Orientation dependence of nickel silicide formation in

Orientation dependence of nickel silicide formation in contacts to silicon nanowires N. S. Dellas, B. Z. Liu, S. M. Eichfeld, C. M. Eichfeld, T. S. Mayer, and S. E. Mohney Citation: Journal of Applied Physics 105, 094309 (2009); doi: 10.1063/1.3115453 View online: http://dx.doi.org/10.1063/1.3115453 View Table of Contents: http://scitation.aip.org/content/aip/journal/jap/105/9?ver=pdfcov Published by the AIP Publishing Articles you may be interested in Very low effective Schottky barrier height for erbium disilicide contacts on n-Si through arsenic segregation Appl. Phys. Lett. 99, 012110 (2011); 10.1063/1.3608159 Controlled large strain of Ni silicide/Si/Ni silicide nanowire heterostructures and their electron transport properties Appl. Phys. Lett. 97, 203110 (2010); 10.1063/1.3515421 Influence of germanium on the formation of Ni Si 1 − x Ge x on (111)-oriented Si 1 − x Ge x J. Appl. Phys. 98, 053507 (2005); 10.1063/1.2034081 Mosaic structure of various oriented grains in CoSi 2 /Si(001) J. Vac. Sci. Technol. B 18, 1953 (2000); 10.1116/1.1305275 Structural and electrical properties of polycrystalline silicon produced by low-temperature Ni silicide mediated crystallization of the amorphous phase J. Appl. Phys. 87, 609 (2000); 10.1063/1.371906

[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP: 192.91.75.29 On: Tue, 29 Sep 2015 13:37:58

JOURNAL OF APPLIED PHYSICS 105, 094309 共2009兲

Orientation dependence of nickel silicide formation in contacts to silicon nanowires N. S. Dellas,1 B. Z. Liu,3 S. M. Eichfeld,1 C. M. Eichfeld,1 T. S. Mayer,2,3 and S. E. Mohney1,3,a兲 1

Department of Materials Science and Engineering, The Pennsylvania State University, University Park, Pennsylvania 16802, USA 2 Department of Electrical Engineering, The Pennsylvania State University, University Park, Pennsylvania 16802, USA 3 Materials Research Institute, The Pennsylvania State University, University Park, Pennsylvania 16802, USA

共Received 9 January 2009; accepted 4 March 2009; published online 4 May 2009兲 The orientation dependence of Ni silicide phase formation in the silicidation of silicon nanowires 共SiNWs兲 by Ni has been studied. SiNWs with a 关112兴 growth direction contacted by Ni pads form ␪-Ni2Si for annealing conditions from 350 to 700 ° C for 2 min. The ␪-Ni2Si has an epitaxial ¯ 兴 and ␪-Ni Si共100兲储 Si共112兲 with the SiNW. On the other hand, orientation of ␪-Ni Si关001兴储 Si关111 2

2

SiNWs with a 关111兴 growth direction react with Ni pads to form NiSi2 with an epitaxial orientation ¯ 0兴储 Si关11 ¯ 0兴 and NiSi 共111兲储 Si共111兲 after annealing at 450 ° C for 2 min. The 关111兴 of NiSi2关11 2 SiNWs were also silicided at 700 ° C for 2 min, forming the low-resistivity NiSi phase. The epitaxial phases identified in the reactions of Ni films with SiNWs suggest that lattice matching at both the silicide/Si growth front and the surface of the original SiNW may play a significant role in determining the first silicide segment to grow. © 2009 American Institute of Physics. 关DOI: 10.1063/1.3115453兴 I. INTRODUCTION

Silicon nanowires 共SiNWs兲 have garnered considerable interest for a variety of applications including field-effect transistors 共FETs兲,1 thin film transistors,2 and chemical and biological sensors.3 Formation of Ohmic contacts to these devices is necessary to achieve optimal transistor and sensor performance. In planar metal-oxide-semiconductor fieldeffect transistors, the use of nickel monosilicide 共NiSi兲 as an Ohmic contact material has recently become attractive.4 One reason for this is that the most favorable silicide phases 共i.e., those with the lowest electrical resistivity兲 in the previously used Ti–Si 共TiSi2兲 and Co–Si 共CoSi2兲 systems are subject to nucleation barriers, and high-resistivity silicide phases remain in small dimensions.5 However, Chang et al.6 showed a change in phase transformation temperatures in Ni/Si contacts, with epitaxial NiSi2 forming at lower temperatures within smaller-sized oxide openings. The reduction in the nucleation temperature was attributed to the large stresses present in small dimensions. These size effects may become more pronounced in the smaller geometries encountered in SiNWs. For this reason, it is important to understand how the size and/or geometry of a SiNW can affect the phase formation during silicidation of SiNWs by Ni. Differences have previously been noted between the reaction of Ni/Si bulk diffusion couples and thin films of Ni on Si.7,8 The most notable difference is the sequential phase formation encountered in thin films versus the simultaneous phase formation observed in bulk diffusion couples. In a bulk diffusion couple, the Ni and Si supplies are large enough that a兲

Electronic mail: [email protected].

