IES84859Y1 - Method and apparatus for measuring the wafer etch rate and etch depth in a plasma etch process. - Google Patents

Abstract

ABSTRACT The present invention relates to plasma etch processes. More particularly, the invention relates to a method and an apparatus for determining the process etch rate and wafer etch depth in a plasma etching process on a semiconductor wafer of a particular wafer batch

Description

Title Method and apparatus for measuring the wafer etch rate and etch depth in a plasma etch process.

Field of Invention The present invention relates to plasma etch processes. More particularly, the invention relates to a method and an apparatus for determining the process etch rate and wafer etch depth in a plasma etching process on a semiconductor wafer of a particular wafer batch.

Background of thgggntion One of the main processes involved in semiconductor manufacturing is the etching of the semiconductor. A typical etch process requires plasma discharge to remove a patterned layer of exposed material on the semiconductor wafer surface. The water may comprise of one or more layers. Where patterned trenches are etched on the Silicon wafer. the process is known as Deep Reactive Ion Etching (IDRIE) or Shallow Trench lsolation (STI).

There are a number of etching processes which are in use by the semiconductor industry. Two commonly used etching tools or reactors for the etching process are the Capacitive Coupled Plasma (CCP) tool, and the Transformer Coupled Plasma (TCP) tool.

The principles of the etching process may be explained with reference to Figures 1 to 3. Figure I shows a cross sectional view of a typical CCP processing tool. A vacuum chamber 10 incorporates a bottom electrode 2, on which the water or substrate 3 is placed, and a top electrode 7. A gas inlet 8 and an exhaust line 9 are also provided.

The chamber also includes a bottom electrode radio frequency (RF) power supply 1.

Figure 2 shows a cross sectional view of a typical TCP processing tool. This processing tool incorporates substantially the same components as the CCP processing tool. but does not include a top electrode. 11 also includes a second RF power supply 12, an antenna 13 and a dielectric window 6. It is customary to place a matching network (not shown) between the RF power supplies 1 and 12 and the powered electrode/antenna. The purpose of the network is to match the power supply impedance, which is typically 500, to the electrodes/antenna impedance.

Typical operation of such tools is explained with reference to Figure 3. in relation to a CCI’ tool. It involves placing a wafer or substrate 3 on the bottom electrode 2. and igniting the plasma by the radio frequency power supply 1 applying a constant amount of energy to the electrode 2 and/or antenna. A constant gas flow of a selection of feedstock gases 1 1 is also provided, which is pumped at a constant throughput into the chamber.

The etch process results in the removal of material from the wafer 3 by sputtering, chemical etch or reactive ion etch. The removed material is then volatised into the plasma discharge 5. These volatile materials are called etch-by-products 4. and, together with the feedstock gases 11, contribute to the chemistry of the plasma discharge 5. The etch—by-products 4 and the gases 11 are pumped away through the exhaust or pumping port 9. The etch process for a TCP tool operates in a similar fashion.

It will be appreciated that it would be highly desirable to be able to measure the plasma etch or material removal rate, so that the etch feature depth can be determined.

This is due to the fact that the depth of the etched patterns is critical for the performance of the electronic devices being constructed from the wafer.

A number of techniques are currently in use to detect the etch rate or etch depth. One such technique described in US Patent number 4367044 is based on refraction. Other techniques involve the use of diffraction (US Patent number 5337144), rellectometry (US Patent number 6939811), and optical emission spectroscopy (OES) (US Patent number 4430 I 5 I ).

Many of these techniques require complicated set ups to be put in place, such as for example the provision of light sources, optical alignment detectors and space about the plasma etching tool. This of course has the undesirable drawback of adding to the cost of the semiconductor manufacture. Furthermore, the techniques are often based on measurements of certain regions of the wafer, which, in some cases, do not account for the centre to edge variation of the etch depth. Finally, some of these techniques depend on the thickness of the mask which is simultaneously etched. It will be appreciated that these techniques have adverse affects on the accuracy of the depth measurements which are problematic in the semiconductor industry.

