Semiconductor Device Fabrication Research Articles

Synthesis, characterization, and thermal properties of novel silicon 1,1,3,3‐tetramethylguanidinate derivatives and use as single‐source chemical vapor deposition precursors

AbstractA series of chemical vapor deposition (CVD) precursors have been synthesized by a single‐step reaction of 1,1,3,3‐tetramethylguanidine and a variety of silicon chlorides. The structures of the 1,1,3,3‐tetramethylguanidinate‐based compounds were verified by 1H NMR, 13C NMR, XPS, EI‐MS, and elemental analysis. The thermal stability, transport behavior, and vapor pressures of these compounds were evaluated by simultaneous thermal analyses (STA). These compounds are highly stable and those in liquid form are very volatile. Silicon carbonitride (SiCN) thin films were prepared by using bis (tetramethylguanidine)‐dimethyl‐silane as the precursor in helicon wave plasma chemical vapor deposition (HWP‐CVD). The properties of the films were investigated by SEM, AFM, and XPS. The results showed that the films have good uniformities, low friction coefficient, and high hardness, enabling the films for fabrication of semiconductor devices.

Nanocatalyst-induced hydroxyl radical (·OH) slurry for tungsten CMP for next-generation semiconductor processing

Chemical mechanical polishing (CMP) is one of the important steps that involves during fabrication of semiconductor devices. This research highlights the importance of tungsten (W) polishing slurries consisting of a novel nonionic, heat-activated FeSi nanocatalyst on the performance of W chemical mechanical polishing. The results obtained from the polishing data showed a higher W removal rate of 5910 A/min with a slurry consisting of FeSi nanocatalyst at a polishing temperature of 80 °C. The increase in W polishing rate using FeSi slurry was explained on the basis of formation of a thicker oxide layer (WO3) due to the interaction between the W surface and hydroxyl radicals (·OH) generated via the reaction between FeSi and hydrogen peroxide at 80 °C. Higher ·OH generation and increase in oxygen depth profile of W film were confirmed by UV–Vis spectrometer and AES analysis, respectively. Compared to Fe(NO3)3 catalyst, the slurry with FeSi showed a higher static etch rate at 80 °C. Potentiodynamic polarization results obtained using FeSi slurry showed thicker WO3 passivation layer as compared to the slurry with Fe(NO3)3. The increase in the polishing rate of W CMP using slurry with FeSi nanocatalyst can be essentially attributed to the generation of much stronger oxidant ·OH due to its increased catalytic effect at a high polishing temperature of 80 °C.

Guest Editorial Special Section on the 2018 SEMI Advanced Semiconductor Manufacturing Conference

The 2018 SEMI Advanced Semiconductor Manufacturing Conference, the 29th since its conception, was held once again in Saratoga Springs, NY, USA, May 1 through May 3, 2018. Saratoga Springs provided a beautiful backdrop for discussion of the many topics central to the continued success of the semiconductor manufacturing industry in areas including Advanced Metrology, Advanced Processes and Materials, Advanced Process Control, Contamination-Free Manufacturing, Cost Reduction, Equipment Reliability and Productivity, Defect Inspection, Environmental, Health and Safety, Factory Automation and Dynamics, Human Resources Management and Development, Lithography Advances and Challenges, Yield Enhancement, and Yield Modeling. Semiconductor device manufacturers, equipment and materials suppliers, academia, and others all came together these three days, as presenters and attendees, to exchange ideas and insights. Throughout the conference, two parallel session tracks provided options to match the interests of each attendee, without being overwhelming.

