`
`Elm Exhibit 2160
`Samsung, Micron, SK hynix v. Elm
`IPR2016-00387
`
`
`
`P~OROCEEDINGS OF THE SYMPOSIUM ON
`
`SILICON NITRIDE AND SILICON
`
`
`DIOXIDE THIN INSULATING FILM~~ .....
`
`Editors
`
`M. Jamal Deen
`Simon Fraser University
`Burnaby, BC, Canada
`
`William D. Brown
`University of Arkansas
`Fayetteville, Arkansas, USA
`
`Kalpathy B. Sundaram
`University of Central Florida
`Orlando, Florida, USA
`
`Stanley I. Raider
`IBM T. J. Watson Research Center
`Yorktown Heights, New York, USA
`
`DIELECTRIC SCIENCE AND TECHNOLOGY
`
` AND ELECTRONICS
`
` DIVISIONS
`
`Proceedings Volume 97-10
`
`THE ELECTROCHEMICAL SOCIETY, INC.,
`10 South Main St., Pennlngton,
` NJ 08534-2896
`
`Elm Exhibit 2160, Page 2
`
`
`
`Copyright
`
` 1997 by The Electrochemical
`All rights reserved.
`
`
` Society, Inc.
`
` Inc.
` Center,
` Clearance
` with Copyright
`This book has been registered
`For further
` information,
` please contact
` the Copyright
` Clearance
` Center,
`Salem, Massachusetts.
`
`
`
`
`Published
`
` by:
`
` Society, Inc.
`The Electrochemical
`10 South Main Street
`Pennington,
` New Jersey 08534-2896,
`
` USA
`
` (609) 737-1902
`Telephone
`Fax (609) 737-2743
`
`e-mail: ecs@electrochem.org
`Web site: http:Hwww.electrochem.org
`
`
`
`ISBN 1-56677-137-4
`
`
`
`Printed in the United States of America
`
`Elm Exhibit 2160, Page 3
`
`
`
`STRESS AND BONDING CHARACTERIZATION
`SILICON DIOXIDE
` FILMS
`
`
`
` OF PECVD
`
` and W. D. Brown
` M. S. Haque,
`H. A. Naseem,
` Engineering,
`High Density Electronics
` Center
` (HiDEC),
` Department
` of Electrical
`University
` of Arkansas,
` Fayetteville,
` Arkansas
` 72701
`
`
`
`The presence of hydrogen-related impurities and the resulting stress instability
`of cheraical vapor deposited (CVD) silicon dioxide films is an important issue
`in microelectronics. By observing the stress behavior and bonding nature of
`oxide films simultaneously, insight is provided into the possible physical
`causes of stress. The bonding nature and stress behavior of relatively low-
`temperature, high-rate deposited silicon dioxide films were investigated.
`Depending on the type and concentration of impurities, both reversible and
`irreversible bond reconstruction were observed upon annealing such films.
`Concomitantly, both reversible and irreversible changes in stress were
`observed in the annealed films.. The reaction of strained Si-O bonds with
`moisture and a corresponding near-neighbor Si-OH formation, along with
`hydrogen-bonded moisture, were found to be primarily responsible for stress
`instability in these films. Moisture was found to play an important role in Si-
`O bond strain relief. Annealing the films in a steam ambient or capping the
`film with a thin moisture barrier material are shown to improve stress stability
`considerably.
`
`INTRODUCTION
`
`
`
`Stress in silicon dioxide films is an important issue in microelectronics and becomes
`more so as the wafer size continues to increase. Oxide films with moderate compressive stress
`are desirable to partially compensate tensile stress in the metal interconnects, thus avoiding
`film cracking. Different types of impurities incorporated during deposition affect the stress
`behavior of these films. R is important, therefore, to identify incorporated impurities and to
`understand their impact on film stress, especially for low-temperature, high-rate deposited
`silicon dioxide films. Excessive stress can cause cracking or delamination of the dielectric
`film and the formation of voids and notches in metal interconnects [1]. To prevent diffusion
`of shallow junctions, interdiffusion of metals in multi-level metailization systems, and hillock
`formation on aluminum metal lines, there has been a continuous move toward low
`temperature processes for dielectric deposition [2].
