three-dimensional atom-probe tomographic studies of...

Three-Dimensional Atom-Probe Tomographic Studies of Nickel Monosilicide/Silicon Interfaces on a Subnanometer Scale

Praneet Adusumillia, Conal E. Murrayb, Lincoln J. Lauhona, Ori Avayuc, Yossi

Rosenwaksc, David N. Seidmana,d*

aDepartment of Materials Science and Engineering, Northwestern University, 2220 Campus Drive, Evanston, Illinois 60208-3108, USA

bIBM T. J. Watson Research Center, Yorktown Heights, New York 10598, USA cSchool of Electrical Engineering, Tel-Aviv University, Tel-Aviv 69978, Israel

dNorthwestern University Center for Atom-Probe Tomography, Evanston, IL 60208-3108 USA

*Email: [email protected]

Three-dimensional atom-probe tomography was utilized to study the distribution of M (M = Pt or Pd) after silicidation of a solid-solution Ni0.95M0.05 thin-film on Si(100). Both Pt and Pd segregate at the (Ni1-xMx)Si/Si(100) heterophase interface and may be responsible for the increased resistance of (Ni1-xMx)Si to agglomeration at elevated temperatures. Direct evidence of Pt short-circuit diffusion via grain boundaries, Harrison regime-B, is found after silicidation to form (Ni0.99Pt0.01)Si. This underscores the importance of interfacial phenomena in stabilizing this low-resistivity phase, providing insights into the modification of NiSi texture, grain size, and morphology caused by Pt. The relative shift in work function between as-deposited and annealed states is greater for Ni(Pt)Si than for NiSi as determined by Kelvin probe force-microscopy. The nickel monosilicide/Si heterophase interface is reconstructed in three-dimensions and its chemical roughness is evaluated.

I. Introduction As complementary metal-oxide-semiconductor (CMOS) device dimensions decrease with each technology generation, new silicide source/drain contacts are developed to retain a low-resistivity silicide phase to meet the demands associated with scaling requirements [1-2]. An important aspect of the development of metallic contacts is materials chemical analyses at a continuously decreasing length scale. The use of x-ray diffraction is difficult because of the limited volume of material available and the compositional analysis capability of analytical transmission electron microscopy is limited to dopant concentration levels of approximately 1 at.% [3]. The recent advent of highly sophisticated commercial three-dimensional (3-D) atom-probe tomographs (APTs), using femto- or picosecond lasers to dissect semiconducting specimens, essentially one atom at a time, has dramatically enhanced our ability to determine subnanoscale chemical compositions [4-6] of all components that comprise an individual field-effect transistor. An important contributing factor to APT analysis is the ability to prepare routinely 3-D APT specimens with the requisite dimensions and shape, which is

ECS Transactions, 19 (1) 303-314 (2009)10.1149/1.3118957 © The Electrochemical Society

303Downloaded 22 Jul 2009 to 129.105.37.231. Redistribution subject to ECS license or copyright; see http://www.ecsdl.org/terms_use.jsp

accomplished employing dual-beam focused-ion-beam (FIB) microscope based methods [7-8]. The atoms that are dissected from an APT specimen are mass-analyzed by a novel combination of time-of-flight mass spectrometer and 2-D detector, consisting of a multi-channel plate (MCP) and a delay-line detector in series. This configuration permits the positions of individual atoms to be reconstructed in 3-D, thereby yielding both the positions of the atoms as well as their chemical identities. High precision characterization using APT gives us an opportunity to study not only the phase formation and evolution but also 3-D dopant distributions, the chemical interface roughness, and inter-diffusion mechanisms, which are all relevant to the fabrication of low-resistivity contacts.