Summary of the invention The present invention, as set out in the appended claims, provides method for detecting the etch rate ofa plasma etch process being performed on a semiconductor wafer, the method comprising the steps of: detecting light being generated from the plasma during the etch process; filtering the detected light to extract modulated light; and processing the detected modulated light to determine the etch rate ofthe etch process.

By detecting the modulated light being emitted from the plasma, a very accurate assessment of etch rate and etch depth of the etch process can be obtained.

The detecting may further comprises the step of filtering the light to detect selected wavelength bands.

The processing may comprise the steps of: converting the detected light into a digital signal; transfonning the digital signal into a frequency domain signal; extracting one or more pre-selected frequencies from the frequency domain signal for use as process monitor signals; generating a plot proportional to the intensity of the process monitor signals over the elapsed time of the etch process, and determining the etch rate from the plot.

The step ofgenerating a plot proportional to the intensity ofthe process monitor signals over the elapsed time ofthe etch process may comprise: calibrating the values ofthe process monitor signals so as to generate converted signal values; and generating a plot ofthe converted signal values over the elapsed time of the etch process.

Preferably, the step of calibrating comprises the multiplication ofa conversion constant to the values ofthe process monitor signals.

The method may further comprise the step of integrating the plot so as to generate a second plot of etch area over elapsed time of the etch process, and determining the etch depth from the second plot.

The method may further comprise the step of generating an indicator when a signal level transition in the second plot matches a stored value representing a target etch depth.

Suitably, the indicator is a visual or an aural indicator that the target etch depth has been reached.

Preferably, the transforming of the digital signal comprises performing a fast fourier transform on the digital signal.

Preferably, the process monitor signals are determined during a test wafer analysis of wafers of the same batch as the wafer.

Preferably, the conversion constant may be determined during a test water analysis of waters of the same batch as the wafer.

The test water analysis ofthe batch may comprise the steps of: detecting modulated light being generated from the plasma ofa test wafer being etched over the duration of an etch process; convening the detected modulated light into digital signals; transforming the digital signals into frequency domain signals; determining the main frequencies of the frequency domain signals; and selecting those main frequencies which are sensitive to changes in the etch rate as the process monitor signals.

The step of selecting those main frequencies which are sensitive to changes in the etch rate as the process monitor signals may comprise the step of: generating electron microscopy images of a set of test wafers over the etching process. measuring the etch rate and etch depth of the etch process as a function oftime from the generated images; and selecting those main frequencies which have values over time which correlate to the measured etch rate and etch depth as the process monitor signals.

Suitably, the method further comprises the step of establishing the linear relationship between the values of the selected process monitor signals over time and the actual etch rate.

Preferably, the established linear relationship is stored as the conversion constant.

The determining the main frequencies comprises the step of determining those frequency domain signals having the higher signal intensity values.

The present invention also comprises a method to determine the process monitor signals and conversion constant for use in a method of detecting the etch rate of a plasma etch process to be performed on a semiconductor wafer from a particular wafer batch, the method comprising the steps of: placing a test wafer ofthe wafer batch in a plasma etching tool and initiating the etch process; detecting modulated light being generated from the plasma of the test wafer over the duration of the etch process; convening the detected modulated light into digital signals: transforming the digital signals into frequency domain signals; determining the main frequencies of the frequency domain signals; selecting those main frequencies which are sensitive to changes in the etch rate as the process monitor signals; establishing the linear relationship between the values of the selected process monitor signals over time and the actual etch rate; and storing the established linear relationship as the conversion constant.

The step of selecting those main frequencies which are sensitive to changes in the etch rate as the process monitor signals may comprise the step of: generating electron microscopy images of the test wafer, measuring the etch rate and etch depth of the etch process as a function of time from the generated images; and selecting those main frequencies which have values over time which correlate to the measured etch rate and etch depth as the process monitor signals. '1" he determining the main frequencies may comprise the step of determining those frequency domain signals having the higher signal intensity values.