`
`Even though PECVD produces films at reasonably low temperatures, the incorporation
`of impurities may make the film properties less than desirable. The most common impurities
`incorporated into silicon dioxide are Si-H and silanol (Si-OH). By choosing appropriate
`
`Electrochemical
`
` Proceedings Volume 97-10
`
`217
`
`Elm Exhibit 2160, Page 4
`
`
`
`deposition conditions, PECVD films can be deposited with compressive stress values varying
`over a wide range [3]. However, the stability of film stress during the several thermal
`excursions normally associated with a fabrication process, as well as during aging is very
`important. The stability of different types of impurities during film aging or annealing needs
`to be fully investigated to gain a better understanding of their effects on stress behavior. This
`will help in the development of a clear cause and effect relationship between the two, which
`may point to ways for more precisely controlling stress in these films.
`
`The effects of annealing and aging on the behavior of CVD oxide films have been
`studied since the 1970s [4,5]. However, the effects of aging and the resulting changes in
`chemical bonding in CVD oxide films have only recently been studied in detail [6]. Even
`though moisture has been reported to be responsible for changes in film stress during aging,
`no detailed study has been reported so far which correlates the change in stress with changes
`in the bonding nature and impurity content of the films [5].
`
`The stress behavior of PECVD oxides vary during annealing and/or aging depending
`on the impurity content in the material. In the work reported here, concurrent FTIR and stress
`studies on low-temperature, high-rated deposited silicon dioxide films were performed in
`order to establish a cause and effect relationship between the two. From the FTIR and stress
`study results, the impurities involved, their impact on stress, and their chemical bonding
`stability during annealing and/or aging were identified. In light of the results of this bonding
`and stress behavior study, different techniques for improving the stability of low-temperature,
`high-rate deposited silicon dioxide films were determined.
`
`EXPERIMENTAL
`
`
`
`A Plasma-Therm model Shuttle-Lock SLR 700 PECVD system was used for all film
`depositions. In this system, the reactant gases are delivered through a showerhead type
`powered electrode. The deposition parameters used are shown in Table 1. P-type, <100>,
`125 mm diameter, 650 ~tm thick silicon wafers were used as substrates for all films on which
`stress measurements were performed. For the FTIR study, all the oxide films were 1 ttm thick
`and were deposited on substrates cut from 75 mm diameter, 360 ~tm thick, double-side
`polished, lightly-doped, p-type silicon wafers. For the specified deposition parameters, the
`deposition rate was about 1000 A/min and changed negligibly with deposition temperature.
`The films were annealed at temperatures in the range of 250-400°C. Unless otherwise noted,
`all annealing was performed in a Labline Instruments oven in a nitrogen ambient at
`atmospheric pressure for 30 minutes.
`
`A Nanospec model CTS 102 system was used to measure film thickness. A Mattsons
`Research Series FTIR system with a resolution of 4 cm"t in the mid-IR range (400-4000 cm-1)
`was used to study the bonding characteristics and impurity content of the material. This
`system uses a deuterated triglycine sulphate (DTGS) detector operating at room temperature
`
`Electrochemical
`
` Proceedings Volume 97-10
`
`218
`
`Elm Exhibit 2160, Page 5
`
`
`
`for data acquisition. The FTIR cell i~ continuously purged with dry N2 gas at a flow rate of
`1 l/rain through a Balston FTIR purge gas generator. A bare silicon substrate was scanned to
`obtain a reference spectrum. Fifty scans were used for both the sample and the reference
`spectra. A Tencor model FLX 2320A stress measurement system employing a dual
`wavelength laser was used to measure the radius of curvature of the wafers before and after
`film deposition. Using the change in radius of curvature, Young’s modulus and Poisson’s
`ratio for silicon and the silicon dioxide film thickness, the stress in the films was calculated
`using the modified Stoney’s equation [7]. The deposited and annealed films were stored in
`a class 100 cleartrcom environment (65°F, 45% RH). The stress behavior, chemical bonding,
`and impurity concentrations were then monitored periodically.
