• Journal of Welding and Joining
• /
• v.13 no.3
• /
• pp.106-115
• /
• 1995

To know the bonding phenomena of Ti/Cu brazed joint, a characteristic of microstructures in bonding interlayer of vacuum brazed pure Ti to Cu has been studied in the temperature range from 1088 to 1133K for various bonding times using Ag-28wt%Cu filler metal. Also intermediate phases formed in bonded interlayer and behavior of layer growth have been investigated. The obtained results in this study are as follows: 1) Liquid insert metal width at the each brazing temperature was proportional to the square root of brazing time, and it was considered that the liquid insert metal width was controlled by the diffusion rate process of primary .alpha.-Cu formed at the Ti side. 2) Intermediate phases formed near the Ti interface were .betha.-Ti and intermetallic compounds TiCu, Ti$_{2}$Cu, Ti$_{3}$Cu, and TiCu. 3) .betha.-Ti formed in Ti base metal durig brazing transformed to lamellar structure, .alpha.-Ti + Ti$_{2}$Cu. The structure came from the eutectoil decomposition reaction in cooling. And the width of .betha.-Ti layer was proportional to the square root of brazing time, and it was considered that the growth of .betha.-Ti layer was controlled by interdiffusion rate process in .betha.-Ti. 4) The layer growth of TiCu, Ti$_{3}$Cu$_{4}$ and TiCu, phases formed near the Ti interface was linerface was linearly proportional to the brazing time, and it was considered that the layer growth of these phases was controlled by the chemical reaction rate at the interface.

• Korean Journal of Materials Research
• /
• v.20 no.6
• /
• pp.338-343
• /
• 2010

As an alternative to the W plug used in MOSFETs, a Cu plug with a NiSi contact using Ta / TaN as a diffusion barrier is currently being considered. Conventionally, Ni was first deposited and then NiSi was formed, followed by the barrier and Cu deposition. In this study, Ti was employed as a barrier material and simultaneous formation of the NiSi contact and Cu plug / Ti barrier was attempted. Cu(100 nm) / Ti / Ni(20 nm) with varying Ti thicknesses were deposited on a Si substrate and annealed at $4000^{\circ}C$ for 30 min. For comparison, Cu/Ti/NiSi thin films were also formed by the conventional method. Optical Microscopy (OM), Scanning Probe Microscopy (SPM), X-Ray Diffractometry (XRD), and Auger Electron Microscopy (AES) analysis were performed to characterize the inter-diffusion properties. For a Ti interlayer thicker than 50 nm, the NiSi formation was incomplete, although Cu diffusion was inhibited by the Ti barrier. For a Ti thickness of 20 nm and less, an almost stoichiometric NiSi contact along with the Cu plug and Ti barrier layers was formed. The results were comparable to that formed by the conventional method and showed that this alternative process has potential as a formation process for the Cu plug/Ti barrier/NiSi contact system.

• Journal of the Korean Ceramic Society
• /
• v.35 no.1
• /
• pp.55-65
• /
• 1998

The oxidation behavior of Ag-36.8a%Cu-7.4at%Ti alloy brazed on Si3N4 substrate was investigated at 400, 500 and 600$^{\circ}C$ in air. Under this experimental condition Si3N4 and Ag were not oxidized whereas Cu and Ti among the brazing alloy components were oxidizied obeying the parabolic oxidation rate law. The activation energy of oxidation was found to be 80kj/ mol which was smaller than that of pure Cu owing to the presence of oxygen active element of Ti. The outer oxide scale formed from the initial oxidation state was always composed of Cu oxides which were known to be growing by the outward diffusion of Cu ions. As the oxidation progressed the concentration gradient occurred due to the continuous consumption of Cu as Cu oxides and consequently build-up of an Ag-enriched layer below the Cu oxides resulted in the formation of multiple oxide scales composed of Cu oxide (CuO) /Ag-enriched layer/Cu oxide (Cu2O) /Ag-enriched layer. Also the inward diffusing of oxygen through Cu oxide and Ag-enriched layers led to the formation of internal oxides of TiO2.

