Volume 64, Issue 2, April 2020 Published by Johnson Matthey © Copyright 2020 Johnson Matthey

Home , Evans (crater), Marmarabirlik, Platinum diselenide

ISSN 2056-5135

Johnson Matthey’s international journal of research exploring science and technology in industrial applications

Johnson Matthey Technology Review is published by Johnson Matthey Plc.

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License. You may share, copy and redistribute the material in any medium or format for any lawful purpose. You must give appropriate credit to the author and publisher. You may not use the material for commercial purposes without prior permission. You may not distribute modiﬁ ed material without prior permission.

The rights of users under exceptions and limitations, such as fair use and fair dealing, are not aﬀ ected by the CC licenses. www.technology.matthey.com www.technology.matthey.com

Johnson Matthey’s international journal of research exploring science and technology in industrial applications

Contents Volume 64, Issue 2, April 2020

101 Guest Editorial: The Importance of Interdisciplinary Science: When Chemistry Needs Physics By Andrew Smith 103 Ab initio Structure Prediction Methods for Battery Materials By Angela F. Harper, Matthew L. Evans, James P. Darby, Bora Karasulu, Can P. Koçer, Joseph R. Nelson and Andrew J. Morris 119 Autothermal Fixed Bed Updraft Gasification of Olive Pomace Biomass and Renewable Energy Generation via Organic Rankine Cycle Turbine By Murat Dogru and Ahmet Erdem 135 In the Lab: Targeting Industry-Compatible Synthesis of Two‑Dimensional Materials Featuring Niall McEvoy 138 Plasma Catalysis: A Review of the Interdisciplinary Challenges Faced By Peter Hinde, Vladimir Demidyuk, Alkis Gkelios and Carl Tipton 148 Nanosurfaces 2019 A conference review by Alistair Kean and Sara Coles 152 Observing Solvent Dynamics in Porous Carbons by Nuclear Magnetic Resonance By Luca Cervini, Nathan Barrow and John Griffin 165 Insights into Automotive Particulate Filters using Magnetic Resonance Imaging By J. D. Cooper, N. P. Ramskill, A. J. Sederman, L. F. Gladden, A. Tsolakis, E. H. Stitt and A. P. E. York 180 Manufacturing and Characterisation of Robot Assisted Microplasma Multilayer Coating of Titanium Implants By D. Alontseva, E. Ghassemieh, S. Voinarovych, O. Kyslytsia, Y. Polovetskyi, N. Prokhorenkova and A. Kadyroldina 192 EuropaCat 2019 A conference review by Andrew Richardson and Katie Smart 197 “Solid-State NMR in Zeolite Catalysis” A book review by Shingo Watanabe 199 Johnson Matthey Highlights 202 A Short Review on Properties and Applications of Zinc Oxide Based Thin Films and Devices By Sumit Vyas 219 Advances in Cold Sintering By Jessica Andrews, Daniel Button and Ian M. Reaney https://doi.org/10.1595/205651320X15828914551601 Johnson Matthey Technol. Rev., 2020, 64, (2), 101–102

www.technology.matthey.com

Guest Editorial The Importance of Interdisciplinary Science: When Chemistry Needs Physics

Introduction mapping how we apply these capabilities to provide customer solutions into our existing markets (pink Johnson Matthey has over 200 years of history, text) or where they may be aligned with global creating sustainable technologies, shaped around drivers and world challenges. Such techniques customers’ needs. Our ambition is to research, can be a valuable tool to help discuss and identify develop and innovate solutions to make the opportunities and needs for an organisation. world cleaner and healthier, today and for future Two areas increasingly dependent on capabilities generations. Much of the underpinning science bridging chemistry and physics are characterisation behind these technologies relies on a knowledge of and modelling of materials and processes and the chemistry and its application. Like most successful development of functional surfaces and coatings. organisations, Johnson Matthey reflects on the These topics feature heavily in this edition of the scientific capabilities that are key to developing these Johnson Matthey Technology Review. For Johnson solutions today but also looks to the future to plan Matthey, characterisation and modelling are key which capabilities will be required to meet future capabilities to help develop new technology. challenges and opportunities. Much of this learning comes from external insight by looking at what is Characterisation, Modelling, Coatings happening both within the markets and scientific and Surfaces disciplines we are familiar with but also in parallel disciplines. Today, our core scientific capabilities can Characterisation provides insights into composition, be grouped into nine key areas covering catalysis, structure and property-performance relationships characterisation and modelling, chemical synthesis, at all length scales. The latter includes in situ materials design and engineering, electrochemistry, and operando analysis, which is important to platinum group metal and specialist metallurgy, understanding how materials may respond in their process optimisation, product formulation, surface intended application. chemistry and coatings. Pulling these together Modelling also encompasses all length scales forms a powerful toolbox to develop solutions for and includes statistical, empirical and physical our customers’ needs. models. Modelling has been used for a long time in When looking beyond these capabilities, one useful chemical engineering to design reactors, systems ‘lens’ to look through is the overlap between scientific and processes. Examples include designing a new disciplines. For Johnson Matthey this might be to reactor for a chemical reaction, an aftertreatment look at the interface between chemistry, one of our system for a vehicle or a process flow sheet for key underpinning strengths, and other sciences. recycling waste materials. More recently, advances For example, the interface between chemistry in modelling are permitting chemists to be and physics; chemistry and biology or with more predictive, to be able to design materials, other enablers such as the digital transformation reactions and their performance with far fewer that is enabling different ways of exploring experiments. For example, in this edition, the science. Following this premise, Johnson Matthey need for computational modelling methods to Technology Review has devoted this issue to focus replace incremental experimental development to on physics and a future edition will look at biology. meet the need to design complex new advanced Figure 1 shows how Johnson Matthey’s core materials is explained (1). science capabilities today may overlap with physics The application of nuclear magnetic resonance and biology. Further insights can then be drawn by (NMR) to characterising activated carbons leads

Fig. 1. Johnson Matthey’s research and development (R&D) drivers and core capabilities to insights into kinetic exchange of solvent areas such as sensing, electronics and renewable molecules (2). The technique makes use of the energy are explored within this edition (3, 4). magnetic shielding properties of the carbon structure to give insights into molecular level Summary mechanisms which can give information to the chemist about where adsorbed species are in the Looking forward, global drivers such as climate material’s structure. These techniques enable the change, the energy transition, population growth industrial chemist to gain a better understanding and longevity and resource challenges will of the materials being used which leads to faster drive the need for new technologies in areas development and better understood technology. such as more sustainable products, low carbon Equally important is the fundamental understanding operations, clean energy and improved health and of new materials and their properties both at the medical care. To meet these challenges chemists atomic and molecular scales which in time can lead will increasingly need to reach out to adjacent to advances in existing or new technology. disciplines to develop innovative solutions. In this Coatings and surface properties is another area edition of Johnson Matthey Technology Review, at the interface of chemistry and physics. Johnson we welcome you to look at some of the advances Matthey has many examples of products which in physics and explore how they are being used to rely on the functionality of particles deposited onto drive forward R&D. a surface. Examples include precious and base metal catalysts, advanced energy materials and ANDREW SMITH medical components. As the coating thickness Johnson Matthey, Blounts Court, Sonning reduces from micron to atomic, the chemist’s Common, Reading, RG4 9NH, UK traditional toolbox to deposit layers of formulated Email: [email protected] slurries, pastes and inks changes towards different deposition techniques such as chemical vapour References deposition (CVD) and physical vapour deposition (PVD). The ability to design and deposit functional 1. A. F. Harper, M. L. Evans, J. P. Darby, B. Karasulu, C. P. Koçer, J. R. Nelson and A. J. Morris, Johnson particles of a controlled size and shape onto a Matthey Technol. Rev., 64, (2), 103 surface can find application in many disciplines 2. L. Cervini, N. Barrow and J. Griffin, Johnson such as transparent or reflective coatings, Matthey Technol. Rev., 64, (2), 152 semiconductor devices, energy harvesting and 3. A. Kean and S. Coles, Johnson Matthey Technol. sensing. Typically, these applications harness a Rev., 64, (2), 148 combination of electronic, optical and chemical 4. S. Vyas, Johnson Matthey Technol. Rev., 64, (2), functionality. Further examples of applications in 202

www.technology.matthey.com

Ab initio Structure Prediction Methods for Battery Materials A review of recent computational efforts to predict the atomic level structure and bonding in materials for rechargeable batteries

Angela F. Harper*, Matthew L. Evans, by predicting novel materials, both in stand- James P. Darby, Bora Karasulu, alone theoretical calculations and in tandem Can P. Koçer with experiments. In this review, we describe a Department of Physics, Cavendish Laboratory, materials discovery framework based on density University of Cambridge, J. J. Thomson Avenue, functional theory (DFT) to predict the properties Cambridge, CB3 0HE, UK of electrode and solid-electrolyte materials and validate these predictions experimentally. First, Joseph R. Nelson we discuss crystal structure prediction using the Department of Materials Science and ab initio random structure searching (AIRSS) Metallurgy, University of Cambridge, 27 Charles method. Next, we describe how DFT results allow Babbage Road, Cambridge, CB3 0FS, UK; us to predict which phases form during electrode Advanced Institute for Materials Research, cycling, as well as the electrode voltage profile and Tohoku University, 2-1-1 Katahira, Aoba, Sendai maximum theoretical capacity. We go on to explain 980-8577, Japan how DFT can be used to simulate experimentally measurable properties such as nuclear magnetic Andrew J. Morris resonance (NMR) spectra and ionic conductivities. School of Metallurgy and Materials, University of We illustrate the described workflow with multiple Birmingham, Edgbaston, Birmingham, B15 2TT, experimentally validated examples: materials for UK lithium-ion and sodium-ion anodes and lithium-ion solid electrolytes. These examples highlight the *Email: [email protected] power of combining computation with experiment to advance battery materials research.

