High power pulsed magnetron sputtering (HPPMS/HIPIMS) has attracted considerable attention from industry due its ability to produce thin films and features of excellent adhesion, superior density, decreased roughness, and extreme conformity. The intense pulsed plasma density provides a large concentration of metal ions that produce high-quality, homogeneous coatings. The high ionization fraction allows for fine control of the sputtered species during deposition, a feature well suited to the needs of future semiconductor fabrication techniques.
One of the most constant and consuming challenges of chip fabrication has been the deposition of material into high aspect ratio trenches. As characteristic dimensions have shrunk, interconnect trenches have become narrower, but, because they must often connect features in different layers of the chip, they are often relatively deep. These interconnect trenches must be coated with a diffusion barrier such as titanium nitride and then metalized. Depositing these materials uniformly into the trench is exceedingly difficult in modern chip designs where aspect ratios can be as high as 7:1. In the past thermal evaporation was sufficient. Physical collimators were used when that failed but these too were soon obsolete. A technique known as Ionized Physical Vapor Deposition (IPVD) is the current standard. In IPVD the sputtered material is ionized and directed into the trench by a voltage bias. In this way the trench can be filled quite uniformly.
There are improvements to be made, however. Generally IPVD requires the use of a secondary plasma source to attain a high ionization fraction of the sputtered material. This introduces added complexity that may negatively affect process scalability. HPPMS may provide a highly scalable means of depositing diffusion barrier coatings and metallic features in the high aspect ratio interconnect trenches because a HPPMS power supply may be connected to any existing deposition system to increase performance.
The modulated pulse power (MPP) technique is a new development in the pulsed plasma field pioneered by CPMI partner Zpulser, LLC. MPP can be used to shape an arbitrary voltage waveform that is applied to the cathode. This unprecedented freedom allows control over pulse duration, intensity, duty cycle, and average power. Voltage oscillations during the 1.0 – 3.0 ms pulse on the order of 30-65 kHz induce instabilities in the plasma discharge that may have a marked effect on the level of ionization within the discharge and distribution of these metal ions. Past investigations of MPP have only revealed time-averaged plasma parameters, but knowledge of what happens during the pulse is required to further understanding of the basic mechanisms involved. Because of our extensive experience with intense, short-lived discharges, CPMI is in a unique position to uncover the underlying physics of this technology.
Researchers at CPMI are currently characterizing MPP discharges from a 1000 square centimeter circular planar rotating magnetron (see video) and evaluating the technique’s potential as a chip processing tool. A gridded energy analyzer and quartz crystal microbalance are used to measure the ionization fraction at the substrate under a variety of deposition conditions. The energy spectrum and flux of these ions can also be monitored using this equipment. Time-resolved plasma properties including saturation current, electron temperature, and density will be measured and mapped over a 3D region between the sputter target and substrate level using a triple Langmuir probe built in house. The effects of pulse duration, current density, pulse shape, switching frequency, and target material on the discharge will be explored. Eventually, characterization of film structure, quality, and uniformity over the width of a 200 mm wafer and across surface features will be performed.
A comparative study will be conducted in which MPP discharges are examined in a more current Novellus INOVA hollow cathode magnetron (HCM) system. Vacuum systems of this type are in common use by Intel and other chipmakers. The HCM is intended to develop a greater ionization fraction of the sputtered species; it is thought that the addition of a HPPMS power supply will greatly increase the effectiveness of this design. Once again, films will be deposited and analyzed with a focus of parameter optimization.
Future industry demands will require a trench metallization technique that is effective, efficient, scalable, and reproducible. It is expected that this work will produce a step of the chip manufacturing process that will be suited to the designs of ~2013.
There is a desire to create highly ionized metal fluxes by utilizing a extremely high-density plasma, facilitating near ideal IPVD required in filling narrow trenches of 32nm or less – realizing performance required for next generation of chips, produced using EUV(13.5nm) light.
A variety of plasma diagnostics can be used to study the detailed influence of parameter variation on the plasmas used for PVD and PECVD on a commercial 200mm INOVA high power (32 kW) hollow cathode magnetron deposition tool. Because of the intense deposition plasma conditions, non-standard geometry, and some non-standard frequencies used, specifically designed diagnostics are preferable to commercial solutions. These diagnostics include a 3-D scanning Langmuir probe, with analysis for magnetized plasmas to find the electron temperature and density. A self-cleaning in-situ plasma cup is being designed. To find the deposition rates and ionization fraction of the incident metal atom species a quartz crystal microbalance combined with electrostatic and magnetostatic filters will be implemented which will allow calculations of ionization fraction and system efficiency.
