Patent Application
20070281478
Certain preferred embodiments of the present invention will hereinafter be described with reference to the accompanying drawings. Referring first to
In a plasma processing chamber in the plasma etching apparatus (in this case, an etching chamber 101), its inner wall can be controlled within a temperature range of from 20 to 100° C. by a temperature control means (not shown). On a top part of the etching chamber 101, the antenna 102 is arranged, and between the etching chamber 101 and the antenna 102, there is disposed a dielectric 103 through which UHF electromagnetic waves can transmit. Connected to the antenna 102 via a waveguide 104 and a matching network 105 is a RF power source 106 which produces UHF electromagnetic waves.
On and around an outer peripheral part of the etching chamber 101, a solenoid coil 107 is arranged to form a magnetic field in the etching chamber 101. Inside the etching chamber 101 and below the antenna 102, a lower electrode 109 is disposed as a pedestal for placing a substrate such as a wafer 108 thereon. To the lower electrode 109, a RF power source 111 located outside the etching chamber 101 is connected via a RF bias matching network 110.
In the plasma etching apparatus constructed as described above, UHF electromagnetic waves outputted from the RF power source 106 are fed into the etching chamber 101 via the matching network 105, waveguide 104 and antenna 102.
On the other hand, a magnetic field is formed in the etching chamber 101 by the solenoid coil 107 arranged around the etching chamber 101. By an interaction between the UHF electromagnetic waves and the magnetic field formed by the solenoid coil 107, etching gas introduced into the etching chamber 101 via a gas feeder (not shown) is efficiently converted into a plasma.
Referring next to
Numeral 119 indicates a database unit which stores criterion values for determining the end of seasoning. Numeral 118 designates a computing unit, which compares the data collected by the collection unit 117 with the corresponding criterion value or values stored in the database unit 119 and then performs computation. As a result, it is possible to determine the end of seasoning conducted subsequent to wet cleaning. Indicated at numeral 120 is a control unit, which controls the etching apparatus on the basis of output signals from the computing unit 118.
A detailed description will hereinafter be made of examples that determine the end of seasoning.
In this example, the intensity of plasma emission is chosen as a parameter to be monitored. Specifically, a plasma emission produced in the etching chamber 101 is guided out through an optical fiber 121, and the guided-out emission is monitored by a photodetector 113.
A fluorine (F)-containing processing gas employed in this example was a mixed gas of SF6 and CHF3, the flow rates of which were 15 mL/min and 112 mL/min, respectively. On the other hand, the processing pressure was 0.4 Pa, the plasma-producing electric power was 800 W, and the RF bias applied to the lower electrode 109 was 15 W.
In
As shown in
If the seasoning is insufficient, in other words, shortly after wet cleaning, there is a high probability that carbon-containing deposits still remain on the inner wall of the processing chamber. The fluorine (F) in the thus-fed mixed gas is, therefore, consumed by the carbon-containing deposits still remaining on the inner wall of the processing chamber. It is, accordingly, considered that the feed rate of fluorine (F) to the wafer decreases to lead to a reduction in the etch rate of polysilicon.
When the seasoning is performed thoroughly, on the other hand, the carbon-containing deposits on the inner wall of the processing chamber are rendered less. It is, hence, considered that the fluorine (F) to be consumed at the inner wall of the processing chamber decreases, the feed rate of fluorine (F) to the wafer increases, and the etch rate of polysilicon increases.
When the transition of changes in the emission intensity ratio (C2/H) of a plasma emission has been monitored and the emission intensity ratio (C2/H) has reached a steady value as illustrated by the curve b in
In this example, a wafer bias voltage Vpp (peak-to-peak voltage) is chosen as a parameter to be monitored.
Three etching apparatus (apparatus 1, apparatus 2, apparatus 3) of the same specification were used. After applying wet cleaning to each etching apparatus, vacuuming was initiated. The vacuuming was continued until the pressure in its processing chamber reached 0.0005 Pa or lower. It was also confirmed that the out-gas rate in the processing chamber was 0.08 Pa·L/sec or lower. Subsequently, while repeatedly performing seasoning by using fresh Si wafers (dummy wafers), the Vpp voltage of the RF bias power source 111 was measured. The Vpp was detected at the RF bias matching network 110.
A plasma was produced, for example, by controlling the plasma-producing power at 600 W and using a mixed gas of Cl2/HBr/O2 as a processing gas, and under the condition that a RF bias was applied at 50 W, processing was performed.
By setting the processing pressure of the three etching apparatus at 1.6 Pa, 1.2 Pa and 0.4 Pa, separately, processing was performed. Changes in Vpp were monitored at each of the processing pressures. The pressure of the processing chamber 101 in each etching apparatus was monitored via a pressure gauge 114 as a sensor.
