Modeling and microstructural characterization of incubation, time-dependent drift and saturation during electromigration in Al–Si–Cu stripes

Resistance data of passivated Al–Si–Cu contact electromigration test structures clearly show three different stages: incubation, time-dependent drift and ultimately saturation. A detailed model describing all three stages was developed and evidenced by a thorough analysis of the tested material. By using this model, a length dependent lifetime can be calculated and data obtained from different test structures can be compared directly with each other. Moreover, it can be predicted that due to the length dependent saturation, lines below a certain length will never reach the failure criterion! Taking this information into account during the design phase leads to very robust interconnects.     

Electromigration behavior of dual-damascene Cu interconnects––Structure, width, and length dependences

Experiments were performed to study the effect of line width and length, and the results revealed interesting differences in electromigration behavior of via-fed upper and lower layer dual-damascene test structures. The observed location of electromigration induced void in upper and lower layer test structures cannot be completely explained by the theory of current gradient induced vacancy diffusion. The electromigration median time to failure (MTF) were found to be dependent upon the line width for the lower layer test structures while it remained unaffected in the case of upper layer test structure. Cu/dielectric cap interface acting as the dominant electromigration path and the current crowding location being near the Cu/dielectric cap interface for lower layer structures due to structural differences, explain this behavior. Similarly, short length upper and lower layer test structures exhibited completely different characteristics. The back stress effect on short lines was evident on both upper and lower layer structures, however, only the upper layer showed two distinct via and line failure mechanisms. These observed effects are specific to Cu dual-damascene structures and can have major technological implications for electromigration reliability assessment.     

Electromigration of Sn–37Pb and Sn–3Ag–1.5Cu/Sn–3Ag–0.5Cu composite flip–chip solder bumps with Ti/Ni(V)/Cu under bump metallurgy

We examine electromigration fatigue reliability and morphological patterns of Sn–37Pb and Sn–3Ag–1.5Cu/Sn–3Ag–0.5Cu composite solder bumps in a flip–chip package assembly with Ti/Ni(V)/Cu UBM. The flip–chip test vehicle was subjected to test conditions of five combinations of applied electric currents and ambient temperatures, namely, 0.4 A/150 °C, 0.5 A/150 °C, 0.6 A/125 °C, 0.6 A/135 °C, and 0.6 A/150 °C. The electrothermal coupling analysis was employed to investigate the current crowding effect and maximum temperature in the solder bump in order to correlate with the experimental electromigration reliability using the Black’s equation as a reliability model. From available electromigration reliability models, we also present a comparison between fatigue lives of Sn–37Pb solder bumps with Ti/Ni(V)/Cu and those with Al/Ni(V)/Cu UBM under different current stressing conditions.     

Electromigration failure distributions of dual damascene Cu /low – k interconnects

The reliability of dual damascene Cu/low – k interconnects is limited by electromigration – induced void formation at vias. In this paper we investigate via void morphologies and associated failure distributions at the low percentiles typical of industry reliability requirements. We show that Cu/low – k reliability is fundamentally limited by the formation of slit – voids under vias. Using experimental and simulation approaches we clarify the practical importance of apparent incubation phenomena associated with this failure mode.     

Impact of O–Si–O bond angle fluctuations on the Si–O bond-breakage rate

We extend the McPherson model for the silicon–oxygen bond-breakage in a manner to capture the impact of the O–Si–O angle fluctuations (typical for amorphous SiO₂) on the breakage rate. In the McPherson model the transition of the Si ion from the 4-fold coordinated position to the 3-fold coordination is considered as rupture of the Si–O bond. We have studied the potential barrier (separating these saddle points) transformation induced by the O–Si–O bond angle variations and found that the secondary minimum occurs at a critical angle of about 107.75°. Since the Si ion “finds” the way corresponding to the highest breakage probability we used the two-dimensional downhill simplex method in order to find the direction of this maximal rate. It was shown that if the O–Si–O angle deviates from its nominal value 109.48° (typical for α-quartz) corresponding to the regular SiO₄ tetrahedron the symmetry aggravates and the secondary minimum is rotated. Calculated dependencies of the breakage rate on the electric field demonstrate the linear slope in the log–lin scale thus reflecting the linear reduction of the activation energy for the bond-breakage vs. field. The family of distribution functions for breakage rate calculated with a fixed step of field shows that the curves do not change their form and are shifted in parallel with the field. This tendency supports the thermo-chemical model for the bond-breakage also in the case of strongly fluctuating O–Si–O angles. As a consequence, dependencies of the mean value of the rate, its standard deviation and the nominal rate (calculated for the angle of 109.48°) have the same slope on a log–lin scale. The wide spread of the breakage rate is reflected by the high value of its standard deviation.     