Based on the plasma diagnostics results, different processes were selected to study the sputtering deposition of thin films. Ta or TaN were deposited in an Ar or Ar/N2 sputtering plasma respectively, on the planar Si wafers and in Si trenches of different aspect ratios. The effects of ionization fraction, pressure among other parameters on the films quality and trench filling conformity are being investigated.
Cleaning Sn off from EUV optics by plasma etching method
Sn is more preferable among fuel materials for EUV light source since Sn has better conversion efficiency than other materials (Xe and Li). However, Sn is a condensable material so that it builds up on the EUV optics (collectors or mirrors). Debris build-up will roughen the surface and accordingly result in reflectivity loss. Therefore, cleaning is very essential step for HVM (high volume manufacture) level Sn EUV source.
The house modified chamber, GALAXY was first used for this study. With chlorine and argon plasma, etch rates of different material were measured and the optimal condition has been found so that Sn can be selectively removed from Ru mirror surface. Using a small mock-up collector in the GALAXY system, cleaning rate dependence on the distance from a plasma source was investigated as well.
Based on the preliminary studies, more study about cleaning and plasma transport are in progress. For the more realistic experiment, a real commercial DPP EUV system (XTS 13-35) is used. By measuring the cleaning rates along the distance from the (cleaning) plasma source using a comparable size of mock-up collector to real ones, we investigate the means of cleaning Sn debris with plasma based method. Understanding plasma transportation through the mirror shells which have different gap width and modeling surface physics between Sn and chlorine plasma is a key contribution to science during this project.
Sputtering High-purity Atomic Deposition Experiment (SHADE)
The SHADE chamber utilizes an ultra-high vacuum environment with a dual magnetron setup to deposit thin films and a secondary RF antenna in order to tailor the deposited film properties. SHADE is a valuable research tool allowing for a variety of sputtering targets to easily be exchanged and a loadlock system to transfer samples for deposition in and out without breaking vacuum.
SHADE is able to reach ultimate pressures in the 10-8 Torr range by using a turbomolecular pump as well as a cryo pump. However, operating pressures are typically between 1 and 20 mTorr with high stability through the use of two mass flow controllers (MFCs). Having two MFCs allow for multiple gases to be run simultaneously, a necessary ability for reactive sputtering.
Samples are transferred in and out of the chamber via a loadlock and gate valve system in order to maintain the integrity of the vacuum and to allow quick turnaround between experimental runs. Samples can be as large as two inches wide and still easily fit in the transfer mechanism. Once loaded into the chamber, the sample holder can be steadily rotated by a stepper motor in order to enhance uniformity during deposition.
Film thickness is monitored during deposition through the use of a dual quartz crystal microbalance (QCM). The QCM consists of two quartz crystals (one shielded and one exposed) which are oscillating at a very high frequency. As the exposed crystal receives deposition, the frequency slows down and the QCM interprets that as a film thickness. The shielded crystal is used as a check that any detected frequency change is due to actual deposition and not temperature change.
The plasma is monitored by an optical emission spectrometer which can detect wavelengths in the range of 300nm to 2000nm. The intensity of specific emission lines are monitored to ensure that the plasma remains stable and can be used to adjust processing parameters to achieve optimal results.
SHADE can be equipped with either two two-inch or three-inch magnetrons, depending on the size of the s
puttering target being used. Whereas the two-inch magnetrons are a balanced closed-field configuration,the three-inch magnetrons can be either balanced or unbalanced and open- or closed-field through the use of interchangeable magnet packs. This allows for flexibility in plasma density and confinement.
When using the three-inch magnetrons, an RF antenna can be installed between them in order to create a secondary plasma besides the sputtering plasma. This secondary plasma increases the ionization of otherwise neutral gas and sputtered species which increases the sputtering rate and adds additional energy to the deposited film without significantly increasing the temperature of the sample. This additional energy allows for better film quality while maintaining a near room-temperature substrate. This high-quality, low-temperature deposition can therefore be used on plastic substrates which would otherwise melt at the higher temperatures required to achieve some high-quality films without a secondary plasma. Adjusting the RF power varies the power being delivered to the thin film and therefore the crystal structure of the film. Typical RF plasma power is 100-300W.
With such a flexible system there are a several areas of research which have been explored. Previous research has developed Gibbsian-segregated alloys for use in semiconductor manufacturing optics. High reflectivity mirrors are needed in order to bounce the extreme ultraviolet light without intensity loss. These mirrors are prone to damage but by incorporating these specialized coatings they can become “self-healing” so that when they are contaminated, the contamination segregates into the bulk of the film while leaving the surface structure intact, and therefore reflective.
