Preface for the Special Topic Collection Commemorating the Career of John Coburn

This Special Collection in the Journal of Vacuum Science and Technology A (JVST A) is compiled to honor the seminal and interdisciplinary contributions of Dr. John Coburn in plasma science and technology and plasma-surface interactions. This collection includes articles on important challenges, new developments, and current understanding in a wide range of topics in plasma science and technology and offers cutting-edge, outstanding research in the many fields John Coburn influenced and participated in. The special collection comprises 26 research articles, three review articles, and one perspective by scientists from around the globe who wish to acknowledge John Coburn's positive influence on the field of plasma science and technology.

John Coburn joined the American Vacuum Society (AVS) in the early 1960s, was AVS President in 1988, was named Honorary Member in 1991, received the John Thornton Award in 1993, and was appointed Fellow in 1994. He served as AVS Treasurer from 1999–2006, Trustee from 1986–1988 and 1995–1997, Director from 1984–1985, and Thin Film Chair in 1983. He is the author of a few AVS monographs and served on many AVS committees over the years. In 1994, the Plasma Science and Technology Division (PSTD) Coburn and Winters Award was created in his honor. John was also a very active Short Course Instructor, teaching his very popular course on plasma and reactive ion etching for AVS for many decades. He was a huge contributor to AVS journals and was very active in the society's Northern California Chapter. John was an influential force and kind mentor at AVS and is greatly missed.

To understand John Coburn's technical contributions, allow me to trace the highlights of his work relating to plasma etching and deposition and give my perspective of its historical development, as I remember it now in 2020. When John first joined my project at IBM in San Jose in 1968, understanding and exploiting the unique properties of sputtered transition metal, alloy, and metal oxide thin films prepared in a variety of low-density plasma environments was the order of the day. This work was motivated by the need to find appropriate techniques to generate magnetic materials in thin film form to achieve much higher magnetic storage density and explore faster magnetization reversal mechanisms. At that time, evaporation and thin film nucleation and growth studies in a UHV environment were already setting high standards of insight. Although in plasmas the rudiments of the momentum-transfer-sputtering erosion process per se were already well established, one's ability to account for the role of collisional processes of neutral and charged particles traversing the plasma from a sputtering target to a substrate in various short mean free path plasma environments was primitive. The identity, concentration, and state of excitation and kinetic energy of all the various species ultimately arriving at the growing film surface were empirical at best, especially in reactive plasmas. Speculation already abounded about various potential synergistic effects among such a “soup” of species arriving at the substrate surface. In particular, the fact that energetic ion bombardment of a growing thin film on a substrate can dramatically change the morphology, crystallography, structure, and composition and, therefore, all resultant properties had been amply demonstrated by us and others by the time John joined us. However, any quantitative insight such as, for example, the kinetic energy spectrum of positive ions under different mean free path plasma conditions arriving at a substrate or that of negative ions from, for example, oxide sputtering targets were speculative at best. Being able to grow stoichiometric metal oxide films at room temperature was intriguing. It elicited questions about the respective role oxygen radicals generated in the plasma and the inevitable low energy (∼20 eV) ion bombardment of a grounded substrate in an rf capacitively coupled diode system. Certainly, no definitive proof of specific mechanisms existed on any of this when John arrived on the scene. John knew nothing about plasmas or thin film physics when he joined my project, but already during his job interview, he showed real interest and asked very pertinent questions, demonstrating that the complexity of these plasmas did not scare him off and that he welcomed the challenge of helping us gain more insight into better plasma diagnostics in the context of sputtered thin film nucleation and growth in various plasma environments.

