Metal ion deposition from ionized mangetron sputtering discharge
`S. M. Rossnagel and J. Hopwood
`
`Citation: Journal of Vacuum Science & Technology B 12, 449 (1994); doi: 10.1116/1.587142
`View online: http://dx.doi.org/10.1116/1.587142
`View Table of Contents: http://scitation.aip.org/content/avs/journal/jvstb/12/1?ver=pdfcov
`Published by the AVS: Science & Technology of Materials, Interfaces, and Processing
`
`Articles you may be interested in
`Evidence of ionized metal clusters in ion plating discharges
`Appl. Phys. Lett. 81, 1405 (2002); 10.1063/1.1501449
`
`Effect of sputter heating in ionized metal physical vapor deposition reactors
`J. Appl. Phys. 87, 7198 (2000); 10.1063/1.372969
`
`Liner conformality in ionized magnetron sputter metal deposition processes
`J. Vac. Sci. Technol. B 14, 2603 (1996); 10.1116/1.588993
`
`Magnetron sputter deposition with high levels of metal ionization
`Appl. Phys. Lett. 63, 3285 (1993); 10.1063/1.110176
`
`Presence of doublyionized metal ions in rare gas glow discharges
`J. Appl. Phys. 56, 872 (1984); 10.1063/1.334027
`
`
`
` Redistribution subject to AVS license or copyright; see http://scitation.aip.org/termsconditions. Download to IP: 98.234.93.94 On: Thu, 18 Dec 2014 18:26:26
`
`TSMC et. al. v. Zond IPR2014-00800
`Page 1 Zond Ex. 2016
`
`

`

`Metal ion deposition from ionized mangetron sputtering discharge
`s. M. Rossnagel and J. Hopwooda)
`IBM Research, Yorktown Heights, New York 10598
`
`(Received 10 September 1993; accepted 6 October 1993)
`
`A technique has been developed for highly efficient postionization of sputtered metal atoms from a
`magnetron cathode. The process is based on conventional magnetron sputtering with the addition of
`a high density, inductively coupled rf (RFl) plasma in the region between the sputtering cathode and
`the sample. Metal atoms sputtered from the cathode due to inert gas ion bombardment transit the rf
`plasma and can be ionized. The metal ions can then be accelerated to the sample by means of a low
`voltage dc bias, such that the metal ions arrive at the sample at normal incidence and at a specified
`energy. The ionization fraction, measured with a gridded mass-sensitive energy analyzer is low at 5
`mTorr and can reach 85% at 30 mTorr. Optical emission measurements show scaling of the relative
`ionization to higher discharge powers. The addition of large fluxes of metal atoms tends to cool the
`Ar RFI plasma, although this effect depends on the chamber pressure and probably the pressure
`response of the electron temperature. The technique has been scaled to 300 mm cathodes and 200
`mm wafers and demonstrated with Cu, AICu, and TifTiN. Deposition rates are equal to or in some
`cases larger than conventional magnetron sputtering. A primary application of this technique is
`lining and filling semiconductor trenches and vias on a manufacturing scale.
`
`I. INTRODUCTION
`
`Sputter deposition, in particular magnetron sputter depo(cid:173)
`sition, is a broadly used process for depositing metal films
`for semiconductor applications, optical coatings, magnetic
`media, hard and decorative coatings, architectural glass, au(cid:173)
`tomobile parts, and so on. Physical sputtering results in the
`emission of mostly neutral atoms from the near surface re(cid:173)
`gion of the bombarded surface with an emission profile
`which is roughly dependent on the cosine of the emission
`angle from the surface normal. This effect, coupled with the
`extended area of most sputtering sources and possibly some
`gas scattering results in a deposition flux which arrives at a
`sample surface from many directions. While this is quite
`useful for forming continuous films over rough topography
`or steps, when applied to higher aspect ratios such as
`trenches or vias, the resultant deposition usually forms an
`enclosed channel or void (Fig. 1).