Room temperature oxidation of GaAs(110) using high translational kinetic energy molecular beams of O2 visualized by STM

This study examines the reactive surface dynamics of GaAs(110) oxidation with molecular oxygen at room temperature over a range of impinging kinetic energies. Visualization of the surface by scanning tunneling microscopy (STM) after exposures to O2 with kinetic energies of 0.4–1.2 eV provides morphological and kinetic data that were obtained utilizing a novel instrument that combines a supersonic molecular beam with an in-line, in-situ STM. Oxidation was found to proceed by two morphologically distinct, competing mechanisms: a spatially homogeneous process with randomly distributed chemisorbed oxygen atoms leading to layer-by-layer oxide growth, and a spatially heterogeneous process with oxides nucleating on structural surface defects and growing vertically and laterally with continued exposure. Both oxidation mechanisms exhibit enhanced reactivity with increasing kinetic energy. Only trace oxidation was observed with O2 kinetic energies below 0.7 eV; a rapid increase in the rate of oxidation from 1.0 to 1.2 eV was found with homogeneous and heterogeneous oxidation proceeding simultaneously until full surface coverage was reached. In addition, the relative rates of the two mechanisms appear to change with O2 kinetic energy: spatially homogeneous oxidation is expected to dominate at lower kinetic energies (<0.7 eV) while the heterogenous growth of oxide islands increasingly dominates with higher kinetic energies (≥1.0 eV). The results obtained in this study conclusively demonstrate that a heterogenous oxidation mechanism is activated on GaAs(110) at high O2 kinetic energies, and reveal that thin oxide layers can be achieved with higher efficiency at room temperature using molecular beams of oxygen. These results provide vital information about the morphological evolution of the surface in conjunction with the overall kinetics, and identify a controlled method of enhanced oxidation at moderate temperatures that could potentially improve abruptness at oxide interfaces and be used in the fabrication of GaAs semiconductor devices.

Group IV Heteroepitaxy for Advanced Electronic Devices Integrated in BiCMOS Technology

Epitaxy is a key process for group IV semiconductor device fabrication. For precisely-controlled fabrication of steep profiles for high performance devices, lowering the thermal budget of the epitaxy process is getting increasingly important. In this paper, we review group IV heteroepitaxy processes used at IHP and their application. In particular, we present SiGe HBT devices using low-temperature Si2H6 based epitaxy process and discuss potential advantages compared to standard SiH4 based process. Additionally, we show a growth technique for high-quality Ge photodiodes with several cycles of annealing by interrupting the Ge growth (cyclic annealing) followed by HCl etching. Furthermore, we discuss the device performance of SiGe/Si multi-quantum well (MQW) microbolometers and a technique to improve the quality of SiGe/Si MQW structures by introducing C delta layer at interface.

Transfer learning based 3D fuzzy multivariable control for an RTP system

Rapid thermal processing (RTP) is an important process in the fabrication of semiconductor devices. It is difficult to achieve temperature uniformity control of the wafer in RTP since the system is a highly nonlinear process with strong spatial distribution. In this study, a transfer learning-based three-dimensional (3D) fuzzy multivariable control scheme is proposed for the temperature uniformity control of an RTP system. In difference to the traditional expert-knowledge based design, a two-level framework of transfer learning methodology is constructed to design the 3D fuzzy multivariable controller (3D FMC) with the help of a multi-output support vector regression (M-SVR). The 3D FMC defines a qualitative spatial fuzzy structure that will be transferred to the M-SVR. On the other hand, the structure parameters of the M-SVR will be learned from data and transferred to design quantitative parameters of the 3D FMC. Under the framework of transfer learning, the control laws (e.g. human control experience) hidden in spatio-temporal data can be extracted and formulated back into multi-output 3D fuzzy rules. The proposed method provides an effective integration of the spatial fuzzy inference and the transfer learning for 3D FLC design. The newly developed method is applied to the temperature uniformity control of a rapid thermal chemical vapor deposition (RTCVD) system at the set temperature 1000K, and the maximum non-uniformity along the wafer radius is close to 1K.

Effect of partial coherence on dimensional measurement sensitivity for DUV scatterfield imaging microscopy.