By early 1969, he had designed and built a bakable UHV diagnostics system capable of sampling and identifying the particle flux incident on a substrate held in a plasma environment of a planar diode sputtering system. In this system, a beam of particles (ions and neutrals) from the discharge was extracted at the substrate plane into a long mean free path environment where it passed through a 90° deflection electrostatic analyzer and into a quadrupole residual gas analyzer. Among early eye-opening results were all the species arriving at the substrate as well as the energy distribution of incident ions as a function of substrate bias during deposition. Later variations on this general sampling diagnostics approach in various plasma configurations (DC; rf; triode; inverted-magnetron-hollow cathode) led us to gaining a wealth of novel insights relevant to several emerging plasma applications elsewhere.

Also, early on, very diverse surprises popped up in our work. One thing we took notice of was the degree to which different high energy Auger neutralized inert gas atoms, inevitably backscattered from the sputtering target, get “trapped” throughout the growing thin film on a substrate held at room temperature, with very significant effects on the morphology and crystallographic orientation and physical properties of the resultant films. Also surprising was the degree to which sputtered neutral metal atoms leaving the target are Penning ionized by metastable Ar* while crossing the plasma. Alternately, practical issues were noticed, such as the degree to which “film contaminating” species are generated at plasma interfaces with the chamber-wall and apparatus-fixtures depending on their potential with respect to the plasma potential. Such contamination subtly influenced magnetic properties of the thin films of interest to us at the time. By the early 1970s, using this same plasma diagnostic approach also led us to a new generally applicable analytical depth-profiling technique of, for example, elemental composition profiles in multilayered thin film assemblies, with what was considered “good” resolution at the time. The influence of the “altered layer at the target” on depth profiling resolution by this technique became a major area of interest.

Harold Winters, who had joined my project several years earlier in 1964, came with an appropriate surface science background, having studied the adsorption resulting from electron impact on gas phase molecules. By the time John arrived, Harold had extended his UHV beam surface studies to other surface processes deemed relevant to plasma processing. For example, he conducted the first systematic experimental investigation of the physical sputtering of chemisorbed gases from a surface. He showed that the sputter yields were unexpectedly large and that the energy dependence deviated from that for elemental materials. In earlier work, we had accumulated indirect evidence for large sputtering yields of a number of chemisorbed gases of growing biased versus unbiased sputtered films in N₂ or O₂/Ar plasma environments. We found that surface adsorption characteristics of a given material greatly influence compound growth and/or gas incorporation. We learned that a functional relationship between the measured nitrogen concentration in the film and partial pressure of nitrogen places metals into three classes. For class-1 materials, the primary incorporation mechanism is the chemisorption of molecular nitrogen, while for class-2 and class-3 materials it is the chemisorption of atomic N at endothermic or activated exothermic sites.

In an ongoing parallel project by me, the mysteries of producing fluorocarbon thin films by plasma polymerization in an inductively coupled rf plasma system had been unfolding. Our initial studies centered around studying electrical conduction mechanisms in these highly cross-linked polymer films, as a function of frequency and temperature. From a synthesis point of view, we had already learned how the rate of polymerization and composition of the polymer films systematically depended on the ratio of unsaturated (CF₂)_n radicals to F radicals in the plasma for different fluorocarbon feed gases. We used optical emission for studies of the plasma and XPS studies for determination of film structure. Eventually, our interest evolved into exploring the use an rf capacitively coupled diode metal target system for producing metal containing composite polymer films and exploring their physical properties. This particular plasma configuration approach arose from an interest in exploring the feasibility of controlling the degree of metal physical sputtering from the high-voltage metal (Au) target and controlling the degree of polymerization and polymer structure at the controlled low voltage biased substrate for various ratios of Argon to fluorocarbon feed gases. A gold (Au) target was used to minimize surface chemistry (poisoning) at the sputtering target thereby preserving the zero valent nature of Au in the growing polymer film on the substrate. In this study, we had to learn such things as how much a mixture of controlled low energy [Ar⁺, CF₃⁺, (CF₂)_n⁺] ions bombarding the growing polymer film surface changed properties of these films.