Portable electronic devices, electric vehicles and 1. Introduction stationary energy storage applications, which encourage carbon-neutral energy alternatives, are The ability to store clean energy is paramount in driving demand for batteries that have concurrently the struggle to decarbonise the global economy; higher energy densities, faster charging rates, the demand for cheaper, higher performance and safer operation and lower prices. These demands more sustainable energy storage technologies can no longer be met by incrementally improving is growing rapidly with the market for electric existing technologies but require the discovery vehicles and distributed energy grids. A key of new materials with exceptional properties. challenge is discovering new battery materials Experimental materials discovery is both expensive which outperform present technologies. However, and time consuming: before the efficacy of a new experimental materials discovery requires battery material can be assessed, its synthesis and extensive amounts of laboratory resources. This stability must be well-understood. Computational makes materials modelling an attractive tool that materials modelling can expedite this process can reduce the cost and time associated with

103 © 2020 Johnson Matthey https://doi.org/10.1595/205651320X15742491027978 Johnson Matthey Technol. Rev., 2020, 64, (2) the discovery process. The effort to accurately The atomic-scale processes in materials are model battery materials has been made possible described by the quantum mechanical time- largely by a quantum-mechanical theory for independent Schrödinger equation, Equation (i): molecules and materials, known as DFT (1, 2). ĤY({Rj},{ri}) = EY({Rj},{ri}) (i) DFT is an ab initio (or first-principles) technique that requires no experimental input to make in which the wavefunction for the set of electrons predictions about materials. By using DFT and nuclei is denoted by Ψ({Rj},{ri}) where R j to understand how a material behaves at the are the positions of the nuclei, ri are the positions atomic level, predictions can be made about its of the electrons and Ĥ is the Hamiltonian of the behaviour as a battery component. system. The energy E obtained from this equation Results from DFT can both guide experimental represents a specific energy level for the system. design and also help to interpret experimental In general, the ground-state energy of the system, results. However, in order to make these E0, is the quantity of interest. The Hamiltonian for predictions, the atomic structure of the material this time-independent equation is Equation (ii): must be known. When this is not the case, crystal 2 2 ħ 2 ħ 2 Ĥ = – ∇j – ∇i + V({Rj},{ri}) (ii) structure prediction (CSP) can be used to search 2Mj 2me Σj Σi for the most likely arrangements of the atoms. Given a crystal structure, it is then possible to The first two terms in Ĥ are the kinetic energy perform theoretical spectroscopy calculations, operators of the nuclei and electrons, and the which can be compared to the experimental third is the potential energy. Nuclei and electrons spectra. Examples include NMR (3), X-ray interact via the Coulomb interaction. Unfortunately, absorption spectroscopy (XAS), electron energy the conventional Schrödinger equation is too loss spectroscopy (EELS) (4, 5), Raman and complicated to solve beyond just a handful of infrared (IR) spectroscopies (6). This is especially particles. Therefore, approximations are required important in the context of battery materials, as in order to solve this equation and obtain the changes in the atomic structure and chemical ground-state energy of the system of interacting bonding during device operation are crucial to electrons and nuclei. Since electrons move on very battery function. fast timescales compared to nuclear motion, the This review provides an overview of DFT and nuclei can be treated as fixed in space while the CSP applied to battery materials modelling and electronic-ground state is computed. This is the highlights recent computational research on Born-Oppenheimer approximation, which results battery anodes and solid electrolytes. Section 2 in a Schrödinger equation for the electrons, in outlines DFT and CSP methods. Section 3 explains which the nuclear positions and charges enter as how experimentally relevant properties of battery parameters only. The underpinning principle of materials can be computed. In Section 4, several DFT, the Hohenberg-Kohn theorem (1), builds from examples of applying these techniques to battery this approximation, providing a theoretical basis materials are discussed, including conversion/ for working not with the wavefunction, but with the alloying anodes, solid electrolytes and anodes for much simpler ground-state electron density, n(r). Na-ion batteries. Figure 1 shows an example of the calculated ground-state electron density of the atoms in a silicon crystal structure, represented by the smooth 2. First Principles Modelling of surface surrounding the atoms. The total energy of Battery Materials a system of electrons and fixed nuclei is a function 2.1 Density Functional Theory of all possible electron density functions. Using the Kohn-Sham ansatz, finding the ground-state electronic density is made computationally feasible DFT calculations have become an important part by expressing it in terms of auxiliary wavefunctions of materials research to discover and explain the which describe a fictitious non-interacting system causes of experimentally observed phenomena of the same density (2). The full expression for the at the atomic scale. They provide insights into ground-state energy E may then be written as the physics and chemistry of materials which aid KS Equation (iii): in further optimisation of materials for a specific 3 application. DFT primarily provides a means for EKS = T[n] + ENN + ∫ d rVext(r)n(r) (iii) calculating the total energy and electron charge 1 n(r)n(r') + d3 rd3r' + E distribution of any configuration of atoms. 2 ∫ |r – r'| XC

(a) (b) GGA is the focus of much of the theoretical work in the field of DFT today, where the ultimate goal is to find an exchange-correlation functional which accurately describes all possible systems (5). Within this framework, total energies, forces,

c equilibrium geometries, elastic behaviour and many c a other properties of interest can be readily and a b b accurately predicted. However, to predict a material’s Fig. 1. Three-dimensional (3D) visualisation of properties using DFT, it is necessary to know how its the electron density for a Si crystal structure. atoms are arranged. Thus, in the following section, The blue spheres in the structure represent Si we describe the method of CSP, which uses DFT to atoms, connected by rods which depict bonds. In generate structures of novel materials. the solid state, structures are periodic, with the basis vectors shown in the bottom left corner (a, b and c). The boundary of the unit cell is shown by 2.2 Crystal Structure Prediction the black box surrounding the atoms: (a) Si shown along the a direction; (b) Si shown along the a* There are multiple materials databases. Some direction. The electron density for this system is contain only the experimental crystal structures depicted using an isosurface within the crystal and and other relevant properties of known materials, a colourmap along the simulation box boundary. while others contain the computed properties of The isosurface is shown in yellow and the boundary box is shown in blue and green, where blue are both known and hypothetical materials. These areas of lower electron density and green are areas can be leveraged to perform CSP. For example, of high electron density known crystal structure prototypes can be decorated with any set of atomic species, resulting where the first term, Tn , is the kinetic energy in new hypothetical materials. The stability and associated with the non-interacting Kohn-Sham synthesisability of these new materials can then be particles; the second term, ENN , is the nuclear- assessed using DFT calculations and by comparing nuclear interaction; and the third term, Vext, is against thermochemical data in the database. the external potential of ion cores in which the Three of the major exhaustive databases of electrons move. The fourth and fifth terms DFT calculations, the Open Quantum Materials represent electron-electron interaction energies. Database (OQMD) (6), the Automatic Flow The fourth term is the exact classical electrostatic (AFLOW) framework for materials discovery (7) energy; the interaction energy of an electron with and the Materials Project (8) have been used to the mean field of all electrons. The fifth term is predict new materials and screen for desired the exchange-correlation energy, which attempts properties using a combination of high-throughput to account for all interactions not accounted for ab initio calculations and, increasingly, statistical within the first four terms. By dividing up the and machine learning approaches. In addition, energy in this way, while the exact exchange- experimentally identified structures are found correlation functional remains unknown, it may be in the Inorganic Crystal Structure Database approximated in various tractable ways. (ICSD) (9) and the Crystallography Open Database

The simplest approximation to EXC is the local (COD) (10). These databases have been used density approximation (LDA), where the exchange- as a starting point for many theoretical studies, correlation energy per particle is taken to be equal leading to several new discoveries in the field of to that of a uniform electron gas of the same energy storage, including identifying SrFeO3-δ as a electron density, at each point in space. Generalised material for carbon capture (11), verifying Li3OCl gradient approximation (GGA) functionals improve as a solid electrolyte with high ion conductivity (12) on the LDA by taking into account both the electron and predicting LiMnBO3 as a Li-ion battery density and the gradient of that density, resulting cathode (13). While these databases are useful for in a more accurate description of exchange and comparisons of known structures and enable the correlation (3). These functionals have limitations; discovery of materials that are based on known most seriously, both electron localisation and crystal structure prototypes, it is likely that new electronic band gaps are underestimated. structures exist which cannot be classified as one So‑called ‘hybrid functionals’ have aimed at semi- of the currently known prototypes. Therefore, it is empirically correcting the electronic band gap (4) necessary to perform CSP in order to explore novel and developing functionals beyond the LDA and phases of materials.

The search for new thermodynamically stable intuitive constraints which reduce the initial materials (those favoured to form during synthesis, search space to the most experimentally relevant when kinetic factors are excluded) using CSP trial structures. This constraint greatly reduces can take one of many approaches (14), but all the size of the search space and makes AIRSS involve a search for the lowest energy minimum applicable to a wide range of systems, including in a high-dimensional configuration space. The those at high pressure (20, 21). These chemical configuration space for a periodic structure with constraints include, for example: the phases of N atoms per unit cell has dimension 3N+3, taking conversion and alloying anodes (22–25) were into consideration the rotational symmetries and constrained by space group symmetries and unit-cell degrees of freedom, whilst the number atomic distances; high pressure phases of ice (26) of local minima in the space scales exponentially were constrained to H2O units; encapsulated with N (15). Ideally, all low-lying minima would nanowires (27) were constrained by rod group be sampled during CSP since metastable phases symmetries; metal-organic frameworks (28) were may be synthesised experimentally, or indeed constrained to molecular building blocks; grain- be thermodynamically stable under different boundary interfaces (29) and point-defects (30) conditions; for example, graphite is the most stable had some atoms fixed to describe the lattice allotrope of carbon under ambient conditions, but and systematically randomised other atoms to diamond can be easily synthesised under high describe interface and defect structures. pressure. Particularly popular approaches to CSP AIRSS explores configuration space using random include the use of evolutionary algorithms to sampling as shown in Figure 2(a) and proceeds ‘breed’ new structures (15) and particle swarm as follows. optimisation (16–18). To search for a new phase with chemical formula,

AIRSS (19) is the focus of this review. Despite AxBy, any number of atoms of element A and B the potential for having a high computational are placed randomly (denoted ‘Randomise’ in cost, AIRSS remains an effective method for Figure 2(a)) into a 3D simulation cell in the ratio structure prediction which allows for a breadth x:y. The cell and atomic positions are allocated such of searching and has proven successful in a that they obey a set of chosen symmetry operations wide range of materials. Beyond the ease of its (a space group in 3D). Further constraints, such as implementation, AIRSS has several advantages. minimum separation between atoms and a feasible Firstly, individual relaxations do not depend on one range for the atomic density of the unit cell, may another, hence all trials can be run concurrently be imposed. These constraints narrow the region making the algorithm trivially parallelisable to of the configuration space of possible structures the largest of supercomputers. Secondly, AIRSS by avoiding regions that describe unrealistic allows for the easy application of chemically arrangements of atoms.

