USO09544987B2
`
`(12) United States Patent
`Mueller et al.
`
`(10) Patent No.:
`(45) Date of Patent:
`
`US 9,544.987 B2
`Jan. 10, 2017
`
`(54) FREQUENCY TUNING FOR PULSED RADIO
`FREQUENCY PLASMA PROCESSING
`
`(56)
`
`References Cited
`
`U.S. PATENT DOCUMENTS
`
`(71) Applicant: Advanced Energy Industries, Inc.,
`Fort Collins, CO (US)
`
`(72) Inventors: Michael Mueller, Loveland, CO (US);
`Myeong Yeol Choi. Fort Collins, CO
`(US)
`
`(73) Assignee: Advanced Energy Industries, Inc.,
`Fort Collins, CO (US)
`
`(*) Notice:
`
`Subject to any disclaimer, the term of this
`patent is extended or adjusted under 35
`U.S.C. 154(b) by 21 days.
`
`(21) Appl. No.: 14/320,268
`
`(22) Filed:
`
`Jun. 30, 2014
`
`(65)
`
`Prior Publication Data
`US 2015/0382442 A1
`Dec. 31, 2015
`
`(2006.01)
`(2006.01)
`
`(51) Int. Cl.
`H05H L/46
`HOI. 37/32
`(52) U.S. Cl.
`CPC ............. H05H I/46 (2013.01); H01J 37/3299
`(2013.01); H01J 37/32155 (2013.01); H01.J
`37/32.174 (2013.01); H01J 37/321.83
`(2013.01); H01J 37/32926 (2013.01); H05H
`2001/4645 (2013.01); H05H 2001/4682
`(2013.01)
`
`(58) Field of Classification Search
`CPC .............. H01J 37/3299; H01J 37/32802; H01J
`37/32174
`See application file for complete search history.
`
`Frequency
`
`No significant
`impedance
`mismatches
`
`
`
`Power
`
`6,383,554 B1* 5/2002 Chang ................. HO1J 37,3299
`427/10
`6,472,822 B 1 * 10/2002 Chen ................. HO1J 37.32082
`315/111.21
`2009/0237170 A1* 9/2009 Van Zyl. HO1J 37.32082
`331, 127
`2013/0214683 A1* 8/2013 Valcore, Jr. ....... HO1J 37.32082
`315/111.21
`8, 2013 Coumou ............... HO3F 1 O211
`330,75
`2014/036.1690 A1* 12/2014 Yamada ............ HO1J 37,32091
`315/111.21
`
`2013/0222055 A1
`
`FOREIGN PATENT DOCUMENTS
`
`WO
`
`2013099.133 A1
`
`T 2013
`
`OTHER PUBLICATIONS
`
`Rabbini, Firoozeh, “International Search Report and Written Opin
`ion re Application No. PCT/US2015/037607, Sep. 21, 2015, p. 5
`Published in: AU.
`
`* cited by examiner
`Primary Examiner — Douglas W Owens
`Assistant Examiner — Pedro C Fernandez
`(74) Attorney, Agent, or Firm — Neugeboren O'Dowd PC
`(57)
`ABSTRACT
`This disclosure describes systems, methods, and apparatus
`for pulsed RF power delivery to a plasma load for plasma
`processing of a Substrate. In order to maximize power
`delivery, a calibration phase using a dummy Substrate or no
`Substrate in the chamber, is used to ascertain a preferred
`fixed initial RF frequency for each pulse. This fixed initial
`RF frequency is then used at the start of each pulse during
`a processing phase, where a real Substrate is used and
`processed in the chamber.
`17 Claims, 10 Drawing Sheets
`
`Frequency
`
`ldeal Tuning Frequency ----------------------
`
`Time
`
`RENO EXHIBIT 2021
`Advanced Energy v. Reno, IPR2021-01397
`
`

