Search

Conformal chemical vapor deposition (CVD) of silicon carbide (SiC) from methyltrichlorosilane (MTS) and hydrogen (H2) onto high-aspect-ratio (HAR; typically >100:1) three-dimensional features has been a challenge in the fabrication of ceramic matrix composites. In this study, the impact of heterogeneous underlayers on the initial nucleation of SiC-CVD was studied using HAR (1000:1) microchannels with a tailored wetting underlayer of Si(100) and dewetting underlayers of thermally formed amorphous silicon dioxide (a-SiO2) and turbostratic boron nitride (t-BN). Incubation periods were distributed in the microchannels on a-SiO2 and t-BN underlayers, with the longest period of 70 min found at the feature-bottom due to a decreased concentration (C) of film-forming species. The longer incubation periods with more dewetting underlayers arose due to demoted initial nucleation. Prolonged incubation at the feature bottom led to poor conformality because thick films had already formed at the inlet when film formation began at the feature bottom. The incubation periods were eliminated by increasing the supply of MTS/H2, in accordance with classical heterogeneous nucleation theory. In the meantime, carbon-rich SiC films formed in the vicinity of dewetting a-SiO2 and t-BN underlayers at the feature bottoms, with greater carbon segregation on more dewetting underlayers. This was probably due to the deposition of pyrocarbons (CH4, C2H2, and/or C2H4) generated from MTS/H2 in the gas phase. Decreasing the temperature (T) from 1000 to 900 °C prevented carbon-rich film formation, and the expected deposition rate of pyrocarbon decreased to 0.6% for the case of CH4. A higher C of MTS/H2 combined with a lower T enabled conformal and stoichiometric film formation on the heterogeneous HAR features.

The kinetics of heterogeneous nucleation during chemical vapor deposition (CVD) is still unclear despite its importance. Nucleation delay is often observed in many CVD processes, which is known as the incubation period ( τi). In this study, the effects of concentration ( C) and sticking probability ( η) of film-forming species on τi were formulated based on our kinetic model. To discuss the kinetics, [Formula: see text] with the rate dimension was used and formulated using C and η. Because η onto heterogeneous surfaces ( ηhetero) is difficult to evaluate, the study was initiated with η onto homogeneous surfaces ( ηhomo), followed by a discussion on its reasonability. The formulation was validated using the experimental dataset for SiC-CVD from CH3SiCl3/H2 onto BN underlayers because CVD involves multiple film-forming species with different ηhomo ranging from 10−6 to 10−2 and thus is a suitable system for studying the effect of ηhomo. High-aspect-ratio (1000:1) parallel-plate microchannels consisting of τi-involving BN and a τi-free Si surface were utilized to separate these film-forming species along the microchannel depth. τi was exceptionally long, up to several hours, depending on the CVD conditions. [Formula: see text] was found to be proportional to C n, where n is the reaction order. n was quantified as ≈1.6, suggesting the initial nucleation was triggered by the impingement of two adspecies in the second order and lowered possibly by the discrepancy between C in the gas-phase and that actually producing adspecies on the surface. [Formula: see text] was also found to be proportional to ηhomo. The exceptionally long τi was likely originated from the significantly lower ηhetero than ηhomo and the higher activation energy for ηhetero than that for ηhomo.

We propose a new, concise method for conformal chemical vapor deposition (CVD) using sacrificial layers (SLs) to fill three-dimensional features with microscopic pores. SLs are porous membranes (e.g., ceramic felts) that filter film-forming species having high sticking-probability (η). CVD processes with multiple film-forming species generally suffer from poor conformality due to preferential film deposition at the inlets of features by the high-η species, such as reactive intermediates. An SL traps such high-η species before they reach the target features and selectively supplies film-forming species with lower η (e.g., source precursors or stable intermediates) that enables conformal film deposition. Here the trapping efficiency of an SL was predicted and a procedure for designing an optimal SL was established. The procedure was demonstrated by CVD of silicon carbide (SiC) with multiple film-forming species of high-η species (η = 8.0 × 10-3) and lower-η species (η = 5.9 × 10-5 and 2.2 × 10-7). The trapping of 99.2% of incident high-η species was achieved with an optimized SL, wherein the deposition rate (m/s) contribution by high-η species declined from 0.546 at the SL inlet to 0.014 at its outlet. Finally, using these optimized SLs, SiC-CVD filling of micron-scale trenches was demonstrated with an aspect-ratio of 16:1.