(a) (b) Randomise According to constraints B A

Relax DFT geometry optimisation AB2 A2B A3B AB

Repeat energy Formation Until structures have been found multiple times x in AxB1–x

Fig. 2. (a) Workflow schematic of the AIRSS method which is used to find the ground-state structures of different materials; (b) example of a convex hull of elements A–B which details how AIRSS can identify a lower energy structure in the A–B phase diagram. Each green circle represents one structure from an AIRSS search, plotted as composition vs. formation energy. The dashed lines represent a convex hull in which the A2B structure is on the hull but no A3B structures have been identified yet. The solid lines represent the convex hull which contains a new structure of A3B, identified from an AIRSS search, which is lower in energy than A2B

The forces on the atoms and stresses on the 3. Calculating Experimentally cell are calculated with DFT and then minimised Observable Properties using the traditional optimisation algorithms (for example, conjugate gradients). This step is Once a structure is obtained, either through CSP or denoted ‘Relax’ in Figure 2(a). The energy of the from a database, it is possible to use DFT to calculate system is used as a metric to gauge how stable the many experimentally observable properties. In structure is. this section, we highlight several methods for Steps 1 and 2 are then repeated several thousand calculating quantities which are experimentally times in order to generate a representative set of relevant to the field of battery research, especially structures in the A-B chemical space. The search regarding electrodes and solid electrolytes. is stopped once the lowest energy structures have been found multiple times. The set of lowest energy 3.1 Theoretical Voltage Profiles structures are the candidates for phases that are likely to form experimentally. The electrochemical voltage profile is the voltage Using the DFT energies, one can construct a signal of the electrode measured (vs. a reference, ‘convex hull’ of all the structures found by AIRSS, usually Li+/Li) as a function of the number of ions as shown in Figure 2(b). The structures AxBy (i.e. charge) stored in the electrode. The phase which are likely to form, must both have a negative transitions, which occur within the electrode during formation energy relative to elemental A and B cycling provide the characteristic shape of the and lie on the convex hull tie-line between A and B voltage profile; two-phase regions show a constant to avoid decomposition into other binary phases. voltage, while solid-solution regions show a sloping This tie-line is shown by the black line connecting voltage. The voltage drop between two phases is the lowest energy structures in Figure 2(b). This proportional to the difference in their free energies figure illustrates the process of constructing a and thus these voltage drops can be computed convex hull using the optimised structures from directly from the free energies of the phases which AIRSS. Suppose at a given point during the AIRSS lie on the convex hull tie-line. The voltage-drop search, the only structures on the convex hull are between two phases with active ion concentrations

AB, AB2 and A2B, connected by the dashed line x1 and x2 is Equation (iv): in Figure 2(b). Subsequently, a novel phase, –qDGrxn A3B, is identified using AIRSS and is found to lie V = (iv) (x – x )F below the existing tie-line. In this case, CSP has 2 1 identified a new ground-state structure which suggests an additional phase, A3B is likely to where q is the charge of the active ion, F is the exist within the A-B phase diagram. Therefore, as Faraday constant and ∆Grxn is the change in shown in Figure 2(b), the hull is reconstructed Gibbs free energy between phases. In practice, to include the phase A3B, rendering A2B unstable, the change in Gibbs free energy in Equation (iv) given that it is now no longer on the convex hull. is approximated by the change in the DFT total Although this example is given for two dimensions energy, under the assumption that entropic (i.e. a binary system containing elements A and contributions will have a minimal effect on the free B) the convex hull construction is generalisable energy differences between phases during cycling. to N dimensions, in which the tie-lines between When studying a phase diagram computationally the lowest energy structures are computed in a there are a finite number of phases on the tie-line, similar manner. thus the profile will not be a continuous smooth In this way, AIRSS enables the prediction of line, but a sequence of two-phase regions with new thermodynamically stable and metastable constant average voltages. Although the profile compounds in a given phase diagram and the will not have the same characteristic curve as an convex hull construction provides a guide to experimental voltage profile, it is still possible to their stability compared to previously known calculate quantities of interest such as theoretical phases, without performing exhaustive chemical capacity, which is calculated from the maximal synthesis. Synthesis experiments can then be difference in active ion concentration between targeted at the most promising compositions and the predicted stable phases. Similarly, the energy characterisation experiments can be guided by density of an electrode is found by integrating the the predicted model structures. voltage profile between the two endpoint phases.

3.2 Computational Nuclear Magnetic the nucleus and its local environment and, when Resonance Spectroscopy referenced against a model nucleus, is referred to as the chemical shift. The observed chemical Beyond calculating the voltage profile, one may shift in most materials is determined by the further validate a crystal structure against experiment nuclear spin interacting with the orbital angular by using DFT to predict its spectroscopic signatures. momentum of paired electrons. In Figure 3, such Many spectroscopic methods, including XAS, EELS a shift is given for the phases of Li-P which form (31) and Raman spectroscopy (32), can be readily during cycling of a Li-ion battery with a phosphorus 31 calculated using DFT to aid characterisation. anode (22). The P chemical shift of each LixPy Solid-state nuclear magnetic resonance (ssNMR) phase is distinct, as shown by the coloured peaks spectroscopy is a tool for investigating the element- in the figure for each compound. specific local structure of materials, even for the Whilst the theory for computing magnetic disordered and dynamic systems present in battery shielding for isolated systems (such as molecules materials (33). Due to the complex structures and and clusters) was developed in the 1960s and processes that arise during battery cycling, the 1970s in the context of quantum chemistry (35), usefulness of NMR spectroscopy can be greatly these methods were not easily extendable enhanced by applying complementary techniques to solids (36). For periodic systems, such as to aid the assignment of spectra to the local battery anodes and cathodes, most modern environment of each nucleus. Theoretical methods implementations of theoretical ssNMR use DFT in DFT are sufficiently mature that the calculation and the gauge including projector augmented of chemical shielding tensors across a diverse wave (GIPAW) approach (37–39). It is not only range of inorganic systems is now routine (34). possible to compute the full chemical shielding NMR spectroscopy involves the precise tensor, but also several other effects that can measurement of the response of nuclei in an applied modify the lineshape of the NMR signal, namely magnetic field to weak oscillating perturbations; for quadrupolar coupling (for spin I�� >12/ nuclei), a given pulse scheme, the frequency of perturbing dipolar coupling (which can be simulated directly oscillations is adjusted until resonance is achieved, from the geometry using for example the at which point a signal is observed. The frequency SIMPSON software package (40)) and J-coupling of this resonance is a cumulative measure of (interaction of electron spins which can probe several competing interactions between the spin of chemical bonds directly) (41).

Li3P 0.84 V 1 1 2 2 LiP 0.9 V 567 1 2 3 4 1 2 3 4 5 6

Li3P7 1.15 V 1 2 3 2,3 1 4 4 5 5 LiP5 1 1 2 2 3 3 4 4

Intensity, absorbance units Intensity, LiP 1.5 V Black P 7

200 150 100 50 0 –50 –100 –150 –200 –250 –300 31P chemical shift, ppm

Fig. 3. Calculated 31P NMR chemical shifts (22) for various thermodynamically stable Li-P compounds found using a combination of data mining and AIRSS. The shifts show a clear trend towards more negative shifts (increased chemical shielding) as the Li content of the structures increases. This is related to the number of nearest neighbour Li ions of each P. These DFT predictions of NMR shifts enable experimentalists to correlate observed shifts with specific local structure environments. Reproduced with permission from the American Chemical Society

3.3 Predicting Transport Properties and robust and thus the commonly used definition with DFT of diffusivity. One can extract the diffusion coefficient DT() Finally, beyond just characterising the static crystal from the gradient of the MSD, given a well- structure of a battery material, it is also possible converged MD trajectory such that the MSD is to predict the dynamics of ions moving through a linear function of time. Here, the slope of the the material, which is especially useful when line of best fit gives the diffusion coefficient D, studying ionic transport in electrodes and solid times twice the dimensionality d of the diffusion electrolytes. The charge and discharge rates are (2dD* ). For ionic diffusion in three dimensions, key performance factors in battery design, defining d = 3. Depending on the level of mobility of ions the time required to fully charge a battery and in the system, good convergence of the MSD of the amount of power it can deliver, respectively. ions may require long trajectories, for example Rate capability is determined by the speed with 50–100 ps, thereby requiring tens of thousands of which the charge carriers can move through the time steps. As each step involves DFT energy or materials. Since both ions and electrons move in force evaluations, AIMD can be a computationally a battery, the rate capability depends on both the demanding process. Two common solutions to electronic and ionic conductivity of the materials. this are: (a) to analyse trajectories obtained at While the electrodes in batteries must be mixed elevated temperatures (500–2000 K) to foster electronic-ionic conductors, the electrolyte must be higher mobility and faster convergence of the electronically insulating. First principles methods, MSD; or (b) to utilise parameterised atomic force- such as DFT, can be used to study both electronic fields to allow faster evaluation of the interatomic conductivity and ionic conductivity of battery forces in the system compared to ab initio methods materials. Electronic conductivity can be assessed like DFT. A drawback of parameterised force-fields from electronic structure calculations (42–44), is non-transferability, so one needs a new set of while ionic conductivity can be calculated using fitted parameters for the specific set of atoms in a ab initio molecular dynamics (AIMD) or the new system. nudged elastic band (NEB) method (45), as The activation energy (Ea) for the ionic transport outlined below. in a given electrolyte or electrode can be obtained The bulk ionic conductivity, σ()T , of a solid from AIMD simulations using the Arrhenius law, electrolyte can be related to diffusion coefficients Equation (vi): via the Nernst-Einstein relation (46) defined as –E /k T D(T) ≈ D0e a B (vi) Equation (v):

2 2 where D0 is the theoretical maximum diffusivity at ne z D(T)HR s(T) = (v) infinite temperature, under the assumption that the kBT diffusion mechanism is not temperature dependent where n is the diffusing particle density, e the and no phase transition occurs. Analysis of the elementary electron charge, z the ionic charge, trajectories from the AIMD simulations can also kB the Boltzmann constant, T the temperature, provide useful information on the crystallographic

DT() the ionic diffusivity and HR the Haven ratio sites with higher occupation probability, while also accounting for the correlated ionic motion. revealing the preferred ionic conduction pathways between these sites (47, 57, 58). 3.3.1 Ab Initio Molecular Dynamics Simulations 3.3.2 Nudged Elastic Band Method

One way to compute ionic diffusivity of a given Another way to obtain ionic diffusivity from first material, using AIMD simulations, combines the principles is with optimisation-based methods, first principles aspects of DFT with the ability of through the exploration of minimum energy molecular dynamics (MD) to model ionic forces paths (MEP) describing a set of predefined ionic and trajectories. Methods to screen the mobility of migration pathways. To this end, the NEB algorithm ions along an MD trajectory include mean square is often used. Other approaches are also available displacement (MSD), mean jump rate (MJR) (47, for transition-state searches, for example the 48), velocity autocorrelation function (VACF) (49– dimer (59), Lanczos (60) and eigenvector-following 52) van Hove correlation function (53, 54) and (EF) (61) methods as well as others (62, 63). others (55, 56). MSD is the most straightforward Specifically, the NEB method computes the MEP (at

0 K) for a predefined route connecting the initial approaches. AIMD, in contrast, would in principle and final states of the motion of a single ion or a work with any concentration of diffusing ions by few, concertedly diffusing ions (45, 64). The ion- readily addressing the self-diffusion limit (67, 68). transport path is divided into intermediate steps Given these tradeoffs, a common practice in (called NEB images), defined by the interpolation the literature is therefore to combine AIMD with of these two end-point states. The NEB images NEB calculations, specifically by identifying the are concurrently optimised by introducing a potential conduction pathways from relatively set of imaginary spring-forces to ensure the shorter AIMD trajectories at a selected, elevated harmonic coupling of the consecutive images and temperature and to probe the MEPs to get Ea and a continuous path on the corresponding high- compute the other properties relevant to the ionic dimensional potential energy surface. Using the transport (57, 58, 69–71). climbing-image NEB that maximises the energy In many cases, as in the high-voltage high- of the saddle point(s) on the MEP, one can also capacity anode material TiNb2O7 (TNO), both locate the transition states, from which activation ionic and electronic conductivities are relevant energies (Ea) are calculated. to the performance of the battery material (72). In solids, the change of entropy during ionic In this case, density of states (DOS) calculations diffusion is usually negligible and thus activation were used to determine that the electron-doped free energies are typically approximated by their TNO is metallic, as compared to the pristine 0 K values. The diffusion rate can then be related TNO. Additional localised electronic states were to the ionic diffusivity in the dilute carrier limit (65) confirmed in AIMD as a result of bond distortions, (i.e. diffusion carriers do not interact) using thus exemplifying the need in this case for both Equation (vii): AIMD and DOS calculations.