In recent years, NiSi has become the material of choice for contacting to Si-based

devices. NiSi is preferred over previously used Ti- and Co-silicides due to a lower temperature of formation, a lower resistivity in narrow dimensions, reduced Si consumption, and formation kinetics that lead to a significantly smoother interface with Si [9]. The most challenging issue in integrating this low-resistivity material is its thermal stability. At elevated temperatures (ca. 700 °C) NiSi degrades either via agglomeration or via phase transformation to the higher resistivity NiSi2 phase. Addition of a transition metal, Pt, Pd or Rh, has been shown to reduce the agglomeration of thin NiSi films and increase the formation temperature of NiSi2 [10], and it has been suggested that changes in the bulk free energy due to entropy of mixing can explain this effect [11-12]. The distribution of these transition metal additions and their role in phase stabilization is, however, not well understood. We are currently studying, utilizing 3-D APT, silicidation reactions between Ni0.95Pd0.05 or Ni0.95Pt0.05 and Si(100) that are initiated by rapid thermal annealing (RTA). We herein report on the 3-D atomic-scale distributions of Pt or Pd atoms in nickel monosilicide thin films on Si(100) using local-electrode atom-probe (LEAP®) tomography. Additionally we report the chemical roughness of the nickel monosilicide/silicon heterophase interface along with work function measurements of the silicide’s top surface using Kelvin probe force-microscopy (KPFM).

II. Specimen Preparation

For Pd-doped silicide specimens, 10 nm Ni0.95Pd0.05 thin-films were deposited by RF magnetron sputtering on p-type Si(100) substrates at a base pressure of 6.6x10-5 Pa after removing the native oxide using a 2% buffered HF solution. The films were subjected to RTA for 30 seconds at ca. 600 °C to form a nickel monosilicide phase. The unreacted NiPd alloy was removed selectively using a mixture of H2SO4 and H2O2 subsequently followed by annealing at 650 °C for 30 min in a N2 ambient. The latter step was performed to approximate the effects of subsequent thermal treatments experienced by nickel monosilicide contacts in the back-end-of-line process. In the case of Pt-doped specimens, 10 nm thick Ni0.95Pt0.05 thin-films were sputter deposited on undoped Si (100) wafers at room temperature at a base pressure of 2x10-5 Pa after the removal of the native oxide. A 30 nm thick Co capping layer was deposited without breaking vacuum to serve as an oxidation barrier. Cobalt was chosen because its evaporation field is similar to that of Ni, facilitating a smooth transition from this capping layer to the experimental region-

ECS Transactions, 19 (1) 303-314 (2009)


of-interest (ROI). These thin-films were subjected to RTA at 420 °C for 5 s to form a low-resistivity nickel monosilicide phase. Specimens for 3-D APT were prepared from blanket structures by employing the ‘lift-out’ technique [8] using a FEI dual-beam focused-ion beam (FIB) microscope and an Omniprobe micromanipulator. The lift-out technique is widely used for thin-film specimens and is illustrated in Fig. 1. A long wedge-shaped specimen is lifted-out using a micromanipulator needle (Fig. 1(a)) and cold-welded in place, using Pt, on a number of doped Si-microposts (Fig 1(c)). These are then sharpened [Fig. 1(d) and Fig. 1(e)] using a Ga+ ion-beam and annular mill patterns to achieve an end radius of less than ~50 nm to obtain the requisite electric field for evaporating ions from the tip of a specimen. The APT analyses were performed at a specimen temperature of 40±0.3 K, a laser pulse energy of 0.5 nJ pulse-1, an evaporation rate of 0.2% ions pulse-1, a gauge pressure of <6.6x10-9 Pa, and a pulse repetition rate of 250 kHz or greater [4].