As readily envisaged from the processing at the processing pressure of 1.6 Pa in
When the peak-peak-voltage Vpp of a RF bias to be fed to a lower electrode has increased to a steady value with the processing time for seasoning, the end of seasoning can be determined accordingly.
As readily envisaged from the processing at the processing pressure of 1.2 Pa or the processing at the processing pressure of 0.4 Pa illustrated in
In this example, the opening of a vacuum valve 115 (VV opening) is chosen as a parameter to be monitored. The vacuum valve 115 controls the degree of vacuuming by a turbomolecular pump 116 such that the pressure in the processing chamber is reduced to a preset pressure.
After applying wet cleaning to each of the three etching apparatus (apparatus 1, apparatus 2, apparatus 3) of the same specification, vacuuming was initiated, for example, at a constant flow rate by the turbomolecular pump 116, and the vacuuming was continued until the pressure in its processing chamber reached 0.0005 Pa or lower. A plasma was then produced, for example, by controlling the internal pressure of the processing chamber at 2.0 Pa, setting the plasma-producing power at 600 W and using a mixed gas of Cl2/HBr/O2 as a processing gas, and under the condition that a RF bias was applied at 50 W, processing was performed. Using Si wafers (dummy wafers) as substrates, the vacuum valve opening required to maintain the internal pressure of the processing chamber at 2.0 Pa in each seasoning was measured.
As readily envisaged from the processing at the processing pressure of 2.0 Pa in
As a reason for the above-described tendency, insufficient seasoning results in the production of out-gas from water still remaining in the processing chamber as a result of wet cleaning or deposits formed through reactions with atmospheric components, and the thus-produced out-gas slightly raises the internal pressure of the processing chamber. When the seasoning becomes sufficient, the remaining water or deposits are removed so that the resulting out-gas decreases. The rise in the internal pressure of the processing chamber is, therefore, inhibited to result in a smaller vacuum valve opening. As indicated by the processing at the processing pressure of 5.0 Pa in
When the opening of a vacuum valve has decreased to a steady value with the processing time for seasoning, the end of seasoning can be determined accordingly.
In this example, the time required until the internal pressure of the processing chamber reached a constant pressure when the residual gas in the processing chamber was discharged by the turbomolecular pump subsequent to seasoning making use of Si wafers (dummy wafers) was chosen as a parameter to be monitored, and the transition of changes in this time was monitored.
After applying wet cleaning to two etching apparatus (apparatus 1, apparatus 2) of the same specification, vacuuming was initiated to reduce the internal pressure of each processing chamber to 0.0005 Pa or lower. After the out-gas rate in the processing chamber at that time was confirmed to be 0.08 Pa·L/sec or lower, seasoning was performed using fresh Si wafers (dummy wafers).
A plasma was then produced, for example, by controlling the processing pressure at 0.4 Pa, setting the plasma-producing power at 600 W and using a mixed gas of Cl2/HBr/O2 as a processing gas, and under the condition that a RF bias was applied at 50 W, processing was performed.
After the seasoning making use of the Si wafers (dummy wafers) was performed, the residual gas in the processing chamber was discharged, for example, at a constant flow rate by the turbomolecular pump 116. The time required to reach a constant pressure (target pressure) at that time was measured.
In this example, the ultimate pressure was set at 0.005 Pa, and the time until that pressure was reached was measured. The internal pressure of the processing chamber was monitored by the pressure gauge 114.
When seasoning making use of Si wafers (dummy wafers) is repeated, the above-described vacuuming time is progressively shortened to become steady in a constant time. This indicates that the conditions within the processing chamber have become steady. When the transition of changes in vacuuming time subsequent to seasoning has been monitored and the time required until the internal pressure of the processing chamber is reduced to a target pressure has decreased to a steady value with the seasoning time, the end of the seasoning can be determined.
In this example, the matching electrostatic capacitance of each capacitor as a component in the matching network 105 (source RF bias matching network) was chosen as a parameter to be monitored, and the transition of changes in the capacitance was monitored.
Such a tendency may presumably take place as will be described hereinafter. Presumably, when seasoning is insufficient, in other words, shortly after wet cleaning, remaining water and deposits formed through reactions with atmospheric components become out-gas components, so that the atmosphere in the processing chamber becomes unstable and the matching capacitance of each capacitor does not remain steady. When seasoning is sufficient, on the other hand, it is presumed that the atmosphere in the processing chamber becomes stable by deposits formed in the course of the seasoning and the matching capacitance of each capacitor hence remains steady. When the transition of changes in the electrostatic capacitance of each capacitor as a component in the matching network 105 upon matching (matching capacitance) has been monitored and the matching capacitance has increased to a steady value with the seasoning time, the end of the seasoning can be determined accordingly.
In this example, the plasma ignition time, which is the time from the application of plasma-producing RF power to the production of a plasma, was chosen as a parameter to be monitored.