2 DEa D = l gfxDν*exp – (vii) ( kBT ) 4. Applications to Modelling Rechargeable Batteries where�λ�is the hop distance between two adjacent sites, g is a geometric factor that depends on the Each of the theoretical methods described in symmetry of the sublattice of interstitial sites, f Section 3 still require a model crystal structure is the correlation factor, xD is the concentration which can be obtained either from CSP or of the diffusion-mediating defects, v* is the experiments. Thus, we establish a workflow entropy difference between the initial and final from prediction to realisation in several simple states, the activation energy Ea is the energy steps. The general outline of this workflow is difference between the initial and final states, kB is to: (a) use AIRSS or another CSP method to Boltzmann’s constant and T is the temperature of search for novel phases; (b) characterise these the simulation. materials using DFT; (c) use DFT to predict Static methods, such as NEB, provide and compare to experimental spectroscopy, computational efficiency over AIMD: NEB requires or AIMD and NEB to predict diffusion pathways only a few hundred DFT steps to converge and is through ionically conducting materials. Large accurate within the regime in which the electronic computational databases can be constructed for structure of the model system does not change with a particular electrode material, where one phase the ionic migration (66). NEB calculations also allow diagram may contain as many calculations as the for quantitative comparison of different migration entire databases mentioned in Section 2.2; the routes. Nevertheless, NEB is less likely to reveal new Python package ‘matador’ (73) has been created conduction mechanisms compared to AIMD, and to perform this high-throughput workflow and the complex cooperative conduction mechanisms automate this database construction from CSP may not be as straightforward to sample with NEB results. The following sections provide examples as with AIMD. Moreover, NEB usually operates in in which this workflow has been successfully the dilute regime (Equation (vii)), where vacancy implemented for anodes and solid electrolytes. defects are manually introduced in the sublattice Whilst this same methodology could be applied of the diffusing ions to have a low diffusion carrier to cathode materials (66), we focus here on concentration and mediate the ionic motion. These anodes as cathodes are typically layered oxides artificial defects not only decrease the accuracy that undergo intercalation reactions where of the simulation models, but also impede the the structure of the host lattice is preserved. integration of the NEB method in high throughput In this case, Li sites within the host can

110 © 2020 Johnson Matthey https://doi.org/10.1595/205651320X15742491027978 Johnson Matthey Technol. Rev., 2020, 64, (2) usually be enumerated and the most probable 0.8 configurations studied using a cluster expansion (23) DFT (66, 74, 75). (79) 25°C (79) 400°C 0.6 4.1 Modelling Conversion and Alloying Anodes for Lithium-ion Batteries 0.4 Voltage, V Voltage, Graphite is ubiquitous in contemporary commercial Li-ion batteries. However, alternative anode 0.2 materials are a highly researched topic, due to graphite’s low capacity (372 mAh g–1) and tendency for Li plating and subsequent dangerous short- 0 1 2 3 4 5 x in Li Sn circuiting due to its low operating voltage (76). x These factors make graphite anodes unattractive Fig. 4. Comparison between theoretical and experimental voltage profiles for the Li-Sn for applications that require high performance and conversion anode. The black line is the theoretical capacity, such as electric vehicles. predicted voltage profile based on the phases that Here we highlight developments in predicting are on the convex hull tie-line (23), which matches high capacity conversion and alloying anodes to well with the experimental results of Wang et al., replace graphite, based on tin (990 mAh g–1) and shown in magenta and green for 25°C and 400°C respectively (79) antimony (660 mAh g–1). Such conversion and alloying anodes undergo a succession of reversible phase transformations during charging and results The Li-Sb phase diagram was found to be in their observed higher capacity retention than somewhat simpler, with only two stable phases other conversion and alloying anodes (77). predicted during cycling: Li2Sb and Li3Sb. Two

Both Sn and Sb were previously employed as competing polymorphs of Li3Sb were found and NMR anodes in Li-ion batteries, showing evidence of calculations were performed on both to provide a conversion reactions, with unknown phases of LixSn signature of each phase to aid the interpretation of and LixSb forming during Li insertion. An AIRSS future experiments. search for the thermodynamically stable phases This work on Li-Sn and Li-Sb anodes provided of both LixSn and LixSb was conducted (23, 78) theoretical confirmation of experimental binary in order to understand the voltage profiles and phases in this family of conversion anodes and reaction mechanisms of these two alloying anodes. allowed for more concrete evidence of the specific

In this case, a new phase Li2Sn was identified by mechanism of Li insertion into these anodes. AIRSS to lie near the convex hull. The resulting Furthermore, this study confirmed the new phase voltage profile is compared with experimental of Li2Sn. measurements in Figure 4. During the cycling process in conversion anodes 4.2 Modelling Lithium Diffusion in such as Sn, the material at the anode undergoes Solid Electrolytes several conversion reactions as Li is inserted (77). In the voltage profile shown in Figure 4, the The electrolyte in a battery forms a conductive black line is constructed from the ground state bridge between the anode and cathode which phases in the Li-Sn system, which were predicted allows ions to move from one electrode to the using AIRSS (23, 78). Each plateau in Figure 4 other without permitting the flow of electrons. represents a two-phase region between one ground Conventional Li-ion battery architectures use a state Li-Sn alloy and another, until a critical point is liquid electrolyte consisting of a Li salt mixture reached at which there is a phase transformation dissolved in an organic solvent. Two prominent (a vertical line) to the next Li-Sn alloy. safety concerns arise from the use of organic The DFT predictions lie within the voltage range of liquid electrolytes (80, 81). The first is that the the experiment and are an accurate match to both organic solvent component tends to be flammable sets of experimental data by Wang et al. (79). In and poses a fire hazard when exposed to air if the many cases, the experimental data has less-sharp battery casing is breached (82). The second is that distinctions between separated phases, due to Li dendrites (83, 84) form, which can eventually reactions which appear to occur gradually rather bridge the gap between the anode and cathode than at a well-defined stoichiometry. resulting in short-circuiting.

All solid-state batteries attempt to solve these quaternary chemical spaces and identified two Li safety issues by replacing the organic electrolyte superionic conductors, Li3Y(PS4)2 and Li5PS4Cl2. solutions with solid equivalents, which exhibit Particularly, Li3Y(PS4)2 is predicted to exhibit a high mechanical strength, suppressing dendrite room-temperature Li+ conductivity of 2.16 mS cm–1, formation, thus enabling the use of the high energy which can be further enhanced with aliovalent density Li-metal anodes (85, 86). Most proposed doping (93). However, these materials are yet to solid electrolytes have sufficient mechanical be synthesised. strength, as demonstrated by high throughput Following the structure prediction of these new screening based on machine learning methods (87). solid electrolyte phases, it is then desirable to A key challenge in developing solid electrolytes use NEB and AIMD simulations to investigate the is finding solids with room temperature (RT) ionic atomistic origins of their ionic conductivity. For conductivities that approach those of their liquid instance, Li-ion transport was elucidated in the counterparts. Among several solid electrolyte sulfide-based electrolytes, Li7P3S11 (99), argyrodite families identified to date, the thiophosphide Li6PS5Cl (48, 53), LGPS (57, 100), Li-Sn-S/Li-Sn- Se ceramics, for example Li2S-P2S5, chemically- (101, 102) and Li-As-S/Li-As-Se alloys (103), doped sulfides, like Li10GeP2S12 (LGPS) (88) Li3PS4 (48, 104, 105), Li4GeS4 (57, 103) as well as and Li9.54Si1.74P1.44S11.7Cl0.3 (89), are known to oxides, for example LLZO (71, 106–108), LiTaSiO5, deliver the highest RT Li-ion conductivities (1.2– LiAlSiO4 (71), Li4SiO4−Li3PO4 solid mixtures (109) 2.5 × 10–2 S cm–1). Sulfides, however, have high and several others. The problem of identifying solid moisture sensitivity and their chemical stability electrolyte candidates for all solid-state batteries against common electrodes is low, thus limiting which are air stable and highly conducting can be their practical use (90). By contrast, oxides like solved using a combination of structure prediction garnets (for example, LixLa3M2O12, where M = techniques and atomistic modelling such as AIMD zirconium, niobium, tantalum) display notably and NEB. higher chemical stability than sulfides but exhibit lower ionic conductivities (91). The latter limitation 4.3 Beyond Lithium: Applying can be partly remedied by a chemical doping with Structure Prediction to Na-ion diverse metals, including aluminium, gallium Batteries and scandium (92). High throughput CSP is useful for exploring So far, the battery materials we have discussed new superior electrolytes with combined high (Sections 4.1 and 4.2) are based on Li-ion conductivity and chemical stability. Various studies chemistry. However, cost and sustainability have performed extensive screening of superionic are driving research efforts into ‘beyond Li-ion’ conductors within databases such as the Materials batteries. The philosophy presented in Section 3, Project (8), searching for phases with good phase using CSP and DFT, is straightforward to extend to stability, high Li+ conductivity, wide band gap and ‘beyond Li-ion’ chemistries. A prominent example good electrochemical stability (12, 53, 93–95). is Na-ion batteries, where Li is replaced with the Various LGPS-derived compositions were more earth-abundant Na. predicted using ab initio calculations through Unlike in Li-ion batteries, graphite shows poor elemental swapping (95), such as Li10(Sn/Si)PS12 capacity for Na, although other carbonaceous and then verified by experimental synthesis and materials offer some promise (110). As such, measurements (96, 97). LiAlSO was discovered the success of future Na-ion batteries will rely solely through structure prediction and proposed on the discovery of new anode materials. There to be a superionic conductor with AlS2O2 layers, are many classes of anode materials which are which facilitate faster movement of Li-ions, low applicable to Na-ion batteries including two- activation barriers and a wider electrochemical dimensional transition metal carbides (111) and window (94). Similarly, Fujimura et al. (98) group V elements (P, As, Sb) (112). Although this presented a high throughput (HT) screening of review focuses on one Na-ion anode material in the chemical phase space for Li3.5Zn0.25GeO4 particular, structure prediction has been used to (LISICON)-type electrolytes. The authors proposed predict phases of each of the anode materials new electrolytes with higher conductivities than in several cases (113–115). In particular, black the parent LISICON material. Later, Zhu et al. (93) P shows a high theoretical capacity for Na of reported a HT screening of the Li-P-S ternary and 2596 mAh g–1, corresponding to the formation of