Immiscible semiconductors are of premier importance since the source of lighting has been replaced by white light-emitting-diodes (LEDs) composed of thermodynamically immiscible InxGa1−xN blue LEDs and yellow phosphors. For realizing versatile deep-ultraviolet to near-infrared light-emitters, Al1−xInxN alloys are one of the desirable candidates. Here we exemplify the appearance and self-formation sequence of compositional superlattices in compressively strained m-plane Al1−xInxN films. On each terrace of atomically-flat m-plane GaN, In- and Al-species diffuse toward a monolayer (ML) step edge, and the first and second uppermost $$\stackrel{-}{1}\stackrel{-}{1}20$$ 1 - 1 - 20 > cation-rows are preferentially occupied by Al and In atoms, respectively, because the configuration of one In-N and two Al-N bonds is more stable than that of one Al-N and two In-N bonds. Subsequent coverage by next $$\stackrel{-}{1}\stackrel{-}{1}20$$ 1 - 1 - 20 > Al-row buries the $$\stackrel{-}{1}\stackrel{-}{1}20$$ 1 - 1 - 20 > In-row, producing nearly Al0.5In0.5N cation-stripe ordering along [0001]-axis on GaN. At the second Al0.72In0.28N layer, this ordinality suddenly lessens but In-rich and In-poor $$\stackrel{-}{1}\stackrel{-}{1}20$$ 1 - 1 - 20 >-rows are alternately formed, which grow into respective {0001}-planes. Simultaneously, approximately 5-nm-period Al0.70In0.30N/Al0.74In0.26N ordering is formed to mitigate the lattice mismatch along [0001], which grow into approximately 5-nm-period Al0.70In0.30N/Al0.74In0.26N {$$10\stackrel{-}{1}2$$ 10 1 - 2 } superlattices as step-flow growth progresses. Spatially resolved cathodoluminescence spectra identify the emissions from particular structures.

A process for activating Mg and its relationship with vacancy-type defects in Mg-implanted GaN were studied by positron annihilation spectroscopy. Mg+ ions were implanted with an energy of 10 keV, and the Mg concentration in the subsurface region (≤ 50 nm) was on the order of 1019 cm−3. After the Mg-implantation, N+ ions were implanted to provide a 300-nm-deep box profile with a N concentration of 6 × 1018 cm−3. From capacitance–voltage measurements, the sequential implantation of N was found to enhance the activation of Mg. For N-implanted GaN before annealing, the major defect species were determined to Ga-vacancy related defects such as divacancy. After annealing below 1000 °C, the clustering of vacancies was observed. Above 1200 °C annealing, however, the size of the vacancies started to decrease, which was due to recombinations of vacancy clusters and excess N atoms in the damaged region. The suppression of vacancy clustering by sequential N-implantation in Mg-implanted GaN was attributed to the origin of the enhancement of the Mg activation.

For solving the energy crisis problem, the exploitation of energy-saving high-power transistors 1) workable at high frequencies and solid-state lighting using highly efficient light-emitting diodes 2) is one of the significant ways. Gallium nitride (GaN) and (B,Al,Ga,In)N alloys are a suitable system for this purpose and several devices have already been put into market.

