KR830000573B1 - Device manufacturing method by plasma etching - Google Patents

Abstract

No content.

Description

Device manufacturing method by plasma etching

The figure is a graph showing the degree of the loading effect, and the horizontal axis shows the number of wafers with the etch rate ratio at 10% and 100% loading rate.

The manufacture of small circuits and circuit elements, such as silicon high density integrated circuits (silicon LSIs), includes a patterning process that removes the multilayer material by etching. The pattern shape is determined by the resist material having the patterned openings arranged above. In the presently widely used technique, the resist pattern 1 is used to expose (a) the non-exposed portion (negative resist) or (b) the exposed portion (positive resist) by an exposure process through a patterned mask and (2) development. It is formed by the process of removing.

As the circuit becomes smaller, it becomes difficult to sequentially overlap the mask patterns. In general, in order to reduce the pattern figure to micrometers or less, it is considered that the maskless method (direct method) in which the patterning is performed directly in the resist layer of the device under manufacture is more advantageous.

The trend toward miniaturization has a significant impact on the device manufacturing method itself. Wet etching, which has been in use for many years and satisfies the 4 micrometer range, is being replaced by the dry method. Plasma etching is a side effect, and dry etching has the advantage of improving resolution, in particular by reducing the tapering, for example due to the lower erosion. Dry etching also has the advantage that the demand for adhesion of the resist is alleviated and the processing of the etching additives is easy.

The plasma etching method is applied satisfactorily for the materials used in the manufacture of silicon integrated circuits (SICs). Currently fabricated device structures include silicon nitride, aluminum, phosphorus-injected amorphous oxide (P glass), polycrystalline silicon (polysilicon), thermally formed silicon oxide gate layer, and pyrolysis-forming silicon nitride by plasma deposition. And each layer of a silicon wafer. Plasma etching of SIC and other integrated circuits may also include boron nitride. Plasma etching (also called plasma stripping) is used to remove this after the masking action of each resist layer is complete. Resist removal sometimes contains quite subtle factors during its removal process. For this reason, the dry method is sometimes used in integrated circuits and other devices that use other technologies in parallel with silicon. Materials to be treated include weak and ferromagnetic materials (permalloy substituted yttrium iron garnet, etc.) optically active materials (lithium niobate, lithium tantalate, etc.), metals and metal compounds.

In the historical background, the plasma etching method, which is rapidly put into practical use and is still widely used, is treated using known CF ₄ -O ₂ as an etchant. This etchant is commercially available from the past and used for the removal of various substances and other substances mentioned above in various reactors. In addition, the CCl ₄ series etching material is also used to remove the above-mentioned materials.

When simultaneously etching a plurality of wafers as a CF ₄ -O ₂ there appears (to loading effect- etching rate is dependent on the size of the features front and back surfaces), well known as the "loading effect" due to the effect ". When using a CCl ₄ This phenomenon occurs for all commonly used materials: The dependence of the etch rate on the surface to be treated leads to low yields for high loadings, which is a significant problem in itself but leads to more non-uniform etching. Non-uniformity between or within the wafer may be ignored, but factors such as (a) the etch rate ratio between the material to be processed and the engine raised below, (b) the necessity of end point detection, and factors related to the method (c) dimension error This may cause a serious problem (increasing the loading rate may increase the etching rate suddenly and make it impossible to control the lower erosion phenomenon within the required etching time). There is).

Attempts were made to mitigate the effect of loading by improving the reactor, but there was no performance. Although the uniformity of the etching can be improved to some extent by the reactor improvement, the non-uniform etching also depends on the loading rate, and still remains a problem in small circuits. For the reactor currently in use, see R.G. Plsen's paper (J. Vac. Sci. Technology 14,266 (1977)).

The loading effect is already known and it is known to be a serious obstacle for certain materials but not for other materials. For example, this effect is disturbed when CF ₄ -O ₂ is used for etching polysilicon, but it is not a problem when used for silicon oxide. The etching of silicon oxide does not cause a problem even if the etchant is changed. It is thought that the loading effect does not matter for this material.