0021-8979/2009/105共9兲/094309/7/$25.00

neither becomes limited. Additionally, the bulk couples are generally held at high temperature for long times, resulting in reaction regions with thicknesses on the order of microns and allowing all thermodynamically stable phases on the phase diagram to grow simultaneously. It has also been observed that the phases grow in thicknesses proportional to their relative interdiffusion coefficients.7 On the other hand, in a reaction of a Ni thin film on a Si wafer, a sequential phase formation occurs. Typically, ␦-Ni2Si initially grows because its interdiffusion coefficient is highest at low reaction temperatures 共⬍400 ° C兲.8 All of the Ni film is consumed in this reaction, and ␦-Ni2Si is the only phase that usually forms. Following the complete consumption of Ni and formation of ␦-Ni2Si, NiSi begins forming via the reaction of ␦-Ni2Si with the underlying Si wafer. NiSi is stable until high temperature 共greater than 700 ° C兲, at which NiSi2 begins nucleating.9 The NiSi and NiSi2 phases have similar free energies of formation; hence, there is only a small driving force for formation of NiSi2, resulting in a nucleation barrier, which explains why NiSi2 usually forms only at high temperatures. Deviations exist from the typical thin film reaction described above, particularly for very thin films. Si orientation effects on Ni silicide formation are also observed in the reactions of extremely thin films of Ni on Si. During silicidation of Ni films that are only ⬃10 nm or less in thickness on 共111兲, 共011兲, and 共001兲 Si wafers, epitaxial NiSi2 formation is reported as the initial phase formed.10,11 Additionally, Teodorescu et al.12 found that on 共111兲 and 共100兲 oriented Si wafers with a thin chemical oxide, epitaxial NiSi2 is the first phase to form at temperatures below 400 ° C when 12 nm thick Ni films are reacted with the Si wafers. NiSi2 has the

105, 094309-1

© 2009 American Institute of Physics


094309-2

J. Appl. Phys. 105, 094309 共2009兲

Dellas et al.

cubic fluorite crystal structure and is lattice matched well to Si with a mismatch of less than 1%. Gibson et al.13 studied the reaction of thin Ni films on 共111兲 and 共100兲 oriented wafers as well. For the 共111兲 Si wafers, they found that ␪-Ni2Si is initially formed at a temperature of 300 ° C, and upon further annealing it is converted to epitaxial NiSi2. The ␪-Ni2Si共001兲 plane is extremely similar to the Si 共111兲 plane. The in-plane bond distances on the 共001兲 plane of ␪-Ni2Si are 3.836 Å with an angle between rows of atoms of 60°. This compares very well with the Si 共111兲 plane that has in-plane spacings of 3.84 Å and an angle between bonds of 60°. Part of the explanation for the early formation of ␪-Ni2Si is that the excellent lattice match minimizes the free energy of the system because in very thin films 共1–2 nm thick兲, a large fraction of atoms initially lie at the Ni/Si interface. However, in a reaction of the same very thin Ni films on 共100兲 oriented wafers, the ␪-Ni2Si phase is not observed because the absence of fourfold symmetry in the hexagonal structure does not allow for a good epitaxial match between ␪-Ni2Si and Si in this particular orientation. In the limited work available on the silicidation of SiNWs by Ni, one finds several differences in phases observed and no evidence of a sequential phase formation. Appenzeller et al.14 reported that Ni2Si 共not identified as ␦ or ␪-Ni2Si or supported by diffraction data兲 formed axially along the SiNW after annealing for 30 s at 280 ° C. Lu et al.15,16 found NiSi under annealing conditions in the range of 500– 650 ° C for Ni silicidation of 关111兴 SiNWs. Sheu et al.17 and Wu et al.18 also observed NiSi. Results obtained by Wu et al. differ from others in that no significant lateral diffusion was observed after silicidation despite the high temperature 共550 ° C兲, long time 共5 min兲, and secondary annealing step carried out at 600 ° C. This lack of lateral diffusion was verified in their work by the fact that the silicide segment length was approximately equal to the patterned Ni stripe deposited on top of the SiNW. The experimental setup of Lu et al. differs from our study and other studies because they reacted Ni nanowires in point contact with SiNWs and observed silicide formation at the ends of the SiNW, not at the Ni/Si contact region. Zhang et al.19 identified predominantly ␦-Ni2Si forming over a range of annealing conditions from 500 to 950 ° C for 30 s using polycrystalline SiNWs, with traces of NiSi, Ni3Si2, and ␪-Ni2Si identified at different temperatures. Additionally, Weber et al.20 reported the formation of single crystalline NiSi2 after siliciding 关110兴 SiNWs by electrolessly deposited Ni pads at an annealing temperature of 480 ° C. These studies further suggest that differences may exist in the phase formation sequence of SiNWs compared to their bulk and thin film counterparts. Here, we study the reaction of a Ni contact pad with an underlying SiNW in a device geometry appropriate for FETs, where electrical measurements and transmission electron microscopy 共TEM兲 can ultimately be performed on a single sample fabricated on an electron-transparent silicon nitride membrane. We report Ni silicidation of 关112兴 and 关111兴 oriented SiNWs with an orientation-dependent silicide phase formation.