The performance of optical devices based on nitride semiconductors operating in the ultraviolet and visible wavelength regimes has been remarkably improved. For example, high-efficiency blue InGaN light-emitting diodes (LEDs) with an external quantum efficiency (EQE) of over 80% was achieved [Y. Narukawa, M. Ichikawa1, D. Sanga, M. Sano, and T. Mukai, J. Phys. D: Appl. Phys. 43, 354002 (2010)], and green laser diodes (LDs) was realized [Y. Enya, Y. Yoshizumi, T. Kyono, K. Akita, M. Ueno, M. Adachi, T. Sumitomo, S. Tokuyama, T. Ikegami, K. Katayama and T. Nakamura, Appl. Phys. Express 2, 082101 (2009)]. Recently, deep-ultraviolet AlGaN LEDs and LDs have also become hot topics. The nitride semiconductors have been actively applied not only to optical applications but also to electronic devices such as high-voltage transistors and high-electron mobility transistors based on their high breakdown voltage and saturation electron velocity. The performance of these optical and electronic devices relies on the crystal growth technology, where quality of GaN freestanding substrate is one of the most important issues. In recent years, it has become common for the "high-quality" GaN freestanding crystals to have a threading dislocation (TD) density of 106 cm-2 or less. TD density of 106 cm-2 corresponds to an average distance between TDs of 10 μm, which means that the TDs are sparser than 1 μm, the order of the carrier diffusion length. In other words, the TDs are no longer dominant in recombination processes of minority carriers (although this is a rather broad argument). In this case, the recombination processes, especially nonradiative recombination process, are likely to be dominated by point defects and impurities rather than by the sparse TDs. In fact, a negative correlation between room-temperature (RT) photoluminescence (PL) lifetime for the near-band-edge (NBE) emission and the concentration of VGaVN (a divacancy consisting of Ga and N vacancies) has been pointed out by a combined study of time-resolved PL measurements, positron annihilation spectroscopy, and theoretical calculations [S. F. Chichibu, A. Uedono, K. Kojima, H. Ikeda, K. Fujito, S. Takashima, M. Edo, K. Ueno, and S. Ishibashi, J. Appl. Phys. 123, 161413 (2018)]. Since PL lifetime measured at RT indicates nonradiative recombination lifetime when the internal quantum efficiency (IQE) of the crystal is low, this negative correlation means that the lower concentration of VGaVN induces larger intensity of the NBE emission. Actually, we have observed PL lifetime exceeding 2 ns at RT in a GaN crystal grown by halide vapor phase epitaxy (HVPE) on m-plane GaN grown by the acidic ammonothermal method as a seed crystal [K. Kojima, Y. Tsukada, E. Furukawa, M. Saito, Y. Mikawa, S. Kubo, H. Ikeda, K. Fujito, A. Uedono, and S. F. Chichibu, Appl. Phys. Express 8, 095501 (2015)]. This is an extremely long value for a GaN crystal, which means that the nonradiative recombination lifetime is long, i.e., the IQE of the NBE emission is large. In this case, the validity of the approximation where the PL lifetime at RT is directly read into the nonradiative recombination lifetime becomes an issue. Therefore, in order to compare the quality of GaN freestanding crystals, it is important to measure the IQE as well as the PL lifetime. The IQE value for the NBE emission in semiconductor crystals is a quantity that represents the balance between radiative and nonradiative recombination processes and is directly related to the concentration of point defects and impurities that form deep levels. Therefore, the IQE can be used as an indicator of not only the energy-saving performance of optical devices but also the performance of GaN crystals as a substrate material in electronic devices. In this presentation, the characterization of semiconductors with direct bandgap by using omnidirectional photoluminescence (ODPL) spectroscopy, a technique that can determine the IQE from the EQE without the need for model calculations, will be reviewed. Using GaN, ZnO, and metal halide perovskites as examples, specific procedures and applications of IQE determination will be discussed. In particular, the ability to measure large sample crystals or wafers is one of the major features of the ODPL spectroscopy, and it is expected to be combined with mapping measurements of the entire surface of semiconductor wafers and various types of nonlinear and microscopic spectroscopy.

Seeded growth of 2-inch-diameter GaN crystals via low-pressure (∼100 MPa) acidic ammonothermal method is demonstrated. Nearly bowing- and mosaic-free GaN crystals exhibiting full-width at half-maximum values for the 0002 X-ray rocking curves below 20 arcsec were achieved on high lattice coherency c-plane SCAATTM seeds with gross dislocation densities in the order of 104 cm−2. The photoluminescence spectra of the grown crystals exhibited a predominant near-band-edge emission at 295 K, of which intensity was one order of magnitude higher than the characteristic deep-state emission bands. A nearly bowing-free 60 mm × 60 mm c-plane GaN crystal was eventually obtained.