John's extensive plasma diagnostic input using his mass spectrometry and ion energy analysis scheme was absolutely crucial in helping us gain more detailed insight into the synthesis of these composite fluorocarbon thin films. For example, clear evidence arose how the energy of incident ions at the substrate influences the polymerization kinetics and subsequent composition of these highly cross-linked polymer films. Changes in the dielectric breakdown behavior reflected these changes. This project eventually led to a rich set of solid-state thin film studies, such as, the dramatic changes of optical, electrical, and mechanical properties associated with the onset of percolation at a unique volume fraction of the metal granules in this inert, dielectric, highly cross-linked polymer film matrix. The viability of “laser beam circuit writing in air” was demonstrated using scanning Excimer-laser-induced coalescence of metal (Au) granules from just below to just above onset of percolation and simultaneous selective complete UV photo-ablation of the polymer using these composite film materials. Bulk conductivity in Au metal lines was demonstrated.

It is perhaps not surprising that by 1976, in our particular working environment, we also recognized the importance of “plasma ion enhanced etching” and “plasma polymerization” in the Si/SiO₂ semiconductor world in the context of microminiaturization issues in integrated circuit fabrication. By 1977, John and I reported our first mass spectrometric diagnostic study of the chemistry in a 13.56 MHZ rf capacitively coupled CF₄ plasma with Si, SiO₂ and Si₃N₄ excitation electrodes, respectively. Supplementing the mass spectra in these studies were simultaneous in-situ measurements of etch rate of both Si and SiO₂ using two quartz microbalances at the grounded electrode. We saw not only how the composition of the large area excitation electrode dominates the chemistry of the plasma, but that the combination of the mass spectra and microbalance results enabled measurement of the magnitude of this effect. Specific key results clearly showed that oxygen in the lattice of the solid (SiO₂) being etched combines chemically with carbon leaving as CO, CO₂ and COF₂, leading to more fluorine in the gas phase relative to the (Si) etching case, where much of the fluorine is consumed in removing the carbon from its surface. The presence of unsaturated carbon-fluorine species in the gas phase is a reliable indicator of polymerization occurring on surfaces not subjected to energetic ion bombardment. Alternately, energetic ion bombardment prevents the accumulation of polymer deposit and enables the etching to proceed in a fluorine deficient plasma. These results were clearly supportive of then-emerging ideas on directionality, selectivity and sidewall blocking of etching micro features in Si/SiO₂-CF₄ plasmas.

In 1977, after numerous discussions, John, Harold and I published our first plasma etching “model” for etching Si and SiO₂ in CF₄ and CF₄/O₂ plasmas. The model provided a framework where key emphasis is given to accounting for active or etching species including their generation rate and, most importantly, the recombination processes of these active species and their recombination rate, the latter, primarily at surfaces exposed to the plasma. The difference of the two defines the composition of the effluent of the system which can be experimentally tested. This model allowed interpretation of a large amount of experimental data obtained from various laboratories elsewhere. We called this paper “Plasma etching - A pseudo-black-box approach”.

About this time, I was enticed into building and managing a much larger Material Science Group covering several other areas of interest. Naturally, my personal research activity for the rest of my career with post-Doc partners narrowed to understanding surface versus bulk magnetic phenomena during sputtered thin film growth, using in situ spin polarized (escape depth limited) electron spectroscopy and Kerr magneto-optics to distinguish surface from bulk magnetism. Dual ion beams were used, one to sputter the target and one to ion bombard the growing thin film at controlled incident ion energies in an UHV system. Auger spectroscopy was used for surface chemical analysis. Understanding magnetic exchange mechanisms across interfaces of sputtered multilayered film assemblies was an overall goal. However, plasma and associated surface science issues remained an integral and “close to my heart” responsibility.

By the late 1970's, phenomenological data of plasma etching obtained by many others working on the Si/SO₂ system had already demonstrated the practical efficacy and promise of this approach with respect to selectivity and directionality. For us, gaining more insight how specific key species interact with the surfaces to be etched was considered the most appropriate contribution we could now make. Fluorine and Cl₂ interacting with Si was on top of our list. Harold and John had already studied some aspects XeF₂ interacting with a Si surface in Harold's UHV beam system. At that time, XeF₂, F₂, or F reacting with a Si surface were, to first order, considered equivalent since chemisorption of all three gases leads to F atoms, which subsequently react to produce etching.