`A second constraint to magnetron sputtering is the diffi(cid:173)
`culty of drawing a significant bias current to a sample during
`sputter deposition. The magnetron plasma is generally lo(cid:173)
`cated close to the cathode/target surface and relatively few
`ions are present in the vicinity of the sample. For many re(cid:173)
`active deposition cases (particularly the nitrides for hard
`coatings), the addition of substrate bias is useful in adding
`the necessary energy to form the compound film. For con(cid:173)
`ventional sputter deposition, it is necessary to heat the
`sample to 300-600 °C to form the nitride phase. Hard coat(cid:173)
`ing applications which could not allow significant heating
`(with substrates such as plastics, tool steels, etc.) led to the
`
`development of the unbalanced magnetron 1,2 which, by suit(cid:173)
`able design of the magnetic field, can allow some of the
`electrons from the primary magnetron discharge to stream
`out towards the sample position. These electrons may either
`induce ions to follow by means of a weak potential gradient,
`
`aiCurrent address: Department of Electrical and Computer Engineering,
`Northeastern University, Boston, MA 02115.
`
`or may also form additional ions in the vicinity of the sample
`which can then be drawn to the sample as a bias current. This
`technology has been demonstrated on a commercial basis,
`but requires the use of large electromagnets to supply the
`magnetic field out to the position of the sample. Unbalanced
`magnetrons are not generally used for semiconductor appli(cid:173)
`cations of nitride films.
`Recently, a technique has been developed which can ad(cid:173)
`
`dress these two problems. 3,4 If a significant fraction of the
`sputtered atoms from the cathode could be ionized in the
`region between the cathode and the sample, then the sput(cid:173)
`tered metal ions could be accelerated to the sample by means
`of a simple de potential. The directionality of the metal ions
`would be well controlled: ions would be accelerated perpen(cid:173)
`dicularly to the sample by the electric field in the plasma
`sheath at the sample surface. This effect has been demon(cid:173)
`strated in a related experiment by Holber et al. using metal
`evaporation into an electron cyclotron resonance (ECR)
`plasma. 5 In addition, the kinetic energy of the ions could be
`controlled at will simply by adjusting the relative potential
`difference between the plasma and the sample. Typically
`each of these effects only requires a simple negative de bias
`on the sample.
`
`II. EXPERIMENTS
`
`Several systems were outfitted with a variety of commer(cid:173)
`cial, planar magnetrons. These range from a 5 em diam cath(cid:173)
`ode operating at a few hundred watts to 30 cm rotating(cid:173)
`magnet planar magnetrons which operate at powers up to 30
`kW. The results described below were all taken with either a
`20 cm planar (fixed magnet) cathode or with the larger, ro(cid:173)
`tating magnet devices. These latter cathodes are used rou(cid:173)
`tinely on semiconductor manufacturing-scale systems such
`as the Varian M/20oo or Applied Materials 5500 integrated
`processing systems. No modifications were made to any of
`the cathodes.
`
`449
`
`J. Vac. ScI. Technol. B 12(1), Jan/Feb 1994
`
`0734-211 X/94/12(1)/449/5/S1.00
`
`©1994 American Vacuum Society
`
`449
`
` Redistribution subject to AVS license or copyright; see http://scitation.aip.org/termsconditions. Download to IP: 98.234.93.94 On: Thu, 18 Dec 2014 18:26:26
`
`TSMC et. al. v. Zond IPR2014-00800
`Page 2 Zond Ex. 2016
`
`

`

`450
`
`S. M. Rossnagel and J. Hopwood: Metal ion deposition
`
`,
`
`,
`, ,
`
`FI(;. I. Sketch of a cross section of a sputter deposited. layered film onto a
`medium aspect ratio feature.
`
`A dense, inductively coupled rf (RFI) plasma was gener(cid:173)
`ated in the region between the sputtering cathode and the
`sample plane (typically a Si wafer mounted on a holder). The
`plasma was generated by applying 13.S6 MHz power to a
`multiple turn coil of water-cooled Cu tubing located inside
`the vacuum chamber. A number of coil designs and locations
`were explored (Fig. 2). Much of this work utilized two to
`three turn coils which had a diameter approximately the
`same as the magnetron cathode and were located~-4 em
`from the cathode surface and 2-3 em from the sample sur(cid:173)
`face. Spiral coils were also used on some of the rotating
`magnet systems. These latter systems required great attention
`to coil symmetry, as the moving magnetic field coupled
`weakly into the RFI plasma, causing variations on the scale
`of the magnet rotation rate.
`
`2-turn r1 coil
`
`or equivalent •
`
`u "
`
`;
`
`200 mm wafer (on
`holder or chuck)
`
`.J
`-L~-._-_-~ ____ -_
`
`t '"OO, •• ~,.
`
`0- -250V, O-SA DC
`
`FIG. 2. Schematic of the experimental design.