`

`U.S. Patent
`
`Jan. 10, 2017
`
`Sheet 1 of 10
`
`US 9,544,987 B2
`
`Bulyoeyy 801
`
`BUWSEldYAOMION
`[eg ||
`
`q|Aouenbe.4jo.Uu0y
`
`
`
` Jaquieyoewseld||||
`
`||
`
`901POL
`
`che
`
`
`
`jeNluyAguanbal4
`
`
`
`s|npo|!sinpoy
`
`
`
`velcock
`
`
`
` WNDID|Bulyoms|bL--eeJ
`
`OIF
`
`L“OlaOb
`
`
`
`80JNOSJeMOd
`
`
`
` ||—--—-—Leite||||Scr|oct!!SOeLO}U|JOS/|Kejdsiq,Loi|
`
`|vib|49||O.U0D
`
`uosuedwoy
`‘bas4
`juewainseayy
`
`3|NPO|
`
`ainpow
`
`BLT
`
`

`

`U.S. Patent
`
`Jan. 10, 2017
`
`Sheet 2 of 10
`
`US 9,544.987 B2
`
`
`
`Kouanbºu
`
`Áouenbeu
`
`

`

`U.S. Patent
`
`Jan. 10, 2017
`
`Sheet 3 of 10
`
`US 9,544.987 B2
`
`
`
`
`
`
`
`--------- - - - ---------------Kouanbºu-, 6u?un L |eep|
`
`

`

`U.S. Patent
`
`Jan. 10, 2017
`
`Sheet 4 of 10
`
`US 9,544.987 B2
`
`
`
`J?MOCH
`
`

`

`U.S. Patent
`
`Jan. 10, 2017
`
`Sheet S of 10
`
`US 9,544.987 B2
`
`500
`
`Set initial frequency to f
`502
`
`
`
`Start of pulse?
`504
`
`Tune frequency to maximize delivered power
`506
`
`
`
`End of pulse?
`508
`
`Set initial frequency to f based on attempt to
`minimize delivered power error function at
`Start of SubSequent pulse as Compared to fin-1
`510
`
`Initial frequency of last
`iteration or average of
`last few iterations = fixed
`initial RF frequency
`
`
`
`
`
`
`
`Iterations
`failing to significantly decrease
`error function?
`512
`
`
`
`FIG. 5
`
`

`

`U.S. Patent
`
`Jan. 10, 2017
`
`Sheet 6 of 10
`
`US 9,544.987 B2
`
`600
`
`
`
`
`
`Set initial RF frequency to fo
`606
`
`Start of pulse?
`602
`
`Tune RF frequency to maximize delivered
`pOWer
`608
`
`End of pulse?
`610
`
`
`
`FIG. 6
`
`

`

`U.S. Patent
`
`Jan. 10, 2017
`
`Sheet 7 of 10
`
`US 9,544.987 B2
`
`700
`
`START
`
`Provide pulsed RF power to plasma load to ignite
`and Sustain plasma in plasma processing chamber
`having a dummy substrate or no substrate
`702
`
`Tune initial RF frequency at start of each pulse
`based on Comparison of characteristic to ideal
`characteristic for initial frequency at start of
`previous pulse
`704
`
`Calibration
`Phase
`
`Processing
`Phase
`
`
`
`
`
`
`
`
`
`
`
`Continued tuning
`providing insignificant
`improvement?
`7O6
`
`YeS
`
`Initial frequency fo identified
`708
`
`Provide pulsed RF power to plasma load to ignite
`and sustain plasma in plasma processing chamber
`having a substrate, where initial frequency at start
`of each pulse is fo
`710
`
`Tune frequency during each pulse to optimize
`delivered power
`712
`
`FIG. 7
`
`

`

`U.S. Patent
`
`Jan. 10, 2017
`
`Sheet 8 of 10
`
`US 9,544.987 B2
`
`
`
`800
`
`COMPUTER SYSTEM
`
`PROCESSOR(S)
`
`MEMORY
`
`NETWORK
`INTERFACE
`
`GRAPHICS
`CONTROL
`
`STORAGE CONTROL KC)
`
`808
`
`y
`
`807
`
`STORAGE
`
`809
`
`OPERATING
`SYSTEM
`
`EXECS
`
`DATA
`
`API
`APPLICATIONS
`
`DISPLAY
`
`INPUT INTERFACE
`
`INPUT
`DEVICE(S)
`
`OUTPUT
`INTERFACE
`
`OUTPUT
`DEVICE(S)
`
`STORAGE DEVICE
`INTERFACE
`
`STORAGE
`DEVICE(S)
`
`STORAGEMEDIUM
`INTERFACE
`
`STORAGE
`MEDIUM
`
`FIG. 8
`
`