`
`RESULTS
`
` AND DISCUSSION
`
`
`
`
`
`
`Stress Study Resnlts
`Figure 1 shows plots of average stress values of oxide films measured immediately
`after deposition
` as a function of film thickness
` for three different
` deposition
` temperatures.
`
`Film stress is observed to decrease with increasing film thickness for all deposition
`
`temperatures
` with the magnitude of the decrease varying inversely with deposition
`temperature. Ideally,
` the film stress should be independent
` of film thickness. The observed
`thickness dependence
` is due to the surface reactivity of the deposited films. That is, the film
`surface behaves differently from the bulk portion of the film. In the stress equation,
` the total
`film thickness is used to extract
` stress values from the measured wafer bow. Thus, the impact
`of the surface effect
` which,
` by definition,
` occurs only at the surface,
` will be less significant
`for thicker
` films than for thinner
` ones. This surface reactivity effect is strongest
` in low-
`temperature
` deposited films, and thus, they exhibit
` the largest
` apparent
` change in stress with
`increasing
` film thickness. For high-temperature
` deposited films, which have a relatively
`denser surface structure,
` the surface effect
` is less dominant
` [8].
`
`
`
`
`
`Figure 2 shows the change in stress (stress value after annealing minus the as-
`deposited stress value) for 1 ttm thick films as a function of annealing temperature. It is
`observed that annealing makes the film stress more tensile. It is further observed that the
`change in stress due to annealing is largest for 250°C deposited films and decreases with
`increasing deposition temperature. Furthermore, the magnitude of the change in stress
`increases with increasing annealing temperature.
`
`Figure 3 shows the total change in stress upon aging (stress value following two
`months of storage after annealing minus stress value immediately after annealing) of 1 ~tm
`thick annealed films as a function of anneal temperature. For comparison purposes, data for
`unannealed samples (stress value at~ two months of storage minus stress value immediately
`aRer deposition) are also shown. For all films, it is observed that film stress becomes more
`compressive after two months of storage. However, the change is essentially independent of
`anneal temperature. It is also observed, for both annealed and as-deposited films, that the
`
`Electrochemical
`
` Proceedings Volume 97-10
`
`219
`
`Elm Exhibit 2160, Page 6
`
`
`
`change is smallest for 350°C deposited films and increases with decreasing deposition
`temperature.
`
`Figures 4(a)-4(c) show the variation in stress with storage time of as-deposited, 250°C
`annealed, and 400°C annealed 1 I~m thick films deposited at three temperatures. For
`unannealed films, the change in stress during storage is highest for the lowest deposition
`temperature and decreases with increasing deposition temperature. Assuming that the change
`in stress during storage is solely due to absorbed moisture, diffusion coefficients for moisture
`were calculated for the unannealed films using the technique suggested by McInerney and
`Flinn [9]. Theoretical diffusion equation curves were fitted to the data points in Fig. 4(a) and
`the diffusion coefficient value for the best fit was extracted.
`
`For all films, a rapid increase in stress values (becoming more compressive) is
`observed during the first few hours of aging with the rate being dependent on film thickness
`and deposition temperature. Since the surface reacts vigorously with the ambient, the stress
`behavior during the first few hours is determined primarily by the surface effect. Also, the
`lower the deposition temperature and/or the smaller the thickness, the higher the rate of
`change in stress during the first few hours and the higher the deviation from the diffusion
`equation fit for the initial portion of the stress versus time plot.