`
`J. Vac. Sci. Technol. e, Vol. 12, No.1, Jan/Feb 1994
`
`450
`
`Several parameters were varied as a means of exploring
`the relative levels of ionization and the resultant film prop(cid:173)
`erties. The magnetron power could be varied up to 10 kW on
`the 20 em cathodes and up to 30 kW on the 30 em cathodes.
`Cathode materials used were Cu, AlCu, and Ti. The rf power
`to the RFI plasma was varied up to 3 kW in each system. The
`sample bias voltage was varied from 0 to - 300 V dc, al(cid:173)
`though for virtually all of the cathode materials used, the
`self-sputter yield effectively exceeded 1,0 at ~ 200 e V ( - 200
`V sample bias). At this point, there is no net deposition (on
`the flat plane parts of a sample) and the original substrate
`surface is sputtered. Finally, the operating pressure could be
`varied from less than 1 to ~SO mTorr. It should be noted that
`these experiments were performed in test chambers of ~50 /
`vol and a base pressure of IO- H Torr. Manufacturing systems
`for semiconductor applications typically would have a base
`pressure of 10- 9 Torr, but generally do not operate at pres(cid:173)
`sures above S mTorr by design.
`To measure the relative ionization level of the metal
`fluxes (to the sample) one system was configured with a
`differentially pumped, gridded energy analyzer in place of
`the sample. The collector of this analyzer contained the ac(cid:173)
`tive area a quartz crystal microbalance. In this way, the
`weight of the depositing flux to the collector surface as well
`as the current could be measured. This allows differentiation
`between the working gas ions (Ar) and the depositing metal
`ions. By suitable biasing of the grids in the energy analyzer,
`ions could be repelled, allowing the neutral metal deposition
`flux to be measured. In this way, the relative ionization level
`of the metal flux could be inferred. This measurement tech(cid:173)
`nique was only useful at low RFI densities. As the applied rf
`power exceeded about SOO W, the plasma density was high
`enough to penetrate the grid structure in the energy analyzer.
`It is estimated that this occurs at about 3 X 1011 ions/cm 3 for
`the 300 mesh grids used. Higher rf powers were monitored
`by means of optical emission spectroscopy (OES) from the
`plasma, observing both metal ion and neutral emission lines.
`
`III. RESULTS AND DISCUSSION
`
`A. Relative ionization
`
`The work of Holber et al. S used ECR ionization of metal
`atoms evaporated from a source which was not in the line(cid:173)
`of-sight of the sample. The observed level of ionization was
`essentially 100%. In the present experiment, the ionization
`fraction was expected to be less than 100% and a function of
`the operating parameters, such as the system pressure, the rf
`power to the RFI plasma, and the injection rate of metal
`atoms into the Ar RFI plasma. In Fig. 3, the relative ioniza(cid:173)
`tion for Al (from an AI cathode) as a function of chamber
`pressure at constant magnetron and RFl power is shown. At
`low pressure (a few mTorr), there is apparently little ioniza(cid:173)
`tion and the deposition flux is almost entirely composed of
`neutral AI. This case would be similar to conventional mag(cid:173)
`netron sputtering and would result in features much like Fig.
`1. As a function of increasing chamber pressure, the relative
`ionization rate climbs and saturates in the 30-40 mTorr
`range at ~80%. This is consistent with the reduction of the
`mean free path for the sputtered atoms from a distance ex-
`
` Redistribution subject to AVS license or copyright; see http://scitation.aip.org/termsconditions. Download to IP: 98.234.93.94 On: Thu, 18 Dec 2014 18:26:26
`
`TSMC et. al. v. Zond IPR2014-00800
`Page 3 Zond Ex. 2016
`
`

`

`451
`
`S. M. Rossnagel and J. Hopwood: Metal Ion deposition
`
`451
`
`1.0
`
`I
`
`250 W RF induc1ion
`
`~o--o
`
`1.0
`
`c: 0.8
`0
`
`~ u.. 0.6
`
`0.4 I-
`
`c:
`..Q
`E
`::::l c:
`'E
`::::l
`~ 0.2 t-
`
`(e)argon(qneon /o~_
`
`0/-
`
`-/-
`
`o
`
`0.0 1--_-'-__ -'-__ ' - -_ - - ' -_ - - '
`o
`40
`50
`20
`30
`10
`Pressure (mTorr)
`
`0.8
`c:
`.2
`~
`It
`0.6
`c:
`..Q
`E
`::::l c:
`'E
`::::l
`C(
`
`0.2
`
`--.