`

`U.S. Patent
`
`Jan. 10, 2017
`
`Sheet 9 of 10
`
`US 9,544.987 B2
`
`305
`
`
`
`J?QUue?O BUuSela
`
`eUuSele+
`
`
`
`
`
`?Inpou
`
`?uÐuuÐInseeW
`
`
`
`?OunOS JONAO)
`
`

`

`U.S. Patent
`
`Jan. 10, 2017
`
`Sheet 10 of 10
`
`US 9,544.987 B2
`
`
`
`000 ||
`
`BULISE|e)
`
`
`
`90.InOS J3/WO).
`
`JOSU?S
`
`?InpOUu
`
`?uÐUuÐInSeÐIN
`
`

`

`US 9,544,987 B2
`
`1.
`FREQUENCY TUNING FOR PULSED RADIO
`FREQUENCY PLASMA PROCESSING
`
`FIELD OF THE DISCLOSURE
`
`The present disclosure relates generally to pulsed radio
`frequency power. In particular, but not by way of limitation,
`the present disclosure relates to systems, methods and appa
`ratuses for frequency tuning a power generation system for
`igniting and Sustaining a plasma load in a plasma processing
`chamber.
`
`BACKGROUND
`
`Plasma processing of a substrate often calls for radio
`frequency (RF) power to be used to Sustain the plasma, and
`sometimes this RF power is provided in a pulsed envelope.
`During each pulse, frequency tuning of the RF power can be
`used to optimize power delivery (e.g., by impedance match
`ing). Existing tuning algorithms call for the tuned frequency
`at the end of a pulse to be used at the start of a Subsequent
`pulse. Yet, the tuned frequency at an end of a pulse is often
`not well-tuned at the start of a Subsequent pulse, leading to
`a large impedance mismatch at a start of many pulses. This
`impedance mismatch can disrupt processing recipes or even
`damage the Substrate being processed.
`
`SUMMARY OF THE DISCLOSURE
`
`10
`
`15
`
`25
`
`2
`optimize delivered power to a plasma load. The method can
`include setting a frequency of pulsed RF power to a fixed
`initial RF frequency at a start of multiple pulses of RF power
`delivered to the plasma load for processing a Substrate. The
`fixed initial RF frequency can be selected by (1) repeatedly
`adjusting an RF frequency at a start of a plurality of
`consecutive pulses of RF power delivered to the plasma load
`when a dummy Substrate is in the plasma processing cham
`ber and (2) selecting the RF frequency as the fixed initial RF
`frequency when the adjusting settles on a steady state RF
`frequency. The adjusting can be performed so as to minimize
`a difference between a characteristic indicative of the pulsed
`RF power and a desired characteristic of the pulsed RF
`power as compared pulse to pulse. The method can further
`include tuning the frequency of pulsed RF power for a
`remainder of each pulse to minimize a difference between a
`characteristic indicative of the pulsed RF power and a
`desired characteristic of the pulsed RF power.
`Other embodiments of the disclosure can be characterized
`as a power delivery system comprising a power source, a
`sensor, and a controller. The power source can be configured
`to provide pulsed RF power to a plasma load via a matching
`network. The sensor can be configured to sample a charac
`teristic indicative of the pulsed RF power indicative of
`delivered power. The controller can be in communication
`with the sensor and the power source and can comprise a
`plurality of logical blocks. The logical blocks can comprise
`a measurement module, an initial frequency comparison
`module, a frequency control module, and an initial fre
`quency identification module. The measurement module can
`receive samples of the characteristic indicative of the pulsed
`RF power from the sensor. The initial frequency comparison
`module can compare samples of the characteristic indicative
`of the pulsed RF power at a start of each of a plurality of RF
`pulses to a desired characteristic of the pulsed RF power at
`a start of each of the plurality of pulses to determine an error