After applying wet cleaning to each of two etching apparatus (apparatus 1, apparatus 2) of the same specification, vacuuming was initiated. The internal pressure of each processing chamber was set at 0.0005 Pa or lower. Subsequently, Si wafers (dummy wafers) were transferred in turn into each processing chamber to repeat seasoning. During that seasoning, the plasma ignition time was monitored by the photodetector 113.
As seasoning conditions, the plasma-producing power at 400 W, a mixed gas of Ar/CF4/CHF3/O2 was used as a processing gas, and a RF bias to be applied to the lower electrode was set at 150 W.
When the transition of changes in the time until the ignition of a plasma from the feeding of plasma-producing RF power to a processing chamber have been monitored and the time has decreased to a steady value with the seasoning time, the end of the seasoning can be determined accordingly.
In this example, an electrostatic chuck voltage (ESC voltage) was chosen as a parameter to be monitored. It is to be noted that an electrostatic chuck voltage is a d.c. voltage generated by an electrostatic chuck power source 112 and is a voltage for electrostatically chuck the wafer 108 on the lower electrode 109. The electrostatic chuck voltage varies depending upon the internal resistance of the electrostatic chuck power source 112 even if the source voltage of the power source remains constant. In other words, the electrostatic chuck voltage varies depending upon fluctuations in an electrostatic chuck current.
After applying wet cleaning to each of three etching apparatus (apparatus 1, apparatus 2, apparatus 3) of the same specification, vacuuming was initiated by the turbomolecular pump 116, and the vacuuming was continued until the pressure in its processing chamber reached 0.0005 Pa or lower. A plasma was then produced, for example, by controlling the internal pressure of the processing chamber at 2.0 Pa, setting the plasma-producing power at 600 W and using a mixed gas of Cl2/HBr/O2 as a processing gas, and under the condition that a RF bias was applied at 50 W, processing was performed.
When with a seasoning time, scattering of an electrostatic chuck voltage to be fed to a lower electrode has been progressively reduced and the electrostatic chuck voltage has increased to a steady value, the end of the seasoning can be determined.
In this example, the flow rate of heat transfer gas to be fed to between a lower electrode and a wafer placed on the lower electrode was chosen as a parameter to be monitored. The flow rate of the heat transfer gas (He) is high shortly after wet cleaning, but becomes lower and steadier as the seasoning becomes sufficient.
After applying wet cleaning to each of three etching apparatus of the same specification, vacuuming was initiated by the turbomolecular pump 116, and the vacuuming was continued until the pressure in its processing chamber reached 0.0005 Pa or lower. A plasma was then produced, for example, by controlling the internal pressure of the processing chamber at 2.0 Pa, setting the plasma-producing power at 600 W and using a mixed gas of Cl2/HBr/O2 as a processing gas, and under the condition that a RF bias was applied at 50 W, processing was performed.
By repeating performing seasoning subsequent to the wet cleaning while using Si wafers (dummy wafers), residual water on a head part of the lower electrode was removed, and further, an outer peripheral part of a surface of the lower electrode was covered with reaction products to restore the wafer chuck force. This can be presumed to lower and stabilize the flow rate of the heat transfer gas. It is to be noted that the transition of changes in the pressure of the heat transfer gas (the pressure on the side of the back of the wafer) can also be used in place of the transition of changes in the flow rate of the heat transfer gas.
When the transition of changes in the flow rate of heat transfer gas to be fed to between a lower electrode and a substrate placed on the lower electrode has been monitored and the flow rate has decreased to a steady value with the seasoning time, the end of the seasoning can be determined.
As has been described above, the time points that the end of seasoning was determined in the respective examples, specifically the time points that:
The end of seasoning can, therefore, be determined by indirectly measuring at least an increase in the internal pressure of the processing chamber (an increase in the time required for vacuuming of the processing chamber) as a result of the occurrence of the above-described out-gas. In each of the above-described examples, the end of seasoning was determined by making use of the tendency that during seasoning, the corresponding specific physical value (parameter) changes with the seasoning time and then becomes steady.
Upon setting up criterions which can be relied upon to determine that the above-mentioned physical values have become steady, the present inventors conducted numerous experiments, and recorded how the physical values changed.
Based on the results of the experiments, it has been found that one or more of the parameters each becomes steady at a constant value whenever seasoning is continued subsequent to wet cleaning and that the remaining one or more parameters each becomes steady at a different values upon each wet cleaning.
Where the value of a parameter is not affected by the wearout or like rate or rates of its relevant component or components, for example, the value at which the parameter becomes steady does not change upon each wet cleaning (for example, Examples 2, 3, 4, 5, 6, 7 and 8).
Where the value of a parameter is affected by the wearout or like rate or rates of its relevant component or components, on the other hand, the value at which the parameter becomes steady differs upon each wet cleaning (for example, Example 1).