The present invention also provides an apparatus for detecting the etch rate ofa plasma etch process being performed on a semiconductor wafer, comprising: means for detecting light being generated from the plasma during the etch process; means for filtering the detected light to extract modulated light; and means for processing the detected modulated light to determine the etch rate of the etch process. ,) 3 The means for detecting may further comprise a means for filtering the light to detect selected wavelength bands.

The means for processing may comprise: a means for converting the detected light into a digital signal; a means for transforming the digital signal into a frequency domain signal: a means for extracting one or more pre-selected frequencies from the frequency domain signal for use as process monitor signals; a means for generating a plot proportional to the intensity ofthe process monitor signals over the elapsed time ofthe etch process; and a means for determining the etch rate from the plot.

The means for generating a plot proportional to the intensity of the process monitor signals over the elapsed time of the etch process may comprise: a means for calibrating the values of the process monitor signals so as to generate converted signal values; and a means for generating a plot of the converted signal values over the elapsed time of the etch process.

The means for calibrating may comprise a means for multiplication ofa conversion constant to the values of the process monitor signals.

The apparatus may further comprise a means of integrating the plot so as to generate a second plot of etch area over elapsed time of the etch process, and a means of determining the etch depth from the second plot.

Preferably, the apparatus further comprises a means of generating an indicator when a signal level transition in the second plot matches a stored value representing a target etch depth.

Preferably, the indicator is a visual or an aural indicator that the target etch depth has been reached. t\) tJ1 The means for detecting may be a photosensitive device.

The means for transforming may comprise a mierocontroller.

The means for transforming may comprise a Field Programmable Gate Array.

The means f or extracting one or more pre-selected frequencies from the frequency domain signal for use as process monitor signals and the means for generating a plot proportional to the intensity of the process monitor signals over the elapsed time of the etch process may comprise a computer.

The means of integrating the plot so as to generate a second plot of etch area over elapsed time of the etch process and the means of generating an indicator when a signal level transition in the second plot matches a stored value representing a target etch depth may comprise a computer.

The present invention also provides an apparatus for determining the process monitor signals and conversion constant for use in detecting the etch rate ofa plasma etch process to be performed on a semiconductor wafer from a particular wafer batch, comprising: a plasma etching tool; a means for detecting modulated light being generated from the plasma of the test wafer over the duration of the etch process; a means for converting the detected modulated light into digital signals: a means for transforming the digital signals into frequency domain signals; a means for determining the main frequencies of the frequency domain signals; a means for selecting those main frequencies which are sensitive to Changes in the etch rate as the process monitor signals; [Q ‘J1 a means for establishing the linear relationship between the values of the selected process monitor signals over time and the actual etch rate; and a means for storing the established linear relationship as the conversion constant.

The means for selecting those main frequencies which are sensitive to changes in the etch rate as the process monitor signals comprises: a means for generating electron microscopy images ofthe test wafer, a means for measuring the etch rate and etch depth of the etch process as a function of time from the generated images; and a means for selecting those main frequencies which have values over time which correlate to the measured etch rate and etch depth as the process monitor signals.

There is also provided a computer program comprising program instructions for causing a computer program to can’y out the above method which may be embodied on a record medium, carrier signal or read-only memory.

The present invention also provides a method for detecting the etch rate ofa plasma etch process being performed on a semiconductor wafer, the etch process generating a plasma sheath proximate the wafer, the method comprising the step of determining the etch rate using substantially only light emitted from the plasma sheath.

The detected light may include both modulated and non-modulated light.

Preferably, the light emitted from the plasma sheath and the remainder of the plasma are detected together, but the etch rate is determined using substantially only light emitted from the plasma sheath. in a further aspect of the invention, a method is provided for monitoring a process parameter of a plasma etch process being performed on a semiconductor water, the method comprising the steps of : detecting light being generated from the plasma during the etch process; filtering the detected light to extract modulated light; and processing the detected modulated light to determine a process parameter the etch process.