`value. The frequency control module can instruct the power
`Source to adjust the initial frequency in order to reduce the
`error value. The initial frequency identification module can
`identify the initial frequency as a fixed initial RF frequency
`when adjustments to the initial frequency at the start of two
`or more pulses fail to result in further reduction of the error
`value.
`
`30
`
`35
`
`40
`
`Exemplary embodiments of the present invention that are
`shown in the drawings are Summarized below. These and
`other embodiments are more fully described in the Detailed
`Description section. It is to be understood, however, that
`there is no intention to limit the invention to the forms
`described in this Summary of the Invention or in the
`Detailed Description. One skilled in the art can recognize
`that there are numerous modifications, equivalents and alter
`native constructions that fall within the spirit and scope of
`the invention as expressed in the claims.
`Some embodiments of the disclosure may be character
`ized as a method of selecting a fixed initial RF frequency for
`each of a plurality of RF pulses provided to a plasma load.
`The method can include providing pulsed RF power to a
`plasma load, where each pulse comprises RF power having
`a controllable frequency. Also, the providing occurs while a
`45
`plasma having the plasma load interacts with a dummy
`substrate in a plasma chamber. The method can further
`include setting an initial RF frequency at a start of each pulse
`to a different frequency from the start of the previous pulse
`Such that the initial RF frequency is tuned, pulse to pulse, to
`minimize a difference between a characteristic indicative of
`the pulsed RF power and a desired characteristic of the
`pulsed RF power. The method further includes tuning a
`frequency of RF power for a remainder of each pulse to
`minimize a difference between a characteristic of the pulsed
`RF power and a desired characteristic of the pulsed RF
`power. The method can further include selecting the initial
`RF frequency as a fixed initial RF frequency for use in a
`processing run using a real Substrate when tuning of the
`initial RF frequency between two or more consecutive
`pulses results in an insignificant improvement in the differ
`ence between the characteristic of the pulsed RF power and
`the desired characteristic of the pulsed RF power.
`Other embodiments of the disclosure may also be char
`acterized as non-transitory, tangible computer readable Stor
`age medium, encoded with processor readable instructions
`to perform a method for frequency tuning a power source to
`
`BRIEF DESCRIPTION OF THE DRAWINGS
`
`Various objects and advantages and a more complete
`understanding of the present invention are apparent and
`more readily appreciated by referring to the following
`detailed description and to the appended claims when taken
`in conjunction with the accompanying drawings:
`FIG. 1 illustrates a power generation system providing
`pulsed RF power to a plasma chamber via a matching
`network, where the pulsed RF power uses a fixed initial RF
`frequency at a start of each pulse.
`FIG. 2 illustrates the impedance mismatch that can result
`when traditional frequency tuning methods are used.
`FIG. 3 illustrates a reduced impedance mismatch at the
`start of pulses where a fixed initial RF frequency is used.
`FIG. 4 shows a tuning regime where only three frequency
`set points per pulse are possible.
`FIG. 5 illustrates a method of frequency tuning RF pulsed
`power to a plasma load during a calibration phase.
`FIG. 6 illustrates a method of frequency tuning RF pulsed
`power to a plasma load during a processing phase.
`
`50
`
`55
`
`60
`
`65
`
`