In the former case (where the value of a parameter is reproduced every time), the end can be determined when the value of the parameter has exceeded (or fallen below) a given threshold. This determination method will hereinafter be called “the level determination method”.
In the latter case (where the value of a parameter differs every time), the end cannot be determined if the value of the parameter at a particular time point is solely used. It is, therefore, necessary to determine the end by checking the rates of changes in the parameter in a plurality of seasoning discharges to be performed. Such a determination has been practiced for years in the determination of the end of etching that makes use of a plasma emission. According to this method, a first or second derivative of the parameter is calculated, and a determination is made based on the rate of its change. This method will hereinafter be called “the differential determination method” irrespective of whether the first derivative or the second derivative is relied upon.
In the case of
In the case of
In this example, the parameters monitored in Examples 1-8 were all monitored as parameters. Described specifically, the parameters monitored in the respective examples can be monitored at the same time during seasoning that makes use of dummy wafers. It is, therefore, possible to determine the end of seasoning by any one of the methods described in Examples 1-8. It is also possible to make a comprehensive determination by using all the parameters. Examples of such a comprehensive determination (Comprehensive Determination Examples 1, 2 and 3 will hereinafter be described with reference to Tables 1, 2 and 3.
Referring to Table 1, scores are firstly allotted to the respective monitored parameters. With respect to each of the monitored parameters, a judgment is next made as to whether the end of seasoning was successfully determined. Only the scores allotted to the parameters each of which allowed to determine the end are cumulatively added.
Described specifically, a score of 10 is allotted to each of the parameters, and the scores allotted to the parameters (Nos. 1, 2, 6, 7 and 8) each of which allowed to determine the end are cumulatively added. As five parameters allowed to determine the end in this example, the cumulative sum is 50. The total score is 80 at the maximum because there are eight parameters to be monitored. When a score of 60 has been achieved out of 80, for example, the seasoning can be determined to be ended at this time point as a comprehensive determination.
Referring to Table 2, weighting factors are allotted to the respective parameters to be monitored. As in the example of Table 1, a score of 10 is allotted to each parameter. With respect to each of the monitored parameters, a judgment is next made as to whether the end of seasoning was successfully determined. Only with the scores allotted to the parameters each of which allowed to determine the end, the weighting factors were multiplied, respectively, followed by a cumulative addition.
Described specifically, a score of 10 is allotted to each of the parameters, and further, the weighting factors shown in Table 2 are allotted. With the scores allotted to the parameters (Nos. 1, 2, 6, 7 and 8) each of which allowed to determine the end, the weighting factors were multiplied, followed by a cumulative addition. As five parameters allowed to determine the end in this example, the cumulative sum is 80. The total score is 115 at the maximum because there are eight parameters to be monitored. When a score of 90 has been achieved out of 115, for example, the seasoning can be determined to be ended at this time point as a comprehensive determination. As the total score may become 115 at the maximum in this example, it is preferred to conduct normalization as needed.
Reference is made to Table 3. According to the above-described level determination method, it is possible to obtain a value at a part-way progress, which indicates what percentage of a target value the current value amounts to.
In this example, to each of those of monitored parameters that permit measuring their levels, a degree of progress (0 to 100%) is allotted depending upon the measured level.
Firstly, weighting factors are allotted to the respective parameters to be monitored. As in the example of Table 1, a score of 10 is allotted to each parameter. With respect to each of the monitored parameters, a judgment is next made as to whether the end of seasoning was successfully determined. Only with the scores allotted to the parameters each of which allowed to determine the end, the weighting factors and the degrees of progress were multiplied, respectively, followed by a cumulative addition. Specifically, a calculation is made in accordance with the following formula: score=10 (score)×(the degree of progress÷100)×the weighting factor. It is to be noted that the designation of “−” as the degree of progress indicates the use of the differential determination method.
The inclusion of weighting factors and progress degrees as described above makes it possible to perform a finer comprehensive determination, can also avoid an unnecessary seasoning time, and can also save dummy wafers to be used in seasoning.
As has been described above, the present invention has a system for monitoring the transition of changes in an etching parameter with time, and performs a comparison between a monitored value of the etching parameter and a preset criterion value and also a computation to control a system that controls an etching apparatus. It is, therefore, possible to determine with high accuracy the end of seasoning after wet cleaning and then, to end the seasoning. As a consequence, the seasoning time can be limited to the processing of as few dummy wafers as needed at minimum, thereby making it possible to shorten the non-operation time of a semiconductor fabrication apparatus and the fabrication cost of semiconductor devices.
This application claims the priority of Japanese Patent Application 2006-152305 filed May 31, 2006, which is incorporated herein by reference.
| Number | Date | Country | Kind |
|---|---|---|---|
| 2006-152305 | May 2006 | JP | national |