In summary, the present invention

The manufacturing method by plasma etching can be configured to reduce the effect of the loading effect. In general, in the plasma etching of the present invention, the dependence of the etching rate on the surface to be treated is reduced, so that the optimum conditions in the preferred embodiment can be substantially independent of each other. In addition, the uniformity of etching (between wafers or in wafers) is also improved.

For ease of explanation, the etching method of the present invention is defined as plasma etching only. However, as detailed later, these methods are sometimes referred to by specific names, for example, reactive ion etching and reactive sputtering. It is known that certain types of known methods do not involve loading effects. The mycology of the present invention therefore relates to a method involving a loading effect. According to the present technology, the field of the present invention is defined as a method involving a loading effect when etching using plasma generated by the use of CF ₄ -O ₂ or CCl ₄ . Although the method of the present invention is not limited to this field, these etchant is particularly useful in the art because it is widely used in silicon integrated circuit technology and applied to the whole material to be plasma etched. In general, the method relates to finding conditions such as reducing the dependence of the etching rate on the surface area.

This is achieved by a system of characteristics as long as the lifetime of an effective etch source is determined by conditions other than etching. In other words, the inherent life of the etching source is shorter than the life due to the etching reaction itself. For this reason, the former is defined as "etching agent intrinsic life" and the latter as "etching life". If the inherent life of the etchant is sufficiently short that the effect of the change in the area to be treated is small, the loading effect is virtually eliminated. For example, the preferable conditions of the maximum loading in the target reactor are defined by etchant intrinsic life ≤ 0.1 x etching life.

Reducing the loading effect leads to an improvement in the uniformity of the etching itself. Optimum uniformity is obtained by setting conditions such as to uniformly arrive the etching source on the target surface. The reduction of the loading effect also contributes to at least the reduction of the bottom erosion effect.

Many factors are involved in optimizing etching, but in a preferred embodiment the inherent lifetime of the etchant is shortened by recombination in the plasma. In this preferred embodiment, the etchant is divided into two parts: (a) an effective etching source and (b) a rebinding agent.

When the present invention is described in detail,

Plasma etching is a general term for a method of removing a base material mainly by a chemical reaction using a reaction source in a plasma state. The method of the present invention differs from these methods in that removal is mainly done by momentum exchange. The latter includes ion milling, various ion etching and sputter etching which do not exhibit the loading effect related to the present invention. Of course, a certain amount of momentum is created by the plasma field itself. Therefore, surface removal is not mainly due to momentum exchange, but it is believed that momentum exchange is occurring and actually contributes to the generation or increase of chemical activity.

The target plasma etching in the present invention is a method in which the loading effect is reduced, which is defined as follows.

Reactive plasma etching means a plasma etching method having characteristics due to the loading effect. In the field to be etched by CF ₄ -O ₂ or CCl ₄ in connection with silicon technology, the object plasma etching is defined as a method which produces a loading effect in the plasma of these etchant. The change in the etching rate of silicon oxide in Cf ₄ -O ₂ is 25% or less, between 100% and 10% loading rate in any apparatus and under the same conditions. Among the SIC-related materials, materials having a loading effect under such conditions include silicon (single crystal or polycrystalline), aluminum, silicon nitride by plasma deposition, silicon nitride by thermal decomposition, boron nitride and various resists.

The loading effect indicates that the etch rate depends on the surface area. Usually it is impossible to completely remove this dependency in practice. However, even if present, this can be allowed to a certain extent. Silicon nitride etching, for example, does not require strictness because the pattern spacing is relatively large, thus allowing a significant loading effect. Although it is not possible to quantify this term, 25% of the etching rate at 10% loading and full loading of the capacity used is acceptable.

(a) Inherent life of etchant

In the case where the surface to be treated does not exist, the average life of the plasma-forming etching source is shown. The end of life results from the recombination of the material formed by the plasma with the material that normally corresponds to the gaseous component introduced into the plasma.

(b) etching life

It shows the lifetime when the etching source in the plasma state undergoes chemical reaction with the surface to be treated. It is convenient to show the etching life at the time of full loading of the reactor, and this is called the minimum etching life.

In the embodiment of the present invention, the intrinsic life of the etchant is always shorter than the etching life and usually becomes 10% or less of the minimum etching life. Embodiments mentioned later match this condition. The operating conditions can be easily measured. Etchant intrinsic life can be measured in a simple way by operating the reactor without the workpiece.