Optical scatterfield imaging microscopy technique which has the capability of controlling scattered fields in the imaging mode is useful for quantitative nanoscale dimensional metrology that yields precise characterization of nanoscale features for semiconductor device manufacturing process control. To increase the sensitivity in the metrology using this method, it is required to optimize illumination and collection optics that enhance scatterfield signals from the nanoscale targets. Partial coherence of the optical imaging system is used not only for enhancing image quality in the traditional microscopy or lithography but also for increasing the sensitivity of the scatterfield imaging microscopy. This paper presents an empirical investigation of the effect of partial coherence on measurement sensitivity using a deep ultraviolet scatterfield imaging microscope platform that uses a 193 nm excimer laser as a source and a conjugate back focal plane as a unit for controlling partial coherence. Dimensional measurement sensitivity is assessed through analyzing scatterfield images measured at the edge area of periodic multiline structures with nominal linewidths ranging 44-80 nm on a Molybdenum Silicide (MoSi) photomask. Intensities scattered from the targets under the illuminations with various partial coherence factors and two orthogonal polarizations are assessed with respect to sensitivity coefficient. The optimization of partial coherence factor for the target dimension is discussed through the sensitivity coefficient maps.

Atomic layer deposition of silicon-based dielectrics for semiconductor manufacturing: Current status and future outlook

The fabrication of next-generation semiconductor devices has created a need for low-temperature (≤400 °C) deposition of highly-conformal (&amp;gt;95%) SiO2, SiNx, and SiC films on high-aspect-ratio nanostructures. To enable the growth of these Si-based dielectric films, semiconductor manufacturers are transitioning from chemical vapor deposition to atomic layer deposition (ALD). Currently, SiO2 films deposited using ALD are already being integrated into semiconductor device manufacturing. However, substantial processing challenges remain for the complete integration of SiNx films deposited by ALD, and there are no known processes for ALD of SiC at temperatures that are compatible with semiconductor device manufacturing. In this focused review, the authors look at the status of thermal and plasma-assisted ALD of these three Si-based dielectric films. For SiO2 ALD, since low-temperature processes that deposit high-quality films are known, the authors focus primarily on the identification of surface reaction mechanisms using chlorosilane and aminosilane precursors, as this provides a foundation for the ALD of SiNx and SiC, two material systems where substantial processing challenges still exist. Using an understanding of the surface reaction mechanisms, the authors describe the underlying reasons for the processing challenges during ALD of SiNx and SiC and suggest methodologies for process improvement. While both thermal and plasma-assisted SiNx ALD processes have been reported in the literature, the thermal NH3-based ALD processes require processing temperatures &amp;gt;500 °C and large NH3 doses. On the other hand, plasma-assisted SiNx ALD processes suffer from nonuniform film properties or low conformality when deposited on high-aspect-ratio nanostructures. In the SiNx section, the authors provide a broad overview of the currently known thermal and plasma-assisted SiNx ALD processes using chlorosilane, trisilylamine, and aminosilane precursors, describe the process shortcomings, and review the literature on precursor reaction pathways. The authors close this section with suggestions for improving the film properties and conformality. In the case of SiC, the authors first outline the limitations of previously reported SiC ALD processes and highlight that unlike SiO2 and SiNx plasma-assisted ALD, no straightforward pathway for low-temperature plasma-assisted growth is currently apparent. The authors speculate that low-temperature ALD of SiC may require the design of completely new precursors. Finally, they summarize the progress made in the ALD of C-containing SiNx and SiO2 films, which may provide many of the benefits of SiC ALD in semiconductor manufacturing. In closing, through this review, the authors hope to provide the readers with a comprehensive knowledge of the surface reactions mechanisms during ALD of Si-based dielectrics, which would provide a foundation for future precursor and process development.

Key role of antibonding electron transfer in bonding on solid surfaces

The description of the chemical bond between a solid surface and an atom or a molecule is the fundamental basis for understanding a broad range of scientific problems in heterogeneous catalysis, semiconductor device fabrication, and fuel cells. Widespread understandings are based on the molecular orbital theory and focused on the degree of filling of antibonding surface-adsorbate states that weaken bonding on surfaces. The unoccupied antibonding surface-adsorbate states are often tacitly assumed to be irrelevant. Here, we show that most antibonding states become unoccupied because the electrons that would occupy these antibonding states are transferred to the lower-energy Fermi level. Such antibonding electron transfer goes beyond molecular orbital theory. It leads to an energy gain that largely controls the trends of surface adsorption strength and can serve as a primary descriptor for bonding on surfaces. This finding is illustrated from the first-principles study of hydrogen adsorption on MoS$_2$ surfaces. A clear linear relationship between the energies of antibonding electron transfer and hydrogen adsorption is identified. The hydrogen evolution reaction on MoS$_2$ is found to originate from the in-gap states induced by sulfur vacancies or edges. The effects of surface inhomogeneity (e.g., defects, step edges) on surface catalysis can be understood from the corresponding different energies gained from antibonding electron transfer. The emerging picture also offers a physically different explanation for the well-known $d$-band theory for hydrogen adsorption on transition metal surfaces.