`----.
`/.----.
`/.
`
`• _ _ e
`
`0.4
`
`•
`
`/
`•
`
`Pressure: 36 mTorr
`(A) 1 kW magnetron
`(II) 2 kW magnetron
`(e) 3 kW magnetron
`
`Y
`0.0 ....... - - - ' - - - - ' - - - - ' - - - - - '
`o
`400
`100
`200
`300
`RF Induction Power (W)
`
`FIG. 3. Relative ionization level for sputtered AI as a function of chamber
`pressure for Ar and Ne working gases using a 20 cm diam planar AlCu
`cathode.
`
`FIG. 5. Relative ionization for sputtered AI as a function of increasing rf
`power and increased metal sputtering rate for a 20 cm dis. AlCu cathode.
`
`ceeding the cathode-to-sample (throw) distance at low pres(cid:173)
`sures to a path length of a few millimeters at the higher
`pressure. Slowing down the sputtered atoms results in a
`greater chance for ionization due to an increased cross
`section6 and a longer residence time in the plasma.
`A similar dependence is observed in the relative metal
`ionization as a function of applied rf power to the RFI
`plasma. At constant magnetron power (constant metal flux)
`and constant pressure (at the high end), the relative ioniza(cid:173)
`tion climbs and saturates as a function of rf power (Fig. 4).
`The saturation is again -80% for these conditions. Measure(cid:173)
`ments at the upper end of this range are somewhat question(cid:173)
`able, because at slightly higher powers the plasma penetrates
`the grids. This would indicate curvature in the sheaths near
`the grids at powers -500 W, which may result in significant
`defocusing of the ions and reduced collection.
`The effect of high rates of metal sputtering, and the sub(cid:173)
`sequent high fluxes of metal atoms entering the RFI plasma,
`is to cool the RFI plasma. This was observed with both the
`energy analyzer as well as OES. The results of increasing the
`
`1.0 r-----,-----,----..,.----,
`
`magnetron power, and hence the metal sputtering rate, is
`shown in Fig. 5, which shows the relative ionization as a
`function of rf power for several magnetron powers. Even
`though the depositing metal flux increases markedly, the
`relative number of ions decreases, consistent with a cooling
`of the electron temperature in the RFI plasma. Optical emis(cid:173)
`sion from the Aralso showed a similar effect (Fig. 6), al(cid:173)
`though some amount of this reduction may also be due to gas
`heating and rarefaction.7 Because of the metal-rich character
`of these plasmas, there was no attempt to make classical
`Langmuir probe measurements of the electron temperature.
`Previous measurements in magnetron plasmas have proven
`difficult due to the deposition of metal on the insulating parts
`of the probe.8
`By using DES and observing both metal neutral and metal
`ion emission, the relative ionization could be qualitatively
`extended to higher rf powers. This is shown in Fig. 7, in
`which the relative ionization inferred from DES is normal(cid:173)
`ized to a low power (200 W) measurement with the electro(cid:173)
`static energy analyzer. This indicates that the cooling effect
`of the metal ion fluxes can be countered somewhat by addi(cid:173)
`tional rf power to the RFI plasma.
`
`~ 150
`~
`~
`
`·f 100
`~
`< 50
`
`Magnetron: 2 kW
`Pressure: 38 mTorr
`(A) Neon
`(II) Argon
`
`I
`...
`100
`300
`200
`RF IndUction Power (W)
`
`400
`
`(.)913nm
`(e) 922 nm
`
`• , .".
`""'3 "'-. '.
`---·---IID -.
`.. -. --. --.
`-----.--._-. - e
`
`..............
`
`500 1000 1500 2000 2500 3000 3500
`DC Magnetron Power (W)
`
`FIG. 4. Relative ionization for sputtered AI at constant pressure as a function
`of rf power into the inductively coupled plasma. The cathode used was a 20
`cm diam AlCu planar cathode at 36 mTorr pressure.
`
`FIG. 6. Optical emission from Ar as a function of increased metal sputtering
`rate (dc magnetron power). Optical emission from the region above sample
`location, 10 crn from the cathode.