Li-M-P-S (where M is a non-redox-active element) Na3P (22). Here, P acts as an alloying electrode,

112 © 2020 Johnson Matthey https://doi.org/10.1595/205651320X15742491027978 Johnson Matthey Technol. Rev., 2020, 64, (2) so its cycling is expected to involve multiple contemporary NMR calculations lack a rigorous phase transformations. For these reasons, there treatment of paramagnetic contributions to the has been recent focus on understanding sodiation isotropic shifts, the chemical shift anisotropies processes in P. were computed for the thermodynamically Applying a combination of AIRSS, data mining (22) accessible range of predicted structures to provide and a genetic algorithm (25), the convex hull of a set of chemical environments to screen against the Na-P system has been mapped out and is experimental measurements. During the reverse shown in Figure 5. The Na-P system contains a cycle when Na is removed from the system, P helices number of stable crystalline phases (coloured black re-formed in a tangled fashion and the original circles in Figure 5) with compositions varying from crystalline P was not recovered. Amorphous phases

NaP7 through Na3P, and the voltage curve derived were encountered experimentally on desodiation from these phases shows good agreement with and, while modelling of amorphous materials experimental measurements (25). In addition is challenging, the local structural features of to these stable phases, there are metastable predicted metastable phases were discovered to be phases lying close to the convex hull across a present even in the amorphous structures. range of compositions. Aside from P, Sn also shows promise as a Na- ion By following the structures which fall on or near battery anode. Sn presents a lower theoretical the convex hull in Figure 5, from least sodiated capacity for Na (847 mAh g–1) but offers better

(pure P) to most sodiated (Na3P), the calculations capacity retention than P (24). The results of an predicted many changes in local structure: the AIRSS search for Na-Sn phases (24), predicted that layered black P is broken upon successive Na insertion of Na into Sn would result in hexagonally insertion, forming P chains and helices, then layered structures NaSn3 and NaSn2, before passing dumbbells, which eventually break apart to form through an amorphous phase of approximate isolated P atoms. These structural motifs are composition Na1.2Sn, after which a solid-solution distinctive and have characteristic NMR signatures, consisting of Sn dumbbells surrounded by Na which can be accurately modelled. In order to ions would form. The final product, 15 Na Sn4, confirm this explicitly, ex situ 31P solid-state NMR contains isolated Sn atoms surrounded by Na. measurements were taken at different points during Importantly, the computational workflow used to both the sodiation and desodiation cycle (25). Since study Li and Na-ion batteries is the same and

P Na 1.28 0 Distance from hull, eV per atom 0.64 –0.1 Phosphorus 0.32 –0.2 NaP7 0.16 Na5P4 –0.3 Na3P11

0.08 Na3P7 Na3P –0.4 Formation energy, eV per atom energy, Formation NaP Na5P4 0.04 –0.5 NaP Na P Composition: increasing Na content 3 0.02

Fig. 5. Convex hull (see Figure 1(b)) of the Na-P system as predicted using DFT through a combined approach using data mining, AIRSS and an evolutionary algorithm (22, 25). The ground state phases are labelled below the green tie line and their chemical compositions are given. The inset figures around the convex hull show the structures of intermediate Na phosphides, which are related to the structure of black P shown in the top left corner. In these structures the orange spheres represent P atoms and the purple spheres represent Na

113 © 2020 Johnson Matthey https://doi.org/10.1595/205651320X15742491027978 Johnson Matthey Technol. Rev., 2020, 64, (2) is equally as applicable to conversion anodes for and Can Koçer acknowledge the Winton Programme other chemistries. for the Physics of Sustainability, University of Cambridge, UK. James Darby acknowledges the 5. Conclusion funding provided by the Sims Fund, University of Cambridge, UK and EPSRC. Andrew Morris and Bora In this review, we have provided an overview of Karasulu would like to acknowledge funding from computational modelling of battery materials EPSRC (EP/P003532/1). The authors acknowledge using DFT, with a focus on cases where the atomic networking support via the EPSRC Collaborative structure of the material is unknown. In these Computational Projects on the Electronic Structure cases, CSP methods are used to find the most of Condensed Matter (CCP9) (EP/M022595/1) and stable arrangements of the atoms during battery NMR crystallography (EP/M022501/1). Computing operation. Once the atomic structure is known, resources on the Tier 1 resource ARCHER were a variety of theoretical spectroscopy and other provided through the UKCP EPSRC High-End modelling techniques can be employed to compare computational consortium (EP/P022561/1) and on these computational results to experiments. These the Tier 2 resources HPC Midlands+ (EP/P020232/1) include the prediction of NMR spectra, the probing and CSD3 (EP/P020259/1). of ionic conductivities using the AIMD or the nudged elastic band method and the construction References of voltage profiles. In this way, CSP combined with chemical synthesis can accelerate battery 1. P. Hohenberg and W. Kohn, Phys. Rev., 1964, research by creating a feedback loop between 136, (3B), B864 experimentalists and theorists. One method for 2. L. J. Sham and W. Kohn, Phys. Rev., 1966, CSP, AIRSS, has been used as a tool to predict new 145, (2), 561 phases in battery electrodes and has been shown 3. J. P. Perdew, K. Burke and M. Ernzerhof, Phys. to be effective both for understanding the atomistic Rev. Lett., 1996, 77, (18), 3865 mechanisms for electrodes and electrolytes which 4. P. J. Stephens, F. J. Devlin, C. F. N. Chabalowski are already in use, and for discovering new and M. J. Frisch, J. Phys. Chem., 1994, 98, (45), chemistries beyond those used in contemporary Li- 11623 ion batteries. 5. N. Mardirossian and M. Head-Gordon, Mol. Phys., By reducing the experimental trial-and-error 2017, 115, (19), 2315 necessary to optimise new battery chemistries, 6. S. Kirklin, J. E. Saal, B. Meredig, A. Thompson, computational modelling has the potential to J. W. Doak, M. Aykol, S. Rühl and C. Wolverton, reduce the time-to-market for novel device npj Comput. Mater., 2015, 1, 15010 chemistries, as well as providing overarching 7. S. Curtarolo, W. Setyawan, G. L. W. Hart, design principles. In addition, CSP, and atomistic M. Jahnatek, R. V Chepulskii, R. H. Taylor, modelling more generally, can now be used to S. Wang, J. Xue, K. Yang, O. Levy, M. J. Mehl, screen for new battery chemistries within the H. T. Stokes, D. O. Demchenko and D. Morgan, application-imposed constraints on performance Comput. Mater. Sci., 2012, 58, 218 and sustainability, with the goal of circumventing 8. A. Jain, S. P. Ong, G. Hautier, W. Chen, the need for unsustainable materials such as W. D. Richards, S. Dacek, S. Cholia, D. Gunter, cobalt. This growing interplay between modelling D. Skinner, G. Ceder and K. A. Persson, APL and experiment will be crucial to meeting energy Mater., 2013, 1, (1), 011002 storage goals required for decarbonisation. 9. M. Hellenbrandt, Crystallogr. Rev., 2004, 10, (1), 17 Acknowledgements 10. S. Gražulis, A. Daškevič, A. Merkys, D. Chateigner, L. Lutterotti, M. Quirós, N. R. Serebryanaya, Angela Harper acknowledges the financial support P. Moeck, R. T. Downs and A. Le Bail, Nucleic of the Gates Cambridge Trust, University of Acids Res., 2012, 40, (D1), D420 Cambridge, UK. Matthew Evans acknowledges 11. C. Y. Lau, M. T. Dunstan, W. Hu, C. P. Grey and the Engineering and Physical Sciences Research S. A. Scott, Energy Environ. Sci., 2017, 10, (3), Council (EPSRC) Centre for Doctoral Training in 818 Computational Methods for Materials Science, UK, 12. A. D. Sendek, Q. Yang, E. D. Cubuk, K.- for funding (EP/L015552/1). Can Koçer would like to A. N. Duerloo, Y. Cui and E. J. Reed, Energy thank the EPSRC for financial support. Angela Harper Environ. Sci., 2017, 10, (1), 306

13. J. C. Kim, X. Li, C. J. Moore, S.-H. Bo, 35. R. M. Stevens, R. M. Pitzer and W. N. Lipscomb, P. G. Khalifah, C. P. Grey and G. Ceder, Chem. J. Chem. Phys., 1963, 38, (2), 550 Mater., 2014, 26, (14), 4200 36. F. Mauri, B. G. Pfrommer and S. G. Louie, Phys. 14. R. Oganov, C. J. Pickard, Q. Zhu and R. J. Needs, Rev. Lett., 1996, 77, (26), 5300 Nat. Rev. Mater., 2019, 4, (5), 331 37. C. J. Pickard and F. Mauri, Phys. Rev. B, 2001, 15. C. W. Glass, A. R. Oganov and N. Hansen, 63, (24), 245101 Comput. Phys. Commun., 2006, 175, (11–12), 38. C. Bonhomme, C. Gervais, F. Babonneau, 713 C. Coelho, F. Pourpoint, T. Azaïs, S. E. Ashbrook, 16. Y. Wang, J. Lv, L. Zhu and Y. Ma, Phys. Rev. B, J. M. Griffin, J. R. Yates, F. Mauri and C. J. Pickard, 2010, 82, (9), 094116 Chem. Rev., 2012, 112, (11), 5733 17. Y. Wang, J. Lv, L. Zhu and Y. Ma, Comput. Phys. 39. J. R. Yates, C. J. Pickard and F. Mauri, Phys. Rev. Commun., 2012, 183, (10), 2063 B, 2007, 76, (2), 024401 18. S. T. Call, D. Y. Zubarev and A. I. Boldyrev, 40. M. Bak, J. T. Rasmussen and N. C. Nielsen, J. Comput. Chem., 2007, 28, (7), 1177 J. Magn. Reson., 2000, 147, (2), 296 19. C. J. Pickard and R. J. Needs, J. Phys.: Condens. 41. S. A. Joyce, J. R. Yates, C. J. Pickard and F. Mauri, Matter, 2011, 23, (5), 053201 J. Chem. Phys., 2007, 127, (20), 204107 20. Y. Li, L. Wang, H. Liu, Y. Zhang, J. Hao, 42. C. P. Koçer, K. J. Griffith, C. P. Grey and C. J. Pickard, J. R. Nelson, R. J. Needs, W. Li, A. J. Morris, Phys. Rev. B, 2019, 99, (7), 075151 Y. Huang, I. Errea, M. Calandra, F. Mauri and 43. C. P. Koçer, K. J. Griffith, C. P. Grey and A. J. Morris, Y. Ma, Phys. Rev. B, 2016, 93, (2), 20103 J. Am. Chem. Soc., 2019, 141, (38), 15121 21. J. R. Nelson, R. J. Needs and C. J. Pickard, Phys. 44. G. K. H. Madsen and D. J. Singh, Comput. Phys. Rev. B, 2018, 98, (22), 224105 Commun., 2006, 175, (1), 67 22. M. Mayo, K. J. Griffith, C. J. Pickard and 45. G. Henkelman, B. P. Uberuaga and H. Jónsson, A. J. Morris, Chem. Mater., 2016, 28, (7), 2011 J. Chem. Phys., 2000, 113, (22), 9901 23. M. Mayo and A. J. Morris, Chem. Mater., 2017, 46. R. J. Friauf, J. Appl. Phys., 1962, 33, (1), 494 29, (14), 5787 47. N. J. J. de Klerk and M. Wagemaker, Chem. 24. J. M. Stratford, M. Mayo, P. K. Allan, O. Pecher, Mater., 2016, 28, (9), 3122 O. J. Borkiewicz, K. M. Wiaderek, K. W. Chapman, C. J. Pickard, A. J. Morris and C. P. Grey, J. Am. 48. N. J. J. de Klerk, I. Rosłoń and M. Wagemaker,