Current research using SHADE involves deposition of transparent conducting oxides (TCOs) such as Indium Tin Oxide (ITO) and Aluminum-doped Zinc Oxide (AZO). TCOs are a class of materials with an increasing presence in digital displays and photovoltaic applications. They allow light to be transmitted through them while still being able to conduct electricity. The most commonly used TCO is ITO but it has the drawback of using Indium, an expensive material with uncertain world reserves. Therefore alternatives are being sought, such as AZO, which can maintain the desired material properties but at a lower price. SHADE is being used to deposit AZO with the secondary RF plasma to achieve low-temperature deposition of high-quality films on flexible plastic substrates. The goal is to develop flexible photovoltaic sheets for low-cost power production.
The goal of PACE is to show how a helium plasma can reduce or eliminate organic contamination from materials used in extreme ultraviolet lithography.
Why PACE Has Taken on This Task
Reliability and speed in computing are two key aspects that continue to drive the economic and personal lives of individuals in today’s society. These two key aspects of computing are achieved through the continued development and refinement of the way in which integrated circuits (IC) are manufactured. In order to obtain more functionality out of a computer chip, more transistors are needed on that chip compared to previous generations. This proves to be a difficult hurdle to overcome, either the features (transistors) need to become smaller to fit more of them into the same amount of real-estate, or the chip needs to become physically larger and thus leads to larger devices. The trend has been to decrease the size of the transistors to increase the number of them per chip, however, this presents many complications for the design and fabrication engineers. In 1965, Gordon Moore made the observation that the number of transistors on a chip would double every two years . To keep the trend now referred to as Moore’s law applicable, the shift to extreme ultraviolet lithography (EUVL) at 13.5 nm is seen as one solution in order to achieve this . The idea of EUVL has led to a huge scientific undertaking by both private industry and Universities in order to produce the concepts and machines capable of making EUVL possible. There have been numerous challenges in both optical design, source design, photoresist sensitivity, and lithographic mask that have stood in the way as hurdles to EUVL’s implementation . The subject of a lot of these hurdles has led to great advances in science and engineering, as well as aided numerous students to partake in research and development for both academic and industrial purposes. EUVL has the potential to be the next large step and revolutionize the way IC’s are manufactured. The ability to manufacture IC’s with a direct pattern technique instead of phase-shift masks or double patterning, as is the current path of IC manufacturing with current 193 immersion technology, will provide IC designers with more tools and abilities in the design of IC’s to further their applicability to the desire for increased computing power. One of the largest hurdles facing EUVL implementation is contamination in the EUVL system, particularly contamination of the photomask in the form of particulate defects and thin film contamination and carbon buildup and contamination of the collector optics. These key aspects need to be addressed and a high volume manufacturing solution determined before EUVL can be successfully implemented.
Contamination and Its Effect on EUVL
Integrated circuits are the combination of miniaturized electronic circuits and passive elements built into the surface of a semiconductor material such as silicon. There are several steps that are required to form an integrated circuit, these steps are: imaging, deposition, and etching . To make a single device, several processing steps focused around the previously mentioned three are conducted. Supplementary steps such as cleaning and planarization are also required to ensure the functionality of devices and to make the processing steps more precise.
As shown in Figure 1.1, there is no pellicle between the mask and the wafer due to the non-transparency of materials to 13.5 nm wavelength light. Also, the mask requires illumination from the side and at an angle due to the inability of a passthrough type design which is also due to the non-transparency of material to the 13.5 nm wavelength. Thus, whereas in an optical lithography system utilizing a transmissive mask, a pellicle can be used to reduce the number of defects that reach the mask, in an EUV system, this is less likely to be the case.
Figure 1.1: A diagram of a particle falling onto an EUV lithography mask. If the particle remains on the mask, then it will block light reflecting off of the mask and appear to be another feature not intended by the mask design which will cause errors and device failures.
The alternating light/dark layers beneath the dense features are the multilayers. The multi-layers are specifically designed to reflect EUV light. NOTE: the size of the wafer/resist are not to size but have been enlarged for clarity and the incoming light is not at as extreme an angle.
EUV Mask Fabrication
There is an additional issue for particle contamination when considering EUV lithography. In forming the alternating multilayers for the reflective properties of the EUV mask, particle control is important. As seen in Figure 1.2, if a particle falls onto the surface during multilayer deposition, the overall smoothness of the mask is affected. Thus, by effective particle and contamination control or an effective cleaning process, the mask smoothness before patterning can be maintained to provide a defect free mask during fabrication.
Figure 1.2: A diagram of a non-patterned EUV mask. To fabricate the mask, a blank 6.35 mm thick quartz blank is used to alternatively deposit the reflective layers. If a particle is present, it will perturb the smoothness and effect the mask properties.
EUV Optics Degradation Due to Carbon Contamination
In addition to the mask degradation due to particulate contamination outlined in section 1.3, the high intensity photon irradiation of other components of the system, namely the EUV mirrors which compose the collector array, can become contaminated. It has been shown that the high energy photon interaction with collector mirrors in the presence of background hydrocarbons in the vacuum of the proposed EUV lithography tool will result in a buildup of carbon layers on the mirror . This resulting carbon layer can be cleaned by atomic hydrogen, plasma etching, or through use of molecular oxygen . However, these cleaning techniques can lead to damage of the surface through either physical damage, chemical attack, or oxidation .