Mechanism of pre-annealing effect on electromigration immunity of Al–Cu line

In this work, we have investigated effects of pre-annealing, which means annealing performed prior to electromigration (EM) test, on EM lifetime of Al–Cu lines. We also investigated the relationships between void formation and size of Cu precipitated area in the line under various pre-annealing conditions. It is found that EM lifetime decreases while the size of the Cu precipitated area increases with lengthening of the pre-annealing period. However, no void is observed after this pre-annealing treatment. The results indicate that the tiny voids generated by formation of Cu precipitation do not move during the pre-annealing period. In the case of EM testing, Cu precipitation occurs followed by void formation at the cathode area, probably due to diffusion of vacancies which are generated by Cu atom movement by electron wind. As a result, resistance of the line increases and eventually it fails completely.It is demonstrated that pre-annealing helps Cu atoms to accumulate at the grain boundary forming the Cu precipitates. However, in samples with no pre-annealing treatment, the accumulation of Cu atoms at the grain boundaries begins just after the start of the EM testing and then the Cu precipitates diffuse toward the anode. Since EM test conditions are the same for samples with and without pre-annealing treatment, the only variation is the incubation time to accumulate Cu atoms at the grain boundaries. This is the reason why EM lifetime of pre-annealed samples is shorter than that of samples with no pre-annealing treatment.     

Study of Ta–Si–N thin films for use as barrier layer in copper metallizations

This work focuses on the deposition process, microanalytical characterization and barrier behaviour of 10–100-nm thick sputtered Ta–Si and Ta–Si–N films. Pure Ta–Si films were found to be already nanocrystalline. The addition of N₂ leads to a further grain fining resulting in amorphous films with excellent thermal stability. According to microanalytical investigations, Ta–Si barriers between Cu and Si with a thickness of only 10 nm are not stable at 600°C. Copper silicides are formed due to intensive Cu diffusion throughout the barrier. In contrast, 10-nm thick nitrogen-rich Ta–Si–N barriers remain thermally stable during annealing at 600°C and protect the Si wafer from Cu indiffusion.     

Electromigration Reliability and Morphologies of 62Sn–36Pb–2Ni and 62Sn–36Pb–2Cu Flip-Chip Solder Joints

The near-eutectic Sn-Pb-Cu and Sn-Pb-Ni ternary solder alloys were developed based on the consideration of strength and fatigue reliability enhancement of solder joints in part via the altering of formation of interfacial intermetallic compounds. In this work, we examine electromigration reliability and morphologies of 62Sn-36Pb-2Ni and 62Sn-36Pb-2Cu flip-chip solder joints subjected to two test conditions that combine different average current densities and ambient temperatures: 5 kA/cm² at 150 degC and 20 kA/cm² at 3 degC. Under the test condition of 5 kA/cm² at 150 degC, 62Sn-36Pb-2Cu is overwhelmingly better than 62Sn-36Pb-2Ni in terms of electromigration reliability. However, under the test condition of 20 kA/cm2 at 30 degC, the electromigration fatigue life of 62Sn-36Pb-2Ni shows a profuse enhancement and exceeds that of 62Sn-36Pb-2Cu. Electromigration-induced morphologies are also examined on the cross sections of solder joints using scanning electron microscopy.     

The influence of the passivation material on stress voiding in Al–Cu alloys

Storage tests have been performed to obtain information on the influence of the passivation material and its geometry on the mechanical reliability of Al–Cu lines. During storage tests, the stress in the lines is tensile and depends on the passivation material and passivation geometry. The passivation was either a SiO₂ layer or a SiN_x layer. In both cases the influence of a conformal and a planarised passivation geometry has been studied. Passivating the lines with a material with a higher stiffness such as SiN_x will increase the stress void density in the lines. Moreover, a conformal passivation layer induces less stress voids in the metal than a planarised passivation, deposited by a Dep/Etch method. The number of stress voids saturates within 24 h at 200°C. However, the voids continue to grow during longer storage times. When the lines are passivated with the lower stiffness material SiO₂, stress voids have not been observed after storage testing.     