`
`JVST B • Microelectronics and Nanometer Structures
`
` Redistribution subject to AVS license or copyright; see http://scitation.aip.org/termsconditions. Download to IP: 98.234.93.94 On: Thu, 18 Dec 2014 18:26:26
`
`TSMC et. al. v. Zond IPR2014-00800
`Page 4 Zond Ex. 2016
`
`

`

`452
`
`results indicate that ~50 eV per sputtered metal atom is re(cid:173)
`quired to compensate for the effect of passing the metal atom
`through the RFI plasma. This is consistent with the minimum
`required energy per atom ohserved in the creation of ions
`with a Kaufman-type ion source.')
`
`B. Film deposition from ions
`
`The principal, expected advantages of the deposition of
`metal films from ions, rather than mainly neutrals, are the
`directionality of the deposition and the ability to reactively
`deposit compound films using the kinetic energy of the de(cid:173)
`positing ion to drive the reaction. A clear example of these
`two effects is in the deposition of TiN films as thin liners or
`diffusion barriers along the sides and bottom of high aspect
`ratio trenches. Thin films of TiN have many interesting char(cid:173)
`acteristics: they are hard, chemically inert, moder<ltely con(cid:173)
`ducting, and have an attractive gold color. Under conven(cid:173)
`tional sputter deposition conditions, the deposition of TiN
`requires either significant sample heating (300-700 0c) or
`measurable sample bias currents (milliamps per square cm)
`which require the use of unbalanced magnetrons. Collimated
`or filtered magnetron sputtering has been used successfully
`to deposit thin TiN films on the sides and bottom of trenches
`with good conformality.lO This process is enh<lnced by the
`much less than 1.0 sticking (or reaction) coefficient for the
`depositing film atoms, which allows some redistribution of
`the atoms within the trench, leading to better than expected
`coverage. However, collimated sputtering can be moderately
`slow and there <Ire concerns with particle formation on the
`collimator, which receives a large deposition flux.
`TiN films were deposited using the present experiment by
`reactively sputtering a Ti cathode in a combination of Ar and
`
`FIG. H. Cross section of high aspect ratio trench which has been lined with
`TiN at room temperature using ionized magnetron sputter deposition.
`
`Flti. 7. Relative ionization level for sputtered Al as a function of rf power tG
`the RFI plasma. The "'lid data points to the left-hand side arc from Pig .. :;
`using Ihe mass-sensilh c energy analyzer. The open data points were taken
`u,ing OES, comparing the intensity of an ionized Al line to the sum of the
`intensity of that ionized line and a nearby AI neutral emission line. The (laS
`data were then normali~cd to the energy analyzer data at 200 W.
`
`The high levels of metal ionization (50%-90%) should
`not be surprising given the nature of the experiment. The RFI
`plasma was generated using Ar, which has an ionization po(cid:173)
`tential of 15.7 eV The electron temperature at these pref;(cid:173)
`sures (30 mTorr) is relatively low (1-2 eV), but it is suffi(cid:173)
`cient to create a high density plasma (>lOii/cmJ). Tht~
`typical metal species used (Cu, Ti, Al) have ionization p~
`tentials in the 5-H e V range, although their cross sections at
`these energies are not well characterized. The introduction of
`a metal atom into the 1-2 eV plasma should result in B
`relatively rapid ionization of the metal atoms. Approximate
`calculations (because of the lack of well defined cross sec(cid:173)
`tions for ionization of the metal species) are consistent with
`the observation of high levels of ionization as the pressure is
`increased. The increase in pressure results in a longer resi(cid:173)
`dence time for the metal atom in the plasma. At low pres(cid:173)
`sures, the sputtered atoms transit the plasma rapidly at sev(cid:173)
`eral e V of kinetic energy.
`The observed cooling of the plasma, as implied by both
`the reduced relative ionization as well as the reduced OES
`signal from the inert gas, is also a fairly obvious result. The
`metal fluxes into these plasmas can be rather significant,
`even at modest magnetron sputtering rates. For example, for
`the sputtering of Cu, the sputter yield is ~ 2.0. At a magne(cid:173)
`iron discharge power of 2 kW, the typical discharge current
`might be 5 A. This results in a net emission rate from the
`cathode of 6X 1014 atoms/s, or the equivalent of overHlO
`sccm. This is somewhat misleading, though, because the
`pumping speed is effectively infinite for these metal atoms.