The next major step was obvious. How does ion bombardment with an energetic Ar⁺ beam modify what they and others had already learned about F atoms interacting with a clean Si surface, especially with respect to etch rate? John and Harold got together and designed a directed beam (Ar⁺/XeF₂) setup, which included a quartz microbalance coated with Si. The parameter control achieved with this directed beam approach in a high vacuum environment allowed the first unambiguous demonstration of ion assisted gas surface chemistry. In particular, this experiment very clearly demonstrated the very large synergistic effect, in that the Si etch rate observed with both the XeF₂ and Ar⁺ beams simultaneously incident on the surface greatly exceeds the sum of the etch rates observed with each beam separately. A subsequent experiment measuring the etch yield of Si atoms per 1 keV incident Ar⁺ as a function of XeF₂ flux showed that a single Ar⁺ is capable of removing 25 Si atoms from the surface and of the order of 100 F atoms, mostly in the form of SiF₄. This very large number clearly emphasized the fact that the chemistry cannot be provided by the incident ion but must come from the XeF₂, i.e., the XeF₂ surface chemistry is somehow enhanced by the incident energetic ion. They had already established that XeF₂ will etch Si at very large rates in the absence of radiation, and the fact that the etch rate is proportional to the flux of XeF₂ on the surface provides good evidence that the adsorption-dissociation process is the rate-limiting step. Once F atoms are present on the surface, the formation and desorption of SiF₄ proceeds instantaneously at room temperature. Therefore, one can conclude that the mechanism responsible for the Ar⁺-enhanced XeF₂-Si reaction rate is an increase in the rate at which XeF₂ dissociatively chemisorbs on Si. They further concluded that the high ion energy ion bombardment induced surface damage is responsible for this enhancement.

In contrast to F, they showed that Cl atoms do not spontaneously form a SiCl₄ molecule at room temperature in the absence of ion bombardment. On the assumption, at the time, that surface penetration is key to the formation of volatile Si-halogen compounds, Paul Bagus, also in our project group, showed in a theoretical study that F atoms can penetrate the Si lattice by going over a relatively small barrier of ∼1 eV, whereas the barrier for Cl penetration is much higher, ∼13 eV. This study was based on calculating the electronic structure using an ab initio cluster representation.

Understanding the many details and implications of these important observations on the three basic steps—adsorption, product formation, and product desorption—on this and a variety of other material systems relevant to this important etching technology now consumed much of John and Harold's research efforts for the rest of their careers.

John and Harold's contributions during the 1980s and early 1990s, as well as those of many others working in the field, are discussed in an excellent Review by Harold and John in Surface Science Reports 14, 161-269 (1992), entitled “Surface science aspects of etching reactions.”

John and Harold's contributions during this latter period were the result of a well-coordinated research effort between the two of them working as a close team. Harold was the more introverted, meticulous researcher with a strong bent in his later years for trying to create a unifying theoretical framework based on the multitude of surface processes he, John, and, by now, many others elsewhere had studied in the context of ion enhanced etching chemistry. John exercised a more global approach. He probably impacted a broader community. He, like Harold, was an excellent experimentalist but also an excellent communicator. He had a knack for honing in on key technical issues and then sharing them in ways tailor-made to the interests of his audience. He was both interested in, and equally effective in, communicating with those confronted with demanding technological issues as with those more dedicated to basic science issues. He understood and respected the fact that they often face different goals and constraints.