• Journal of Powder Materials
• /
• v.6 no.4
• /
• pp.270-276
• /
• 1999

Ti-Ni-Cu alloy powders were fabricated by ball milling, and the properties of these powders were characterized. Mixed 50Ti-(50-x)Ni-xCu powders of 5 to 10at.%Cu composition were milled for 100 hours using SUS 1/4" balls in argon atmosphere. Ball to powder ratio was 20:1 and rotating speed was 100 rpm. Tensile strength, microstructure and phase transformation of ball milled Ti-(50-x)Ni-xCu powders were studied. After 100 hours milling, Ti, Ni and Cu elements were alloyed completely and an amorphous phase was formed. Amorphous phase was crystallized to martensite(B 19') and austenite(B2) after heat treatment for 1 hour at $850^{\circ}C$. As the Cu contents were increased, tensile strength of extruded 6061Al/TiNiCu was decreased, and B19'martensite phases In the TiNi particles were the causes of high tensile stress of extruded 6061Al/TiNiCu.NiCu.

• Korean Journal of Materials Research
• /
• v.5 no.2
• /
• pp.169-177
• /
• 1995

The diffusion barrier property of 100-nm-thick titanium nitride (TiN) film between Cu and Si was investigated using sheet resistance measurements, etch-pit observation, x-ray diffractometry, Auger electron spectroscopy, and transmission electron microscopy. The TiN barrier fails due to the formation of crystalline defects (dislocations) and precipitates (presumably Cu-silicides) in the Si substrate which result from the predominant in-diffusion of Cu through the TiN layer. In contrast with the case of Al, it is identified that the TiN barrier fails only the in-diffusion of Cu because there is no indication of Si pits in the Si substrate. In addition, it appears that the stuffing of TiN does not improve the diffusion barrier property in the Cu/TiN/Si structure. This indicates that in the case of Al, the chemical effect that impedes the diffusion of Al by the reaction of Al with $TiO_{2}$ which is present in the grain boundaries of TIN is very improtant. On the while, in the case of Cu, there is no chemical effect because Cu oxides, such as $Cu_{2}O$ or CuO, is thermodynamically unstable in comparison with $TiO_{2}$. For this reason, it is considered that the effect of stuffing of TiN on the diffusion barrier property is not significant in the Cu/ TiN/Si structure.

• Korean Journal of Materials Research
• /
• v.11 no.5
• /
• pp.361-366
• /
• 2001

It is difficult to obtain high Cu nucleation density and continuous Cu films in Cu-MOCVD without cleaning the TiN substrate prior to Cu deposition. In this study effects of plasma precleaning on the Cu nucleation density were investigated using SEM, XPS, AES, AFM analyses. Direct plasma pretreatment is much more effective than remote plasma pretreatment in enhancing Cu nucleation. Cleaning effects are enhanced with increasing the rf-power and the plasma exposure time in hydrogen plasma pretreatment. The mechanism through which Cu nucleation is enhanced by plasma pretreatment is as follows: Hydrogen ion\ulcorner in the hydrogen plasma react with TiN to form Ti and $NH_3$ Cu nucleation is easier on the Ti substrate than TiN substrate.

• Journal of Korean Society on Water Environment
• /
• v.21 no.1
• /
• pp.84-91
• /
• 2005