FIG. 1. 共Color online兲 Schematic diagram of process flow used to fabricate TEM samples. The process begins with 共a兲 a Si wafer coated with 100 nm of SiNx. 共b兲 Ag electrodes are then patterned and deposited for electrofluidic alignment of the SiNWs. 共c兲 The Ag alignment pads are etched away, and 共d兲 Ni pads are patterned and deposited to contact the aligned SiNWs. Finally, 共e兲 a window is RIE etched in the back side of the wafer, and 共f兲 the Si is etched through with KOH to the front side SiNx layer.

II. EXPERIMENTAL

The SiNWs were grown via the vapor-liquid-solid growth technique21 using gold as a catalyst. The 关112兴 SiNWs were 20 ␮m long with an average diameter of 50 nm and lightly doped p-type with a B共CH3兲3 / SiH4 flow ratio of 2 ⫻ 10−5 during growth. The 关111兴 SiNWs were grown using SiCl4 without intentional doping, with average diameters of 45 nm. Any effect of B doping on the silicidation would be unexpected because it has been previously demonstrated that there is no significant effect at annealing temperatures greater than 350 ° C.22 TEM samples were fabricated 共schematic diagram shown in Fig. 1兲 by starting with a 3 in. diameter, 350 ␮m thick, double side polished Si wafer that was lightly doped p-type with B 共␳ ⬎ 1 ⍀ cm兲. The wafer was coated with 100 nm of low pressure chemical vapor deposited 共LPCVD兲 SiNx 关Fig. 1共a兲兴. LPCVD conditions were a deposition temperature of 820 ° C, pressure of 300 mTorr, and NH3 and SiH2Cl2 gases at flow rates of 180 and 40 SCCM 共SCCM denotes cubic centimeter per minute at STP兲. The LPCVD SiNx was used as an etch stop on the front side of the wafer and acts as a hard etch mask for KOH etching on the back side. The front side of the wafer was patterned using photolithography to define NW alignment pads 关Fig. 1共b兲兴. 80 nm of Ag 共Alfa Aesar 99.999% purity兲 was e-beam evaporated in an Edwards FL400 e-beam evaporation system with a base pressure of 2 ⫻ 10−7 Torr. The SiNWs were next aligned to the Ag pads using electric-field assisted assembly.23 Another photolithography step was done to pattern contact windows to the aligned SiNWs. Before Ni deposition, Au etchant 共Transene gold etchant type TFA兲 was used to remove Ag in the contact area 关Fig. 1共c兲兴, and a 25 s etch in 10:1 共NH4F : HF兲 buffered oxide etch 共BOE兲共JT Baker兲 was used to remove the native oxide on the SiNWs in the contact


094309-3

Dellas et al.