As one method, the remaining time of an etching source may be measured by turning off a power supply, for example by measuring the absorption amount of CW laser beam of a suitable wavelength. As an alternative method, chemical titration may be performed at least two points downstream of the reactor. Experiments showing the loading effect and the like show that the etching life under conditions under which the actual etching rate is obtained is about 10 milliseconds or more. Therefore, under the conditions of the present invention, the intrinsic life of the etchant is 1 millisecond or less. Considering the measurement error and the variation according to the measurement technique used, the inherent life of the etchant in the preferred embodiment is defined to be 0.1 millisecond or less. This average value is considered to be a feature of the present invention which significantly reduces the deleterious effects of loading rate compared to the above known art. Preferred methods are conditions such that the intrinsic life of the etchant does not exceed about 0.01 milliseconds. In fact, this figure corresponds to an experimental system with a loading effect of less than 1% (compared to full loading and 10% loading in certain reactors).

In most cases short intrinsic lifetimes are achieved by the selection of gas adjustments introduced into the plasma, and such compositions are chosen to promote recombination. In a preferred embodiment, the recombination agent is chemically different from an etching source or an etching source generating material. However, useful systems often have the same etchant and rerelease.

As mentioned above the surface area dependence of the etching rate is detrimental on its own but according to the invention this dependence is reduced and can be substantially eliminated in some cases. Uniformity can be achieved by appropriate etching conditions if the loading effect is substantially eliminated. The most important condition is the flow pattern, which is usually determined by the design of the reactor

Deira in the examples is obtained in a parallel plate reactor that generates a radial flow as described in US Pat. No. 3,757,333. In implementing the present invention as such an apparatus, sufficient etching uniformity is obtained at least for the current LS1 technology.

About the etchant composition,

In the examples, shortening of the intrinsic life of the etchant is achieved by recombination at the site serving as the inlet gas. This site may be inert and may only occur in the discharge region, may be the same as or different from the etching source, and is usually etched by recombination with the etching source at an average time of 0.1 milliseconds or less (preferably 0.01 milliseconds). Relatively low or preferred sex forms non-etchable materials. This reaction product may be the same as or different from the initial reactant. Etchant strains of the examples include CF ₃ Cl, CF ₃ Br, C ₂ F ₆ -Cl ₂ . Mass spectrometry and other techniques show that the effective etch sources for these strains are Cl, Br, and Cl in atomic form, respectively. Therefore, in the two examples, the recombination agent is a CF ₂ or CF ₃ derivative, and also in the third example, carbon fluoride (CF ₃ product, CF ₃ or CF ₃ derivative of the same family). The fourth system, C ₂ F ₆ (which is mainly academic interest because of low etching speed) is thought to produce CF ₃ in the plasma state and is thought to involve recombination between etching sources. In the fifth system using BCl ₃ -Cl ₂ , the presence of atomic chlorine is recognized by luminescence analysis, which is thought to contribute at least to the etching source. The rebinding agent is also present in the main plasma and is thought to be BCl ₃ or other derivative. For the C ₂ F ₆ -Cl ₂ line, studies were made by changing the relative amounts of the two substituents and argon substitution of C ₂ F ₆ . In the latter case, Cl + Cl = Cl ₂ is considered to be the actual recombination mechanism in the fact that the loading effect does not change under the same conditions. However, in the former experiment, carbon fluoride is also believed to contribute to recombination.

It is usually desirable to see the recombinant and the etchant as chemically different from consideration other than the loading effect. By separating the two materials, flexibility to bind to factors such as etch rate, etch shape, etc. is obtained (US Pat. App. No. 529549, issued March 31, 1978). This describes the type in the plasma. Thus, a single gas CF ₃ Cl generates chemically different recombinants and etch sources, with an atomic ratio of 1: 1, which is suitable for many etching methods, but with chlorine or carbon fluoride (Cl ₂ or C _2). This ratio can be changed by the addition of F ₆ ) Experiments have shown that all of the effective etching sources are Cl (generated from CF ₃ Cl or Cl ₂ ).