In 1993, after 25 years, John retired from the Almaden IBM Research Center, San Jose. He then spent one year as a Senior Distinguished US von Humboldt Scientist at the Fraunhofer Institut in Freiberg, Germany, studying dry etching of III-V heterojunctions. He then worked part time as a Senior Research Associate for a number of years with David Graves's group at the University of California, Berkeley. He continued to focus on plasma diagnostics and especially surface etching mechanisms, extending them to a broader class of materials including photoresists.

John passed away in November of 2018 and Harold in 2016. I am proud to have been their friend and colleague throughout all these years. They are greatly missed by all of us.

It is now 2020. Naturally, much has changed as the various articles in this collection clearly demonstrate. We have, in fact, arrived in the atomic scale era. In the context of today's manufacturing needs, we need to define the shape, sharpness, and precision of features at technology nodes with sub-10 nm dimensions. This has led to Atomic Layer Etching (ALE) as the most advanced etching technique in production today. Although disregarded as too slow in the 1990s, the more recent adoption of ALE has made impressive progress in this regard. Leveraging plasma has made ALE several orders of magnitude faster than the earlier approaches.

The basic concept behind ALE is to break down the etch into two or more individually controlled, self-limiting surface reaction steps that remove material only when run in sequence. One then switches back and forth between the steps until the desired number of atomic layers are removed. ALE can be utilized in either directional or isotropic etching by employing proper surface modification and product removal steps. This approach clearly makes understanding and controlling the chemistry and physics of the individual steps easier, in contrast to conventional continuous plasma etching where many competing plasma/surface interactions are at play simultaneously. In ALE, targeting one individual surface reaction per step, atoms wide, directs our thinking more than ever toward understanding the underlying surface mechanisms.

Judging from the papers in this collection, several other key characteristics inherent in the ALD and ALE approaches effectively resolve problems encountered in earlier plasma deposition and etching procedures. In fact, one of the hallmarks of ALE is maintaining smoothness of the etched film, since the surface morphology replicates downward with each removed “monolayer.”

One article reports about an order of magnitude smoother surfaces for ALE relative to high ion energy RIE. Others report subtler consequences, such as the specific choice of the plasma chemistry in an inductively coupled plasma reactor greatly affecting the surface smoothness. The example given is much greater smoothness resulting in etching Si in a SF₆ plasma versus in an NF₃ plasma. Furthermore, the addition of only 10% SF₆ to a pure NF₃ plasma produced a much higher reaction probability than in a pure NF₃ plasma. Experimental evidence is given that this surprising enhancement of reaction probabilities is probably due to adsorbed sulfur on Si acting as a catalyst and that this catalytic effect of S is ascribed to an enhanced F sticking probability and/or a decreased desorption rate on a surface covered with S.

Several other articles show that plasma polymerized films still play an important role in resolving selectivity issues. One of the articles points out that combining substrate-selective deposition of the polymer with etching opens a new processing window for selective ALE, as demonstrated in the particular case of selective removal of HfO₂ over Si.

Another paper addresses one of the most challenging etches requiring both directionality and selectivity, e.g., contact etching in transistors. Preferential ALD deposition of a specific plasma polymerized fluorocarbon film is used to protect SiN in the SiO₂:SiN system.

In another ALE paper, an SiO₂ surface is modified in a CF₄/NH₃ plasma forming ammonium fluorosilicate plasma and then removed by a thermal treatment with a lamp.

In yet another paper, the authors report atomic layer etching of SiO₂ using a steady-state Ar plasma, periodic injection of a defined number of unsaturated fluorocarbon molecules synchronized with plasma-based Ar⁺ bombardment. This leads to chemical modification of the SiO₂ surface. Using the synchronized low ion energy bombardment, <30 eV, facilitates etching of the SiO₂. The authors report the temporal variation of the chemically enhanced etch rate of SiO₂ as a function of fluorocarbon coverage using such a low Ar⁺ energy.

Yet another dimension of polymers addresses plasma chemistries with various levels of UV emission and their effect on a known set of photoresist polymers. Identifying specific photon-induced modifications on photoresist polymers can help detect UV/VUV emission in the plasma and decouple ion and photon effects on these polymer materials.