`
`In Fig. 2, the change in stress (during annealing) as a function of anneal temperature
`is largest for films deposited at the lowest temperature. Also, for 250°C deposited films, the
`change is almost independent of anneal temperature, whereas, for 300 and 350°C deposited
`films, the change is larger for higher annealing temperatures. In Fig. 3, however, the change
`in stress caused by moisture absorption from the ambient is essentially independent of anneal
`temperature for all three deposition temperatures. Also, the change in stress during annealing
`is much higher than during aging for high temperature deposited films which were annealed
`at higher than 350°C. Thus, a reversible stress behavior for low-temperature deposited,
`annealed films is observed irrespective of anneal temperature, whereas, an irreversible stress
`behavior for high-temperature deposited films annealed at temperatures higher than 350°C.
`The cause of this stress behavior will become clearer in the next section where results on the
`behavior of different impurities in these films during annealing and aging are discussed.
`
`FTIR Study Results
`Figure 5 shows FTIR transmittance spectra of unannealed silicon dioxide films
`deposited at three temperatures. Several impurity-related peaks are observed, along with
`characteristic Si-O peaks (stretching bending and rocking modes at approximately 1070, 810,
`and 450 cm1, respectively). Multiple (and broad) peaks observed in the 3300-3700 c~l
`range have been linked to silanols (Si-OH) and hydrogen-bonded moisture [10]. The peaks
`at 880 and 2270 cmt are referred to as Si-H bending and stretching modes, respectively [11].
`It should be noted that the Si-OH related impurity band in the 3300-3700 cm"t range has the
`highest intensity for films deposited at 250°C and that the intensity decreases with increasing
`deposition temperature. On the other hand, the Si-H related peaks observed at 880 and 2270
`
`Electrochemical
`
` Proceedings Volume 97-10
`
`220
`
`Elm Exhibit 2160, Page 7
`
`
`
`cm"~ have their highest intensity for films deposited at 350°C and decrease with decreasing
`deposition temperature.
`
`A significant difference in the Si-O stretching peak position and FWHM (determined
`from absorbance spectra) of as-deposited films was observed for film deposited at the three
`different temperatures. It is clearly seen that the higher the deposition temperature, the lower
`the peak position and the larger the FWHM. In fact, lower Si-O stretching peak positions,
`larger FWHMs, and higher Si-H bond densities in films deposited at higher temperatures are
`all linked to silicon richness of the film. These conclusions were verified by results from
`direct measurement techniques, such as XPS, and by refractive index values [12]. It should
`be noted that the stretching and bending vibration peaks of Si-H groups around 2270 and 880
`cm"~ confirms that hydrogen is present only as HSiO3 [11]. Thus, the films studied in this
`work are slightly silicon-rich, near-stoichiometric silicon dioxide, especially ones deposited
`at the higher temperatures.
`
`The presence of strained bonds and impurities also cause PECVD silicon dioxide film
`properties to change with aging. The effects of aging on the characteristic Si-O and impurity
`peaks of these films were studied. With increasing aging time, the Si-O stretching peak shifts
`to higher wavenumbers and its FWHM decreases for films deposited at all three temperatures.
`Both the peak position shift and the FWHM decrease are largest for films deposited at 250°C.
`
`Figures 6(a) and 6(b) compare the silanol bands in unannealed films deposited at 250
`and 350°C, respectively, with aging time as a parameter. With increasing storage time, the
`Si-OH peak (3300-3700 cm"1) broadens to the lower wavenumber side and its intensity
`increases. It is clearly observed that the lower the deposition temperature, the wider and more
`intense the Si-OH band becomes with increasing aging time. Similar Si-OH bond broadening
`with aging has been reported by Pliskin [10] and Theii et al. [6]. This broad band is due to
`the fact that Si-OH can exist in different forms. The strained Si-O bonds react with moisture
`and form near-neighbor Si-OH. The near-neighbor silanol pairs are hydrogen-bonded to each
`other. On the other hand, the hydrogen bonding interaction between an OH group and a
`bridging O from a nearby Si-O-Si group is much smaller. The isolated silanols, incorporated
`from the deposition chemistry, are too far apart to undergo hydrogen bonding interaction.