`

`US 9,544,987 B2
`
`3
`FIG. 7 illustrates a method of frequency tuning RF pulsed
`power to a plasma load during both a calibration phase and
`a processing phase.
`FIG. 8 shows a diagrammatic representation of one
`embodiment of a computer system.
`FIG. 9 illustrates another power generation system pro
`viding pulsed RF power to a plasma chamber via a matching
`network, where the pulsed RF power uses a fixed initial RF
`frequency at a start of each pulse.
`FIG. 10 illustrates yet another power generation system 10
`providing pulsed RF power to a plasma chamber via a
`matching network, where the pulsed RF power uses a fixed
`initial RF frequency at a start of each pulse.
`
`5
`
`4
`tuning during each pulse (after the initial frequency) allows
`the tuned frequency to approach the ideal tuning frequency,
`but the initial impedance mismatch is less than ideal.
`FIG. 3 illustrates a reduced impedance mismatch at the
`start of pulses where a fixed initial RF frequency is used. The
`pulse on and off states can be seen in the lower of the two
`plots while the frequency at which the RF power is being
`delivered at any point in time is seen in the upper of the two
`plots. In the first pulse, the frequency is tuned in a generally
`downward trending fashion Such that the frequency f is
`being used when the pulse turns off. Unlike the prior art,
`instead of using f as the initial RF frequency for the next
`pulse, f is again used, and this pattern repeats for the initial
`RF frequency of each pulse. In this way, a fixed initial RF
`frequency, for can be used for each pulse, where the fixed
`initial RF frequency, f, has been preselected as an optimal
`average starting frequency for pulses. As seen, the result is
`that the initial RF frequency, f is much closer to the ideal
`tuning frequency at a start of each pulse than if the tuned
`frequency at an end of a previous pulse was carried over to
`the start of the next pulse (as in FIG. 2). This causes far less
`impedance mismatch at the start of pulses and hence more
`consistent and accurate processing runs.
`FIGS. 2 and 3 illustrate frequency tuning where rapid
`changes in frequency are possible. In other situations fre
`quency tuning may be slower or the pulse envelopes may be
`faster, such that only a handful of frequency set points per
`pulse is possible. For instance, FIG. 4 shows a tuning regime
`where only three frequency set points per pulse are possible.
`Yet, the use of a fixed initial RF frequency, f at a start to
`each pulse, prevents any significant impedance mismatches
`at the start of each pulse (similar to FIG. 3). Hence FIGS. 3
`and 4 show that the herein described systems and methods
`can be applied to any system regardless as to the rapidity of
`frequency tuning.
`FIG. 1 illustrates a power generation system providing
`pulsed RF power to a plasma chamber via a matching
`network, where the pulsed RF power uses a fixed initial RF
`frequency at a start of each pulse. The power generation
`system 100 can include a power source 110 providing pulsed
`RF power. Optionally, the power source 110 can provide RF
`power and an optional switching circuit 116 can pulse the RF
`power to generate pulsed RF power. The frequency of the
`RF, the frequency of the pulses, and a duty cycle of the
`pulses, among other things, can be controlled by a controller
`114 (e.g., via one or more logical blocks). Further, the
`controller 114 can carry out frequency tuning of the pulsed
`RF power during each pulse. The controller 114 can also
`determine a fixed initial RF frequency, f, and instruct the
`power source 110 to use this fixed initial RF frequency at a
`start of each pulse.
`In particular, a sensor 112 can sample a characteristic of
`the pulsed RF power (e.g., reflected power, delivered power,
`load impedance, plasma density, etc.) on either side of the