III. Transition metal segregation at the silicide/silicon interface After silicidation and subsequent annealing, formation of the (Ni0.98Pd0.02)Si phase is observed in the Pd-doped specimen. Fig. 2 displays the 3-D LEAP tomographic results in two representations. First Fig. 2(a) presents a 2-D projection of the full 3-D reconstruction for an analysis volume of 30x30x60 nm3. Fig. 2(b) presents a proximity histogram (or proxigram), which is a 3-D non-linear composition profile created by calculating the average chemical composition of a series of isoconcentration surfaces as

Fig. 1: The different stages of dual-beam FIB microscopy based sample preparation from blanket wafer specimens: (a) Lift-out of a wedge using a micromanipulator; (b) A silicon micropost array; (c) Cold welding the wedge to the micropost on two sides using ion-beam deposited Pt; (d) A populated micropost; (e) Sharpening of a specimen using annular mill-patterns and stream files; and (f) The final tip with a radius of curvature ca. 50 nm. This was accomplished using a FEI Helios dual-beam FIB microscope at Northwestern University in the central facility called NUANCE.



these surfaces are advanced into the film [13]. These figures demonstrate that the distribution of Pd is uniform within the silicide film [14]. We therefore conclude that the thermal processing format was sufficient to achieve the equilibrium distribution of Pd. Furthermore, the sum of the Pd and Ni concentrations is 50 at.%, suggesting that the Pd atoms are sitting substitutionally on the Ni sublattice of NiSi as expected; Pd and Ni monosilicides are known to be mutually soluble [10]. The experimental results are also in agreement with first-principles calculations performed using the Vienna ab-initio simulation package [15]. Interestingly, the proxigram in Fig. 2(b) also reveals the segregation of Pd atoms at the NiSi/Si heterophase interface as indicated by the narrow Pd peak localized at this interface.

The thermodynamic driving force for Pd segregation is the decrease in the interfacial Gibbs free energy associated with the (Ni0.98Pd0.02)Si/Si(100) heterophase interface. The segregation of a solute species at a heterophase interface is quantified by the Gibbsian interfacial excess of solute, �s, which is given by:

( )/ 1n os o

n n

C C CA� ��

−Γ = −� ; [1]

and An = Nn/��n. Where An is the effective area of a slice at proximity n, Nn is the number of atoms in the slice, �n is the shell thickness, and � is the ideal atomic density of (Ni0.98Pd0.02)Si. The proxigram yields a discrete count of the atoms in the vicinity of this heterophase interface, thus permitting a direct determination of �s [16]. The Gibbsian interfacial excess of Pd at this heterophase interface is determined to be 3.4±0.2 atoms

Fig. 2: (a) Two-dimensional projection of a three-dimensional reconstruction of LEAP tomographic data showing the distribution of elements (colored points) and the isoconcentration surface (shaded dark blue sheet) that defines the heterophase interface. Nickel, Si and Pd atoms are displayed in green, blue and red, respectively. Palladium atoms are enlarged and only 10% of the Si atoms are shown for clarity. Within the NiSi phase, the Pd is observed to be distributed uniformly. The Ni atoms on the top are from a protective capping layer. (b) Proxigram (proximity histogram) displaying the concentrations of Ni, Si, and Pd versus thin-film depth. Concentrations were calculated for a series of isoconcentration surfaces moving into the thin film; the NiSi/Si heterophase interface provides an example of an isoconcentration surface [14].



nm-2. Similarly, for the case of the platinum-doped nickel monosilicide, Pt segregates at the (Ni0.99Pt0.01)Si/Si(100) interface, which has a Gibbsian interfacial excess of Pt of 0.88±0.01 atoms nm-2. The addition of small concentrations of a transition metal, M, (where M = Pt or Pd) to Ni films on Si(100) is known to have two beneficial effects on the nickel monosilicide phase formed employing RTA [10]. First, the agglomeration of thin (Ni1-

xMx)/Si films under further thermal processing is diminished compared to pure NiSi. Second, the nucleation and growth of NiSi2 occurs at higher temperatures. The agglomeration of thin, uniform pure NiSi films on Si(100) is driven by a decrease in the total interfacial Gibbs free energy that is achieved by reducing the contact area between the two phases. The enhanced resistance of (Ni1-xMx)Si to agglomeration has been attributed to the reduction in the nickel silicide/Si interfacial free energy as a result of segregation of M at that interface [10]. The segregation of Pt or Pd at the (Ni1-

xMx)Si/Si(100) interface reduces its interfacial Gibbs free energy by 8.5 mJ m-2 or 43.3 mJ m-2 [17], respectively. The quantitative measurement of Gibbsian interfacial excesses of M and the concomitant decrease in the interfacial free energy provides direct evidence in support of this argument.