`However, depending on the hydrogen bonding interaction, the characteristics of the vibration
`modes of the Si-OH groups will changed and the vibration energy will vary accordingly.
`Moisture can also hydrogen bond to existing OH in the film. It is important to note that
`surface-adsorbed and hydrogen-bonded moisture are primarily responsible for the surface
`effect.
`
`Thus, a considerable shift in the Si-O characteristic peak position, a decrease in peak
`FWHM, and a concomitant increase in the near-neighbor Si-OH bond concentration is
`observed with increasing aging time. The Si-O peak shift is related to near-neighbor Si-OH
`formation during aging because the strained Si-O-Si bonds react with atmospheric moisture
`and convert to near-neighbor Si-OH pairs [6]. The low wavenumber side of the Si-O
`
`Electrochemical
`
` Proceedings Volume 97-10
`
`221
`
`Elm Exhibit 2160, Page 8
`
`
`
`stretching peak is associated with strained bonds, so, as more and more strained Si-O bonds
`are converted to near-neighbor Si-OH, the FWHM decreases and the Si-O peak moves to
`higher wavenumbers. A relatively larger decrease in FWHM and shift in the Si-O stretching
`peak correlates with higher near-neighbor silanol formation in low-temperature deposited
`films.
`
`Annealing and subsequent aging effects on the Si-O characteristic peak position and
`the FWHM of the Si-O stretching peak were also studied. FTIR spectra of films were taken
`within an hour of deposition/annealing. Subsequent FTIR spectra of these films were then
`taken every few days during the aging process. Comparing the before and after annealing
`values, it was observed that annealing at any temperature causes a considerable shift in the Si-
`O peak position and a decrease in the FWHM. During aging of annealed films, a further shift
`in Si-O peak position and decrease in the FWHM were observed. For low-temperature
`deposited films, the shift in the Si-O peak and the decrease in the FWHM were the largest,
`although the lower the deposition temperature, the less dependent these parameters were on
`the anneal temperature. However, for higher annealing temperatures, the change in these
`parameters during aging were somewhat smaller.
`
`A considerable reduction in the intensity of the near-neighbor Si-OH band with
`annealing was observed for films deposited at all three temperatures. During annealing of the
`films, near-neighbor Si-OH is converted to Si-O with water as a by-product. It appears likely
`
`that the newly formed Si-O bonds return to their previously strained bonding configuration,
`although the decrease in FWHM and shift of the Si-O stretching peak to a higher wavenumber
`with annealing confirm that bond strain is actually relieved. It should also be noted that, with
`increasing aging time, some of the remaining strained bonds react with moisture and give rise
`to a small increase in near-neighbor Si-OH. Furthermore, films deposited at the lower
`temperature show a larger near-neighbor Si-OH buildup than do films annealed at higher
`temperatures. A considerable reduction in Si-H and isolated Si-OH peak intensity was
`observed for annealing at and above 350°C. However, once these impurities are annealed out,
`unlike near-neighbor Si-OH, they cannot form again in annealed films during aging~ Their
`concentration in unannealed films also does not change during aging.
`
`These FTIR results suggest that both reversible and irreversible bond reconstruction
`phenomena occur in PECVD silicon dioxide films. During annealing, the films lose Si-OH
`(near-neighbor and isolated) and/or Si-H bonds depending on the annealing and deposition
`temperatures. During aging, the films primarily become repopulated by near-neighbor Si-OH
`which occurs through the reaction of strained Si-O bonds with adsorbed atmospheric moisture.
`This effect is observed to be smaller for higher deposition and/or anneal temperatures. From
`these stress results, it can be concluded that the low-temperature deposited films exhibit
`reversible stress behavior, whereas, high-temperature deposited films show irreversible stress
`behavior during annealing and subsequent aging.
