IE84329B1 - A method for transferring process control models between plasma processing chambers - Google Patents
A method for transferring process control models between plasma processing chambers Download PDFInfo
- Publication number
- IE84329B1 IE84329B1 IE2004/0568A IE20040568A IE84329B1 IE 84329 B1 IE84329 B1 IE 84329B1 IE 2004/0568 A IE2004/0568 A IE 2004/0568A IE 20040568 A IE20040568 A IE 20040568A IE 84329 B1 IE84329 B1 IE 84329B1
- Authority
- IE
- Ireland
- Prior art keywords
- chamber
- fault
- process control
- sensor
- fingerprint
- Prior art date
Links
- 210000002381 Plasma Anatomy 0.000 title claims description 25
- 238000004886 process control Methods 0.000 title claims description 20
- 239000011159 matrix material Substances 0.000 claims description 15
- 230000001131 transforming Effects 0.000 claims description 8
- 238000002474 experimental method Methods 0.000 claims description 5
- 238000000034 method Methods 0.000 description 43
- 235000012431 wafers Nutrition 0.000 description 29
- 230000004044 response Effects 0.000 description 17
- 238000001514 detection method Methods 0.000 description 11
- 238000004519 manufacturing process Methods 0.000 description 10
- 239000007789 gas Substances 0.000 description 7
- 230000035945 sensitivity Effects 0.000 description 7
- XUIMIQQOPSSXEZ-UHFFFAOYSA-N silicon Chemical compound [Si] XUIMIQQOPSSXEZ-UHFFFAOYSA-N 0.000 description 4
- 229910052710 silicon Inorganic materials 0.000 description 4
- 239000010703 silicon Substances 0.000 description 4
- 230000002596 correlated Effects 0.000 description 3
- 230000000875 corresponding Effects 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 230000004048 modification Effects 0.000 description 3
- 238000006011 modification reaction Methods 0.000 description 3
- 230000003287 optical Effects 0.000 description 3
- 238000005070 sampling Methods 0.000 description 3
- 239000004065 semiconductor Substances 0.000 description 3
- 239000000758 substrate Substances 0.000 description 3
- 230000035693 Fab Effects 0.000 description 2
- 230000002159 abnormal effect Effects 0.000 description 2
- PIGFYZPCRLYGLF-UHFFFAOYSA-N aluminum nitride Chemical group [Al]#N PIGFYZPCRLYGLF-UHFFFAOYSA-N 0.000 description 2
- 230000003292 diminished Effects 0.000 description 2
- 238000005530 etching Methods 0.000 description 2
- 150000002500 ions Chemical class 0.000 description 2
- 239000002184 metal Substances 0.000 description 2
- 230000003449 preventive Effects 0.000 description 2
- 235000008536 Capsicum baccatum var pendulum Nutrition 0.000 description 1
- 240000000866 Capsicum baccatum var. pendulum Species 0.000 description 1
- 235000018306 Capsicum chinense Nutrition 0.000 description 1
- 230000036809 Fabs Effects 0.000 description 1
- 238000004164 analytical calibration Methods 0.000 description 1
- 230000002238 attenuated Effects 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 230000001808 coupling Effects 0.000 description 1
- 238000010168 coupling process Methods 0.000 description 1
- 238000005859 coupling reaction Methods 0.000 description 1
- 230000003247 decreasing Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 230000004907 flux Effects 0.000 description 1
- 238000004868 gas analysis Methods 0.000 description 1
- 238000010965 in-process control Methods 0.000 description 1
- 230000001939 inductive effect Effects 0.000 description 1
- 238000001459 lithography Methods 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 238000005086 pumping Methods 0.000 description 1
- 238000004611 spectroscopical analysis Methods 0.000 description 1
Description
PATENTS ACT, 1992
2004/0568
A METHOD FOR TRANSFERRING PROCESS CONTROL MODELS
BETWEEN PLASMA PROCESSING CHAMBERS
SCIENTIFIC SYSTEMS RESEARCH LIMITED
A method for transferring process control models
between plasma processing chambers
BACKGROUND OF THE INVENTION
FIELD OF THE INVENTION
The present invention relates to a method for transferring
process control models between plasma processing chambers.