High Stability of Epitaxial Graphene on a SiC Substrate

The effects of strong‐acid treatment on an epitaxial graphene film on a SiC substrate are investigated to confirm its stability and compatibility with conventional semiconductor device fabrication processes. An epitaxial graphene film is treated with a strong acid in the form of piranha solution (H2O2 + H2SO4), which is conventionally used in washing processes for the silicon‐based technology. Raman spectroscopy, Hall measurements, and contact angle measurements are carried out before and after piranha treatment. Raman mapping results show no drastic changes before and after piranha treatment. In particular, the D band is not observed after the piranha treatment. From Hall measurements, the electron mobility slightly increases from 920 to 1420 cm2 V−1 s−1 after five piranha treatments. The contact angle is almost constant before (72.8°) and after five piranha treatments (75.2°). These results indicate that the epitaxial graphene film is quite stable under piranha treatment.

(Invited) Atomic Layer Processing for Advanced Electronics

With devices and interconnects now being scaled to below the limits of even the most advanced lithographic patterning, dimensional control requires new thin-film processes with atomic-scale control of thickness, uniformity, morphology, and material and electrical properties at all levels, from the atomic scale to the equipment scale. With ever increasing complexity in process technologies, chemically selective and/or area-selective processes can be an efficient way to meet future manufacturing requirements. Novel process technologies are needed for deposition, etch and cleans with atomic level precision to enable area-selective-processing at the structure and material level [1]. In this tutorial we will review and discuss trends in Atomic Layer Processing technologies for advanced semiconductor device manufacturing. The topics and technologies explored in this presentation include atomic layer deposition (ALD), atomic layer etching (ALE), selective deposition and etching, advanced surface preparation, EUV lithography, and self-aligned and multiple patterning schemes among others. Applications include critical material and manufacturing challenges for advanced Logic and Memory devices, including high-k oxides and metals, interconnects, capacitors, as well as Patterning, including Multi Patterning, Fully-Self-Aligned patterning and EUV [2]. A key point of this presentation will be to educate attendees how these advanced process technologies can be leveraged synergistically to deliver power, performance, area and cost scaling for future devices. Examples will be given so that attendees can understand how advanced process technologies will be used in future device manufacturing as well as the benefits and tradeoffs in their use.

Free-standing and ultrathin inorganic light-emitting diode array

We report on the fabrication and characteristics of an individually addressable gallium nitride (GaN) microdisk light-emitting diode (LED) array in free-standing and ultrathin form. A high-quality GaN microdisk array with n-GaN, InGaN/GaN quantum wells and p-GaN layers was epitaxially grown on graphene microdots patterned on SiO2/Si substrates. Due to the weak attachment of the graphene microdots to the growth substrate, a microdisk array coated with a polyimide layer was easily separated from the substrate using mechanical or chemical methods to form an ultrathin free-standing film. Individually addressable microdisk LEDs were created by forming thin metal contacts on the p-GaN and n-GaN surfaces in a crossbar configuration. Each microdisk LED that comprised an ultrahigh resolution array of 2500 pixels per inch was found to be uniquely addressable. The devices in free-standing form exhibited stable electrical and optoelectronic characteristics under extreme bending conditions and continuous operation mode despite the absence of a heat dissipating substrate. These results present promising approaches for the fabrication of high-quality inorganic semiconductor devices for ultrahigh resolution and high-performance flexible applications.