Chem. Soc., 2017, 139, (21), 7273 Chem. Mater., 2016, 28, (21), 7955 25. L. E. Marbella, M. L. Evans, M. F. Groh, J. Nelson, 49. H. Hu, H.-F. Ji and Y. Sun, Phys. Chem. Chem. K. J. Griffith, A. J. Morris and C. P. Grey, J. Am. Phys., 2013, 15, (39), 16557 Chem. Soc., 2018, 140, (25), 7994 50. J. VandeVondele, M. Krack, F. Mohamed, 26. J. M. McMahon, Phys. Rev. B, 2011, 84, (22), M. Parrinello, T. Chassaing and J. Hutter, Comput. 220104 Phys. Commun., 2005, 167, (2), 103 27. P. V. C. Medeiros, S. Marks, J. M. Wynn, 51. H. van Beijeren and K. W. Kehr, J. Phys. C: Solid A. Vasylenko, Q. M. Ramasse, D. Quigley, J. Sloan State Phys., 1986, 19, (9), 1319 and A. J. Morris, ACS Nano, 2017, 11, (6), 6178 52. K. Ghosh and C. V. Krishnamurthy, Phys. Rev. E, 28. J. P. Darby, M. Arhangelskis, A. D. Katsenis, 2018, 98, (5), 052115 J. Marrett, T. Friscic and A. J. Morris, ChemRXiv 53. Z. Deng, Z. Zhu, I.-H. Chu and S. P. Ong, Chem. Prepr., 2019 Mater., 2017, 29, (1), 281 29. G. Schusteritsch and C. J. Pickard, Phys. Rev. B, 54. L. Van Hove, Phys. Rev., 1954, 95, (1), 249 2014, 90, (3), 35424 55. A. Van der Ven, H.-C. Yu, G. Ceder and 30. A. J. Morris, C. J. Pickard and R. J. Needs, Phys. K. Thornton, Prog. Mater. Sci., 2010, 55, (2), 61 Rev. B, 2008, 78, (18), 184102 56. R. Gomer, Rep. Prog. Phys., 1990, 53, (7), 917 31. E. W. Tait, L. E. Ratcliff, M. C. Payne, P. D. Haynes and N. D. M. Hine, J. Phys.: Condens. Matter, 57. Y. Wang, W. D. Richards, S. P. Ong, L. J. Miara, 2016, 28, (19), 195202 J. C. Kim, Y. Mo and G. Ceder, Nature Mater., 32. S. Baroni, S. de Gironcoli, A. Dal Corso and 2015, 14, (10), 1026 P. Giannozzi, Rev. Mod. Phys., 2001, 73, (2), 515 58. A. Vasileiadis, B. Carlsen, N. J. J. de Klerk and 33. O. Pecher, J. Carretero-González, K. J. Griffith M. Wagemaker, Chem. Mater., 2018, 30, (19), and C. P. Grey, Chem. Mater., 2017, 29, (1), 213 6646 34. S. E. Ashbrook and D. McKay, Chem. Commun., 59. G. Henkelman and H. Jónsson, J. Chem. Phys., 2016, 52, (45), 7186 1999, 111, (15), 7010

60. R. Malek and N. Mousseau, Phys. Rev. E, 2000, 83. E. Eweka, J. R. Owen and A. Ritchie, J. Power 62, (6), 7723 Sources, 1997, 65, (1–2), 247 61. L. J. Munro and D. J. Wales, Phys. Rev. B, 1999, 84. K. J. Harry, D. T. Hallinan, D. Y. Parkinson, 59, (6), 3969 A. A. MacDowell and N. P. Balsara, Nature Mater., 62. A. Heyden, A. T. Bell and F. J. Keil, J. Chem. Phys., 2014, 13, (1), 69 2005, 123, (22), 224101 85. S. Yu, R. D. Schmidt, R. Garcia-Mendez, E. Herbert, N. J. Dudney, J. B. Wolfenstine, 63. R. A. Olsen, G. J. Kroes, G. Henkelman, J. Sakamoto and D. J. Siegel, Chem. Mater., A. Arnaldsson and H. Jónsson, J. Chem. Phys., 2016, 28, (1), 197 2004, 121, (20), 9776 86. C. Monroe and J. Newman, J. Electrochem. Soc., 64. G. Henkelman and H. Jónsson, J. Chem. Phys., 2005, 152, (2), A396 2000, 113, (22), 9978 87. Z. Ahmad, T. Xie, C. Maheshwari, J. C. Grossman 65. R. Kutner, Phys. Lett. A, 1981, 81, (4), 239 and V. Viswanathan, ACS Cent. Sci., 2018, 4, (8), 66. A. Urban, D.-H. Seo and G. Ceder, npj Comput. 996 Mater., 2016, 2, 16002 88. N. Kamaya, K. Homma, Y. Yamakawa, 67. A. Van Der Ven, J. C. Thomas, Q. Xu, B. Swoboda M. Hirayama, R. Kanno, M. Yonemura, and D. Morgan, Phys. Rev. B, 2008, 78, (10), T. Kamiyama, Y. Kato, S. Hama, K. Kawamoto 104306 and A. Mitsui, Nature Mater., 2011, 10, (9), 68. A. Van der Ven, G. Ceder, M. Asta and 682 P. D. Tepesch, Phys. Rev. B, 2001, 64, (18), 89. Y. Kato, S. Hori, T. Saito, K. Suzuki, M. Hirayama, 184307 A. Mitsui, M. Yonemura, H. Iba and R. Kanno, 69. J. Kang, H. Chung, C. Doh, B. Kang and B. Han, Nature Energy, 2016, 1, (4), 16030 J. Power Sources, 2015, 293, 11 90. Y. Zhu, X. He and Y. Mo, ACS Appl. Mater. 70. X. He and Y. Mo, Phys. Chem. Chem. Phys., 2015, Interfaces, 2015, 7, (42), 23685 17, (27), 18035 91. R. Chen, W. Qu, X. Guo, L. Li and F. Wu, Mater. 71. X. He, Y. Zhu and Y. Mo, Nature Commun., 2017, Horiz., 2016, 3, (6), 487 8, 15893 92. V. Thangadurai, S. Narayanan and D. Pinzaru, 72. K. J. Griffith, I. D. Seymour, M. A. Hope, Chem. Soc. Rev., 2014, 43, (13), 4714 M. M. Butala, L. K. Lamontagne, M. B. Preefer, 93. Z. Zhu, I.-H. Chu and S. P. Ong, Chem. Mater., C. P. Koçer, G. Henkelman, A. J. Morris, M. J. Cliffe, 2017, 29, (6), 2474 S. E. Dutton and C. P. Grey, J. Am. Chem. Soc., 94. X. Wang, R. Xiao, H. Li and L. Chen, Phys. Rev. 2019, 141, (42), 16706 Lett., 2017, 118, (19), 195901 73. M. Evans, ‘Matador’, Rev. 063ab7ba, 2016: 95. S. P. Ong, Y. Mo, W. D. Richards, L. Miara, https://github.com/ml-evs/matador (Accessed H. S. Lee and G. Ceder, Energy Environ. Sci., on 19th February 2020) 2013, 6, (1), 148 74. J. M. Sanchez, F. Ducastelle and D. Gratias, Phys. 96. P. Bron, S. Johansson, K. Zick, J. Schmedt auf der A: Stat. Mech. Appl., 1984, 128, (1–2), 334 Günne, S. Dehnen and B. Roling, J. Am. Chem. 75. B. Puchala and A. Van der Ven, Phys. Rev. B, Soc., 2013, 135, (42), 15694 2013, 88, (9), 094108 97. A. Kuhn, O. Gerbig, C. Zhu, F. Falkenberg, 76. Y. Liu, Y. Zhu and Y. Cui, Nature Energy, 2019, J. Maier and B. V. Lotsch, Phys. Chem. Chem. 4, (7), 540 Phys., 2014, 16, (28), 14669 77. N. Loeffler, D. Bresser, S. Passerini and M. Copley, 98. K. Fujimura, A. Seko, Y. Koyama, A. Kuwabara, Johnson Matthey Technol. Rev., 2015, 59, (1), I. Kishida, K. Shitara, C. A. J. Fisher, H. Moriwake 34 and I. Tanaka, Adv. Energy Mater., 2013, 3, (8), 78. M. Mayo, J. P. Darby, M. L. Evans, J. R. Nelson 980 and A. J. Morris, Chem. Mater., 2018, 30, (15), 99. I. H. Chu, H. Nguyen, S. Hy, Y. C. Lin, Z. Wang, 5516 Z. Xu, Z. Deng, Y. S. Meng and S. P. Ong, ACS 79. J. Wang, I. D. Raistrick and R. A. Huggins, Appl. Mater. Interfaces, 2016, 8, (12), 7843 J. Electrochem. Soc., 1986, 133, (3), 457 100. Y. Mo, S. P. Ong and G. Ceder, Chem. Mater., 80. J.-M. Tarascon and M. Armand, Nature, 2001, 2012, 24, (1), 15 414, (6861), 359 101. A. Al-Qawasmeh, J. Howard and N. A. W. 81. B. Kang and G. Ceder, Nature, 2009, 458, (7235), Holzwarth, J. Electrochem. Soc., 2017, 164, (1), 190 A6386 82. C. Arbizzani, G. Gabrielli and M. Mastragostino, 102. J. A. Brant, D. M. Massi, N. A. W. Holzwarth, J. Power Sources, 2011, 196, (10), 4801 J. H. Macneil, A. P. Douvalis, T. Bakas,