Brief Description of the Drawings The invention will be more clearly understood from the following description of an embodiment thereof, given by way of example only, with reference to the accompanying drawings, in which:- Fgure l is a cross sectional view of typical CCP processing tool; Figure 2 is a cross sectional view of a typical TCP processing tool; Figure 4 is a diagram of one embodiment ofthe components involved in the implementation ofthe present invention; Figure 5 details the process flow of one embodiment of the present invention; Figure 6 details further steps of the process flow of Figure 5', Figure 7a details an exemplary etch rate plot of the present invention; Figure 7b details an exemplary etch depth plot ofthe present invention: Figure 8 details the process flow ofthe first steps in determining the optimum process monitor signals for a particular wafer batch; Figure 9 shows an example voltage waveform generated from the detection of modulated light; and Figure 10 shows the FFT waveform generated from applying the FFT to the waveform ofFigure 9.

Detailed Description of the Invention The present invention provides a method for monitoring a plasma reactor during the etch process and detecting the etch rate and etch depth of the wafer being etched. ln order to understand the principles behind the present invention, the chemical reactions which occur during the etch process should be appreciated. During the etching ot‘ a wafer, modulated light of a certain amplitude is emitted by the plasma.

The amplitude of the modulated light is related to the etch rate. Accordingly, the amplitude may be seen to vary, such as for example when one layer of the water has been removed and the etch process begins on a layer underneath, made of a different material.

One of the main sources for excitation of atoms or molecules in the discharge is electron impact excitation. These excitations are directly proportional to the electron density. The excitation of atoms and molecules is time uniform in the plasma bulk. where the electron density is time uniform. On the other hand, the electron density in the plasma sheaths, ie. the region between the plasma and the electrode/wafer, as indicated by 4 in Figures 1 to 3, is highly modulated at the driving radio frequency of the etch tool.

The excited species emit light via spontaneous emission with a characteristic decay rate. The excited species can also emit radiation through stimulated emission from the radio frequency cycle. In general, the plasma emission is directly proportional to the number density of species in an excited state. lfthe density of the species in excited states is modulated, it is expected that the light emission will be modulated in a similar fashion. This gives rise to a non—modulated or DC emission component, together with an additional component, which is modulated at the driving radio-frequency. The modulated light is that light which exhibits a periodic temporal intensity variation at a particular frequency.

Etch by—products resident near the wafer surface are more likely to be excited by the electrons, as the local by-product density is higher in the plasma sheath region. Since the electrons are strongly modulated in the plasma sheath regions, the light from these regions will be highly modulated and the modulation will be correlated with the driving radio frequency.

Due to the fact that the modulated light emission corresponds to light emitted significantly by excited etch-by products at “the sheath” region above the wafer or substrate, it will be appreciated that any variation in the speed at which material is being removed from the surface of the wafer (which corresponds to a change in the etch rate) will be also seen as a change in the modulated light emissions. Therefore. the niodulated light is ideal for use in etch rate and depth monitoring. in a single frequency etching tool, it is expected that the modulated light will correspond to the driving radio frequency and harmonics. But in dual frequency systems, it is probable to find light modulated at the mixed up products of the two driving frequencies, as well as at the radio frequencies themselves and their harmonics.

The optical sensor of the present invention detects this plasma light modulation. The detected plasma light modulation is then used in order to determine the etch rate and etch depth. As the modulated light is substantially in the plasma sheath, the invention therefore involves determining the etch rate and etch depth by using substantially only light emitted from the plasma sheath.

Figure 4 shows a diagram of one embodiment of the components involved in the implementation of the present invention. A plurality of sensors 14 provide for the detection of plasma light from the plasma 15 located in the etching tool (etching tool not shown). The sensors 14 can take the form of photo—diodes or photo multiplier tubes. In order to successfully detect the plasma light modulation, the sensors should have fast response times. A plurality of optical filters 16 may be used in conjunction with the sensors 14, each filter adapted to detect a particular optical wavelength band, the filters located between the sensors and the plasma. The optical filters have the effect of narrowing the input light to the sensor to bands a few nanometres wide centred at specific wavelengths, so as to select light from certain species in the plasma, such as for example reactants or etch-by-products. This has the effect of removing unwanted wavelength bands. The filters therefore allow the real time monitoring of specific optical lines, enabling the classification of plasma chemistry at the sheath.