Design Considerations for Robust Manufacturing and High Yield of 1.2 kV 4H-SiC VDMOS Transistors

Production yield is a major factor for semiconductor device manufacturing. To produce high performance devices cost efficiently, it is important to know the process windows of the implemented production technology. This can influence the yield in different ways. One of the critical steps is the photolithography. In this work the impact of misalignment within the technological limits is analyzed and discussed. 4H-SiC VDMOS Transistors were produced and the electrical characteristics were compared with the overlay accuracy of the devices. Small change in channel length can lead to large impact on the electrical characteristic. Especially when the channel length reaches values near to the critical length for short channel effects (SCEs), small overlay inaccuracies influence the electrical characteristic of the devices in an increasing manner. Different cell designs were analyzed regarding their robustness to misalignment.

Epitaxial growth and characterization of high quality In2O3 films on a-plane sapphire substrates by MOCVD

High quality epitaxial indium oxide (In2O3) films were prepared on the a-plane sapphire [Al2O3 (110)] substrates at 570–690 °C by the metal-organic chemical vapor deposition (MOCVD) technique. The properties of structure, morphology and photoelectricity for the obtained films were studied at length. The film deposited at 650 °C showed the best crystalline quality with an out-of-plane epitaxial relationship of In2O3 (111)∥Al2O3 (110). This film also had the highest Hall mobility of 42 cm2V−1s−1 with a lowest resistivity of 5 × 10−3 Ω cm and a lowest carrier concentration of 3 × 1019 cm−3. All the samples possessed high average transparencies (over 83%) in the visible region and wide optical band gaps (∼3.70 eV). The In2O3 films with high crystalline quality as well as satisfactory photoelectrical characteristics can be extensively used in the manufacture of various semiconductor devices.

Effect of bistrifluoromethane sulfonimide treatment on nickel/InAs contacts

In this work, the effects of bis(trifluoromethane) sulfonimide (TFSI) (a superacid) surface treatment on electrical properties of nickel/InAs contacts and InAs material are investigated with transmission line measurements (TLMs). After TFSI treatments on bare InAs TLM patterns, the contact resistance between InAs and Ni contact increased. While for InAs TLM structure passivated with a ZrO2 thin film, the total resistance of InAs decreased drastically after TFSI treatments. X-ray photoelectron spectroscopy measurements revealed that the TFSI reduced the indium oxide at the InAs/ZrO2 interface. This novel passivation method can be adopted in III–V semiconductor device fabrication process to reduce contact resistance and, therefore, enhance device performances, providing a new guidance for III–V fabrication process technology.

Recognition and Classification of Wire Bonding Joint via Image Feature and SVM Model

Recognition and classification for wire bonding joint are important to quality assurance in semiconductor device manufacturing. In this paper, a precision recognition and classification system for bonding joint of ultrasonic heavy aluminum wire based on image feature and support vector machine (SVM) is presented. This system consists of feature extraction from images and classification model. In feature extraction, image processing algorithms including Canny edge extraction, histogram equalization, and image morphology closed operation are utilized to extract and locate a joint contour in a complicated background image. In the classification model, the principal component analysis (PCA) is employed to visualize, reconstruct, and reduce the images data dimension for less computation time. The SVM-based model is chosen as the classifier to identify and recognize joint types. The Gauss-radial basis function (RBF) kernel function is adopted in SVM, and its optimal parameters are determined by cross-validation. In the experiment, 588 bonding images are used to implement in this recognition and classification system. The results prove that the classification accuracy for wire bonding joint based on image feature, PCA, and SVM can achieve to 97.3%, and the computation time can be reduced significantly.

(Invited) Laser Thermal Annealing for Low Thermal Budget Applications: From Contact Formation to Material Modification

Pulsed Laser thermal annealing at the nanosecond timescale was successfully introduced into semiconductor device manufacturing in the late 2000s for high volume manufacturing of sensitive 3D architectures such as vertical Si-based Insulated Gate Bipolar Transistors, SiC-based vertical power diodes and backside illuminated CMOS imaging sensors. It is now on the verge of being successfully integrated as process of record for key annealing process steps in next generation CMOS and memory devices manufacturing. This invited paper presents recent experimental and simulation work involving sub-µs Laser annealing for a wide range of applications, from contact formation to material modification and 3D sequential integration.