In many systems suitable for practicing the present invention, the active material acts as an etchant and recombinant. In many cases, however, mainly the etchant and the major rebinding agent are chemically different. In addition, in many preferred systems, the material does not exhibit a loading effect. However, since the required etching characteristics are not obtained, this fact is often of little practical value. For example C ₂ F ₆ does not show a loading effect in most cases, but the etch rate is too low for practical use. Chlorine itself is an etchant with no loading effect, but the edging rate is excessive and uncontrollable for thin film technology. The etch rate of Cl ₂ can be lowered to an acceptable level, for example by dilution with argon, which is generally consistent with the elements of the present invention. Unlike carbon halides, the dilution gas does not act as a rebinding agent. Halogen-carbon based recombinants are shaped by the preferred reaction in the vicinity of the etch wall, as described above in U.S. Patent Application No. 929549, allowing mechanisms to form very few vertical walls of the lower erosion shape. Such a function can also be seen in other halides such as, for example, BCl ₃ , an aluminum etchant.

For some etching sources of the present invention, the composition can be determined. When the main etching source is obtained by adding an atomic chlorine generator in the gas mixture, it is reasonable to set the ratio of etching agent and recombination atoms in the range of 1% to 95% for the gas mixture introduced in the plasma. This positive limit has been found to give good etching under suitable conditions. The use of a single compound such as CF ₃ Cl gives a useful technique for good line in plasma. The ratio in this case is 50% at the time of introduction, and inevitably gives the same ratio even in the plasma. The ratio in this case is 50% at the time of introduction, and inevitably gives the same ratio even in the plasma. This is a useful etchant from various viewpoints including a loading effect but does not exhibit ideal anisotropic action under normal conditions. As described in U.S. Patent Application No. 929549, the optimum ratio in the etch shape is achieved by the vapor of CF ₃ (eg by introduction of molecular state C ₂ F ₆ ).

It is also known that the composition may be changed by other factors. For example, it may be diluted to stabilize the plasma state. In addition, helium, when added to CF ₃ Br, prevents plasma non-uniformity and thereby prevents non-uniformity caused by it.

In order to obtain an optimal etch, it is desirable to reduce the power, for example, to limit the range of the plasma, but the loading effect is usually relatively independent of conditions such as power pressure. Under conditions where the loading effect is recognized, it has been found that the loading effect can be reduced while lowering the temperature. The influence of this comparative hardness is considered to be due to the increase in the etching life.

In a preferred embodiment, recombination sites present in the gas line are used. This fact is strongly demonstrated by spectroscopic analysis of the exhaust gas, but this does not mean that recombination is not excluded from the selective occurrence of solid surfaces, ie resists and / or treated surfaces. In fact, thermal recombination inevitably occurs on solid surfaces that become heat sinks. If some or all of the recombination occurs on a solid surface, it is believed that there are only other mechanisms involved, such as adsorption (which leads to trapping of the etch source) and the active surface and, in particular, the resist surface. It is thought to cause.

The main object of the present invention is related to the loading effect. As described above, the composition and other etchant conditions may be adjusted to meet other purposes. Such conditions assume that the layer to be treated is I micron or less, such as a fine circuit or a small device. Etch rates in such situations are on the order of 100 to 2,000 angstroms per minute. Lower speeds are less productive and difficult to control. The plasma output may be defined as about 100 to 5,000 W (0.05 to 5 W / cm, preferably 0.1 to 1 W / cm) for a 16 inch reactor. If the layer thickness is several microns or more, the upper limit may be exceeded because the etching rate is increased.

The lower limit usually corresponds to the lowest allowable etch rate. The pressure is usually in the range of 0.005 to 1 Torr (5 to 1.000 μm Hg). The lower limit enters the region of reactive ion etching, and at low pressure, lattice breakage may occur. If the upper limit is exceeded, it becomes difficult to limit the plasma range and the uniformity of etching is lost.