With respect to (Al,In,Ga)N epilayers and quantum wells, threading dislocations (TDs) have long been believed to as the principal limiting factor for the internal quantum efficiency of the near-band-edge emission. However, the realization of low TD density GaN and AlN substrates and (Al,In,Ga)N layers enabled investigating the roles of point defects and impurities without interferences by TDs, and vacancy-complexes have been revealed to act as origins of major Shockley- Read-Hall (SRH)-type nonradiative recombination centers (NRCs) in GaN. Accordingly, the concentration of NRCs (NNRC) must be decreased in both optical devices and power-switching electronic devices. Here we show the results of positron annihilation and time-resolved luminescence measurements on n- and p-type GaN, AlN, and Al0.6Ga0.4N alloys to reveal the origins of major intrinsic SRH-NRCs and to obtain their capture coefficients for minority carriers. For unintentionally doped and doped n-type GaN, divacancies comprising of a Ga-vacancy (VGa) and a N-vacancy (VN), namely VGaVN, are assigned as major SRH-NRCs with a hole capture-coefficient (Cp) of 6×10-7 cm3s-1. For Mg-doped ptype GaN epilayers grown by metalorganic vapor phase epitaxy (MOVPE), VGa(VN)2 are assigned as major NRCs with electron capture-coefficient (Cn) of 8×10-6 cm3s-1. For Mg-implanted GaN, VGaVN are the dominant NRCs right after implantation, and they agglomerate into (VGaVN)3 clusters with Cn of 5×10-6 cm3s-1 after high-temperature annealing. Since AlN films grown by MOVPE usually contain vacancy-clusters comprising of an Al-vacancy (VAl) such as VAl(VN)2-3, complexes of a cation-vacancy and a few VNs may be the major NRCs in AlN and Al0.6Ga0.4N alloys.

Epitaxial growths of NiO thin films were realized on (0001) sapphire and (100) MgO substrates by using a reactive RF magnetron sputtering method. The NiO epilayers grown on a (0001) sapphire exhibited the (111)-oriented double-domain structure, which comprised of a triangular and its inverted triangular grains. Meanwhile, the NiO epilayers on a (100) MgO exhibited the (100)-oriented single-domain structure, which comprised of quadrangular grains. The observed grain structures most likely reflect the growth planes of respective NiO epilayers, and, mixed crystals of NiO and MgO were present near the interface. Therefore, a (100) MgO substrate is suitable for obtaining a single-domain NiO epilayer, whereas a (0001) sapphire substrate is suitable for obtaining a NiO epilayer without interdiffusion between NiO and sapphire. These NiO epilayers will be expected for applying the physical properties evaluation using photoluminescence or Hall measurements, and the fabrication of electrical or optical devices.

To accelerate the development of GaN power devices, reproducible fabrication of p-type GaN (p-GaN) segments by ion-implantation (I/I) that enables selective-area doping is preferred. In this Letter, the results of time-resolved photoluminescence (PL) and spatially resolved cathodoluminescence (CL) measurements on p-GaN fabricated by I/I are presented. The p-GaN samples of a 220-nm-deep box-shaped Mg concentration profile ([Mg] = 3×1018 cm−3) were fabricated by sequential I/I of Mg and N followed by atmospheric-pressure post-implantation annealing (PIA). During PIA, Mg redistribution was guided by N-I/I-induced vacancies. The intensities of PL peaks originating from Mg acceptors were increased by increasing the PIA temperature (Ta), indicating progressive activation of Mg and simultaneous decrease in the concentration of nonradiative recombination centers (NRCs). As a result, a record-long room-temperature PL lifetime for the near-band edge emission of Ga-polar Mg-implanted p-GaN (1.4 ps) was obtained when Ta = 1300 °C. Because the species of major vacancy-type defects in the present sample was determined as vacancy clusters comprised of few Ga-vacancies (VGa) and N-vacancies (VN) such as (VGaVN)2–3, the NRC concentration is roughly estimated at around 1017 cm−3 provided that (VGaVN)2–3 are major NRCs with the electron-capture-coefficient of 7×10−6 cm3 s−1. Cross-sectional CL linescan images revealed certain activation of Mg from the surface to the leading edge of the box-shaped [Mg] profile, while NRCs remained at the depths right below the box-shaped profile. The results support advantages of using the vacancy-guided Mg redistribution technique for controlling the activated [Mg] profile and realizing p-type Ga-polar GaN:Mg segments by I/I.

GaN is one of the promising candidates for the use in high-power electronic devices 1) operating at high frequencies, and normally-off GaN-based transistors on freestanding (FS) GaN substrates with low specific on-state resistances (~1.0 mΩ•cm 2) and high off-state breakdown voltage (>1.7 kV) have been demonstrated. 2 – 4) One of the challenging issues for producing such devices at low cost is the control of conductivity type and conductivity at designated segments using an ion-implantation (I/I) technique. Especially, p-type doping by Mg-I/I has been difficult 5 – 8) because donor-type defects introduced by I/I and/or donor impurities such as O or Si diffused from the protective overlayer during post-implantation annealing (PIA) 7) likely compensate holes.

