thickness such that its thickness will be the same as the thickness of the gate insulating film 17. It is of course possible to increase the thickness of the film as mentioned in the proviso of Embodiment 1. When B.sup.+ ions are implanted after the floating gate 6 has been formed, high impurity concentration regions 15 and 16 will be formed in the silicon substrate 1 in the portions of the regions 12 and 13 which have not been covered with the floating gate 6. Then, the second insulating film 9 and the control gate 10 are formed in the same manner as in Embodiments 1 and 2.

As the area of thin gate insulating film decreases, the capacitance formed between the floating gate and the channel is decreased, and the capacitance formed between the control gate and the floating gate plays an increasing role. The resulting advantage is that the speed of the writing and erasing operations is increased.

In Embodiment 1, it was mentioned that the floating gate 6 needs not necessarily ride on both the source region 2 and the drain region 3. Here, it is possible to positively provide a gate oxide on an offset region which has not been covered with the floating gate 6 on the channel region. FIG. 10 is a cross-sectional view showing this embodiment. The gate insulating film 17 of an ordinary thickness (50 to 120 nm) is provided on the channel region on the side of the drain. According to this embodiment, the floating gate 6 is riding on a portion of the insulating film on the source region 2 and a portion of the gate insulating film 17.

With the thus constructed memory element, the switching operation can be effected between the memory region having the floating gate 6 and the drain region owing to the region of MOS construction having the gate insulating film 17. That is, if a threshold voltage of the region of MOS construction having the gate insulating film 17 is set to be of a small positive value, the switch can be rendered on or off by a positive voltage or zero voltage applied to the control gate 10. Owing to the above operation, it is allowed to select specific lines or rows of the memory array.

Although the foregoing embodiments have dealt with the n-channel elements, it is of course possible to reverse the conductivity type. Further, although the foregoing description has dealt with the cases in which the floating gate is located on the outermost side while the ions are implanted to form the high impurity concentration layer, it is also possible to form an insulating film of a thickness of about 50 nm on the whole surface of the chip and implant the ions from the upper surface of the insulating film. In this case, the region for implanting ions is determined by the floating gate; the energy for implanting ions should simply be increased in relation to the thickness of the insulating film. When a photoresist film is to be used as a mask 7, the insulating film should be made of silicon nitride or silicon oxide such that the chip will not be contaminated by the photoresist agent. The insulating film may be retained even after ions have been implanted, so that it is used as a second gate insulating film between the control gate and the floating gate.