region. The sample was loaded into the e-beam evaporation system within 5 min of the BOE and pumped down to a base pressure of 2 ⫻ 10−7 Torr. 80 nm of Ni 共Alfa Aesar 99.995% purity兲 was deposited, and liftoff was done in acetone immediately after the wafer was removed from the vacuum chamber 关Fig. 1共d兲兴. Next, photolithography on the back side of the wafer was used to pattern windows directly below the Ni pads and SiNWs on the front side. The patterned photoresist was used as an etch mask for reactive ion etching 共RIE兲. The SiNx in these patterned areas was etched by RIE in a Plasma Therm 720 system to open windows to bare Si for KOH etching 关Fig. 1共e兲兴. RIE etch conditions were a power of 300 W, pressure of 10 mTorr, and CF4 and O2 at flow rates of 45 and 5 SCCM, respectively. The remaining photoresist on the back side of the wafer was removed with acetone. The final step included putting the wafer in a KOH etch jig 共AMMT Single 3 holder兲 to seal and protect the front side patterning of the wafer from KOH. The wafer was etched in 30 wt % KOH 共JT Baker兲 at 90 ° C until the SiNx window on the front side of the wafer was reached 关Fig. 1共f兲兴. In order to obtain high-resolution TEM images, the SiNx window was further thinned to 60 nm by an additional RIE etch. The etch conditions were the same as listed above except that the power was reduced to 25 W to maintain a smooth surface on the back side of the SiNx window. Samples were annealed in a rapid thermal annealing furnace 共AG Associates Heat Pulse 610兲 ex situ in ultrahigh purity Ar at 300– 700 ° C for 2 min. TEM and electron energy loss spectroscopy 共EELS兲 analysis of the samples were performed using a JEOL EM-2010F field-emission TEM/ scanning TEM with a Gatan double-tilt holder. Samples fabricated using the above process were glued to 3 mm Cu rings with M-Bond 200 cyanoacrylate epoxy cured at room temperature to properly fit into the TEM sample holder.

III. RESULTS AND DISCUSSION A. Ni silicidation of †112‡ oriented SiNWs

After samples were annealed at 300 ° C for 2 min, no visible reaction between the Ni and SiNW is observed. After annealing at 350 ° C, Ni silicide formation down the wire is observed with an average length of 50 nm. It is expected that Ni will diffuse into the SiNW and form an axial silicide segment along the SiNW because Ni is the dominant diffusing species in the phases in the Ni–Si system.24 Further Ni silicide formation can be identified at 400 ° C by the image contrast change along the SiNW, as shown in Fig. 2共a兲. The selected area electron diffraction 共SAED兲 patterns from the silicide 关Figs. 2共b兲 and 2共c兲兴 show that the silicide formed is hexagonal ␪-Ni2Si, the same phase observed at 350 ° C. When possible, two patterns were taken by tilting to different zone axes, specifically the 关401兴 and 关201兴 directions, and the angle tilted matched the angle between these two directions in the hexagonal system. It should be noted that ␪-Ni2Si is a metastable phase under these annealing conditions and should not be observed according to the phase diagram until temperatures exceeding 825 ° C.25 The ␪-Ni2Si nanowire segment is single crystalline and has a slightly larger diam-

J. Appl. Phys. 105, 094309 共2009兲

FIG. 2. 共a兲 TEM image showing sample annealed at 400 ° C for 2 min and 共b兲 SAED patterns from samples annealed at 400 ° C for 2 min identified as ␪-Ni2Si by the 关201兴 and 共c兲关001兴 zone axis diffraction patterns.

eter than the original diameter of the unconverted SiNW, which is expected because of the volume increase due to the formation of ␪-Ni2Si. Figure 3 shows a TEM image of a sample which was annealed at 300 ° C for 2 min and then at 450 ° C for 2 min, which are the conditions that were recently used to form a Ni silicide Schottky contact to a SiNW.26 The 300 ° C preannealing does not cause significant silicidation axially down the SiNW, and it was observed in the present work that similar results were obtained using a single anneal at 450 ° C for 2 min. It can be seen that silicidation occurs at 450 ° C and the silicide travels on average 500 nm, compared to an average of 150 nm at 400 ° C. From Fig. 3 it is apparent that the NW can be divided into three sections, each identified by a change in NW diameter. Section I is the unconverted SiNW, which has a diameter of 59 nm. The SAED pattern shows that the growth direction of the SiNW was 关112兴. Section II has been converted into ␪-Ni2Si, as again confirmed by SAED. Section II has slightly larger diameter 共65 nm at the maximum or a 10% increase兲 compared to the unconverted

FIG. 3. TEM image showing samples annealed at 300 ° C for 2 min and ¯兴 then 450 ° C for 2 min, with an epitaxial relation of ␪-Ni2Si关001兴储 Si关111 and ␪-Ni2Si共100兲储 Si共112兲. The nanowire is split into three regions: region I is the original SiNW, region II is ␪-Ni2Si, and region III is an unidentified Ni-rich silicide phase.


094309-4

Dellas et al.

J. Appl. Phys. 105, 094309 共2009兲

FIG. 4. 共a兲 TEM image showing overview of a sample annealed at 700 ° C for 2 min and 共b兲 higher magnification image. Diffraction patterns 共c兲 and 共d兲 from the main silicide segment 共trunk兲, where 共c兲 and 共d兲 show extra spots from microtwinning and 共e兲 shows the diffraction pattern from a branch, showing that the branch is the same ␪-Ni2Si phase as the main part of the nanowire, but without the extra spots from the microtwins.