PRIOR ART
The manufacture of integrated circuits is a detailed process
requiring many complex steps.
(or fab)
A typical semiconductor
manufacturing plant can require several hundred
highly complex plasma processing chambers to fabricate
intricate devices such as microprocessors or memory chips.
These fabs typically construct these devices on a substrate
of silicon, known as a silicon wafer.
A single wafer,
containing many such similar integrated circuits, often
requires over 200 individual steps to complete the
manufacturing process. These steps include lithographic
patterning of the silicon wafer to define each device,
etching lines to create structures and filling gaps with
metal or dielectric to create the electrical device of
interest. From start to finish the process can take weeks
to complete.
On each chamber the wafer is processed according to some
recipe, which is controlled by the tool operator. This
recipe includes input parameter settings such as process gas
flow rates,
chamber pressure, substrate/wall temperatures,
RF power settings on one or more power generators, recipe
time, inter-electrode spacing, etc. This is the case for
all plasma processing tools, such as etch, deposition, etc.
The wafer will undergo very many plasma process steps before
completion. Each step contributes to the overall product
yield; a fault at any one step may destroy potential
product.
Fig. 1 shows a typical plasma process reactor. It includes
a plasma chamber 1 containing a wafer or substrate 2 to be
processed. A plasma is established and maintained within
the chamber by an RF power source 3. This source generally
has real impedance which must undergo a transformation to
match that of the complex plasma load. This is done via
match network 4. Power is coupled to the plasma chamber,
typically by capacitive or inductive coupling, through an
electrode 8. Process gases are admitted through gas inlet 7
and the chamber is maintained at a desirable pressure by a
pump 11 pumping through gas exhaust line 10. A throttle
valve 9 may be used to control pressure. The plasma permits
effective manufacture of for example, semiconductor devices,
by changing gas chemistry. Gases such as C12, used to etch
silicon and metal, for example, are converted into reactive
and ionized species. Etching of the very fine geometry used
to fabricate semiconductor devices is made possible by the
reactive gases, ions and electrons of the plasma.
Referring again to Fig. 1, an RF sensor 5 is used to measure
the RF electrical power signal in the complex post—match
electrical line. A Fourier Transform is performed in data
collection electronics 6 using a sampling technique which
extracts the Fourier components of the voltage and current
and the phase angle between these vectors. This data
sampling should have sufficiently high resolution to
determine the Fourier components across a very large dynamic
range. Suitable techniques for high resolution sampling and
measurement of the Fourier components are described in US
Patent 5,808,415.
The output of the data collection electronics 6 is connected
to a controller 12 which may be a computer or other system
which uses the signals to yield information about and/or
control the plasma process.
The Fourier components are very sensitive to plasma events.
The wafer fabrication process involves running entire
batches of wafers with similar plasma process recipes to
ensure reliable volume production. If the plasma process on
each wafer is the same, then the measured Fourier components
will reflect this. Any change in the plasma process will be
registered by change(s) in the Fourier components.
Key goals in a manufacturing plant are to maximise line
yield (the percentage of wafers successfully processed) and
die yield (the number of fully functioning devices on each
wafer). Various process control mechanisms are used to
optimise the performance of each tool to meet these
objectives. One approach to process control is to apply a
fault detection and classification (FDC) scheme to a
fingerprint obtained from a sensor, such as the RF sensor
described above.
the
An FDC scheme can be implemented as follows. First,
state of the process is measured using one or more sensors.
The sensor data can be multidimensional data from a single
sensor (e.g., Fourier components of the RF source 3 obtained
from the sensor 5, as in the embodiment to be described), or
data from a set of sensors, but in either case the data must
be sensitive to hardware and process changes. The important
criterion is that the sensor data has sufficient dimensions
to permit a plurality of different fingerprints to be
defined for a respective plurality of different fault
conditions. The RF sensor described above fulfils these
requirements." As used herein, a “fingerprint” is a set of
sensor data which defines a particular state of the
equipment - thus a fault fingerprint means a set of sensor
data defining the state of the equipment in a fault
condition. As wafers are processed through the chamber, the
fingerprint is analysed to determine if the process is in an
abnormal state. If such a state is detected, wafer
processing is halted until the problem is resolved. This is
the fault detection step. The time taken to solve the
problem and restore the tool to production can be reduced by
further classifying the abnormal process state against a
historical record of known fault conditions to determine if
a similar fault has re—occurred. This is the fault
classification step.