A signal conditioning block 17 receives the output data from the sensors 14. At the signal conditioning block 17, the detected light signals from the sensors 14 are conditioned and digitised. In one embodiment of the invention, the conditioning is carried out by a transimpedance amplifier and a programmable voltage amplifier. The transimpedance amplifier converts the signals from the sensors to voltage signals. while the voltage amplifier amplifies these voltage signals. The amplified voltage signals are digitised by an analog to digital converter (ADC). ln a preferred embodiment of the invention, the ADC operates at frequencies up to 70 Mllz. A processor 18 provides for the processing of the digital signals into the format required in order to enable the etch rate and depth to be estimated by the computer (PC) 19.

The processor may be any suitable processing device, such as a micro—controller or a Field Programmable Gate Array (TFPGA). The computer 19 provides for the further processing of the processor output signal to determine the etch rate and depth of the etching process, and to generate an indicator when a preset etch depth is reached.

Figure 5 details the process flow of one embodiment of the present invention. ln step 1. light is generated from the plasma of a wafer of a particular batch which is to be etched in an etching tool. The optical sensors continuously detect the modulated light emitted from the plasma sheath and the non~modulated light from the remainder of the plasma (step 2). The light may be additionally filtered to only detect light of particular optical wavelength bands. In step 3, the detected plasma light modulation signals are processed in real time using an etch rate and depth algorithm. This algorithm determines the etch rate and when a desired etch depth has been reached. An indicator is then generated when the depth has been reached.

The process flow can be broken down into a number of further steps, which are described in more detail below in relation to Figure 6. The etch process is started in step 1. In step 2a, the modulated plasma light of different optical wavelength bands is detected by the optical sensors. The non-modulated light may also be detected. The light is converted to a voltage signal by the transimpedance amplifier, and then subsequently amplified by the voltage amplifier (step 2b). The amplified voltage signal is then digitised by the ADC to provide a digital signal (step 2c). A Fast Fourier transform filter in the processor transforms the digital signal into the frequency domain by calculating a FFT of the digital signal (step 2d).

Steps 2a to 2d are repeated approximately two thousand times, and the resulting set of Fl-"Ts averaged to generate a sample FF'l‘ (step 2e). It should be noted that the entire averaging process only takes about 250ms. This sample FFT is recorded by the computer (step 3). ln step 4, the data values of the one or more frequencies of the sample FFT which have been pre—selected to act as process monitor signals are extracted. These process monitor signals have been selected to be those signals which will provide the most accurate assessment of the etch rate and depth of the etching process. The selection of the process monitor signals is carried out during test wafer analysis, details of which will be described later. It is therefore through the monitoring of the data values of these process monitor signals that the etch rate may be evaluated, and by which a determination may be made as to whether the required etch depth has been reached.

It will be appreciated that the above described steps have provided for the filtering of the detected light to extract modulated light from the plasma light, which could have included both modulated and non—modulated light, and the subsequent monitoring of pre-selected modulated light signals in order to determine the etch rate and depth. The process then moves to step 5.

In step 5, a plot proportional to the intensity of the process monitor signals over the elapsed time of the etch process is generated using the steps outlined below, lfa single frequency has been selected as a process monitor signal, the data values for that frequency which have been extracted from sample FFT values which have already been generated over the elapsed time of the etch process are first calibrated. this calibration involves the multiplication of a conversion constant to each data value, in order to generate a converted signal value, which, when plotted over the time of the etch process, provides the actual etch rate of the etching process. The conversion constant represents the relationship between the process monitor signal and the actual etch rate.

The correlation between the values of the process monitor signal and the actual etch rate is established during test wafer analysis which has previously been carried out. and the conversion constant is then stored in the computer. This process is described later.

Once the conversion is performed, a plot of the converted process monitor signal versus time is generated in real time, as shown in Figure 7a. This plot corresponds to the etch rate of the etching process. Therefore the etch rate of the etch process can be determined from this plot.

Where there is more than one frequency selected as a process monitor signal, the time evolution proportional to the intensity of the various frequency components may be combined as a single plot, using multivariate analysis (MVA) techniques.