SiNW, but the diameter is close to the unconverted SiNW diameter at the growth front. By comparing the SAED patterns between the silicide and Si, the epitaxial orientation relationship between the ␪-Ni2Si and SiNW is found to be ¯ 兴 and ␪-Ni Si共100兲储 Si共112兲. Section III ␪-Ni2Si关001兴储 Si关111 2 is close to the Ni pad and has a much larger diameter 共72 nm兲. This section was not found in the samples annealed at lower temperatures. Unfortunately, the SAED patterns 共not shown兲 did not provide conclusive phase identification. The patterns obtained from these regions do have large lattice spacings that rule out all but the Ni31Si12 and Ni3Si2 phases. Of these phases, Ni31Si12 seems more likely because an intermediate phase between the pure Ni pad and ␪-Ni2Si should not have a composition more Si rich than the ␪-Ni2Si phase. From the electron diffraction data we have identified the ¯兴 epitaxial orientation relationship of ␪-Ni2Si关001兴储 Si关111 储 and ␪-Ni2Si共100兲 Si共112兲. Interestingly, for the 共100兲 ␪-Ni2Si and 共112兲 Si planes, there is a very good lattice match, and therefore it is not surprising that the planes coincide at the interface between the silicide segment and the SiNW. For ␪-Ni2Si, the 共100兲 in-plane bond distances are 2.474 and 3.836 Å with a bond angle of 90°, which compares well with that of the 共112兲 Si in-plane bond distances of 2.35 and 3.840 Å with a bond angle of 90°. This results in a ⫺5.1% mismatch in one direction and 0.1% mismatch in the other direction. Additionally, the cross section of a 关112兴 oriented wire is nearly rectangular with the two long sides being bound by 共111兲 type planes and the other two by a combination of 共110兲 and 共311兲 type planes.27 These surfaces surrounding the SiNW are important to recognize because they will be the point of initial contact between the Ni film and SiNW. As mentioned above, ␪-Ni2Si is nearly perfectly matched to the Si 共111兲 surface in that the 共001兲 in-plane

spacing is 3.836 Å compared to 3.840 Å for the Si 共111兲 in-plane spacing, and both have bond angles of 60°. Thus, an epitaxial arrangement is favorable over much of the original circumference of the SiNW as well. Although there is not a good lattice match for the 共110兲 and 共311兲 planes, these planes comprise a smaller fraction of the overall surface area of the original SiNW compared to the 共111兲 planes. Samples were also annealed at temperatures of 500, 600, and 700 ° C for 2 min each. In each case, the phase observed was again ␪-Ni2Si, consistent with the samples annealed at lower temperatures. However, after annealing at 700 ° C for 2 min, branches were observed to form from the main ␪-Ni2Si nanowire trunk, which was identified by SAED 共Fig. 4兲. Using SAED, the branches were also observed to be the same phase as the main silicide segment, ␪-Ni2Si. However, extra diffraction spots were also observed in the SAED patterns from the ␪-Ni2Si silicide segments. These extra spots are attributed to microtwinning, which can be observed from the inset in Fig. 4共b兲 and can be caused by large compressive stresses. The direction along which the microtwins formed and the additional spots observed in the diffraction pattern are consistent when the microscope’s rotation calibration is considered. Interestingly, the native oxide thickness may not be uniform around the SiNW. As observed by TEM, branches form at locations where the silicide has broken through the oxide, providing further evidence that the ␪-Ni2Si phase is under a large compressive stress. The large indiffusion of Ni compared to the negligible outdiffusion of Si and increase in volume of the ␪-Ni2Si phase compared to the original Si volume are all reasons for compressive stress to build in the nanowire. The mechanism for formation of these branches may be similar to the formation of Sn whiskers encountered in electronic packaging.28 In any case, it is


094309-5

Dellas et al.

J. Appl. Phys. 105, 094309 共2009兲

FIG. 5. 共a兲 Ni-silicided 关111兴 oriented SiNW after annealing at 450 ° C for 2 min and 共b兲 HRTEM image of silicide/ SiNW interface. The inset FTs are ¯ 0兴 NiSi zone axis identified as the 关11 2 ¯ and the 关110兴 Si zone axis with epitax¯ 0兴储 Si关11 ¯ 0兴 ial relations of NiSi2关11 and NiSi2共111兲储 Si共111兲.

a significant phenomenon because in the use of SiNWs for electronic devices, these branches could cause shorts between two neighboring devices. B. Ni silicidation of †111‡ oriented SiNWs