A method for learning the profile of sensor data associated
with a specific type of fault is described in US Patent
6,441,620 and is explained using the following simple
example.
A designed experiment (DOE) is carried by varying three
process inputs: power, pressure and electrode spacing. The
DOE design is shown in Fig. 2. In addition, the experiment
includes a fault condition whereby a process kit part of the
wrong dimension is installed in the chamber. The process
kit is a part of the chamber hardware that is replaced
during preventive maintenance. Installing an incorrect part
alters wafer processing conditions which can cause wafers to
be scrapped. An RF impedance sensor is used to measure the
subsequent changes in the plasma chamber.
Fig.
shows how three of the sensor outputs, A1 (RF voltage
fundamental), A2 (first harmonic of RF voltage) and A3 (RF
phase fundamental), respond to changes in power and
electrode spacing on a given chamber. Sensor outputs Al and
A2 respond in a similar fashion to changes in power and
spacing, but sensor output A3 responds differently. Thus,
especially when all sensor outputs are taken into account, a
change in process power will be different and
distinguishable from a change in electrode spacing. If many
of the hardware conditions and process inputs are changed in
a design of experiments then a comprehensive tool profile
comprising a set of sensor responses for all hardware and
process input changes can be established, with each fault
condition having its own unique fingerprint. A collection
of sets of sensor data defining respective different fault
conditions is referred to hereafter as a “fault library”.
Each fault-defining set of sensor data in the fault library
is preferably recorded as a set of differences from the
relevant sensor outputs when the process is in a known good
state.
To perform fault detection and classification, changes in
sensor outputs are recorded for each wafer as it is
processed. The difference between the current sensor
outputs, and the sensor outputs when the process was in the
known good state, is compared to the fault library. A match
is considered to have been found if the fingerprint of the
wafer is well correlated to a fault fingerprint in the fault
library. Typically, the user is presented with a chart
which shows how the current wafer correlates to all
fingerprints in the library. An example of such a chart is
shown in Fig. 4. This indicates that the current wafer is
well correlated to the process kit fault fingerprint in the
library, and the user can have good confidence that the FDC
system has found an accurate match.
The magnitude of the change can be used to perform fault
detection. If the change exceeds a user-specified
threshold, a fault detection decision is reported to the
user. Fig 5 shows the magnitude of change for the a
sequence of wafers processed with the wrong process kit part
installed. These wafers are above a fault detection
threshold which has been set so as not to cause any false
alarms on known good wafers.
A typical wafer fab has many plasma chambers dedicated to
each step of the process and each chamber requires a fault
library. In some cases, it may not be possible to copy the
fault library learned on one chamber directly to all other
nominally identical chambers, and use it for fault detection
and classification. Fig. 6 shows what happens when an
identical process kit fault to that induced in the chamber
referred to in relation to Figs. 3 to 5, Chamber A, is
induced on a second chamber, Chamber B, nominally identical
to Chamber A, where the fault library has been copied
directly from Chamber A. The fault detection signal has
been reduced by a factor of 5. The reason for the reduced
sensitivity can be seen in Fig. 7 which shows the
corresponding correlation information. Identical faults
induced on Chamber A and Chamber B are not well correlated.
Thus, this approach to process control has two limitations.
Decreased sensitivity: The robustness of the FDC system
depends on a large signal-to-noise ratio for each fault
compared to the baseline of normal behaviour. Reducing the
signal—to—noise ratio of the FDC system can lead to faults
being missed, with a subsequent impact to line yield, die
yield and cost.