Fig. 3: Two-dimensional top view projections of 3-D APT data showing distribution of elements in the silicide portion of the sample. Clockwise from top: (a) Schematic showing a 5 nm slice selected for projection: (b) 0 < x < 5 nm; (c) 5 < x < 10 nm; and (d) 10 < x < 15 nm. An analysis volume of 50x50x5 nm3 is displayed in each of these frames. Nickel, Si, and Pt are displayed in green, blue and red, respectively. Only 20% of the Ni and Si atoms are shown for the sake of clarity and Co is not displayed since its concentration is less than 3 at.%.



IV. Platinum Short-Circuit Diffusion via Grain Boundaries

Fig. 3 displays the top view of a 50x50x5 nm3 analysis volume, within the nickel

silicide region of the film at varying depths from the (Ni1-xPtx + Co)/(Ni0.99Pt0.01)Si interface. Each dot represents a single atom in the reconstruction: Ni, Si, and Pt are displayed in green, blue, and red, respectively. Cobalt is not displayed since its concentration is less than 3 at.%. Platinum atoms are seen decorating the GBs of the (Ni0.99Pt0.01)Si phase at a high concentration. Fig. 4(a) displays a side view of the Pt distribution associated with (Ni0.99Pt0.01)Si grains in a 50x5x30 nm3 sub-volume of the total reconstructed volume exhibiting the monosilicide portion of the film. Fig. 4(b) displays 1-D composition profiles of Pt and Ni atoms in a cylindrical analysis volume, placed ca. 14 nm below the (Ni1-xPtx + Co)/(Ni0.99Pt0.01)Si interface, which is perpendicular to the GBs of three (Ni0.99Pt0.01)Si grains, Fig. 4(a). The much higher Pt concentration in the GB region compared to the bulk of the grains implies that short-circuit diffusion is occurring via the GBs of the (Ni0.99Pt0.01)Si thin film. GB diffusion of Pt has recently been suggested as occurring in the metal-rich Ni2Si phase [18]. A concomitant decrease in the Ni concentration profile is observed in the GB region suggesting that Pt is substituting for Ni, Fig. 4(b). The Pt concentration profile is indicative of Harrison’s type-B regime for short-circuit diffusion along high-diffusivity paths [19-23]. In this case, the type-B regime is characterized by: (i) volume diffusion from the unreacted Ni0.99Pt0.01 thin film; (ii) volume diffusion laterally from the GBs into the grains in the (Ni0.99Pt0.01)Si region; and (iii) short-circuit diffusion along the GBs, as illustrated in Fig. 5.

Platinum is found to accumulate at the (Ni1-xPtx+Co)/(Ni0.99Pt0.01)Si interface,

which acts as a source for both volume and GB diffusion, Fig. 4(a). The cylindrical analysis volume, Fig. 4(a), is placed at a depth greater than the root-mean-square (RMS) diffusion distance (~3 nm) from the surface. The extent of the lateral Pt concentration profile across the (Ni0.99Pt0.01)Si grains is dependent on grain size. For smaller grains, with diameters less than ca. 15 nm, the Pt concentration profiles from adjacent GBs overlap, Fig. 4(b) (center grain). The high spatial-resolution available enables direct, precise, and accurate measurements of the lattice- (Dl) and GB- (Dgb) diffusivities at 420 oC. No Pt concentration gradients are found along the GBs, implying that Dgb, is significantly greater than Dl. By plotting the log10(Pt concentration) in the grains vs. the square of the distance, we obtain a Dl of 4.2±1.2x10-19 m2 s-1 averaged over 10 grains [24]. Utilizing the Whipple-Suzuoka analysis [25-26] at 420 °C the value of �Dgb is 7.3 x10-26 m3 s-1, where δ is the GB width. Taking � to be 0.5 nm implies that Dgb = 1.5±0.2x10-16 m2 s-1, which is ~350 times greater than Dl. The diffusivity measurements are affected by the simultaneous reaction between Ni and Si to form the nickel monosilicide phase, which is not taken into account in our analyses.