S. W. Martin, M. D. Gross and J. A. Aitken, Chem. C. Masquelier and M. S. Islam, J. Am. Chem. Mater., 2015, 27, (1), 189 Soc., 2015, 137, (28), 9136 103. A. Al-Qawasmeh and N. A. W. Holzwarth, 110. M. A. Reddy, M. Helen, A. Groß, M. Fichtner and J. Electrochem. Soc., 2016, 163, (9), A2079 H. Euchner, ACS Energy Lett., 2018, 3, (12), 104. N. D. Lepley, N. A. W. Holzwarth and Y. A. Du, 2851 Phys. Rev. B, 2013, 88, (10), 104103 111. M. A. Hope, A. C. Forse, K. J. Griffith, 105. N. J. J. De Klerk, E. Van Der Maas and M. R. Lukatskaya, M. Ghidiu, Y. Gogotsi and M. Wagemaker, ACS Appl. Energy Mater., 2018, C. P. Grey, Phys. Chem. Chem. Phys., 2016, 1, (7), 3230 18, (7), 5099 106. K. Meier, T. Laino and A. Curioni, J. Phys. Chem. C, 2014, 118, (13), 6668 112. S. M. Beladi-Mousavi and M. Pumera, Chem. Soc. 107. R. Jalem, Y. Yamamoto, H. Shiiba, M. Nakayama, Rev., 2018, 47, (18), 6964 H. Munakata, T. Kasuga and K. Kanamura, Chem. 113. P. Bhauriyal, A. Mahata and B. Pathak, J. Phys. Mater., 2013, 25, (3), 425 Chem. C, 2018, 122, (5), 2481 108. F. A. García Daza, M. R. Bonilla, A. Llordés, 114. L. Shi, T. S. Zhao, A. Xu and J. B. Xu, J. Mater. J. Carrasco and E. Akhmatskaya, ACS Appl. Mater. Interfaces, 2019, 11, (1), 753 Chem. A, 2016, 4, (42), 16377 109. Y. Deng, C. Eames, J.-N. Chotard, F. Lalère, 115. S. Kirklin, B. Meredig and C. Wolverton, Adv. V. Seznec, S. Emge, O. Pecher, C. P. Grey, Energy Mater., 2013, 3, (2), 252

The Authors

Angela Harper is a second year PhD student in Physics in the Theory of Condensed Matter Group at the University of Cambridge, UK. She earned her BS in Physics at Wake Forest University, USA and her MPhil in Physics at the University of Cambridge. Angela’s PhD is focused on understanding the interfaces of Li-ion battery materials using DFT and crystal- structure prediction. In addition to her passion for studying materials with applications in green energy, she is also interested in mentoring students and encouraging women and underrepresented students especially to pursue careers in science.

Matthew Evans is a final year PhD student in the Theory of Condensed Matter Group at the Cavendish Laboratory, University of Cambridge. He obtained an MPhil in Scientific Computing at the University of Cambridge, following an MPhys in Physics with Theoretical Physics from the University of Manchester, UK. His research involves crystal structure prediction for beyond-Li battery electrodes and methods of materials discovery more generally. Matthew is an active practitioner of open source software and open science; he is the author and maintainer of two Python packages for materials science, ‘matador’ (for high-throughput computation and reproducible analysis) and ‘ilustrado’ (evolutionary algorithms for structure prediction) and has contributed to the CASTEP DFT code, the OptaDOS package and the Open Databases Integration for Materials Design (OPTiMaDe) specification for interoperation of materials databases.

James Darby is in the final year of his PhD studies in the Theory of Condensed Matter Group at the University of Cambridge. Prior to this, he studied Natural Sciences, also in Cambridge, where he obtained an MSci. His current work focuses on the application of symmetry constraints during crystal structure prediction and how such constraints may be ‘tweaked’ to minimise computational effort.

Bora Karasulu is a research associate in the Physics Department, University of Cambridge. His current research focuses on the computational (ab initio) material design for the next- generation all-solid-state batteries towards sustainable energy technologies. Previously, he was a research associate at the Eindhoven University of Technology (TU/e), The Netherlands, working on the first-principles modelling of the surface chemistry underlying the atomic layer deposition of metals on various substrates. He received his PhD in Computational Chemistry at the Max Planck Institute for Coal Research, Muelheim Ruhr, Germany, addressing the bio-enzymatic processes catalysed by flavoproteins.

Can P. Koçer is a PhD student in the Theory of Condensed Matter Group at the Cavendish Laboratory, University of Cambridge. He obtained his BA and MSc degrees in Natural Sciences from the University of Cambridge. His research is in the area of first-principles modelling of electronic, structural and dynamic properties of transition metal oxide materials, specifically for battery electrodes. Most recently, he has been working on complex oxides of early transition metals for high-rate anode applications.

Joseph Nelson is a research associate in the Department of Materials Science and Metallurgy, University of Cambridge, and an Advanced Institute for Materials Research (AIMR) Joint Center Scientist at Tohoku University, Japan. His current research is focussed on developing techniques to visualise and ‘navigate’ materials structure space, drawing on methods in applied mathematics. Previously, he was a research associate in the Department of Physics, University of Cambridge, using first-principles modelling to predict crystal structures and NMR spectra in battery materials. His obtained his PhD in Physics from the University of Cambridge, applying simulation to study the properties of materials subject to extreme pressures, in particular high temperature superconducting hydrides.

Andrew Morris is a Senior Birmingham Fellow at the University of Birmingham, UK. His research interests include the prediction of both the atomic structure and the spectroscopic signatures of new energy materials, with a focus on close collaboration with experimental groups. He is an author of the OptaDOS package for predicting the electron energy loss and optical spectra of materials from DFT calculations.

www.technology.matthey.com

Autothermal Fixed Bed Updraft Gasification of Olive Pomace Biomass and Renewable Energy Generation via Organic Rankine Cycle Turbine Green energy generation from waste biomass in the Mediterranean region

Murat Dogru*, Ahmet Erdem 1. Introduction Environmental Engineering Department, Gebze Technical University, Gebze, Kocaeli, 41400, Modern energy sources, mainly fossil fuels, Turkey are being used inefficiently at a high rate with concern of exhaustion. At the same time, there *Email: [email protected] is growing comprehension and recognition of greenhouse gas emissions, climate change and environmental pollution issues which have drawn Waste biomass, a renewable resource, is a worldwide attention to renewable power sources. reasonable choice for green clean power generation Recent environmental and energy policies support using advanced thermal treatment technologies investigations to increase the use of renewable such as gasification. In this research, dried-densified sources to reduce fossil fuel use and decrease olive pomace residues from olive oil production environmental impacts. As a renewable source, have been applied as biomass feedstock in a new biomass is an attractive feedstock for decentralised gasification process for synthesis gas (syngas) power generation. The European Union (EU) generation using a 500 kg h–1 throughput capacity is increasingly highlighting the objectives of autothermal modified updraft gasifier system. decreasing emissions of greenhouse gases and The product syngas generation rate is found to be enlarging the portion of renewable energy sources approximately 2.5 Nm3 kg–1 of olive pomace, with and hence using waste biomass as a valuable a calorific value (CV) between 5.0 MJ Nm–3 and resource. A significant amount of renewable energy 7.0 MJ Nm–3. More than 85% of carbon in pomace is derived from biomass feedstock. The renewable is converted to produced syngas by the gasification electricity provided from biomass feedstock is system. The gasification reactor generates syngas assumed to be around 14% of the entire renewable which passes through a specially designed swirl hot electricity production by 2030 in the Eurozone (1). gas burner and is then burned directly in a thermal In general, fossil-based fuels are the primary oil boiler retrofitted to an organic Rankine cycle feedstock for fuels and power sources on our planet. (ORC) turbine generator. As a result, the produced The use of biomass feedstock for energy production syngas at around 350°C is directly combusted with can reduce the consumption of fossil-based fuel tars so that a great deal of chemical energy loss is and contribute to decreasing the emissions of prevented. The thermal oil heater has a thermal greenhouse gas (2). Biomass materials constitute energy capacity of 1.77 MWh. The generated 1.6 the most significant proportion of carbonaceous MWh thermal energy from the thermal oil heater is waste materials. As an alternative form of energy, transferred to the ORC turbine to generate 240 kW the use of waste biomass feedstock to form electrical power. fuel sources is most welcome and appreciated

119 © 2020 Johnson Matthey https://doi.org/10.1595/205651320X15746781209529 Johnson Matthey Technol. Rev., 2020, 64, (2) because of regulations and legislation. Biomass is gasifier reactor, is later conducted to drying zone, abundant and widely available in nature. Biomass pyrolysis, reduction and oxidation processes, can provide constant power besides generating respectively. The updraft gasifier reactor as given other types of products. Consequently, biomass in Figure 1 shows the gas generated proceeding waste is considered to be a clean energy source upwards. Syngas, which is the main product of the and one of the alternatives to fossil-based fuels gasification process, flows through the gas exit at for the future. Biomass feedstock residues mostly the upper section of the gasifier. In a typical air-fed comprise wastes of forestry, agriculture and the gasifier, syngas is a mixture of a flammable gas food processing industry. such as hydrogen, carbon monoxide (CO), methane

Olives are a significant agrarian product. The (CH4), some types of tars and non-flammable inert world’s largest producers are Spain, Italy, Greece, gases like carbon dioxide (CO2) and nitrogen. The Turkey, Portugal, Tunisia, Morocco and Syria. 70% variety of syngas from gasification of biomass is of the world capacity of olive pits is produced influenced by aspects such as process parameters, within the EU countries. The rest of the world biomass specifications and design of the gasifier produces the remaining 30% (3). From the olive reactor. The features of biomass that have to be industry, the most critical and massive wastes dealt with in gasification are physical and chemical are olive pomace formed during oil production. A structures, such as density, elemental composition, small volume of pelleted olive pomace residue is fixed carbon (FC), volatile matter (VM), moisture and burnt; however, this feedstock can lead to several ash content. Controlling parameters in gasification complications in combustion boilers such as are equivalence ratio, temperature conditions and slagging, agglomeration and formation of clinkers feedstock throughput rate. The produced syngas (4, 5). Many studies have been conducted to date can be directly burned as a fuel without cooling on the combustion of olive wastes. In comparison, at atmospheric pressure in gas burners; there is there are insufficient observations available on no requirement for syngas treatment, reinforcing olive pomace use in gasifier systems. the efficiency of gasification. The syngas burner Several thermochemical conversion technologies design is a critical part since the syngas has a high can be applied for power generation from waste tar content. A properly designed burner helps the biomass. However, gasification is a convenient produced gas to burn efficiently. choice because it supplies higher efficiencies Feedstock reliability is vital in gasification compared to combustion or pyrolysis (6, 7). systems to accomplish sustained flow through Biomass gasification is a thermochemical the gasifier and to supply consistently produced conversion process that uses limited oxygen at high gas composition and higher heating value (HHV) temperature conditions to transform the solid form for the upstream power conversion. Additionally, of biomass into gas, volatile organic compounds densification of the feedstock reduces the gasifier (such as tars) and a small volume of ash and char. size, while the size and shape of the intensified The gas produced from the gasifying process agent biomass reduce fluctuations in produced gas. Fuel