Reduced accuracy of classification: The trustworthiness of
fault classification information depends on achieving a good
match to a previously learned pattern when a similar fault
re-occurs. If the same fault re-occurs but does not trigger
a reasonably close match to a pattern in the fault library,
a user's confidence in the FDC system is diminished.
An alternative strategy would be to build a new copy of the
fault library on each chamber. This method also has a major
limitation. As each fault is required to be learned through
a designed experiment carried out while the chamber is not
running production material, this method would require
significant downtime and is very costly. This is especially
true for those experiments which require chamber hardware to
be temporarily modified, as is the case with the process kit
fault described above.
It is also possible to construct a fault library using only
those sensor parameters which are known to respond similarly
across different chambers. This approach would also lead to
reduced sensitivity as other parameters that contain useful
information about a fault condition are not used.
Therefore, there is a need for a method for manipulating a
fault library in some fashion so that it can be used on
multiple chambers without sacrificing sensitivity.
SUMMARY OF THE INVENTION
According to the present invention there is provided a
method of transferring a multi-variate process control model
from one RF~powered plasma processing chamber (the reference
chamber) to another nominally identical RF—powered plasma
processing chamber (the target chamber), comprising the
steps of:
(a) determining a multi-variate process control model
on the reference tool based upon sensor data,
(b) taking a fingerprint of the reference chamber by
running a designed experiment,
(c) taking a fingerprint of the target chamber by
running the same designed experiment,
(d) determining the relationship between the reference
and target chambers by comparing the results of the designed
experiments and Calculating a transform matrix representing
differences between the two chambers, and
(e) transforming the process control model from the
reference chamber to the target chamber using the transform
matrix obtained in step (d).
While in general the reference and target chambers would be
different, in one embodiment the reference and target
chamber are one and the same chamber and the transformation
method is used to update the process control model on the
chamber.
BRIEF DESCRIPTION OF THE DRAWINGS
Embodiments of the invention will now be described, by way
of example, with reference to the accompanying drawings, in
which:
Fig. 1 depicts a typical plasma processing chamber.
Fig. 2 describes the scheme for a designed experiment (DOE)
to build a fault detection and classification (FDC) model.
Fig. 3 shows the response of three sensor outputs to
variations in two process inputs.
Fig. 4 is a chart showing the degree of correlation between
a fault wafer and each of the faults in a fault library.
Fig. 5 is fault detection chart showing the magnitude of
change for a fault compared to normal wafers.
Fig. 6 shows the reduced sensitivity to a process kit error
when a fault library is copied directly from one chamber to
another.
Fig. 7 shows the diminished degree of pattern matching when
a fault library is copied directly from one chamber to
another.
Fig. 8 illustrates the difference in sensor response for the
same fault condition on two different chambers.
Fig. 9 shows the similarity in sensor response for the same
fault condition measured using two different sensors on the
same chamber.
Fig. 10 describes a simplified design of experiments used to
characterise the differences between chambers.
Fig. 11 shows the improved sensitivity when a transformed
fault library is used.
Fig. 12 shows the improved correlation when a transformed
fault library is used.
Fig. 13 is a flow chart illustrating an embodiment of the
invention.
DETAILED DESCRIPTION OF A PREFERRED EMBODIMENT
As described above, a library of fault fingerprints cannot
be always be transferred from one chamber to another due to
differences in the fingerprint recorded on each chamber.
This is illustrated in Fig. 8, which shows the response of
sensor output A7 (l4“‘harmonic of current) to an electrode
spacing change on two chambers running nominally the same
recipe. The data is recorded by Sensor 1 on Chamber A and
by Sensor 2 on Chamber B. Note that the response is very
different so that an electrode spacing fault learned on
Chamber A could not be used to detect the same fault on
Chamber B, if the profile of the fault includes sensor
parameter A7.
There are techniques available to compensate for the
difference in response across multiple sensors, a process
known as instrument standardization. Wang et al.
(“Multivariate Instrument Standardization”, Yongdong Wang et
al, Analytical Chemistry, vol. 63, pp2750-2756, 1991)
describe a range of techniques, which typically involve
collecting sensor responses to a range of stimuli, and using
the results to factor out sensor differences. Instrument
standardization is used to transform data from multiple
sensors to behave as it were collected on a single reference
sensor. This allows models built using data collected on
one sensor to be applied to data collected on other sensors.