`
`The observed film stress behavior can be explained by considering four different types
`
`Electrochemical
`
` Proceedings Volume 97-10
`
`222
`
`Elm Exhibit 2160, Page 9
`
`
`
`of traps, namely 1 ) surface-related, reversible traps (active below 125 ° C), 2) low temperature
`active, reversible, near-neighbor silanol-reiated traps, 3) high temperature active, irreversible,
`isolated silanol traps, and 4) Si-H related traps. Si-H and isolated silanol-related traps are
`generated during film deposition. Thus, the measured stress values of the as-deposited films
`consist of contributions from all four types of traps. The hydrogen-bonded moisture trap
`formation saturates after a few hours, whereas, near-neighbor Si-OH trap formation can
`continue over months, and plays the most important role during long term aging. During
`annealing~ depending on the annealing temperature, both irreversible (isolated Si-OH and Si-
`ll related) and reversible (hydrogen-bonded water and near-neighbor Si-OH related) traps
`may be emptied. Emptying of these traps makes the film stress less compressive (or more
`tensile). The presence and density of these different types of traps depends on the deposition
`temperature. Reversible traps can be emptied by annealing at lower temperatures, whereas,
`emptying of irreversible traps requires higher annealing temperatures. Thus, the change in
`stress caused by annealing depends on both the film deposition and annealing temperatures.
`However, during aging of annealed films, only the reversible traps reappear and cause the film
`stress to become more compressive with time.
`
`Effect
`
`
` Relief
` on Si-O Bond Strain
` of Moisture
`From the results of the chemical
` bonding and stress change study, it was determined
`that annealing
` of low-temperature
` deposited PECVD silicon dioxide films at atmospheric
`pressure in a nitrogen ambient
` improves their stress stability during subsequent
` aging. In fact,
`films annealed at temperatures
` even lower than the deposition
` temperature
` (at and above
`250°C)
` show better stability during subsequent
` aging after annealing than as-deposited
` films.
`
`To gain a better understanding
` of the mechanism(s)
`
` involved in the improvement
` of film
`stability due to low temperature
` annealing,
` the effects of annealing on films under different
`ambient
` and at different
` stages of aging were studied. The results of the study suggest
` that
`the reversible reaction of moisture with strained Si-O bonds is responsible
` for the
`improvement
`
` in film stability. Thus, moisture,
` present
` in the annealing ambient,
` adsorbed on
`the film surface,
` and diffused into the film, plays a very important
` role in the relief
` of Si-O
`bond strain during low temperature
` annealing.
`
`
`
`
`
`
`
`Two sets of experiments were performed to confirm the role of moisture in silicon
`dioxide film stability. The firs~ set involved the annealing of samples for thirty minutes in the
`deposition chamber. For some samples, the samples were lef~ in the chamber at the deposition
`temperature and pressure in flowing N2 gas for thirty minutes. Other samples were removed
`from the chamber immediately after the deposition and then re-loaded after two hours of
`aging in room ambient. They were then annealed in the PECVD chamber under conditions
`identical to those of the first set of samples.
`
`The second set of experiments involved the annealing of samples at atmospheric
`pressure at the deposition temperature in an oven in either a steam/nitrogen or a nitrogen
`ambient for thirty minutes. For the nitrogen/steam ambient case, nitrogen, bubbled through
`a quartz flask containing boiling water, was introduced into the oven. Other samples,
`
`Electrochemical
`
` Proceedings Volume 97-10
`
`223
`
`Elm Exhibit 2160, Page 10
`
`
`
`deposited during the same run, were first aged in room ambient conditions (65°F, 45% RH)
`for various lengths of time and then annealed at atmospheric pressure in a pure nitrogen
`ambient for thirty minutes. FTIR spectra of the samples were taken before and after annealing
`and at different stages of aging. Table 2 shows the characteristic peak positions and FWHMs
`of the Si-O stretching peak of samples annealed at 250°C under different ambient. For
`comparison, the characteristic peak position and FWHM of a thermally-grown, ideal silicon
`dioxide film are approximately 1085 and 80 cm"1, respectively. All the PECVD as-deposited
`samples (sample A) have a considerable amount of bond strain and exhibit a lower peak
`position and higher FWHM. With suitable activation, however, these values tend to approach
`the ideal values of thermally-grown silicon dioxide. The relative change in these values with
`annealing and/or increasing aging time is an indication of the film stability - the smaller the
`change, the better the stability. It should be noted that silicon richness causes an effect similar
`to bond strain. However, previous XPS, FTIR, and refractive index studies indicated that
`films deposited at 250°C in this study were minimally silicon rich. So, all films discussed in
`this section were deposited at 250°C.