It should be noted that the process monitor signals will remain constant where the plasma is removing the wafer material continuously during the etch process, and at a constant rate. it will be appreciated that when the process monitor signals remain constant, there will be a linear relationship between the area and time.

The area underneath the plot of Figure 7a is directly proportional to the etch depth.

‘Therefore, in order to determine the etch depth, an evaluation of the area underneath the plot is required to be performed. In step 6, a numerical integration of the etch rate signal is carried out in order to calculate the current etch depth. Figure 7b shows a to (J1 graphical representation of the etch depth calculation. Therefore the etch depth can be determined from the plot of Figure 7b. 'l‘he plot of Figure 7b is then analysed to determine whether the target etch depth has been reached for the etch process. In one embodiment of the invention, this is achieved by determining whether a signal level transition on the etch depth plot matches a stored signal level value which represents the target etch depth. The target etch depth is a requirement of the process for the particular semiconductor device in production, and is typically specified by the original designer of the process.

If the signal level transition matches the target value for the etch depth, the process moves to step 7. If a match is not found, the process flow returns to step 2, provided that the etch process has not already been completed.

In step 7, an indicator is generated by the computer that the target etch depth in the etch process has been reached. In one embodiment of the invention, the indicator generated by the computer is a visual or aural indicator. In another embodiment of the invention, the indicator is a control signal for the etching tool to stop the etch process.

It will be understood that the processor could perform a number of alternative tasks once the required etch depth has been reached, depending on a users requirements for the etch process.

Other numerical techniques could equally well be used instead of Fourier analysis to determine the etch rate/depth.

In order to be able to accurately detect the etch rate and determine the etch depth for a particular wafer, it is necessary to first select the most suitable process monitor signals for monitoring the etch rate and depth. In the case of the present invention, this involves determining which ofthe frequencies ofthe modulated light are most suitable to act as monitor signals for the etch rate. In reality, each wafer batch has its own unique characteristics. Accordingly, prior to being able to determine the etch rate and l\) U! depth of the etch process for Waters of a particular wafer batch, it is necessary to carry out advance preparation, by performing an analysis of each individual wafer batch. to select the most appropriate frequencies which should be monitored in order to enable the etch rate to be determined for waters from that particular batch. This is carried out through tcst wafer analysis of the batch. Fttrtlierinore, where there is more than one layer, the values ofthe process monitor signals for each layer may not necessarily be the same, as every layer produces different etch by products, which affect the discharge in different ways. Accordingly, the test wafer analysis needs to be carried out for each wafer layer.

The process of selecting the optimum process monitor signals is described below using an implementation perfomied through Fourier analysis. However, as previously advised, it should be appreciated that a number of other numerical techniques could equally well be used instead of Fourier analysis.

The first few steps to determine the optimum process monitor signals are identical to those performed during the etch rate and depth monitor technique described above.

However, for ease ofunderstanding, they are briefly described below again.

Figure 8 details the process flow of determining the optimum process monitor signals for a particular wafer batch. In step 1, a test wafer of the batch is placed in the etching tool and the etching process begun. In step 2a, light from the plasma is detected by the sensors, and the light signal is converted to a voltage signal. This light may include both modulated and non-modulated components. The voltage signal is then amplilied (step 2b). ln step 20, the voltage signal is digitised and input to the processor. The processor transforms the digitised voltage signal into the frequency domain using the Fast Fourier Transform to provide a FFT (step 2d).

Steps 2a to 2d are repeated approximately two thousand times, and the resulting set of FFT averaged to generate a sample FFT (step 2e), which is recorded by the computer (step 2t). It should be noted that the entire averaging process only takes about 250ms.

Steps 2a to 2f are repeated over time until the etch process is complete. At this stage. the processor will have recorded a set of sample FFT covering the duration of the entire etch process of the test wafer. Once the process is complete. the generated sample FFT waveform is ready to be examined to determine the optimum frequencies for use as process monitor signals for monitoring the etch rate and depth for that particular wafer batch.