The observation of short-circuit diffusion via GBs sheds light on the kinetics of diffusion and has important implications for phase formation and phase retention. Since very little Pt is found inside the (Ni0.99Pt0.01)Si grains, we can conclude that for the short annealing times (5 s) employed for source/drain contact fabrication, entropy of mixing arguments cannot explain the increased resistance to the formation of NiSi2. The presence of high-diffusivity GB paths makes it easier for Pt to diffuse to the nickel monosilicide/Si interface even for short annealing times. This may be responsible for the



reported modification of the texture of NiSi thin-films with Pt additions [9,11,27-29]. In light of the discovery of Pt short-circuit diffusion along GBs, the observed modification of the NiSi grain size [30] and morphology may be explained by solute drag [31] on GBs due to Pt.

Fig. 5: Schematic depicting: (a) volume diffusion into the bulk from the unreacted Ni1-xPtx metal on top; (b) volume diffusion laterally from the grain boundaries; and (c) short-circuit diffusion along the grain boundaries.

Fig. 4: (a) 2-D vertical slice of 3-D LEAP tomographic data at a depth of ca.14 nm. An analysis volume of 30x50x5 nm3 containing 2.88x105 atoms is displayed with the same coloring scheme for atoms as in Fig. 1. Only 10% of Ni and Si atoms are displayed for clarity. (b) 1-D composition profile of Pt and Ni within the 4x4x48 nm3 cylindrical volume element placed horizontally across three GBs of the (Ni0.99Pt0.01)Si grains as shown in (a).



V. Characterizing the Buried Interface As feature widths in integrated circuits decrease, the interfacial properties of the NiSi/Si interface will play an even greater role in determining the total resistance of the contact [32]. Thus, it is important to follow the temporal evolution of this heterophase interface during nickel silicide formation. Sheet resistance measurements, being sensitive to the reduction in the current-carrying cross-sectional area, provide indirect evidence of agglomeration of thin-films. In contrast, 3-D LEAP tomography, by allowing us to reconstruct this buried interface in 3-D, offers a highly direct method for evaluating agglomeration in a thin-film by following the morphological evolution of the interface. Additionally, it provides the ability to study phase transformations taking place concurrently at the interface. A silicon isoconcentration surface (defined at the mid-point of the Si concentration on either side of the interface) was used to define the interface between (Ni1-xMx)Si and the Si substrate and its chemical RMS roughness was evaluated (Fig. 6). The chemical RMS roughnesses of the (Ni0.98Pd0.02)Si/Si(100) and (Ni0.99Pt0.01)Si/Si(100) interfaces are 0.8 and 1.5 nm, respectively. We did not detect any indications of nickel silicide agglomeration. This direct subnanoscale measurement represents an improvement over other methods, such as light scattering, in which the smoothness of a nickel silicide/silicon interface is averaged over significantly larger sample regions of 0.5 μm or more [33].