`
`For samples annealed in the deposition chamber before breaking vacuum (sample B),
`it is observed that the peak positions and FWHM values are essentially identical to those of
`as-deposited samples (sample A), whereas, films which were aged for 2 hours in room
`ambient before being annealed under the same conditions (sample C) exhibited a considerable
`change in these parameters. However, as indicated in Table 2, during two hours of aging, the
`FWHM and the Si-O characteristic peak positions moved only slightly, indicating some small
`improvement
` in bond strain even before annealing. The improvement in film quality is
`greater for samples annealed in a nitrogen ambient at atmospheric pressure (sample D).
`However, the greatest improvement in stability due to annealing is observed for samples
`annealed in a steam environment at atmospheric pressure (sample E). The effects of an
`oxygen annealing ambient was also investigated. The samples were annealed in the
`deposition chamber in oxygen gas at the deposition pressure without breaking vacuum
`(sample F). Not much change in the parameters was observed as seen in Table 2. Samples
`deposited at 350°C were also annealed under different ambient and similar results were
`obtained. However, these samples were annealed at 350°C instead of 250°C.
`
`It is worth mentioning that the samples which were annealed in the chamber
`immediately after deposition were further annealed in an oven at atmospheric pressure in
`flowing N2 for half an hour. A considerable shift in the Si-O characteristic peak and a
`decrease in the FWHM of the Si-O stretching peak were observed as a result of this anneal.
`The shift and the decrease were identical to those observed for as-deposited samples annealed
`under the same conditions. These data strongly suggest that moisture present in the annealing
`ambient and/or adsorbed by the silicon dioxide films after they are exposed to room ambient
`plays an important role in the improvement of film stability with annealing.
`
`Si-OH formation during aging of films annealed in different ambient was also studied.
`It was observed that annealing at 250°C reduces the near-neighbor Si-OH broad band
`
`Electrochemical
`
` Proceedings Volume ~97-10
`
`224
`
`Elm Exhibit 2160, Page 11
`
`
`
`(centered at 3350 cm-l), whereas, the isolated Si-OH band at 3650 cul1 was not affected.
`Ftathermore, annealing in steam did not increase the magnitude of the near-neighbor Si-OH
`band but, in fact, resulted in the largest decrease compared to annealing in other ambient.
`During aging, samples annealed in steam (sample E) exhibited the smallest near-neighbor
`silanol formation and the samples annealed in nitrogen in the chamber immediately after
`deposition (sample B) exhibited the largest Si-OH formation. Since the formation of Si-OH
`during aging depends on the concentration of strained Si-O bonds present in the films, the best
`and worst film stability is observed for samples E and B, respectively. As seen in Table 2,
`annealing in steam produces the largest change in Si-O peak position and FWHM, and moves
`these values closer to those of thermal oxide, whereas, annealing without moisture present
`produces no change. Fu~ermore,
` during aging of these samples, the smallest change in Si-O
`peak position and FWHM was observed for samples annealed in steam, whereas, samples
`annealed under vacuum exhibited no improvement. These results clearly indicate that the
`largest improvement in film stability is achieved by annealing in steam. The bond strain
`relieving phenomenon during annealing in steam can be explained as follows. The strained
`Si-O bonds can react more efficiently with moisture at elevated temperatures in a moisture-
`saturated ambient and thus, most of the strained bonds form near-neighbor Si-OH. However,