Angle-dependent energy shifts in the near-band edge emission peak were observed at room temperature in planar ZnO microcavities (MCs) fabricated by a top-down process that simultaneously maintains high radiative performance of a ZnO active layer and high reflectivity of wide-bandwidth distributed Bragg reflectors (DBRs). An approximately 2λ-thick ZnO active layer with a thickness gradient less than 10 nm/mm (i.e., an angle of gradient less than 2 arc sec) across the entire 10 × 5 mm2 device area was formed by thinning a bulk single crystal of ZnO grown by the hydrothermal method, whose typical threading dislocation densities were lower than 102 cm−2. The DBRs consisting of 10 and 12 pairs of SiO2/ZrO2 multilayers with a thickness gradient of 0.2 nm/mm/pair were deposited by using the nearly surface-damage-free reactive helicon-wave-excited-plasma sputtering method as the top and bottom mirrors, respectively. The quality factor of a passive cavity consisting of the same DBR stacks was in the range between 670 and 720 for the areal size of 1 mm in diameter. Angle-resolved photoluminescence spectra of the ZnO MCs measured at different positions with a macroscopic spot size of 80 μm in diameter exhibited distinct emission from the lower branch of cavity polaritons with apparent detunings ranging from −40 meV to 40 meV at room temperature.

Magnesium (Mg) ion implantation into gallium nitride (GaN) at 1000 °C is proposed. Since ion implantation and annealing occur simultaneously in high-temperature implantation, it is considered that Mg ions can be introduced at an appropriate position upon their implantation. GaN vertical diodes implanted with Mg ions were fabricated and current–voltage measurements were performed. As a result, clear rectifying properties were confirmed in a sample that was implanted with Mg ions at 1000 °C and annealed after implantation. However, the sample subjected to Mg ion implantation at RT and annealed after implantation showed no clear rectification. These results show that high-temperature implantation of Mg ions at 1000 °C reduces Mg ion implantation-induced damage and simultaneously activates Mg ions upon their implantation.

Seeded ammonothermal growths of a few-mm-thick GaN crystals on a 2 inch diameter c-plane and a 45 mm long m-plane GaN wafers were carried out by using an NH4F mineralizer in a 60 mm diameter Ag-lined autoclave. As a result of dynamic control of the temperature profile, low dislocation density and nearly bowing-free m-plane GaN was grown: i.e. the full-width at half-maximum values for the X-ray rocking curves of the 10 2 reflections were smaller than 28 arcsec and the radius of curvature was estimated to be 1460 m. In addition, its low temperature photoluminescence spectrum exhibited free and neutral donor-bound exciton emission peaks.

We demonstrate the growth of sub-10-nm-thick continuous Cu films using chemical vapor deposition (CVD) for next-generation Cu interconnects for ultra-large-scale integration (ULSI). The thickness of such films is equivalent to that of Cu during coalescence, and optimized operating conditions and substrate materials are required to form high-density nucleates. Ru was used as an underlayer, and the time evolution of nucleation and grain growth were studied with systematically varied conditions using two Cu precursors: conventional β-diketonato and newly developed amidinato precursor compounds. The revealed geometry of the initial nano-scale Cu grains prior to coalescence suggests the required nucleate density for 7-nm-thick continuous film growth, and which was 2.4 × 1011 /cm2. The maximum nucleate density was achieved with the lowest deposition temperature and highest precursor concentration for both precursors; i.e., 6.9 × 1011 /cm2 for β-diketonato at 100◦C, and 4.6 × 1011 /cm2 for amidinato at 150◦C. A 10-nm-thick continuous Cu film was formed using amidinato under the optimized conditions. Furthermore, the framework used in this study to enable a high nucleate density suggests that it is possible to form thinner (4 nm∼) Cu films using amidinato. Because of the inherent good step coverage of CVD, this process is a promising candidate for next-generation ULSI Cu interconnects. © The Author(s) 2015. Published by ECS. This is an open access article distributed under the terms of the Creative Commons Attribution 4.0 License (CC BY, http://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse of the work in any medium, provided the original work is properly cited. [DOI: 10.1149/2.0061508jss] All rights reserved.