Potential of amorphous Mo–Si–N films for nanoelectronic applications

The properties of amorphous metallic molybdenum–silicon–nitrogen (Mo–Si–N) films were characterised for use in nanoelectronic applications. The films were deposited by co-sputtering of molybdenum and silicon targets in a gas mixture of argon and nitrogen. The atomic composition, microstructure and surface roughness were studied by RBS, TEM and AFM analyses, respectively. The electrical properties were investigated in the temperature range 80 mK to 300 K. No transition into a superconductive state was observed. Nanoscale wires were fabricated using electron beam lithography with their properties measured as a function of temperature.     

Electromigration Reliability and Morphologies of Cu Pillar Flip-Chip Solder Joints with Cu Substrate Pad Metallization

The Cu pillar is a thick underbump metallurgy (UBM) structure developed to alleviate current crowding in a flip-chip solder joint under operating conditions. We present in this work an examination of the electromigration reliability and morphologies of Cu pillar flip-chip solder joints formed by joining Ti/Cu/Ni UBM with largely elongated ∼62 μm Cu onto Cu substrate pad metallization using the Sn-3Ag-0.5Cu solder alloy. Three test conditions that controlled average current densities in solder joints and ambient temperatures were considered: 10 kA/cm² at 150°C, 10 kA/cm² at 160°C, and 15 kA/cm² at 125°C. Electromigration reliability of this particular solder joint turns out to be greatly enhanced compared to a conventional solder joint with a thin-film-stack UBM. Cross-sectional examinations of solder joints upon failure indicate that cracks formed in (Cu,Ni)₆Sn₅ or Cu₆Sn₅ intermetallic compounds (IMCs) near the cathode side of the solder joint. Moreover, the ~52-μm-thick Sn-Ag-Cu solder after long-term current stressing has turned into a combination of ~80% Cu-Ni-Sn IMC and ~20% Sn-rich phases, which appeared in the form of large aggregates that in general were distributed on the cathode side of the solder joint.     

High-Performance Al–Sn–Zn–In–O Thin-Film Transistors: Impact of Passivation Layer on Device Stability

We fabricated high-performance thin-film transistors (TFTs) with an amorphous-Al–Sn–Zn–In–O (a-AT-ZIO) channel deposited by cosputtering using a dual Al–Zn–O and In–Sn–O target. The fabricated AT-ZIO TFTs, which feature a bottom-gate and bottom-contact configuration, exhibited a high field-effect mobility of 31.9 $ hbox{cm}^{2}/hbox{V}cdothbox{s}$, an excellent subthreshold gate swing of 0.07 V/decade, and a high $I_{{rm on}/{rm off}}$ ratio of $≫hbox{10}^{9}$, even below the process temperature of 250 $^{circ}hbox{C}$. In addition, we demonstrated that the temperature and bias-induced stability of the bottom-gate TFT structure can significantly be improved by adopting a suitable passivation layer of atomic-layer-deposition-derived $hbox{Al}_{2} hbox{O}_{3}$ thin film.      

Effects of thermal annealing on the structural properties of sputtered W–Si–N diffusion barriers

W–Si–N thin films were deposited via rf-magnetron sputtering from a W₅Si₃ target in Ar/N₂ reactive gas mixtures over a large range of compositions, obtained by varying the partial flow of nitrogen within the reaction chamber. The samples of each set were then thermally annealed in vacuum at different temperatures up to 980 °C.Film composition was determined by Rutherford backscattering spectrometry (RBS), surface film morphology by scanning electron microscopy (SEM), micro-structure by transmission electron microscopy (TEM), vibrational properties by FT-IR absorption and Raman scattering spectroscopy, and electrical resistivity by four-point probe measurements.Independently of the deposition conditions, all the as-deposited films have an amorphous structure, while their composition varies, showing an increase of Si/W ratio from 0.1 up to 0.55 when the nitrogen concentration in the films increases from 0 to 60 at%. Thermal treatments in vacuum induce an important loss of nitrogen in the nitrogen-rich samples, especially at temperatures higher than 600 °C. Samples with high nitrogen content preserve their amorphous structure even at the highest annealing temperature, despite the chemical bonding ordering observed by means of FT-IR measurements. Raman spectroscopy of as-deposited films rich in nitrogen suggests the presence of an important amorphous silicon nitride component, but fails to detect any structural rearrangement either within the composite matrix of film or within silicon nitride component. Segregation of metallic tungsten was detected by TEM in the annealed sample with lowest nitrogen content (W₅₈Si₂₁N₂₁). Finally, the resistivity of the films increases with the N content, while the loss of nitrogen accompanies the decrease of resistivity especially of samples with high nitrogen content.     