VI. Work Function Measurements

Kelvin probe force-microscopy (KPFM) [34-43] was used to measure the work function (WF) of the fabricated nickel silicide contacts. The KPFM measures the contact potential difference, CPDV , defined as:

( ) qV StCPD /ϕϕ −−≡ ; [2]

Fig. 6: Reconstruction of the silicide-Si interface based on a Si isoconcentration surface: (a) (Ni0.98Pd0.02)Si/Si(100); and (b) (Ni0.99Pt0.01)Si/Si(100). The root-mean-square chemical roughnesses of the interfaces are 0.8 nm and 1.5 nm, respectively. Note the different scales for (a) and (b).



where q is the elementary charge on an electron, tϕ is the WF of the tip, and Sϕ is the WF of the sample. The measured work functions were calibrated against that of highly oriented freshly peeled pyrolytic graphite (HOPG), with a known work function of 4.6 eV. The KPFM setup was based on a commercial atomic-force microscope (Dimension 3100, Veeco Inc.), where the electrostatic force is measured in the lift mode [34]. All samples were selectively etched using a solution of H2O2 and H2SO4, to remove the cap and unreacted Ni0.95Pt0.05 alloy on top. A typical simultaneous measurement of both the topography [Fig. 7(a)] and the contact potential difference [Fig. 7(b)] are shown; in general, the WF of both NiSi and

NiSi(Pt) is extremely uniform (±15 meV) over large

areas. The WF of the (Ni1-xPtx)Si contacts is found to be 4.57±0.015 eV following RTA processing. A similar measurement on the as-deposited samples yielded a value of around 3.85±0.015 eV. This increase of WF would reduce the Schottky barrier height of the contact to p-Si and the corresponding contact resistance. In comparison to an unalloyed NiSi sample, the relative shift in WF between the as-deposited and annealed states is greater for Ni(Pt)Si (0.72±0.015 eV) than for NiSi (0.67±0.015 eV). We also note that the segregation of Pt to the silicide/silicon interface modifies the barrier profile in a manner that has not, to our knowledge, been included in theoretical models of contact resistance. The correlation of chemical composition profiling with electrical WF measurements represents a powerful tool for investigating contact materials.

VII. Conclusions In conclusion, we have shown the 3-D atomic-scale distributions of transition metal elements M, (M = Pt or Pd) in alloyed nickel monosilicide thin-films. These alloying additions are found to segregate at the (Ni1-xMx)Si/Si(100) heterophase interface and might be responsible for the increased resistance of these thin-films to agglomeration at elevated temperatures. In case of Pt-doped nickel monosilicide films, direct evidence of short-circuit diffusion via GBs, Harrison regime-B, is provided. Using precise APT measurements, a Dl of 4.2±1.2x10-19 m2 s-1 and a Dgb of 1.5±0.2x10-16 m2 s-1 were measured. Owing to the large difference in these diffusivities, Pt is found at the (Ni1-

xMx)Si/Si(100) interface even after short annealing times, 5s. This provides valuable mechanistic insights into how Pt short-circuit diffusion via GBs may affect nickel silicide grain size as well as the texture and morphology of the grains. Additionally, the (Ni1-

xMx)Si/Si(100) heterophase interfaces were characterized in 3-D and the chemical RMS roughnesses evaluated. High resolution WF measurements of the silicide top surface have been demonstrated employing Kelvin probe force-microscopy. The relative shift in

Fig. 7: Kelvin probe force-microscopy (KPFM) measurements of the nickel monosilicide’s top surface in Ni(Pt)Si. Both (a) topographic and (b) contact potential difference are measured simultaneously.



WF between as-deposited and annealed states is greater for Ni(Pt)Si (0.72±0.015 eV) than for NiSi (0.67±0.015 eV). WF measurements in conjunction with 3-D atomic-scale chemical analyses employing an atom-probe tomograph enable a more comprehensive approach to the development of silicide contacts to Si-based devices.

Acknowledgments This research is supported by Semiconductor Research Corporation, Task No. 1441.001. Atom-probe tomographic measurements were performed at the Northwestern University Center for Atom-Probe tomography (NUCAPT), which is managed by Research Assistant Prof. D. Isheim. The 3-D LEAP tomograph was purchased with funding from the NSF-MRI and ONR-DURIP programs. The FEI dual-beam FIB was used in the EPIC facility of the NUANCE Center at Northwestern University. The Kelvin probe force-microscopy measurements were performed at Tel-Aviv University. Profs. Y. C. Kim and H. D. Lee are thanked for providing the Pd-doped specimen and Dr. R. Alvis is thanked for help with specimen preparation.