The drive to continuously downscale Cu interconnects in ultra-large-scale integrated (ULSI) devices requires strategic improvements in materials and their design processes. For example, development of thinner single-layer barrier/liner materials is desired, because Cu line widths approaching the mean free path greatly increase RC signal delays. We designed Co films with the addition of W [Co(W)] for use as a single-layer barrier/liner material using principles from surface engineering and metallurgy. Dupre's equation was used to evaluate the adhesion of metals to Cu, which suggested Co as the main component of our proposed single-layer barrier/liner material. Metallurgical analysis suggested that the addition of W would lead to nanostructural improvements and increased barrier performance in Co films. The Co–W phase diagram suggested that W would segregate at grain boundaries, thereby improving barrier performance due to grain-boundary stuffing. Amidinato, metallocene, and carbonyl Co and W precursors were evaluated in chemical vapor deposition (CVD) and atomic layer deposition (ALD) processes for the formation of Co(W) films. Oxygen and halogen inclusions were undesirable, because they were predicted to increase the resistivity and cause deviation from the Co–W binary system. To reduce the prevalence of grain boundaries through formation of an amorphous structure, sequential feeding of Co and W precursors in an ALD process was developed for each set of precursors. Oxygen-free Co(W) CVD and ALD processes were achieved using amidinato and metallocene precursors, both of which led to stuffed grain-boundary structures. ALD-Co(W) films exhibited amorphous structures with sufficient barrier performance and low resistivity, which was consistent with our material and process design.

For accelerating the development of GaN power-switching devices, current knowledge on the origins and dynamic properties of the major intrinsic nonradiative recombination centers (NRCs) in Mg-doped GaN (GaN:Mg) are reviewed, as lightly to heavily doped p-type planar GaN segments are required but certain compensating defects including NRCs hinder their formation. The results of complementary time-resolved photoluminescence and positron annihilation spectroscopy measurements on the epitaxial and ion-implanted GaN:Mg formed on low dislocation density GaN substrates indicate the following: major intrinsic NRCs are the clusters of Ga vacancies (VGas) and N vacancies (VNs), namely VGa(VN)2 in the epitaxial GaN:Mg and (VGa)3(VN)3 in the ion-implanted GaN:Mg after appropriate thermal annealings. The minimum electron capture-cross-sections of VGa(VN)2 and (VGa)3(VN)3 are commonly the middle of 10−13 cm2 at 300 K, which is approximately four times larger than the hole capture-cross-section of the major intrinsic NRCs in n-type GaN, namely VGaVN divacancies, being 7 × 10−14 cm2.

Cobalt film with tungsten [Co(W)] has the potential to be an effective single-layered barrier/liner in interconnects, due to its good adhesion with Cu, low resistivity compared with TaN, and comparable barrier properties to TaN. To reduce the resistivity of Co(W), oxygen-free Co(W) was fabricated from oxygen-free precursors, bis(N-tert-butyl-N � -ethylpropionamidinato) cobalt and bis(tert-butylimino)bis(dimethylamino) tungsten, by chemical vapor deposition (CVD) and atomic layer deposition (ALD). Our results showed that W addition improved the barrier properties of both CVD-Co(W) and ALD-Co(W) against Cu diffusion. The diffusion coefficient of Cu in Co(W) was reduced by W addition. The activation energy of Cu in CVD-Co(W) and ALD-Co(W) was 1.5 eV and 1.7‐1.8 eV, respectively; whereas that in CVD-Co was 1.0 eV. Improvement in the barrier properties of Co(W) was attributed to the amorphous-like structure created by W. High-resolution transmission electron microscopy and multivariate spectral analysis of the nanoscale properties of Co(W) revealed that the improvement in the barrier properties was due to the self-assembled segregation of W atoms at the grain boundary. The segregated W atoms in both CVD-Co(W) and ALD-Co(W) films may act as a stuffing material. These results suggest Co(W), especially ALD-Co(W), as a promising material for Cu-interconnects in future ultra-large-scale integrated circuits.

Photoluminescence (PL) spectra of (000 1 ¯ ) N-polar p-type GaN fabricated by using the sequential ion-implantation of Mg and H with subsequent high temperature annealing exhibited the near-band-edge (NBE) emission at 300 K. The longest PL lifetime ( τ PL ) for the NBE emission of the sample with Mg and H concentrations of 1 × 1019 and 2 × 1020 cm−3, respectively, annealed at 1230 °C was 18 ps at 300 K. This value is almost comparable to that of the (0001) Ga-polar p-type Mg-doped GaN (p-GaN:Mg) homoepitaxial film of the same Mg concentration. By correlating τ PL and the concentration of major vacancy-type defects quantified using positron annihilation spectroscopy, the electron capture-cross-section (σn) of the major nonradiative recombination centers (NRCs), namely, clusters of Ga vacancies (VGas) and N vacancies (VNs) such as (VGa)3(VN)3, is estimated at a few times 10−13 cm2. This σn value is also comparable to that of the major NRCs in p-GaN:Mg epilayers, namely, VGa(VN)2 or VGa(VN)3, although the clustering sizes of the defects are different. These σn values are commonly larger than the hole capture-cross-section (σp = 7 × 10−14 cm2) of the major NRCs, VGaVN divacancies, in n-type GaN.