Tuning the Bandgap Character of Quantum‐Confined Si–Sn Alloyed Nanocrystals

Nanocrystals in the regime between molecules and bulk give rise to unique electronic properties. Here, a thorough study focusing on quantum‐confined nanocrystals (NCs) is provided. At the level of density functional theory an approximate (quasi) band structure which addresses both the molecular and bulk aspects of finite‐sized NCs is calculated. In particular, how band‐like features emerge with increasing particle diameter is shown. The quasiband structure is used to discuss technological‐relevant direct bandgap NCs. It is found that ultrasmall Sn NCs have a direct bandgap in their at‐nanoscale‐stable α‐phase and for high enough Sn concentration (≈41%) alloyed Si–Sn NCs transition from indirect to direct bandgap semiconductors. The calculations strongly support recent experiments suggesting a direct bandgap for these systems. For a quantitative comparison many‐body GW + Bethe–Salpeter equation (BSE) calculations are performed. The predicted optical gaps are close to the experimental data and the calculated absorbance spectra compare well with the corresponding measurements.     

Growth-direction-dependent characteristics of Ge-on-insulator by Si–Ge mixing triggered melting growth

The lateral liquid-phase epitaxy of Ge-on-insulator (GOI) using Si seeds has been investigated as a function of the Si-seed orientation and the growth direction. Giant single-crystalline GOI structures with ∼200 μm length are obtained using Si(1 0 0), (1 1 0), and (1 1 1) seeds. The very long growth is explained on the basis of the solidification temperature gradient due to Si-Ge mixing around the seeding area and the thermal gradient due to the latent heat around the solid/liquid interface at the growth front. In addition, growth with rotating crystal orientations is observed for samples with several growth directions. The rotating growth is explained on the basis of the bonding strength between lattice planes at the growth front. This rotating growth does not occur in any direction for (1 0 0) orientated seeds. Based on this finding the mesh-patterned GOI growth with a large area (250 μm × 500 μm) is demonstrated.     

Evaluation of performance–reliability trade-offs in a Si–Ge BiCMOS process using fast wafer level techniques

Fast wafer-level reliability (fWLR) techniques are successfully implemented in order to investigate several gate oxide reliability–performance tradeoffs that affect the architecture of a high speed BiCMOS process. Fast feedback of device and reliability parameters is required during process development in order to avoid failures during process qualification. This study highlights some performance–reliability tradeoffs that had to be overcome during the development of a modern BiCMOS process.     

Analysis of Ti–Si–N diffusion barrier films obtained by r.f. magnetron sputtering

Thin films of Ti–Si–N are deposited by r.f. magnetron sputtering in a Ar/N₂ gas mixture. The magnetron discharge is operated at 10 mTorr with 5 and 10% N₂ in the gas mixture and r.f. powers ranging from 100 to 200 W. The composition and electrical resistivity of the thin films were determined by energy dispersive X-ray spectroscopy and the four-point probe method, respectively. The structure of the films was determined by high-resolution transmission electron microscopy. The Ti–Si–N films were either amorphous or contained cubic TiN nanosized grains in an amorphous phase. The diffusion barrier properties of 10-nm thick film between Cu and Si were studied from 500 to 700°C. The highest failure temperature of 650°C was obtained for Ti_37.5 Si₂₇ N_35.5 which. contains 4-nm TiN crystallites in an amorphous phase.     