`matching network 104 or at the plasma chamber 108 when
`a dummy Substrate or no Substrate is in the plasma chamber
`108 and the plasma 106 is ignited (during a calibration
`phase). The sensor 112 can provide these samples to a
`measurement module 118 of the controller 114, and the
`measurement module 118 can prepare the samples for an
`initial frequency comparison module 120 (e.g., by trans
`forming the samples into a different data form, e.g., an
`analog-to-digital converter). The initial frequency compari
`son module 120 can compare the samples of the character
`istic of the pulsed RF power at a start of each of a plurality
`of RF pulses to a desired characteristic of the pulsed RF
`power at a start of each of the plurality of pulses and via this
`
`DETAILED DESCRIPTION
`
`15
`
`The word “exemplary' is used herein to mean “serving as
`an example, instance, or illustration.” Any embodiment
`described herein as “exemplary” is not necessarily to be
`construed as preferred or advantageous over other embodi- 20
`mentS.
`For the purposes of this disclosure, “pulsed RF power is
`RF power having a power envelope that is pulsed. In other
`words, multiple periods of RF power are delivered, followed
`by a period of no power, followed by another set of period 25
`of RF power (or a pulse).
`To avoid the impedance mismatch that the prior art sees
`at the start of many pulses, this disclosure describes systems,
`methods, and apparatus where an initial RF frequency at the
`start of each pulse is dictated not by the tuned frequency at 30
`an end of a preceding pulse, but by a fixed value that remains
`the same from pulse to pulse. Further, this initial RF
`frequency can be determined prior to a processing run using
`a dummy substrate or other means. While this fixed initial
`RF frequency will still see some impedance mismatch at the 35
`start of each pulse due to unavoidable variations at the start
`of any pulse, the mismatch will on average be far less than
`seen in the prior art (e.g., compare FIGS. 2 and 3).
`In one embodiment, a method is disclosed for identifying
`an initial RF frequency that maximizes delivered power to 40
`the plasma load at a start of each pulse, as identified during
`a calibration phase, where a dummy Substrate, or no Sub
`strate, is in the plasma processing chamber. The calibration
`phase can involve traditional frequency tuning after a start of
`each pulse, but at the start, the frequency can be tuned from 45
`pulse to pulse in order to minimize reflected power or
`optimize some other characteristic indicative of delivered
`power. This initial RF frequency can then be used as the
`initial RF frequency for every pulse during a processing
`phase—or an actual run where a real Substrate is processed 50
`in the plasma processing chamber. In both the calibration
`and processing phases, after the initial RF frequency is set,
`traditional frequency tuning as understood by those of skill
`in the art is implemented for a remainder of each pulse.
`FIG. 2 illustrates the impedance mismatch that can result 55
`when traditional frequency tuning methods are used. The
`pulse on and off states can be seen in the lower of the two
`plots. The frequency at which the RF power is being
`delivered at any point in time is seen in the upper of the two
`plots. In the first pulse, the frequency is tuned in a generally 60
`downward trending fashion such that the frequency f is
`being used when the pulse turns off. The next pulse sees f
`as the initial RF frequency. Yet, for one of numerous reasons,
`an ideal tuning frequency (as shown by the dotted lines) is
`at a much higher frequency, and so starting at f leads to a 65
`large impedance mismatch (just as starting at f at the next
`pulse leads to a large impedance mismatch). Subsequent
`
`