US Patent 5,549,677 describes instrument standardization
techniques that are specific to optical spectrometers, in
particular a “direct standardization” method which is
referred to further below.
It is useful to determine if the difference in Chamber
response shown in Fig. 8 is due to a variation in the
sensors. Fig. 9 shows the result of an experiment where the
two sensors were installed on Chamber A. In this case, both
sensors have nearly identical responses to an electrode
spacing change. The sensor-swap experiment confirms that
the chamber-to-chamber response variation is not due to the
sensor but rather due to differences in the wafer processing
chamber itself. This could be due differences in how the
chambers were manufactured, assembled or calibrated.
Therefore, applying instrument standardization will not
solve the problem, as the cause of variation does not lie
with the sensor.
As can be seen in Fig. 8, the chamber-to-chamber variation
has the effect of changing the slope of the response of
parameter A7 to an electrode spacing fault.
In effect, the
response has been rotated. More generally, the response
could be rotated and scaled (either amplified or
attenuated). It is possible to characterise the difference
between chambers by measuring the amount of scaling and
rotation that occurs.
We have discovered that by estimating the scaling and
rotation differences, a compensation technique can be used
such that a fault library built on one chamber, Chamber A,
can be used on another nominally identical chamber,
Chamber B.
The method used is to run the same DOE, such as
that shown in Fig. 10, on Chamber A and on Chamber B. This
DOE is designed to be run with minimal impact to chamber
productivity, and is considerably simplified compared to the
original full DOE used to establish the initial fault
library on Chamber A. The simplified, or “short”, DOE is
typically a subset of the original full DOE but this is not
necessary. Typically, a short DOE can be performed in hours
compared to days for a full DOE. The difference in response
to the DOE on two chambers can be estimated as described
below.
The DOE results can be represented as a matrix whose
dimensions are nxm, where n is the number of runs in the
DOE, and m is the number of Fourier components measured by
the RF sensor.
Because of chamber differences, the DOE results from each
chamber are not identical, but they can be related to each
other as
DOE3 = DOEA T
where DOEAis an nxm matrix representing the DOE results on
chamber A and DOEBis the corresponding set of results from
chamber B. T is a transform matrix relating the results of
the two DOE’s. Re—arranging the equation to solve for T
yields
T = DOEA” DOEB
where DOEX is the pseudo-inverse of DOEAnatrix, calculated
using a suitable technique. T is a unique least—squares
solution.
The transform matrix T captures the difference between the
chambers. In this case, the transform captures rotation and
scaling differences between sensor responses on different
chambers, but in principal could be extended to capture
other differences such as differences in the coefficients of
non-linear models.
The transform matrix can be used to transform any data or
multi—variate model from one chamber so it appears to have
been measured on a different chamber. This is a direct
transformation method which is superficially similar to the
direct standardization technique in US Patent 5,549,677 but
with key differences. Firstly, it is not applied to correct
for sensor variation; rather to it is intended to remove
differences in the signal output from two wafer processing
chambers in the case where the sensors are matched.
Secondly, it is applied to RF rather than optical
spectroscopy data. Finally, its field of application is in
process control rather than instrument calibration.
Each fault which has been learned on Chamber A can be
represented as an lxm vector where m is the number of
Fourier components measured by the RF sensor. All fault
fingerprints learned on Chamber A can be combined into a
single fault library which can be represented as a matrix
dimensioned pxm, where p is the number of faults in the
library and m is the number of components.
As described above,
this library can not always be deployed
on Chamber B directly. However, if it can be transformed as
follows
FLB = FLA T
where FLBis a transformed fault library, T the transform
matrix and FLAis the full fault library available on
Chamber A. The FL3 library will now behave as it were
learned on Chamber B.