The SiNWs with a 关111兴 growth direction were also silicided by the method described above with 80 nm of Ni and annealing conditions of 450 and 700 ° C for 2 min. Annealing at 450 ° C for 2 min again results in a silicide forming axially down the SiNW. However, in the case of the 关111兴 oriented SiNWs, the silicide phase is identified as NiSi2 by SAED. Figure 5 shows a high resolution transmission electron microscope 共HRTEM兲 image of a silicided SiNW with insets of digital Fourier transforms 共FTs兲 from the silicide segment and the SiNW. From the FTs we can determine that an epitaxial orientation exists between the NiSi2 segment and SiNW. The epitaxial relation is identified as ¯ 0兴储 Si关11 ¯ 0兴 and NiSi 共111兲储 Si共111兲. From the HRNiSi2关11 2 TEM image in Fig. 5共b兲 we can also deduce that the NiSi2 / Si interface is atomically abrupt. Similar to the case of the inplane spacings being well matched for the 共100兲 ␪-Ni2Si and 共112兲 Si planes, the in-plane spacings of the 共111兲 NiSi2 and 共111兲 Si planes are also very well lattice matched. The inplane mismatch between the atomic spacings on the 共111兲 NiSi2 and 共111兲 Si planes is only 0.9%. In addition to having a good lattice match along the growth direction, there is lattice matching to the surface of the SiNW as well, which is what was originally contacted by the Ni pad. The cross section of a 关111兴 growth direction SiNW is hexagonal, being bound on all six facets by 兵112其 type planes.29 The in-plane spacings for NiSi2 along the 共112兲 direction are 3.80 and 2.33 Å and have only a 0.9% mismatch to the in-plane Si spacings of 3.84 and 2.35 Å. To corroborate the SAED and FTs, EELS spectra were collected on Ni-silicided SiNWs with 关112兴 and 关111兴 growth directions. Cheynet and Pantel30 reported a shift to higher energies with increasing Si content in the Ni L2,3 edge of EELS spectra for thin film Ni silicides. Since absolute energy positions are difficult to monitor because of instabilities and drift effects on the EELS spectrometer, spectra were collected from the Ni pad as well as the silicide segments. The Ni L2,3 edge from the Ni pad was used as a standard, and the shifts were measured from the Ni pad Ni L2,3 edge since the Ni pad is common to both types of samples 共diffraction patterns from the Ni pad confirmed that the Ni pad had not

reacted with the underlying SiNx window兲. EELS spectra from Ni, ␪-Ni2Si, and NiSi2 are shown in Fig. 6. We observe a shift in energy of 1 eV for ␪-Ni2Si relative to Ni, and a shift of 2.5 eV for NiSi2 relative to Ni representing a net shift of 1.5 eV between the two silicide phases observed. The shift in the Ni L2,3 edge to higher energies is caused by the reduced hybridization of Ni d states with Si p states, causing Si-rich silicides to have the Ni L2,3 edge at higher energies. This is in agreement with and corroborates our analysis from the SAED and FT patterns. Lastly, the SiNWs with 关111兴 growth directions were also silicided at 700 ° C for 2 min. In this case, the SAED shows the formation of the low-resistivity NiSi phase. Figure 7 shows a TEM image of the silicided 关111兴 oriented SiNW. In contrast to the SiNWs with 关112兴 growth directions silicided at 700 ° C, no branches are observed to form. Both of the ratios of the volume of NiSi2 and NiSi to Si are smaller than that for ␪-Ni2Si to Si. This suggests that less compressive stress would build up in the constrained nanowire geometry and explains why no branches emerge from the silicided 关111兴 SiNWs. This formation of NiSi in the SiNWs with 关111兴 growth directions is also consistent with what is observed in silicidation of 共111兲 oriented Si wafers by very thin Ni films. In the planar silicidation case, NiSi2 is observed to form epitaxially at approximately 400 ° C, and upon continued reaction at increased time and temperature, the NiSi2 is converted to NiSi. NiSi remains stable until temperatures above 700 ° C, where NiSi2 begins nucleating. We suspect that the same may occur in the nanowire silicidation; however, the SiNx membranes used to fabricate the TEM samples could not be annealed above 700 ° C without cracking. In comparison of our results to others, the only previous studies that explicitly mention the orientation of the SiNW and identify the silicide phase that forms are those by Lu et al.,15,16 Wu et al.,18 and Weber et al.20 Lu et al. used 关111兴 oriented SiNWs in point contact with NiNWs and found NiSi to form over a range of annealing temperatures of 500– 650 ° C. It is also worth mentioning that important differences may exist in the reaction of NiNWs with SiNWs compared to a Ni pad with SiNWs. In fact, in the NiNW/ SiNW reaction, the silicide forms at the ends of the SiNW, not at the NiNW/SiNW interface. This differs from our work and others who have observed the silicide to form at the interface between a Ni pad and SiNW. Our initial annealing


094309-6

J. Appl. Phys. 105, 094309 共2009兲

Dellas et al.