`

`US 9,544,987 B2
`
`5
`
`10
`
`15
`
`25
`
`30
`
`40
`
`5
`comparison, the initial frequency comparison module 120
`can determine an error value. During the calibration phase,
`a frequency control module 122 of the controller 114 can
`instruct the power source 110 to adjust the initial RF
`frequency at a start of each of a plurality of pulses based on
`the error value from the initial frequency comparison mod
`ule 120. In particular, the frequency control module 122 can
`instruct the power source 110 to adjust the initial RF
`frequency at a start of each of a plurality of pulses, where the
`adjustments are tailored to minimize the error value.
`As the error value decreases, an initial frequency identi
`fication module 124 eventually determines that further
`adjustments to the frequency of the power source 110 are
`failing to result in further reduction of the error value, and
`identifies the latest initial RF frequency as the fixed initial
`RF frequency to be used during an actual processing run
`(during a processing phase). In one embodiment, the initial
`frequency identification module 124 can identify when
`adjustment to the initial RF frequency at the start of two or
`more consecutive pulses fails to result in further reduction in
`the error value, and can then identify the last initial RF
`frequency as the fixed initial RF frequency for use during the
`processing phase. In some embodiments the initial fre
`quency identification module 124 determines when two or
`more consecutive adjustments to the initial RF frequency
`fail to result in at least a threshold decrease in the error
`value, and then identifies a last initial RF frequency, or an
`average of a last few initial RF frequencies, as the fixed
`initial RF frequency for use during the processing phase. In
`yet another embodiment, the initial frequency identification
`module 124 determines that the initial RF frequency has
`reached a steady state, such that further adjustments to the
`initial RF frequency for consecutive pulses is not resulting
`in significant changes to the initial RF frequency or that a
`magnitude of adjustments to the initial RF frequency from
`35
`pulse to pulse are consistently below a threshold (e.g., 1% of
`a magnitude of the initial RF frequency). For instance,
`where ten consecutive pulses each see less than 1% adjust
`ments to the initial RF frequency, then the initial RF
`frequency can be deemed to have reached a steady state, and
`hence can be identified as the fixed initial RF frequency for
`use during the processing phase.
`Once the fixed initial RF frequency is identified, a sub
`strate can be loaded into the plasma chamber 108 and the
`controller 114 can cause the power source 110 to begin
`Supplying power for the processing phase. In particular, the
`controller 114 can instruct the power source 110 to start each
`pulse using the fixed initial RF frequency, and can then
`perform real-time frequency tuning for the remainder of
`each pulse.
`In some embodiments, the controller 114 and/or the
`sensor 112 can be in communication with an optional
`display 126 and an optional user interface 128. The sensor
`112 can send samples of the pulsed RF power, or any
`characteristic indicative of delivered power (e.g., voltage
`and current or plasma density, to name two examples), to the
`display 126 for presentation to a user. Alternatively, the
`controller 114 can send this same information to the display
`126 for presentation to a user. Alternatively, the controller
`114 can receive the samples form the sensor 112 and convert
`the samples to a data stream that is more easily presentable
`to a user and then send this data stream to the display 126
`for presentation to a user.
`Such displaying of information can merely inform a user,
`or can be used to take further action. For instance, the user
`can use the user interface 128 to provide inputs for control
`ling the power source 110.
`
`55
`
`45
`
`50
`
`60
`
`65
`
`6
`FIG. 9 illustrates a power generation system 900 provid
`ing pulsed RF power to a plasma chamber 908 via a
`matching network 904, where the pulsed RF power uses a
`fixed initial RF frequency at a start of each pulse. The power
`generation system 900 includes a controller 914 and logic
`blocks as described in FIG. 1. The power generation system
`900 further includes a switching circuit 916 that takes the
`AC power from the power source 910 and converts it to
`pulsed RF power. A sensor 912 can sample a characteristic
`indicative of delivered power and provide samples of the
`characteristic to either the system 900 (e.g., the controller
`914) or a display 926 coupled to or including a user interface
`928 (e.g., a keyboard, mouse, or touchscreen, to name a
`few). As seen, the sensor 912 can be arranged external to the
`power generation system 900, although this is not required,
`and as seen in FIG. 10, a sensor 1012 can be part of a power
`generation system 1000 and can sample the characteristic
`from within the system 1000.
`In some embodiments, the sensor 112,912, 1012 (and the
`display 126,926, 1026) can be replaced with an oscilloscope
`(or an oscilloscope can include the sensor 112,912, 1012) to
`provide feedback to a user regarding the characteristic
`indicative of delivered power (e.g., reflected power). A user
`may use readings visible on the oscilloscope to manually
`tune the initial RF frequency until the user identifies a
`frequency to use as the fixed initial RF frequency during the
`processing phase. In another embodiment, the sensor 112,
`912, 1012 is not in direct communication with the controller
`114 (e.g., the line connecting sensor 112,912, 1012 with the
`controller 114, 914, 1014 is dotted/dashed and hence
`optional), and therefore manual analysis of data from the
`sensor 112,912, 1012 is made followed by manual instruc
`tions being provided to the controller 114,914, 1014 to
`adjust the initial RF frequency or select a last initial RF
`frequency (or an average of a last number of initial RF
`frequencies) as the fixed initial RF frequency. In other
`words, where the sensor 112, 912, 1012 is not in direct
`communication with the controller 114, 914, 1014, user
`analysis and control may be required. User inputs to control
`the initial RF frequency during a calibration phase or to set
`the fixed initial RF frequency during a processing phase can
`be made via the user interface 128,928, 1028.
`FIG. 5 illustrates a method of frequency tuning RF pulsed
`power to a plasma load during a calibration phase. The
`method 500 provides pulsed RF power to a plasma load
`where the RF has a controllable frequency. As this method
`500 involves a calibration phase, the pulsed RF power is
`provided to a plasma load where the plasma operates on a
`dummy Substrate or where no Substrate is in the plasma
`chamber. The method 500 sets an initial RF frequency (f)
`at a start of a pulse (Block 502) so that the initial RF
`frequency is applied at a start of a pulse. Once a start of a
`pulse is identified (Decision 504), the method 500 tunes the
`RF frequency to maximize delivered power or some other
`characteristic indicative of delivered power (Block 506)
`until an end of the pulse (Decision 508). After the pulse has
`ended, the method 500 again sets the initial frequency to a
`different frequency (f) in an attempt to minimize an error
`function for delivered power (e.g., a difference between
`actual and desired delivered power) (Block 510). Power is
`applied at this new frequency, f, at the start of the next pulse
`(Decision 504), and then the frequency is tuned to maximize
`delivered power or another characteristic indicative of deliv
`ered power for a remainder of the next pulse (Block 506). At
`the end of this next pulse (Decision 508), the method 500
`again adjusts the initial frequency in an attempt to further
`minimize the error function for delivered power (Block
`
`