`However, since the majority of the metal atoms can be ob(cid:173)
`served to be ionized, it can be assumed that they are also
`rapidly absorbing energy and emitting photons. This effect
`can be qualitatively addressed by measuring the amount of
`additional rf power needed to recover the same ion saturation
`current to the sample as the metal flux is increased. The
`
`J. Vac. Sci. Technol. B, Vol. 12, No.1, Jan/Feb 1994
`
` Redistribution subject to AVS license or copyright; see http://scitation.aip.org/termsconditions. Download to IP: 98.234.93.94 On: Thu, 18 Dec 2014 18:26:26
`
`TSMC et. al. v. Zond IPR2014-00800
`Page 5 Zond Ex. 2016
`
`

`

`453
`
`S. M. Rossnagel and J. Hopwood: Metal Ion deposition
`
`453
`
`N2 , with the cathode operated in the nitride mode. The films
`were deposited on Si wafers clamped to a water-cooled
`stage. It is estimated that the wafer temperature does not
`exceed 75°C during the deposition. The energy of the de(cid:173)
`positing ions was varied from 0 to 100 e V. At the lower end
`(0-10 e V), the films were bronze in color with resistivities in
`the 200 pI! cm range. Films from 20 to 50 eV were a
`brighter, yellow gold, with resistivities in the 75 pI! cm
`range, and films at higher ion energies were roughly equiva(cid:173)
`lent, but were characteristic of highly stressed films. Peeling
`was common at thicknesses over 700 A, and depositions on
`circuit topography features delaminated upon cleaving.
`Films deposited at 20-40 eV onto high aspect ratio trench
`structures (down to 1500 A width) appeared to be conformal
`to aspect ratios of ~5-6, with thicknesses at the bottom of
`the trench about 1/2 the thickness of the deposition on the
`flat area between trenches (Fig. 8).
`
`IV. SUMMARY
`A technique has been developed and demonstrated for the
`high rate deposition of metal and compound films from
`metal-rich plasmas fed by means of sputtering. High levels
`of metal ionization have been observed, and the dependence
`of the relative ionization of chamber pressure, sputtering
`rate, and RFI power are qualitatively understood.
`The primary semiconductor applications for this technol(cid:173)
`ogy are the deposition of high conductivity metal layers into
`high aspect ratio vias and trenches and the lining of these
`
`features by the reactive deposition of nitrides. The ability, in
`the latter case, to deposit high quality films at room tempera(cid:173)
`ture is also attractive, not only for the semiconductor appli(cid:173)
`cations, but also for hard or decorative coatings on fragile
`substrates. In addition, this technology allows new work in
`the field of energetic deposition of a variety of materials.
`Previously, with the exception of filtered arc deposition, the
`direct control of the energy of the depositing particles was
`not possible in any large-scale deposition technology. It is
`expected that control of the deposition energy in the sub- or
`low-sputtering regime will lead to interesting insight on the
`materials properties (stress, defect structure, adhesion, etc.)
`of metal and compound films. Preliminary results with
`highly stressed AlCu films show unique stress-relief features
`which differ from most other sputter deposited films.
`
`lB. Windows and N. Savvides. J. Vac. Sci. Technol. A 4, 196 (1986).
`lw. D. Sproul, P. J. Rudnick, M. E. Graham, and S. L. Rohde, Surf.
`Coatings Technol. 43/44, 270 (1990).
`3M. S. Barnes, J. C. Forster, and J. H. Keller, U.S. Patent No.5, 178,739,
`Jan. 12, 1993.
`4S. M. Rossnagel and J. Hopwood, Appl. Phys. Lett. 63, 3285 (19931.
`5w. Holber, J. S. Logan. H. 1. Grabarz, J. T. C. Yeh, 1. B. O. Caughman, A.
`Sugarman, and F. E. Turene, J. Vac. Sci. Technol. 11,2903 (19931.
`hR. S. Robinson, J. Vac. Sci. Technol. 16. 185 (19791.
`7S. M. Rossnagel, 1. Vac. Sci. Technol. A 6, 11) (1988).
`RS. M. Rossnagel and H. R. Kaufman, J. Vac. Sci. Technol. A 4, 1982
`(19861.
`"H. R. Kaufman (private communication).
`lOR. Joshi, 1. 1. Cuomo, G. W. Dibello, and S. M. Rossnagel, IBM Tech.
`DiscI. Bull. 35, 456 (1992).
`
`JVST B • Microelectronics and Nanometer Structures
`
` Redistribution subject to AVS license or copyright; see http://scitation.aip.org/termsconditions. Download to IP: 98.234.93.94 On: Thu, 18 Dec 2014 18:26:26
`
`TSMC et. al. v. Zond IPR2014-00800
`Page 6 Zond Ex. 2016
`
`