The first step in the selection of the optimum frequencies ofmodulatcd light for use as process monitor signals in respect of all of the wafers of the batch involves the determination ofthe main frequency components ofthe sampled FFT.

Figures 9 and 10 describe how the main frequency components can be determined.

Figure 9 shows an example voltage waveform generated from the detection of modulated light. It will be appreciated that this waveform contains more than one frequency plus noise. Figure 10 shows the FFT waveform generated from applying the FFT to this voltage waveform. This is a plot of intensity versus frequency. In this example it can be clearly seen that there are four peaks, each below 100 MHz. These peaks indicate the frequency signals that are contained in the waveforin. with the height of the peaks indicating the relative intensity oftheir corresponding frequencies in the waveform. It will be appreciated therefore that the main frequency components correspond to the peaks in the sampled FFT wavefonn i.e. those frequency domain signals having higher signal intensity values.

Once the main frequency components are established, those frequencies from the main frequency components which have a time signal which satisfies two conditions must be found. The first condition is that the time signal is steady. The first condition is based on the knowledge that the etch rate should be constant. The second condition is that the time signal is sensitive to small etch rate changes. The second condition is imposed to ensure that the one or more process monitor signals are truly correlated to the etch rate.

In general. it can be assumed that the etch rate through each individual layer (in the case where there is more than one layer present) is approximately constant. While etching a layer, minor variations in the etch rate may occur. as the etch rate is not perfectly constant throughout the process. Small changes in the etch rate may also be caused by small drifts in the etching process. However, large variations in the etch rate are more likely associated with etching layer transitions (endpoint) or variations in the process control parameters; such as for example changes in power, pressure. gas flow or mixture.

The second condition is tested by analysing test wafer images in conjunction with the values obtained for the main frequency components, and determining which of the main frequencies over the time of the etch process exhibit values which most closely correlate to the actual etch rate determined from the test wafer images. as explained below.

The test wafer images may be obtained using any of the techniques known in the art.

One such technique involves placing a first test wafer in the etching tool and running the etch process until a predetermined time period has elapsed. The test wafer is then removed from the etching tool and the state of its surface examined by slicing the wafer. A second test wafer is then placed in the etch tool, and the etch process run until a second predetermined time period has elapsed, with the second time period being greater than the first time period (which is typically a few seconds more than the first time period). The second test wafer is then removed and its surface examined.

This process is repeated on further test wafers from a set of test wafers from the batch, each wafer from the set being of the same quality and possessing the same characteristics, until the predetermined time period exceeds the time taken for the etch depth to be reached for that particular wafer batch. This process can be repeated for several batches of wafers of same quality and characteristics, with the testing operation run on every batch with small changes in the tool operating parameters.

Once all of the test wafers from the set have been placed in the etching tool, Scanning Electron Microscopy (SEM) images for every single wafer are generated. Other (7) imaging techniques could also be used. such as for example an Atomic Force Microscopy (AFM) technique, The images reveal the time evolution of the process. lt will be appreciated that although technically it is not the time evolution of the process of a single water, it is accepted that the results should reflect the time evolution of a single wafer. given that the set of waters have all been prepared in a similar fashion prior to the processing. From the SEM images. it is possible to measure the etch rate and depth as a function oftime.

These test wafer images pennit the calculation of the etch rate and depth as a function oftime. The time signals for the main frequencies detected by the optical sensor that have values which best correlate to the test wafer results for etch rate and depth are then selected for use as the process monitor signals.

When a single frequency signal is selected as a process monitor signal. the process is monitored based on this single signal. Alternatively, if more than one frequency is selected as process monitor signals, then the signals can be combined using Multi- Variate Analysis techniques (MVA) to output a single combined time process signal to be used to determine the etch rate and depth. A typical MVA technique that may be used here is Principal Component Analysis (PCA). As in the ease of a single process monitor signal, the etch rate and depth value is based on the combined time process signal.