FIG. 6. 共Color online兲 EELS spectra showing the Ni L2,3 edge for pure Ni; ␪-Ni2Si and NiSi2 showing a shift from pure Ni at the lowest energy to higher energy from ␪-Ni2Si and NiSi2 at the highest energy.

condition of 450 ° C is also lower than the lowest temperature studied by Lu et al., which was 500 ° C. Conversely, the work by Wu et al. shows NiSi formation on a 关112兴 oriented SiNW, differing from the ␪-Ni2Si phase we identified to form after annealing from 350 to 700 ° C. However, the annealing and processing conditions of Wu et al. differ significantly from our work as well. Wu et al. annealed at a temperature of 500 ° C for 5 min, then etched the Ni pad and proceeded with a secondary anneal at 600 ° C for an unspecified time period. They also used a very narrow Ni line to contact the SiNW and saw negligible lateral diffusion from the Ni source, contrary to the large lateral diffusion we and others have observed when using a large Ni pad to supply the silicidation reaction. We suspect that in the case of a thin Ni strip contacting the SiNW, the supply of Ni is cut off before enough Ni is able to diffuse into the SiNW to grow a long axial segment. The combination of the Ni supply being cut off during the reaction, and also complete removal of the Ni by etching followed by a secondary anneal, makes it difficult to compare the reaction products encountered in each of the experiments. Weber et al. studied the silicidation of 关110兴 SiNWs and found that NiSi2 formed after annealing at 480 ° C. The cross section of a 关110兴 SiNW is triangular and bound by 共111兲

planes.27 Both the ␪-Ni2Si and NiSi2 phases are well lattice matched to 共111兲 Si, and the ␪-Ni2Si phase is actually slightly better matched to Si on this plane, so one might expect it to be found in these conditions given our findings. However, the growth direction of the SiNW is 关110兴. There is a poor lattice match between the 共110兲 and any orientation of the ␪-Ni2Si phase. On the other hand, NiSi2 is matched with only a 0.9% mismatch in plane with 共110兲 Si. It seems from these relations identified that both the initial SiNW interface that the Ni supply is in contact with and the growth front of the silicide are important factors in determining silicide phase formation in SiNWs. IV. CONCLUSIONS

We have found that the reaction products of Ni thin films on SiNWs depend on the original growth direction of the SiNWs. For the case of 关112兴 oriented SiNWs, we observe the formation and stability of ␪-Ni2Si at annealing temperatures from 350 to 700 ° C for 2 min. After annealing at 700 ° C for 2 min, we find the formation of small branches that emerge from the main silicide segment due presumably to a large compressive stress built up in the silicided nanowire. Silicidation of 关111兴 oriented SiNWs results in formation of NiSi2 after annealing at 450 ° C for 2 min. Lastly, NiSi was found to form after annealing at 700 ° C for 2 min, but only in SiNWs with a 关111兴 orientation. From the initial epitaxial phases identified in the reactions of Ni films with SiNWs, it appears that interfacial energy reduction plays a significant role in phase transformations in SiNWs. ACKNOWLEDGMENTS

FIG. 7. TEM image of SiNW silicided by Ni pad at 700 ° C for 2 min. Inset shows SAED pattern of the NiSi 关103兴 zone axis.

The authors are grateful for financial support from the ARO 共Contract No. W911NF0510334兲 and NSF 共Contract No. ECS-0609282兲 and the use of facilities at the PSU NSF NNIN site under Grant No. 0335765. They also appreciate


094309-7

helpful suggestions of Professor Elizabeth Dickey and the Si nanowires provided by Professor Joan Redwing. 1

J. Appl. Phys. 105, 094309 共2009兲

Dellas et al.