Complementary time-resolved photoluminescence and positron annihilation measurements were carried out at room temperature on Mg-doped p-type GaN homoepitaxial films for identifying the origin and estimating the electron capture-cross-section ( σ n ) of the major nonradiative recombination centers (NRCs). To eliminate any influence by threading dislocations, free-standing GaN substrates were used. In Mg-doped p-type GaN, defect complexes composed of a Ga-vacancy (VGa) and multiple N-vacancies (VNs), namely, VGa(VN)2 [or even VGa(VN)3], are identified as the major intrinsic NRCs. Different from the case of 4H-SiC, atomic structures of intrinsic NRCs in p-type and n-type GaN are different: VGaVN divacancies are the major NRCs in n-type GaN. The σ n value approximately the middle of 10−13 cm2 is obtained for VGa(VN)n, which is larger than the hole capture-cross-section (σp = 7 × 10−14 cm2) of VGaVN in n-type GaN. Combined with larger thermal velocity of an electron, minority carrier lifetime in Mg-doped GaN becomes much shorter than that of n-type GaN.

The effect of multiple film-forming species on the film quality during chemical vapor deposition (CVD) of SiC from CH3SiCl3/H2 was examined by separating each species using high aspect-ratio (AR) parallel-plate microchannels. Profiles of the chemical and physical properties of the grown SiC films, including the composition, crystallinity and surface roughness, were characterized along the microchannel depth. The exceptionally high AR, which was typically more than , allowed film growth with a variety of film-forming species that significantly changed with depth. We were thus able to examine the effect of each film-forming species on film growth. The exceptionally large depth, which was typically centimeter-scale, allowed depth profile analysis using various kinds of film characterization techniques having large detection areas. For CVD SiC from CH3SiCl3/H2, multiple film-forming species with different sticking probabilities of 8 × 10−7, 1 × 10−4, and 2 × 10−2 provided almost the same film quality.

The shrinkage of device dimension based on scaling rule is an effective method to realize high-performance and low-power consumption of ULSIs. However, it also increases complexity of ULSI device structure. For example, the half pitch of the Cu interconnects used in ULSIs are expected to be reduced to 15 nm or less with aspect ratio of 3 or higher. Three-dimensional FinFET are commonly selected for high-performance and low-power consumption devices. The memory capacitor in DRAM uses cylindrical shape with the diameter of 20 nm or less and the aspect ratio of 100. These structures require a novel deposition method to form metallic and dielectric thin films with excellent thickness uniformity in a high-aspect ratio features. Atomic layer deposition (ALD) is a suitable method for these applications. ALD has a good advantage in that the process is driven by the chemical saturation of surfaces with the precursor. ALD process allows for conformal coatings of surfaces of nanoscale structure even with a high aspect ratio. Useful precursors for metal-ALD are, however, limited. Precursors have to meet the following requirements: first, they should have sufficient vapor pressure at moderate temperatures. The precursors should be stable unless they are exposed to a reactant. The precursors should not, however, contain oxygen or a halogen, to avoid these elements being included in the deposited metallic film. Therefore, it is preferable to choose oxygen-free and/or halogen-free precursors. These aspects reduce the options of precursors available for use in ALD. Metallocenes have a long history in organometallic chemistry owing to their versatility. Some of the metallocenes have been used as catalysts or reducing agents in synthetic chemistry. Therefore, several metallocenes are reported and commercially available. ALD using metallocene is, however, generally quite difficult to achieve, despite metallocene having an appropriate vapor pressure. The carbon-metal bond in metallocenes is relatively more stable than that in other metalorganic compounds, making it difficult to form high quality metallic films. Hot-wire-assisted ALD (HW-ALD) has been identified as a successful method to form high-quality metallic films using metallocene and NH3. NH2 radical formed by cracking NH3 using hot wire at temperature over 1300°C promotes the dissociation of the carbon-metal bond in metallocene during the reducing period. This is examined by elucidation of the gas-phase kinetics and estimation of the surface reactions by quantum chemical calculations. Co, Co(W)-alloy, Ni, and Ru films were successfully formed by HW-ALD. Ni films formed from (C5H5)2Ni contain only several percent of carbon and Co films had carbon impurity under detection limit of carbon in XPS measurement. We have examined the work function of Ru films obtained by HW-ALD using CV measurement of MIS structure to confirm that carbon is not so much contained in the deposited films. Co(W)-alloy films formed by HW-ALD was examined as the Cu-diffusion barrier layer using atom probe tomography (APT). We could confirm that Co(W)-alloy film can surpress the Cu diffusion effectively, if W content is over 20 at%. In summary, high-quality metallic films with less impurities and uniform step coverage on high-aspect ratio feature were obtained by HW-ALD.