For the example,

For comparison purposes, the examples were carried out under the same conditions and under the same conditions. In the experiments of the example, a radial parallel tubular reactor of 18 kPa (45.6 cm) was used. The apparatus had two parallel horizontal hollow electrodes in a Pyrex vacuum vessel. A high frequency power of MHz was applied, the lower plate was grounded and the workpiece was allowed to be discharged, the discharge occurred in the range of 0.1 to 1 Torr, and the etchant gas flowed continuously in the discharge zone. Discharged by a two-stage machine pump with a discharge capacity of 25 cubic feet / minute (0.708 cubic meters / minute) Hot water (80 ° C) passes through the positive electrode for several minutes before the first pump operation to minimize water vapor condensation during loading The vessel was then opened, the workpiece was placed for the sample stage, the vessel was then closed and the pump was operated at the time the pressure reached about 30 micrometers Hg. The cold water was circulated (25 ° C.) The pumping operation was continued while the pressure was introduced until the pressure was several micrometers Hg The etching conditions were normally maintained as follows.

0.1 to 1.0 Torr

Flow rate 20 to 200 SCCM (cm 3 / min room temperature equivalent)

High frequency output 100 to 2,000 watts

Electrode spacing 7-30mm

Board temperature 25 to 30 °

[Examples 1 to 5 (this embodiment constitutes curve 10 of the drawing)]

Single crystal silicon was etched with CF ₃ Br-30% He and the conditions were 500W output pressure 0.5T, electrode spacing 30mm, sample temperature 25C, flow rate 175SCCM, wafer diameter 3 inches (7.6cm). The number of wafers and the obtained etching rate are as follows.

Wafers having the same surface composition and dimensions were etched with dilutionless CF3Cl. The conditions were 200 W, 0.3 Torr electrode spacing 30 mm, sample zone temperature 25 DEG C, and flow rate 200 SCCM.

[Examples 10-17 (As a known technique, this example constitutes curve 12)]

This data is a control and shows the results of experiments in which the same monocrystalline silicon layer was etched as CF ₄ -8% O ₂ . The loading effect obtained is shown in the table below. The conditions were 300W output, 0.3 Torr electrode spacing 30mm sample temperature 100 ℃, flow rate 150SCCM.

[Examples 18-19]

(Etched by C ₂ F ₆ -Cl ₂ on a 3 inch wafer (7.6 cm) with phosphorus-injected polycrystalline silicon layer), reactor conditions were 400W of output, 0.35 Torr pressure, 30mm electrode spacing, and sample temperature 25 ℃ , 15% Cl ₂ + C ₂ F _{6 An amount of} 85% of the stream was 175 SCCM.

[Examples 20 and 21]

(Etching 90% Cl ₂ : 10% C ₂ F ₆ )

[Example 22-25]

As a special case, the composition of the introduced gas was BCl ₃ 95% + Cl ₂ 5%, and a 4000-mm thick alloy layer of aluminum-copper 4% was formed on a 3 inch diameter (7.6 cm) wafer. The output was 600W, the pressure was 0.1 Torr and the other conditions were the same as the other examples.

The loading effect of the present invention refers to the dependence of the etching rate on the total processed area. The area of each wafer in each tank of an Example is the same. In the first three sets, the wafer is not equipped with a mask. In Examples 18-21, masks are set up for standard patterning between wafers. The last set of wafers is also masked.

Claims (1)

At least one operation is required to fabricate at least one device, and the device to be manufactured has at least one portion of the etched surface, the etched surface being exposed to a plasma atmosphere in the apparatus, the plasma being a gas between the two electrodes. It is formed by applying a high frequency electric field to a phase material, It has a configuration showing a loading effect when the etching target surface is etched by a plasma passing through a gas mixture composed mainly of CF ₆ and O ₂ or CCl ₄ , The loading effect is defined by a change in etch rate of at least 25% when the loading rate of the device varies from 10% to 100%, and the plasma is etched mainly by the reaction of the etched surface with the main etching source. In the device manufacturing method by etching, The gaseous material is composed of active chemically identifiable chemically in the plasma in two, the first component acting as the main etching source reacts with the surface to be etched to remove the surface material, The second component, which acts as a recombination source, combines with the unreacted main etch source to define the average intrinsic lifetime of the main etch source in the plasma, By plasma etching, the amount of recombination source is maintained at a level sufficient to reduce the average intrinsic life of the main etching source to a value not exceeding 1/10 of the average etching life due to chemical reaction with the etching target surface. Device manufacturing method.

Images