Y. Cui, Z. Zhong, D. Wang, W. U. Wang, and C. M. Lieber, Nano Lett. 3, 149 共2003兲. 2 X. Duan, C. Niu, V. Sahi, J. Chen, J. W. Parce, S. Empedocles, and J. L. Goldman, Nature 共London兲 425, 274 共2003兲. 3 Y. Cui, Q. Wei, H. Park, and C. M. Lieber, Science 293, 1289 共2001兲. 4 C. Lavoie, F. M. d’Heurle, C. Detavernier, and C. Cabral, Jr., Microelectron. Eng. 70, 144 共2003兲. 5 A. Lauwers, M. de Potter, O. Chamirian, R. Lindsay, C. Demeurisse, C. Vrancken, and K. Maex, Microelectron. Eng. 64, 131 共2002兲. 6 C. S. Chang, C. W. Nieh, and L. J. Chen, Appl. Phys. Lett. 50, 259 共1987兲. 7 J. H. Gülpen, A. A. Kodentsov, and F. J. J. van Loo, Z. Metallkd. 86, 530 共1995兲. 8 F. M. d’Heurle, in Diffusion in Ordered Alloys, edited by B. Fultz, R. W. Cahn, and D. Gupta 共The Minerals, Metals and Materials Society, Warrendale, PA, 1993兲, p. 185. 9 M.-A. Nicolet and S. S. Lau, in VLSI Electronics Microstructure Sciences, edited by N. G. Einsprich and G. B. Larrabee 共Academic, New York, 1983兲, Vol. 6, Chap. 6, p. 343. 10 R. T. Tung, J. M. Poate, J. C. Bean, J. M. Gibson, and D. C. Jacobson, Thin Solid Films 93, 77 共1982兲. 11 R. T. Tung and F. Schrey, Appl. Phys. Lett. 55, 256 共1989兲. 12 V. Teodorescu, L. Nistor, H. Bender, A. Steegen, A. Lauwers, K. Maex, and J. Van Landuyt, J. Appl. Phys. 90, 167 共2001兲. 13 J. M. Gibson, J. L. Batstone, R. T. Tung, and F. C. Unterwald, Phys. Rev. Lett. 60, 1158 共1988兲. 14 J. Appenzeller, J. Knoch, E. Tutuc, M. Reuter, and S. Guha, Tech. Dig. -

Int. Electron Devices Meet. 2006, 1. K. C. Lu, K. N. Tu, W. W. Wu, L. J. Chen, B.-Y. Yoo, and N. V. Myung, Appl. Phys. Lett. 90, 253111 共2007兲. 16 K.-C. Lu, W.-W. Wu, H.-W. Wu, C. M. Tanner, J. P. Chang, L. J. Chen, and K. N. Tu, Nano Lett. 7, 2389 共2007兲. 17 J. T. Sheu, S. P. Yeh, C. H. Lien, and S. T. Tsai, Jpn. J. Appl. Phys., Part 1 45, 3686 共2006兲. 18 Y. Wu, J. Xiang, C. Yang, W. Lu, and C. M. Lieber, Nature 共London兲 430, 61 共2004兲. 19 Z. Zhang, J. Lu, P.-E. Hellstrom, M. Ostling, and S.-L. Zhang, Appl. Phys. Lett. 88, 213103 共2006兲. 20 W. M. Weber, L. Geelhaar, E. Unger, C. Cheze, F. Kreupl, H. Riechert, and P. Lugli, Phys. Status Solidi B 244, 4170 共2007兲. 21 K.-K. Lew and J. M. Redwing, J. Cryst. Growth 254, 14 共2003兲. 22 Y.-L. Jiang, A. Agarwal, G.-P. Ru, X.-P. Qu, J. Poate, B.-Z. Li, and W. Holland, Appl. Phys. Lett. 85, 410 共2004兲. 23 P. A. Smith, C. D. Nordquist, T. N. Jackson, T. S. Mayer, B. R. Martin, J. Mbindyo, and T. E. Mallouk, Appl. Phys. Lett. 77, 1399 共2000兲. 24 W. K. Chu, S. S. Lau, J. W. Mayer, H. Müller, and K. N. Tu, Thin Solid Films 25, 393 共1975兲. 25 E. B. Badriev, O. S. Petrushkova, and L. A. Panteleimonov, Vestn. Mosk. Univ., Ser. 2: Khim. 15, 367 共1974兲. 26 S. M. Woodruff, N. S. Dellas, B. Z. Liu, C. M. Eichfeld, T. S. Mayer, J. M. Redwing, and S. E. Mohney, J. Vac. Sci. Technol. B 26, 1592 共2008兲. 27 C.-P. Li, C.-S. Lee, X.-L. Ma, N. Wang, R.-Q. Zhang, and S.-T. Lee, Adv. Mater. 共Weinheim, Ger.兲 15, 607 共2003兲. 28 K. N. Tu, Phys. Rev. B 49, 2030 共1994兲. 29 H. Schmid, M. T. Björk, J. Knoch, H. Riel, W. Riess, P. Rice, and T. Topuria, J. Appl. Phys. 103, 024304 共2008兲. 30 M. C. Cheynet and R. Pantel, Micron 37, 377 共2006兲. 15