Growth kinetics,properties, performance,and stability of atomic layer deposition Zn–Sn–O buffer layers for Cu(In,Ga)Se2 solar cells

A new atomic layer deposition process was developed for deposition of Zn–Sn–O buffer layers for Cu(In,Ga)Se₂ solar cells with tetrakis(dimethylamino) tin, Sn(N(CH₃)₂)₄, diethyl zinc, Zn(C₂H₅)₂, and water, H₂O. The new process gives good control of thickness and [Sn]/([Sn] + [Zn]) content of the films. The Zn–Sn–O films are amorphous as found by grazing incidence X‐ray diffraction, have a high resistivity, show a lower density compared with ZnO and SnO_x, and have a transmittance loss that is smeared out over a wide wavelength interval. Good solar cell performance was achieved for a [Sn]/([Sn] + [Zn]) content determined to be 0.15–0.21 by Rutherford backscattering. The champion solar cell with a Zn–Sn–O buffer layer had an efficiency of 15.3% (V_oc = 653 mV, J_sc(QE) = 31.8 mA/cm², and FF = 73.8%) compared with 15.1% (V_oc = 663 mV, J_sc(QE) = 30.1 mA/cm², and FF = 75.8%) of the best reference solar cell with a CdS buffer layer. There is a strong light‐soaking effect that saturates after a few minutes for solar cells with Zn–Sn–O buffer layers after storage in the dark. Stability was tested by 1000 h of dry heat storage in darkness at 85 °C, where Zn–Sn–O buffer layers with a thickness of 76 nm retained their initial value after a few minutes of light soaking. Copyright © 2011 John Wiley & Sons, Ltd.     

A Li–Al–O Solid‐State Electrolyte with High Ionic Conductivity and Good Capability to Protect Li Anode

Using a solid‐state electrolyte (SSE) to stabilize the Li metal anode is widely considered a promising route to develop next‐generation high energy density lithium batteries. Here, a new polycrystalline aluminate‐based SSE (named Li–Al–O SSE) with good capability is introduced to protect Li metal. The SSE is formed on the Li metal surface via a chemical reaction between LiOH and triethylaluminum (TEAL) with the existence of LiTFSI‐based electrolyte. It is a continuous film that consists of polycrystalline LiAlO₂, Li₃AlO₃, Al₂O₃, Li₂CO₃, LiF, and some organic compounds. Such Li–Al–O SSE possesses a room‐temperature ionic conductivity as high as 1.42 × 10^?4 S cm^?1. Meanwhile, it effectively protects the Li anode from the corrosion of H₂O, O₂, and organic solvent, and suppresses the growth of Li dendrite. With the protection of the Li–Al–O SSE, the cycle life of Li|Li symmetric cell and Li|O₂ cell is substantially elongated, indicating that the SSE exhibits an excellent protective effect under both inert and oxidizing circumstances.     

Effects of fourth alloying additive on microstructures and tensile properties of Sn–Ag–Cu alloy and joints with Cu

The effects of the fourth elements, i.e., Fe, Ni, Co, Mn and Ti, on microstructural features, undercooling characteristics, and monotonic tensile properties of Sn–3 wt.%Ag–0.5 wt.%Cu lead-free solder were investigated. All quaternary alloys basically form third intermetallic compounds in addition to fine Ag₃Sn and Cu₆Sn₅ and exhibit improved solder structure. The precipitates of Sn–3Ag–0.5Cu (–0.1 wt.%X; X=Ni, Ti and Mn) alloy are very fine comparing with the other alloys. The effective elements for suppressing undercooling in solidification are Ti, Mn, Co and Ni. All quaternary bulk alloys exhibit similar or slightly larger tensile strengths; especially Mn and Ni can improve elongation without degrading strength. The interfacial phases of Sn–3Ag–0.5Cu (–0.1 wt.%X; X=Fe, Mn and Ti)/Cu joints are typical Cu₆Sn₅ scallops. Sn–3Ag–0.5Cu (–0.1 wt.%X; X=Ni and Co)/Cu joints form very fine Sn–Cu–Ni and Sn–Cu–Co scallops at interface. The Cu/Sn–3Ag–0.5Cu–0.1Ni/Cu joint exhibits improved tensile strength prior to thermal aging at 125 and 150 °C. The fracture surface of Cu/Sn–3Ag–0.5Cu/Cu joint exhibits mixture of ductile and brittle fractures, while Cu/Sn–3Ag–0.5Cu (–0.1X; X=Ni and Co)/Cu joints exhibit only brittle fracture at interface. The Sn–3Ag–0.5Cu–0.1Ni alloy is more reliable solder alloy with improved properties for all tests in the present work.