`

`US 9,544,987 B2
`
`7
`510). This series of operations continues to repeat until
`further iterations fail to result in significant reduction to the
`error function (Decision 512), at which point the initial
`frequency can be identified as the fixed initial RF frequency
`that is to be used during the processing phase (Block 514 and
`see FIG. 6).
`In some embodiments the tuning for a remainder of each
`pulse (Block 506) can be automated while in others it can be
`manual. Automated tuning involves a sensor providing
`samples of a characteristic, and a controller tuning the RF
`frequency based on the samples of the characteristic. Manual
`operation might see a user observing the samples (e.g., on a
`display in communication with the sensor or on an oscillo
`Scope directly monitoring the characteristic with or without
`a separate sensor) and manually adjusting the initial RF
`frequency at a start of each pulse according to the data that
`is visible to the user. A user cannot tune as rapidly as an
`automated system, so the user may not adjust the initial RF
`frequency every pulse, but instead may only be able to adjust
`the initial RF frequency every thousand pulses. The method
`will still be effective, and once a fixed initial RF frequency
`is manually identified, this frequency can be used as a
`setpoint for a processing phase.
`FIG. 6 illustrates a method of frequency tuning RF pulsed
`power to a plasma load during a processing phase. The
`method 600 involves providing pulsed RF power to a plasma
`load, where each pulse comprises RF power having a
`controllable frequency, and where a plasma having the
`plasma load processes a Substrate in a plasma chamber (e.g.,
`the processing phase). The method 600 sets a fixed initial RF
`frequency to f (Block 606), a frequency that is fixed for the
`start of all pulses. After the start of a pulse (Decision 602),
`the method tunes the RF frequency to maximize delivered
`power or a characteristic indicative of delivered power
`(Block 608) until an end of the pulse (Decision 610). Once
`the pulse has ended, the method resets the initial RF fre
`quency to fo and repeats for the next pulse. In this way, a
`fixed initial RF frequency, f is set at the start of each pulse
`and is fixed from pulse to pulse regardless as to the fre
`quency tuning that occurs during a remainder of each pulse.
`FIG. 7 illustrates a method of frequency tuning RF pulsed
`power to a plasma load during both a calibration phase and
`a processing phase. The method 700 involves two phases: a
`calibration phase and a processing phase. The calibration
`phase identifies a fixed initial RF frequency, f for pulsed
`RF power, while the processing phase uses the fixed initial
`RF frequency, f, at a start of each pulse to process a
`substrate. The method 700 starts by providing pulsed RF
`power to a plasma to ignite and Sustain the plasma in a
`plasma processing chamber having a dummy Substrate or no
`substrate (Block 702) (a start of the calibration phase). The
`method 700 then tunes the initial frequency over the course
`of multiple RF pulses by repeatedly adjusting the RF fre
`quency at a start of a plurality of consecutive pulses of RF
`power delivered to the plasma load (Block 704). The adjust
`ing is performed to optimize a characteristic indicative of
`delivered power at a start of each pulse as compared from
`pulse to pulse. When continued tuning ceases to provide
`significant improvements to the optimization (Decision
`706), then the method 700 identifies this frequency as the
`fixed initial RF frequency, f (Block 708).
`The method 700 then enters the processing phase, where
`a Substrate is loaded into the plasma processing chamber and
`an actual processing run is carried out on the Substrate. The
`method 700 provides pulsed RF power to the plasma load to
`ignite and Sustain the plasma with the Substrate in the
`chamber, where the initial RF frequency at a start of each
`
`40
`
`45
`
`8
`pulse is the fixed initial RF frequency, fo. The remainder of
`each pulse can involve tuning the frequency of the RF power
`to minimize a difference between a characteristic of the
`pulsed RF power and a desired characteristic of the pulsed
`RF power.
`In each of the figures, the use of dotted or dashed lines as
`compared to Solid lines, indicates optional components or
`features. At the same time, any components or fea