Figure 11 shows the results of a fault detection experiment
where a fault library learned on Chamber A is first
transformed before being tested on Chamber B. Comparing
Fig. 11 and Fig. 6, sensitivity has been increased by using
the transformation process. Similarly, Fig. 12 shows that
the accuracy of the fault matching process (essential for
reliable classification) has also been enhanced. Recalling
that the fault tested on Chamber B is being compared to a
pattern learned on Chamber A, the accuracy of the pattern
match is now as good as if the fault were learned on the
Chamber B directly.
Fig. 13 is a flow diagram of an embodiment of the invention
for deploying fault libraries or other multi—variate process
control models from chamber to chamber. A full DOE is run
on one chamber,
called the reference chamber, to build a
fault library with multi-variate fingerprint data, step 100.
A subset of the full DOE that can be run with minimal impact
to chamber productivity is run on reference chamber, step
102, and on the chamber where the model is to be
transferred, called the target chamber, step 104. As
mentioned, the DOE in steps 102 and 104 is typically a
subset of the DOE used to build the full fault library on
the reference chamber. The transform matrix T is
calculated, step 106. A transform of the fault library from
the reference chamber to the target chamber is then
calculated using T, step 108.
The chamber transformation technique can be used to copy any
multi—variate process control model from one chamber to
another. The transform matrix T is calculated in a similar
manner, but rather than copy a fault library between
chambers, other process control models can be used. For
example, published US patent application 10/295,350
describes a method for designing a model of sensor
parameters that is predictive of die yield. This model
could be a linear model of sensor parameters such as
YA = 0LAl + BA2 + xA3 + 6A4
where y is the model
output indicative of a yield parameter
such as etch rate or critical dimension, Al-A4 are sensor
parameters and aJi%5 are the model parameters. If this
model were developed on Chamber A it could be transferred to
other chambers using transform technique so that
YB=YAT
is the corresponding model for Chamber B. In this example, T
would be dimensioned as a 4x4 matrix, as just 4 sensor
parameters are used. In general, any multi-variate model can
be transferred in this way.
It is also possible to learn the transform matrix T other
than by running a designed experiment which requires wafers
to be processed in the chamber. For example, the chambers
could be fingerprinted using a plasma—less test, whereby a
fingerprint is recorded with RF power applied to the chamber
but without igniting a plasma. This does not require a
wafer to be present in the chamber.
A transform can be created between any pair of nominally
identical chambers, allowing models to be transferred at
will across an entire production line.
The transform can also be used to update a fault library on
the same chamber, where the chamber has been significantly
altered by a major preventive maintenance activity or
hardware retrofit. In this case, the transform T, accounts
for the modification to the same chamber over a period of
time as shown below.
FLAis a fault library learned on Chamber A, and DOEAis
short DOE on Chamber A. Chamber A is then modified so that
its response is altered. If DOEy is a DOE run after the
modification, then the transform T can be learned as
T=DOEA# DOEA»
and used to generate a fault library FLy that compensates
for the chamber modification as follows
FLA» = FLAT
Although the foregoing has described an embodiment where the
sensor outputs are Fourier components of the RF source, as
derived from the single RF sensor 5, the initial full DOE as
well as the subsequent short DOE’s, can use the data from
non-RF sensors, or from a combination of RF and non—RF
sensors. For example, ion flux sensors can be used, as well
as optical sensors and sensors for residual gas analysis.
The invention is not limited to the embodiments described
herein which may be modified or varied without departing
from the scope of the invention.
Claims (11)
1. A method of transferring a multi-variate process control model from one RF-powered plasma processing chamber (the reference chamber) to another nominally identical RF- powered plasma processing chamber (the target chamber), comprising the steps of: (a) determining a multi-variate process control model on the reference chamber based upon sensor data, (b) taking a fingerprint of the reference chamber by running a designed experiment, (c) taking a fingerprint of the target chamber by running the same designed experiment, (d) determining the relationship between the reference and target chambers by comparing the results of the designed experiments and calculating a transform matrix representing differences between the two chambers, and (e) transforming the process control model from the reference chamber to the target chamber using the transform matrix obtained in step (d).
2. The method of claim 1, wherein step (a) is performed using a designed experiment and steps (b) and (c) are performed on a simpler designed experiment than that used for step (a).