$TiO_2-catalyst$ suspensions work efficiently in Photocatalytic oxidation (PCO) for wastewater treatment. Nevertheless, once photocatalysis is completed, separation of the catalyst from solution becomes the main problem. The PCO of Cu(II)-EDTA was studied to determine the reusability of the titanium dioxide catalyst. Aqueous solutions of $10^{-4}M$ Cu(II)-EDTA were treated using illuminated $TiO_2$ particles at pH 6 in a circulating reactor. $TiO_2$ was reused in PCO system for treatment of Cu(II)-EDTA comparing two procedures: reuse of water and $TiO_2$ and reuse of the entire suspension after PCO of Cu(II)-EDTA. The results are as follows; (i) Photocatalytic efficiency worsens with successive runs when catalyst and water are reused without separation and filtration, whereas, when $TiO_2$ is separated from water, the reused $TiO_2$ is not deactivated. (ii) The $TiO_2$ mean recovery (%) with reused $TiO_2$ was 86.4%(1.73g/L). Although the mean initial degradation rate of Cu(II)-EDTA and Cu(II) was lower than that using fresh $TiO_2$, there was no significant change in the rate during the course of the three-trial experiment. It is suggested that Cu(II)-EDTA could be effectively treated using an recycling procedure of PCO and catalyst recovery. (iii) However, without $TiO_2$ separation, the loss of efficiency of the PCO in the use of water and $TiO_2$ due to Cu(II), DOC remained from previous degradation and Cu(II)-EDTA added to the same suspension was observed after 2 trials, and resulted in the inhibition of the Cu(II)-EDTA, Cu(II) and DOC destruction.

• Journal of the Korean Ceramic Society
• /
• v.36 no.8
• /
• pp.835-842
• /
• 1999

Si3N4 -Si3N4 joints were made using Ag-Cu-Ti and Ag-Cu-In-Ti via brazing method and the change in joint strength was investigated after heat treatment at $400^{\circ}C$ or $650^{\circ}C$ for up to 2000h. The initial strength of as-brazed joints with Ag-Cu-In-Ti was lower but the reduction of the strength was less dramatic than that with Ag-Cu-Ti. The joints made of a new brazing alloy Au-Ni-Cr-Mo-Fe which is developed for high temperature applications were heat-treated at $650^{\circ}C$ for 1000h. As the heat treatment time increased the bond strength increased. The results of the joining system with Mo or Cu interlayer showed that the strength of the joint with Mo interlayer was higher but the system incurred problems in joint production Also it was found from oxidation experiment that Ti and In affected the oxidation resistance of brazing alloy.

• Korean Journal of Materials Research
• /
• v.13 no.5
• /
• pp.338-342
• /
• 2003

Adhesion between Cu and low-k films has been investigated. Low-k films deposited using a mixture of hexamethyldisilane(HMDS) and Para-xylene had a dielectric constant as low as 2.7, showing the thermally stable properties up to $400^{\circ}C$. In this study, Ti glue layer, boron dopant, and $N_2$plasma treatment were used to improve adhesion property of between Cu and low-k films. Ti glue layer slightly improved adhesion property. After $N_2$plasma treatment, the adhesion property was significantly improved due to the increased roughness and the formation of new binding states between Ti and plasma-treated PPpX : HMDS. However, $300^{\circ}C$ annealing of $N_2$plasma treated sample caused the diffusion of Cu into the PPpX : HMDS, degrading the low-k properties. In the case of Cu(B)/Ti/PPpX : HMDS, the adhesion was remarkably increased. This enhanced adhesion was attributed to formation of Ti-boride at the Cu-Ti interface. It is because the formed Ti-boride prevented the diffusion of Cu into the PPpX : HMDS and the Cu-Ti reaction at the Ti interface.

• Journal of the Microelectronics and Packaging Society
• /
• v.19 no.3
• /
• pp.37-41
• /
• 2012

The wafer level stacking with Cu-to-Cu bonding becomes an important technology for high density DRAM stacking, high performance logic stacking, or heterogeneous chip stacking. Cu CMP becomes one of key processes to be developed for optimized Cu bonding process. For the ultra low-k dielectrics used in the advanced logic applications, Ti barrier has been preferred due to its good compatibility with porous ultra low-K dielectrics. But since Ti is electrochemically reactive to Cu CMP slurries, it leads to a new challenge to Cu CMP. In this study Ti barrier/Cu interconnection structure has been investigated for the wafer level 3D integration. Cu CMP wafers have been fabricated by a damascene process and two types of slurry were compared. The slurry selectivity to $SiO_2$ and Ti and removal rate were measured. The effect of metal line width and metal density were evaluated.