References

[1] S. L. Zhang and M. Östling, Crit. Rev. Sol. Stat. Mater. Sci., 28 (1), 1 (2003). [2] J. A. Kittl, A. Lauwers, O. Chamirian, M. Van Dal, A. Akheyar, M. De Potter,

R. Lindsay, K. Maex, Microelec. Eng., 70, 158 (2003). [3] D. E. Newbury, D. B. Williams, Acta. Mater., 48, 323 (2000). [4] D. N. Seidman, Ann. Rev. Mater. Res., 37, 127 (2007). [5] T. F. Kelly, M. K. Miller, Rev. Sci. Inst., 78, 031101 (2007). [6] T. F. Kelly, D. J. Larson, K. Thompson, R. L. Alvis, J. H. Bunton, J. D. Olson

and B. P. Gorman, Ann. Rev. Mater. Res., 37, 681 (2007). [7] M. K. Miller, K. F. Russell, K. Thompson, R. L. Alvis, D. J. Larson, Micro.

Microanal., 13 (6), 428 (2007). [8] K. Thompson, D. Lawrence, D. J. Larson, J. D. Olson, T. F. Kelly, B.

Gorman, Ultramicroscopy, 107, 131 (2007). [9] C. Lavoie, C. Detavernier and P. Besser, in Silicide Technology for Integrated

Circuits, L. J. Chen, Editor, p. 95, IEE, London (2004). [10] C. Lavoie, C. Detavernier, C. Cabral Jr., F. M. d’Heurle, A. J. Kellock, J.

Jordon-Sweet, J. M. E. Harper, Microelec. Eng., 83, 2042 (2006). [11] D. Mangelinck, J. Y. Dai, J. S. Pan, S. K. Lahiri, Appl. Phys. Lett., 75, 1736

(1999). [12] J. F. Liu, J. Y. Feng, J. Zhu, Appl. Phy. Lett., 80, 270 (2002). [13] O. C. Hellman, J. A. Vandenbroucke, J. Rüsing, D. Isheim, D. N. Seidman,

Micro. and Microanal., 6, 437 (2000). [14] Y. C. Kim, P. Adusumilli, L. J. Lauhon, D. N. Seidman, S. Y. Jung, H. D.

Lee, R. L. Alvis, R. M. Ulfig, J. D. Olson, Appl. Phy. Lett., 91, 113106-1 (2007).

[15] Z. Mao and D. N. Seidman, Unpublished data. [16] O. C. Hellman and D. N. Seidman, Mater. Sci. Eng. A, 327, 24 (2002). [17] B. W. Krakauer and D. N. Seidman, Acta Mater., 46, 6145 (1998). [18] B. Imbert, C. Guichet, S. Bonnetier, S. Zoll, M. Juhel, M. Hopstaken, P.

Clifton, O. Thomas, Microelec. Eng., 84, 2523 (2007). [19] L. G. Harrison, Trans. Farad. Soc., 57, 1191 (1961).



[20] A. P. Sutton and R. W. Balluffi, in Interfaces in Crystalline Materials, p.467-481, Oxford Univ. Press, New York, (1995).

[21] G. Martin, B. Perraillon, in Grain-Boundary Structure and Kinetics, p. 239, ASM, Ohio, (1979).

[22] E. W. Hart, Acta. Metallurgica, 5, 597 (1957). [23] J. Philibert, in Atom movements: Diffusion and mass transport in solids, p.

249, les editions de physique, France (1991). [24] P. Adusumilli, L. J. Lauhon, D. N. Seidman, C. E. Murray, O. Avayu, Y.