3. The method of claim 2, wherein the designed experiment performed in steps (b) and (c) is a subset of the designed (a). experiment performed in step
4. The method of claim 2, wherein the designed experiment performed in steps (b) and (c) is performed without a plasma in the chamber.
5. The method of any preceding claim, wherein the fingerprint in steps (b) and (c) is recorded using at least an RF impedance sensor.
6. The method of any preceding claim, wherein the fingerprint in steps (b) and (c) is recorded using at least one sensor other than an RF impedance sensor.
7. The method of any preceding claim, wherein the fingerprint in steps (b) and (c) is recorded using more than one SEDSOI‘ .
8. The method of any preceding claim, wherein the sensor data comprises a plurality of Fourier components of the RF power source driving the chamber.
9. The method of any preceding claim, wherein the process control model is a fault library-
10. The method of any one of claims 1 to 8, wherein the process control model is a yield-prediction model.
11. The method of any preceding claim, wherein the reference and target chamber are one and the same chamber and the transformation method is used to update the process control model on the chamber.
Priority Applications (3)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
IE2004/0568A IE84329B1 (en) | 2004-08-26 | A method for transferring process control models between plasma processing chambers | |
PCT/EP2005/006372 WO2006021251A1 (en) | 2004-08-26 | 2005-06-14 | A method for transferring process control models between plasma procesing chambers |
US11/157,644 US7113842B2 (en) | 2004-08-26 | 2005-06-21 | Method for transferring process control models between plasma processing chambers |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
IE2004/0568A IE84329B1 (en) | 2004-08-26 | A method for transferring process control models between plasma processing chambers |
Publications (2)
Publication Number | Publication Date |
---|---|
IE20040568A1 IE20040568A1 (en) | 2006-03-08 |
IE84329B1 true IE84329B1 (en) | 2006-09-06 |
Family
ID=
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US6441620B1 (en) | Method for fault identification in a plasma process | |
US7062411B2 (en) | Method for process control of semiconductor manufacturing equipment | |
US6238937B1 (en) | Determining endpoint in etching processes using principal components analysis of optical emission spectra with thresholding | |
EP1210724B1 (en) | Method of determining etch endpoint using principal components analysis of optical emission spectra | |
US5841651A (en) | Closed loop adaptive control of spectrum-producing step using neural networks | |
EP1639632B1 (en) | Method of monitoring a processing system | |
US7113842B2 (en) | Method for transferring process control models between plasma processing chambers | |
US20030113945A1 (en) | Disturbance-free, recipe-controlled plasma processing system and method | |
US20070055403A1 (en) | Methods of and apparatuses for maintenance, diagnosis, and optimization of processes | |
US6826489B2 (en) | Fault classification in a plasma process chamber | |
KR20080113352A (en) | Transforming metrology data from a semiconductor treatment system using multivariate analysis | |
EP1029341A1 (en) | Monitor of plasma processes with multivariate statistical analysis of plasma emission spectra | |
US6781383B2 (en) | Method for fault detection in a plasma process | |
Hong et al. | Neural-network-based sensor fusion of optical emission and mass spectroscopy data for real-time fault detection in reactive ion etching | |
US6789052B1 (en) | Method of using control models for data compression | |
Baek et al. | An effective procedure for sensor variable selection and utilization in plasma etching for semiconductor manufacturing | |
Thieullen et al. | Application of Principal Components Analysis to improve fault detection and diagnosis on semiconductor manufacturing equipment | |
IE84329B1 (en) | A method for transferring process control models between plasma processing chambers | |
KR102373933B1 (en) | Diagnostic system for diagnosing semiconductor processing equipment and control method thereof | |
IE20030437A1 (en) | A method for process control of semiconductor manufacturing equipment | |
IE83802B1 (en) | A method for fault detection in a plasma process | |
IE83920B1 (en) | A method of fault detection in manufacturing equipment | |
Hong et al. | Incipient Fault Detection of Reactive Ion Etching Process | |
IE20020107A1 (en) | Fault classification in a plasma process chamber | |
IE83492B1 (en) | Fault classification in a plasma process chamber |