The next step in the test water analysis involves calibrating the frequencies selected to act as process monitor signals to the etch rate. This calibration consists of determining a value for a conversion constant between the actual etch rate (estimated from the water analysis) and the frequencies selected to act as process monitor signals over the course of the etch process. This involves establishing the linear relationship between the values of the selected frequency or MV A signal, in the case of more than one useful frequency, over time and the actual etch rate. This is calculated by dividing the measured etch rate (after water analysis) by the signal value of the selected frequencies. This constant therefore converts the signal value (in arbitrary units) to the actual etch rate (typically micron/min). Once the relationship is determined, this conversion constant is recorded. This constant is required, as previously explained, in order to convert the values which will be obtained from the process monitor signals over time when the technique ofthe present invention is being carried out to represent the actual etch rate. It should be noted that this constant is particular to a given water batch process, and will not convert correctly the signal to the etch rate it‘ the quality or characteristics ofthe water or the process parameters are varied. in the final step in the test wafer analysis process, the computer is programmed to monitor the selected one or more frequencies determined during the test water analysis as process monitor signals for determining the etch rate and etch depth. The computer is also programmed with the recorded conversion constant. Finally, the computer is also programmed with a target etch depth value. This value is that value desired for the depth of the etch on the wafer layer, and is set by the process designer in view of the semiconductor device which is being manufactured on a particular water.

As previously noted, where the etching process is to be carried out on more than one layer. the values obtained for the process monitor signals for each layer may not necessarily be the same. Accordingly, the test wafer analysis process should be repeated for each layer individually.

Once the above described preparation has been completed, the etch rate and depth in the etch process for any layer of a wafer from the analysed batch can be monitored.

This is achieved by placing any of the wafers from the batch into the etching tool, and following the steps of the invention as explained previously with reference to Figure It will be appreciated that the method and apparatus of the present invention can be used in Capacitive Coupled Plasma (CCP) tools, Transformer Coupled Plasma tTCP) tools and any other variation of these.

The embodiments in the invention described with reference to the drawings comprise a computer apparatus and/or processes performed in a computer apparatus. llowever, the invention also extends to computer programs, particularly computer programs stored on or in a carrier adapted to bring the invention into practice. The program m ay be in the form of source code, object code, or a code intermediate source and object code, such as in partially compiled form or in any other form suitable for use in the implementation of the method according to the invention. The carrier may comprise :1 storage medium such as ROM, e.g. CD ROM, or magnetic recording medium, eg a floppy disk or hard disk. The carrier may be an electrical or optical signal which may be transmitted via an electrical or an optical cable or by radio or other means.

The invention is not limited to the embodiments hereinbefore described but may be varied in both construction and detail. The words “comprises/comprising" and the words “having/including” when used herein with reference to the present invention are used to specify the presence of stated features, integers, steps or components but does not preclude the presence or addition of one or more other features, integers, steps. components or groups thereof.

It is appreciated that certain features ofthe invention, which are, for clarity, described in the context of separate embodiments, may also be provided in combination in a single embodiment. Conversely, various features of the invention which are, for brevity, described in the context ofa single embodiment, may also be provided separately or in any suitable sub-combination.

Claims (5)

Claims

1. A method for detecting the etch rate of a plasma etch process being performed on a semiconductor wafer, the method comprising the steps of : detecting light being generated from the plasma during the etch process; filtering the detected light to extract modulated light; and processing the detected modulated light to determine the etch rate of the etch process.

2. The method of Claim 1, wherein the detecting further comprises the step of filtering the light to detect selected wavelength bands.

3. The method of Claim 1 or Claim 2, wherein the processing comprises the steps of: convening the detected light into a digital signal; transforming the digital signal into a frequency domain signal; extracting one or more pre-selected frequencies from the frequency domain signal for use as process monitor signals; generating a plot proportional to the intensity of the process monitor signals over the elapsed time of the etch process; and determining the etch rate from the plot.

4. The method of Claim 3, wherein the step of generating a plot proportional to the intensity of the process monitor signals over the elapsed time of the etch process comprises: calibrating the values of the process monitor signals so as to generate converted signal values; and generating a plot of the converted signal values over the elapsed time of the etch process.

5. A method substantially as described herein, with reference to, and as illustrated in,