Another paper deals with the difference between photoassisted etching (PAE) and ion assisted etching (IAE) of Si in a high-density chlorine plasma. Precisely controlled ion energy distributions were generated by applying pulsed negative DC bias on the conductive sample stage. XPS spectra revealed that the surface layer under PAE conditions had a significantly lower chlorine content, composed of only SiCl. Under IAE conditions, however, silicon dangling bonds Si•, SiCl₂, and SiCl₃ were found on the surface, in addition to SiCl, with a relative abundance of SiCl > SiCl₂ > SiCl₃. The absence of higher chlorides and Si• under PAE conditions suggested that vacuum ultraviolet photons and above threshold-energy ions IAE interact with the surface very differently. Perhaps the most succinct description of the impact of John's work on the development of ALE is made by one of the authors in this collection, who says “Fine control of the ion energy and neutral-to-ion ratio could be the gateway of reactivity control.”¹ 

These samples of papers within this collection clearly demonstrate the breadth and level of sophistication that has been reached commensurate with the needs of contemporary technologies operating at the subnanometer level. It is clear that John's extensive plasma diagnostics work and his and Harold's studies on the surface science leading to a better understanding has been significant. This is true especially of ion enhanced erosion processes such as those encountered in reactive sputtering and, of course, in more recent years, in ion enhanced chemical etching relevant to the Si microminiaturized world. I can only imagine how much fun they would now have being part of the ALD and ALE world. It is also interesting that plasma polymerization continues to play an integral part in some of these processes. John's plasma diagnostics work in that area was certainly invaluable to us.

It is difficult to exaggerate the magnitude of the effect John Coburn had on my research group and on me personally. He joined us as a Senior Research Associate in 1994 after his sabbatical year in Germany following his retirement from IBM Almaden in 1993. He was an active member of the group for over 20 years. He typically joined us on the day of our weekly group meetings and would always spend part of the day talking with all the students, postdocs, and visitors. He brought with him from IBM the vacuum beam apparatus that he had built there. I would never have been able to design, build, and operate this system without John, not to mention his central role in helping to interpret the results. This apparatus was the basis of three Ph.D. dissertations, focusing on many different aspects of radical- and ion-surface interactions.

John designed and helped the students build two other major experimental systems: an inductively coupled plasma system with dual mass spectrometers (among other instruments) and a second, larger vacuum beam apparatus. The former system was used by five students and the latter by three. It needs hardly be stated that without John, none of these systems would have been possible in my lab. His insights and experience helped us immensely. John also generously helped keep the group running during my travels abroad and sabbaticals.

Although John was not a modeler, he made many vitally important contributions to our plasma and plasma-surface modeling projects. His basic instincts and intuitions about the fundamental physics and chemistry of the plasma were invariably correct. The combination of modeling and experiments, so important for the value of the modeling, would not have been possible without John.

Yet, in spite of John's massive scientific contributions to the group, I think it was John's unique personality and nature that had the biggest effect. The university lab experience might be thought of as a kind of apprenticeship, where students learn by observing and doing science. It is hard to imagine a better role model than John. In addition to his superb scientific skills and knowledge, John was an unusually generous and thoughtful person. He was both enthusiastic and honest, and he had an underlying gentleness. He was also funny, with a seemingly endless supply of jokes and anecdotes. It was terrific fun to hang around with John.

I had the amazing good fortune to also have Harold Winters and Dave Fraser, after their retirements from IBM and Intel, respectively, as part of our group for many years. John, Harold, and Dave were truly a dream team! I think both Harold and Dave were motivated to join us in part because John had joined us some years earlier, and they saw how well it worked for all of us. These senior, highly experienced, and quite famous scientists had an enormous influence on the students and others in the group. We learned from them that it is possible to be highly respected professionally while exemplifying remarkable human decency. John Coburn, in particular, set a high standard on all of the most important metrics.