A method is proposed to fabricate a high-aspect-ratio (HAR) microchannel with a microscopic gap and AR of more than 1000:1 applicable to a test structure for kinetic analysis of chemical vapor deposition (CVD). It has a parallel-plate structure and is formed concisely by sticking a planar Si substrate and a patterned Si or silicon-on-insulator (SOI) substrate fabricated by single-step etching, by clamping them. The resulting feature exhibits a uniform gap and smooth surface morphology along its depth. When CVD is conducted into this HAR microchannel, the sticking probability (η) of film-forming species can be detected by analyzing the film thickness gradient. The use of a microchannel with an AR of 1000:1 enables the elucidation of η values from 1 down to the order of 10−7. A kinetic analysis of SiC-CVD from methyltrichlorosilane and H2 is thus performed. It is found that conformal SiC-film growth occurred inside the HAR microchannel, where one of the film-forming species is revealed to have extremely low η around 10−6. The present study demonstrates that the developed HAR microchannel is a solution to access the overall reaction kinetics of CVDs, including film-forming species with extremely low η.

ALD-Co(W) was found to have a potential to replace the conventional PVD-Ta/TaN bi-layer in further shrinking interconnects as a single-layered barrier/liner material. We could confirm good barrier property of CVD/ALD-Co(W) film by BTS-TVS method after 350 °C annealing. ALD-Co(W) showed lower resistivity of 60 µΩ-cm and good adhesion to Cu. Complete trench filling with Co(W) followed by Cu seed deposition was demonstrated. These properties were confirmed to be derived from W stuffing into grain boundaries of oxygen-free ALD-Co(W) films. We would like to suggest ALD-Co(W) as a next-generation barrier/liner layer for future development.

Erratum: Self-Assembled Nano-Stuffing Structure in CVD and ALD Co(W) Films as a Single-Layered Barrier/Liner for Future Cu-Interconnects [ECS J. Solid State Sci. Technol., 2, P471 (2013)] Hideharu Shimizu,a,b Akihito Kumamoto,b Kohei Shima,b Yoshihiko Kobayashi,a Takeshi Momose,b Takeshi Nogami,c and Yukihiro Shimogakib aTaiyo-Nippon Sanso Corporation, Tsukuba, Ibaraki 300-2611, Japan bSchool of Engineering, The University of Tokyo, Bunkyo, Tokyo 113-8656, Japan cIBM in Albany Nano Science & Technology Research Center, IBM Corporation, Albany, New York 12203, USA

We used atom probe tomography (APT) to study the use of a Cu(Mn) as a seed layer of Cu, and a Co(W) single-layer as reliable Cu diffusion barriers for future interconnects in ultra-large-scale integration. The use of Co(W) layer enhances adhesion of Cu to prevent electromigration and stress-induced voiding failures. The use of Cu(Mn) as seed layer may enhance the diffusion barrier performance of Co(W) by stuffing the Cu diffusion pass with Mn. APT was used to visualize the distribution of W and Mn in three dimensions with sub-nanometer resolution. W was found to segregate at the grain boundaries of Co, which prevents diffusion of Cu via the grain boundaries. Mn was found to diffuse from the Cu(Mn) layer to Co(W) layer and selectively segregate at the Co(W) grain boundaries with W, reinforcing the barrier properties of Co(W) layer. Hence, a Co(W) barrier coupled with a Cu(Mn) seed layer can form a sufficient diffusion barrier with film that is less than 2.0-nm-thick. The diffusion barrier behavior was preserved following a 1-h annealing at 400 °C. The underlayer of the Cu interconnects requires a large adhesion strength with the Cu, as well as low electrical resistivity. The use of Co(W) has previously been shown to satisfy these requirements, and addition of Mn is not expected to deteriorate these properties.