How PACE Works
When a plasma is formed, energetic ions and electrons are created in the plasma and are typically what are important to the engineer creating the plasma. However, there is a strong energetic neutral component in plasmas too in the form of metastable atoms. Metastable atoms are those that are stuck in a quantum state and are forbidden through conservation of momentum from decaying to the ground state (such as 2s to 1s transitions being forbidden). Thus, any electron entering the 2s state will be hypothetically `stuck.’ In general, the lifetime of these metastable atoms is low, so they don’t often play a significant role. There are two metastable states in helium, singlet 2s and triplet 2s, with 20.616 eV and 19.820 eV of energy each as seen in Figure 3.1 [28, 29] with lifetimes on the order of seconds. Helium plasma has been used for the current work due to its low sputtering threshold on lithographic material (due to its low mass), in order to achieve particle removal. Under some conditions of prior work, namely the electrostatic removal process associated with PACE, a pulsed DC bias was applied to the sample . This same type of processing has been carried forward to the current work due to the pulsed power supply that was previously used able to draw more current. The pulse typically operated between +10 volts and -70 volts at a frequency of 100 Hz at a duty cycle of 90 % positive. This leads to sputtering of the sample at -70 volts for 10 % of the processing time. Section 3.1 outlines the expected removal of three material types from pure sputtering alone which when compared with the preliminary results, does not explain the removal mechanism. Through optical emission spectroscopy, it is possible to monitor the 501.6 nm and 388.9 nm transition that populate the 2s singlet and 2s triplet state respectively. As shown in Figure 3.2 and Figure 3.3, one is able to affect the density of the metastable states in the plasma through both bias to the sample as well as overall plasma source power.
Figure 3.1: A diagram of the energy levels of the helium atom. The two circled levels represent the metastable energy levels which are quantum mechanically forbidden from decaying to the ground state energy due to the conservation of angular momentum. Diagram from Sasaki et. al. 
Sputtering of Material
Sputtering yields for incident ions on a variety of surfaces can be determined through simulation via SRIM/TRIM . Figure 3.4 shows the results of this simulation, and specific sputtering yields for 70 eV helium ions are given in Table 3.1 for various materials. The ion current being drawn through the sample is measured to be 1.6 Ǻcm-2. Read More>>
An extreme ultraviolet light source is investigated at the University of Illinois. The source is a z-pinch plasma using Xenon gas with a short pulse width (~1 ms). As the plasma compresses, high energy photons in the extreme ultraviolet range are released and available for EUV lithography. The light emission is followed by ejection of multiply charged ions (+8-+10 Xe) which can significantly damage nearby mirror surfaces. After the fast ions pass, a slower moving cloud of ions and neutrals including Xenon and electrode material spread from the source causing continued coating and sputtering of the mirror surfaces.
The XCEED experiment serves to characterize the ejecta from a commercial extreme ultraviolet light source, investigate the damage mechanisms that affect nearby optics lifetime, and evaluate debris mitigation techniques that may be used to increase the lifetime of these mirror optics.
Characterization of the ejecta is performed through several diagnostics. A spherical sector energy analyzer (ESA) is used to diagnose fast ion species by energy-to-charge ratio using ion time of flight (ITOF) analysis. This instrument is capable of characterizing up to 15 keV ions emitted from the source. For neutral characterization, a set of Burle microchannel plates is placed directly in line with the pinch debris and a deflecting potential is used to divert ions from the detector. Signals from particle incidence on a Faraday Cup are also observed.
The effects of particle flux on mirror samples is investigated through exposure experiments and surface analysis. Samples are placed at normal and grazing incidences to the incoming particles and exposed for varying timeframes to the source. Photodiodes measure reflectivity degradation over time. The samples are removed and examined using Scanning Electron Microscopy (SEM), Atomic Force Microscopy, and other methods to analyze surface quality and particle deposition.
Recent modifications of the XCEED system also allow for the investigation of laser assisted discharge produced plasmas (LADPP) with a solid Sn electrode as the EUV fuel source. The laser ablates Sn atoms into the pinch produced plasma, where is it becomes highly ionized and emits a spectrum of light. A small portion of this light is transmitted in the EUV spectrum. The shift to Sn is ongoing within industry due to the increase in EUV emission conversion efficiency seen by Sn over Li and Xe. Laser ablation is carried out in the modified XCEED system using an Nd:YAG laser (100 Hz, 325 mJ/pulse, ~10ns pulse, capable of frequency doubling to 532nm and 266nm).