Time-Resolved Electron Density Measurement Characterization of E–H-Modes for Inductively Coupled Plasma Instabilities

Inductively coupled plasma sources driven by RF power at low-pressure regimes are well adopted for high-volume manufacturing of semiconductor devices. One vexing challenge to the utility of these plasma processing reactors is the existence of the E–H-mode transition. Industry notably avoids the process region associated with this transition, where plasma instabilities and bimodal power coupling prohibit reliable RF power delivery. One plasma instability detailed in this paper is associated with a hysteresis in coupled RF power (current) varying for the E-mode, or weakly capacitive coupling to the plasma, in comparison to the stronger current coupling in the H-mode, where inductive coupling is preferentially dominant. As a result, approximately two orders of magnitude of electron density is relinquished in this transition region from serving industrial manufacturing processes. We characterize the plasma parameter variation through the E-mode to H-mode with a time-resolved measurement of the electron density. Electronegative chemistries are incorporated into our experimental setup. The experimental scheme serves to evaluate RF power delivery and ameliorate its coupling through the transition region. We seek to extend this paper to adopt more efficient power coupling for toroidal plasma sources.

(Invited) Superconformal Electrodeposition

State of the art manufacturing of semiconductor devices involves electrodeposition of copper for device wiring and chip stacking. The success of the electroplating process depends on electrolyte additives that affect the local deposition rate to yield void-free superconformal, or bottom–up filling of trenches and vias that comprise 3D interconnects. Two distinct processes associated with the prerefential filling of recessed surface features, namely, curvature enhanced accelerator coverage and critical phenomena related to supressor induced negative differential resistance. Quantitative chemical process models based on the above processes have been shown to be effective in describing the filling of high aspect ratio features that range from nanometers (on-chip wiring1) to micrometers (through-silicon-vias2) to millimeters (printed circuit boards) in scale. Nevertheless, much remains to be known about the molecular nature of the additives and the competitive, co-adsorption dynamics involved in both suppression and acceleration of metal deposition. Accordingly, a variety of surface science tools, such as scanning tunneling microscopy (STM) and surface-enhanced infrared absorption spectroscopy (SEIRAS) are being used to investigate additive function. The lecture will detail experiments that reveal the important role of adsorbate water interactions, i.e. hydrophilicity, in the design and operation of additives used for superconformal film growth. The talk will conclude with a discussion of the lesson learned and their application to the deposition of other metals.

Use of TRIDYN and medium energy ion scattering to calibrate an industrial arsenic plasma doping process

Plasma doping ion implantation (PLAD) is becoming increasingly important for enabling the manufacture of advanced semiconductor devices. In this study, a VIISTA PLAD implanter was used to implant planar 300 mm Si wafers with As/7 keV from an arsine containing plasma with a total ion fluence of 1 × 1016 ions/cm2. The wafers then underwent a wet chemical clean and anneal to mimic a full industrial process flow. The effects of each process step were measured using cross-sectional TEM images, TEM/energy dispersive spectroscopy measurements, and medium energy ion scattering (MEIS). The PLAD implantation process was modeled using dynamic trim (TRIDYN), a dynamic, binary collision approximation model that accounted for the interactions between wafers and the ions and neutrals produced by the PLAD implanter. MEIS spectra were analyzed to extract elemental concentration depth profiles using POWERMEIS guided by the outputs of the TRIDYN model. The input fluxes of the TRIDYN model were calibrated such that the predicted TRIDYN and MEIS profiles were self-consistent. Combining the different analysis techniques and considering elemental concentrations alongside a TRIDYN model enabled magnitudes of ion and neutral fluxes of Si, As, and H to be proposed, and the relative importance of direct implantation and ion beam mixing during the PLAD implant to be revealed. This, in turn, led to proposals for the sources of the ion and neutral species, the importance of Si neutrals originating from the plasma chamber over those originating from the Si bulk in the “deposited” layer being of particular interest. Following the evolution of the as-implanted profiles through the wet clean and anneal steps gave insights into how the PLAD implant affected the results of the full process flow.