Rosenwaks, Submitted for publication. [25] R. T. Whipple, Phil Mag., 45, 1225 (1954). [26] T. Suzuoka, J. Phys. Soc. Japan, 19, 839 (1964). [27] C. Detavernier, A. S. Ozcan, J. Jordan-Sweet, E. A. Stach, J. Tersoff, F. M.

Ross, C. Lavoie, Nature 426, 641 (2003). [28] C. Detavernier, J. Jordon-Sweet, C. Lavoie, J. Appl. Phys., 103, 113526

(2008). [29] C. Detavernier, C. Lavoie, Appl. Phys. Lett. 84, 3549 (2004). [30] H. Akutsu, H. Itokawa, K. Nakamura, T. Iinuma, K. Suguro, H. Uchida, M.

Tada, Mater. Res. Soc. Symp. Proc., 1070, 1070-E02-09 (2008). [31] S. G. Kim, Y. G. Park, Acta. Mater. 56, 3739 (2008). [32] A. Lauwers, J. A. Kittl, M. J. H. Van Dal, O. Chamirian, M. A. Pawlak, M. de

Potter, R. Lindsay, T. Raymakers, X. Pages, B. Mebarki, T. Mandrekar, K. Maex, Mater. Sci. & Eng. B, 114, 29 (2004).

[33] C. Lavoie, C. Cabral, F. M. d'Heurle, J.M.E. Harper, Def. Diff. For., 194, 1477 (2001).

[34] Y. Rosenwaks, R. Shikler, Th. Glatzel, S. Sadewasser, Phy. Rev. B, 70, 085320 (2004).

[35] S. Sadewasser, Th. Glatzel, R. Shikler, Y. Rosenwaks, M. Ch. Lux-Steiner, Appl. Surf. Sci., 210, 32 (2003).

[36] Th. Glatzel, S. Sadewasser, R. Shikler, Y. Rosenwaks, M. Ch. Lux-Steiner, Mater. Sci. Eng. B, 103, 138 (2003).

[37] A. Schwarzman, E. Grunbaum, E. Strasburg, E. Lepkifker, A. Boag, Th.Glatzel, Z. Barkay, M. Mazzer, K. Barnham, and Y. Rosenwaks, J. Appl. Phys., 98, 84310-4 (2005).

[38] S. Saraf, A. Schwarzman, Y. Dvash, S. Cohen, D. Ritter and Y.Rosenwaks, Phys. Rev. B., 73, 35336 (2006).

[39] S. Shusterman, A. Reizman, Y. Paltiel, A. Sher, A. Schwarzman, E.Lepkifker and Y. Rosenwaks, Nanoletters, 7, 2089 (2007).

[40] S. Yogev, Y. Levine, M. Molotskii, and Y. Rosenwaks, J. Appl. Phys., 103, 64107 ( 2008).

[41] I. Magid, L. Burstein, O. Seitz, L. Segev, L. Kronik, and Y. Rosenwaks,J. Phys. Chem. C., 112, 7145 (2008).

[42] E. Strassburg, A. Boag, and Y. Rosenwaks, Rev. Sci. Inst., 76, 83705 (2005). [43] S. Saraf, M. Molotskii, and Y. Rosenwaks , Appl. Phys. Lett., 86, 172104

(2005). [44] O. Cojocaru-Miredin, D. Mangelinck, K. Hoummada, E. Cadel, D. Blavette,

B. Deconihout, C. Perrin-Pellegrino, Scrip. Mater., 57, 373 (2007). [45] K. N. Tu, R. Rosenberg, Thin Solid Films, 13, 163 (1972).



[46] J. Demeulemeester, D. Smeets, C. V. Bockstael, C. Detavernier, C. M. Comrie, N. P. Barradas, A. Vieira, A. Vantomme, Appl. Phys. Lett., 93, 261912 (2008).



three-dimensional atom-probe tomographic studies of...

Documents