(air, O2, steam, enriched air) is used to create flow also affects the subsequent quality of the the proper operating conditions. For instance, the products (12). lower heating value (LHV) of the produced gas Updraft gasifiers are identified as counter-current must be between 4 MJ Nm–3 and 6 MJ Nm–3 when reactors since oxidising agent passes upwards and air is applied as the gasification agent and has a the feedstock flows downwards under gravitational –3 much higher value when O2 (40 MJ Nm ) or steam force. These types of gasifiers are considered (12 MJ Nm–3 to 18 MJ Nm–3) are applied (8). As appropriate for fuels with relatively high ash and compared to combustion, gasification processes water content, have a high thermal efficiency due are more efficient and effective at generating to low exit gas temperature and have a low ash combined electrical power and thermal energy (9). carryover due to the filtering effect of the fuel However, some factors, such as reactor design bed (13–15). In a fixed bed updraft gasifier, the parameters, feedstock properties (moisture entrance point for the gasifying agent is at the content, particle size and ash) and gasifier reactor bottom section and for the feedstock, it is at the top operating performance conditions (temperature, section. Updraft gasifier reaction sections such residence time and equivalence ratio) affect gasifier as the drying section, pyrolysis section, flaming efficiency (10, 11). pyrolysis (partial oxidation) and gasification zones In updraft gasification, biomass waste feedstock, take place in a sequence in autothermal gasification which is delivered from the upper part of the systems. These zones reach different temperature

Solid H H fines C O

H H C H H 2 5

Feedstock flow H H H H Hot syngas H C H H C H Syngas: H C H C H H H H H Feedstock: H C H CH , CO H C H C C H H 4 Energy H C H H H O biomass H C H H C and H2 generation

H 1 H

H H C H H H C C C H 3 H C C H H H C H H H C H H H Fines Pyrolysis zone H2O and O2 C O 4 O reacts with C

O C O O C H H

Char zone gasification O C H Charcoal Reactor C H C O igniter O O C O Gasification O H H C agent Gasification agent C O O O H (air) H (water)

C C C C

Screening

6 Char discharge Char and residuals

Fig. 1. Schematic illustration of the updraft gasification reactor. Numbers denote the sequence

conditions in the flow of the gaseous product. As turbine based on biomass gasification have not a result of diversity in temperature and reaction been used so far and practically no references can zone sequence in the fixed bed updraft reactor, the be identified in that field (17–19). ORC turbines are performance of the gasifier is affected by design advanced energy generation machinery, based on and operating parameters (16). organic substances with favourable thermodynamic Biomass-based cogeneration processes are properties as working fluids: pressure and low becoming increasingly prevalent and several critical temperature conditions, low viscosity, small studies summarise what has been accomplished specific volume, high thermal conductivity and in this field. Some researchers have investigated surface tension. The main advantage in handling the usefulness of using biomass in combined heat organic working fluid is less need for heat for fluid and power (CHP) plants. Furthermore, most of evaporation compared to water; thus, ORC turbines these researchers have concentrated on methods operate at lower pressures and temperatures than that combine biomass incineration with ORC the conventional steam process (20–22). These turbines and few researchers have considered techniques provide a performance of about 15% in the probability of combining biomass gasification electricity and 60–70% in heat (23). processes. Comparing various biomass incineration and gasification systems, the gasification process 1.2 Aims of the Study was superior to incineration processes, both techno- economically and in terms of the performance The present pilot-scale research study was of the power conversion process. Despite these implemented in an autothermal updraft gasifier advantages, CHP systems operating via ORC reactor with a throughput capacity of 500 kg h–1.

The primary purpose of this study is to inspect and and applied in this experimental pilot-scale setup. to analyse the experimental data obtained including The thermal capacity of the gasifier designed gas concentration, temperature profiles, mass and is 2.20 MWh when the proper biomass is used energy balance. Syngas LHV, carbon conversion in this unique system. The design of the system and energy output via the ORC turbine are further and operation conditions of the updraft gasifier presented and discussed. Another goal of this work require the understanding of biomass feedstock is to demonstrate the possibility of using pelletised characteristics. Properties of biomass such as shape, olive pomace in cogeneration systems based on size, composition and water content are significant the gasification process. parameters that need to be considered prior to The evaluation of state of the art affirms, the design of a gasifier. Operation parameters therefore, that the combination of two unique such as feeding rate, gasification temperature and technologies, i.e., biomass gasification and ORC air:fuel ratio need to be measured as well. All these turbine, which are both technologies in progress, parameters play a crucial role in the performance can be considered as an original approach. of the gasifier in terms of gasification efficiency and Currently, the original combined biomass fixed bed quality of the gas produced during operation. autothermal updraft gasification and ORC turbine pilot-scale plant are both in operation. 2.1 Gasification Parameters In summary, this study was carried out using a pilot-scale (500 kg h–1) gasification plant consisting The autothermal fixed bed gasifier reactor gasifies of an autothermal updraft gasifier, hot syngas a maximum of 500 kg h–1 biomass feedstock. burner where the raw produced syngas was not This amount of gasified biomass approximately cooled or treated prior to the specially designed supplies 1250 Nm3 h–1 of produced gas to the syngas burner on the thermal oil heater which runs thermal oil heater. After the gasification process, in an ORC turbine at an output capacity of 240 kW almost 10% of the gasified biomass comes out as electrical power. The excess 1.36 MWh of thermal char, a byproduct from the reactor. Gasification heat is used for evaporation of the blackwater, a of the pelleted olive pomace is carried out in an harmful byproduct of the olive oil production facility. air-blown (680 Nm3 h–1 at the maximum load) The most important properties of pelleted olive updraft gasifier system operating slightly under pomace that are known to impact the gasification atmospheric pressure. systems are water content, shape and size, bulk The gasification reactor was built using a 6 m long and total density, chemical composition (i.e., reactor made of SUS 306 stainless steel with a ultimate and proximate analysis) and the HHV. 1.5 m diameter. The reactor body is well insulated This paper will focus on the production of power to prevent any significant heat losses. A basic plan and heat from the gasification of olive pomace in a of the autothermal updraft gasification system is pilot-scale autothermal fixed bed updraft gasifier. shown in Figure 2. The objectives of the study are to: The updraft reactor is made of a cylindrical-conical • Evaluate the performance of the updraft gasifier reaction vessel. The fixed bed gasifier structure is using dried and pelleted olive pomace as a fuel cylindrical with a feed rate of about 500 kg h–1. for proof of concept The biomass is conveyed from the main hopper to • Determine the fuel and char rates, gas the upper part of the reactor using a motorised compositions and CV of the gas produced by elevator and screw feeder. Fuel is admitted at the the gasification of pelleted olive pomace in an top with a screw conveyor and proceeds by gravity autothermal updraft gasifier down through inside the unit. • Generate the fundamental energy and mass For the start-up, primary air is used to light the balance data and diagram for the gasifier and biomass. Then, primary blower air is adjusted ORC turbine system to maintain the desired temperatures. Once the • Assess the feasibility of operating an ORC preferred temperature of the reactor is achieved turbine using the product gas in a thermal oil (about 900°C in that case), the moving grate is heater. activated and frequency of the feeder is regulated to stabilise the feeding rate required for olive 2. Experimental Setup pomace. The produced syngas exits the reactor at around 350°C through the channel. Typically, it A pilot-scale autothermal updraft gasifier with a requires around 1 h or 2 h to stabilise the operating capacity of 500 kg h–1 has been specially designed conditions with respect to gasifier temperatures.

Syngas F3 burner T6 SP P3 T8 Main T5 P2 F2 hopper P1 syngas M Electricity ORC T7 Generator turbine

generation Cyclone T4 Feedstock Gasifier T3 T9 Thermal oil reactor heater T2 T10 T11 T1

M VFD VFD M T12 FD air fan FD air fan

Stack Heat Exchanger ID stack fan

Particulates and clinker Char and box soot box

Fig. 2. Process flow diagram of the gasification system (thermocouple (T), pressure transmitter (P), flow meter (F), syngas sample collection port (SP)). Letters followed by numbers indicate the sequence of the instrument (for example (P2) stands for pressure transmitter number 2 and (T10) stands for thermocouple number 10)

Table I Gasification Reactions Heterogeneous gas-solid reactions –1 Boudouard reaction C + CO2 = 2CO, ∆H298 K = 172.5 kJ mol (i) –1 Water gas reaction C + H2O = H2 + CO, ∆H298 K = 131.3 kJ mol (ii) –1 Methane formation C + 2H2 = CH4, ∆H298 K = –74.9 kJ mol (iii) Homogeneous gas-gas reactions –1 Water gas shift reaction CO2 + H2 = CO + H2O, ∆H298 K = –41.2 kJ mol (iv) –1 Reforming CH4 + H2O = CO + 3H2, ∆H298 K = 201.9 kJ mol (v) –1 2CO + 2H2 = CH4 + CO2, ∆H298 K = –247.3 kJ mol (vi) Methane formations –1 CO2 + 4H2 = CH4 + 2H2O, ∆H298 K = 164.7 kJ mol (vii)

All parameters are kept constant for at least an pyrolytic molecules oxidise in the gas phase to hour for the analysis of the produced gas. form CO2 and H2O. The thermal energy, which is In the gasification (reduction) zone, with a high transferred to and used in other regions, is supplied amount of thermal energy from the oxidation by the exothermic reactions in this oxidation zone. region below, a number of endothermic reactions The temperature of the partial oxidation zone is take place between the gases and the char between 900°C and 1100°C. The basic gasification including steam, obtaining a large amount of H2 process is described by the simplified chemical and CO, along with CH4 gases. For instance, the formulas in Table I (Equations (i)–(vii)) (24). incandescent char in the gasification region reacts The moving grate inside the reactor is shown with CO2 gas that should be in the temperature in Figure 3. Transferring the char is possible by range of 700°C to 850°C and the char volume agitating the grill. Char movement causes a loss shrinks as it delivers C atoms to the CO2 to in pressure over the char bed at this stage. When convert CO. the pressure drop across the oxidation zone in the In the partial oxidation region, the gasifying reactor exceeds a threshold, the system activates agent is provided at the bottom of the reactor and the moving grate. The ash and char are removed is dispersed via movable grates to the pyrolysed at the bottom of the reactor by a screw conveyor. char. Not only the incandescent char but also The gas generated in the reactor is then taken out pyrolytic products such as partially oxidised from the top of the gasifier by repulse and pressure heavy hydrocarbons (tars) enter that region. The force of the induced draft (ID) fan and the forced

insulated thermal oil boiler. The pumps circulate thermal oil in the heated coils through the ORC turbine to generate 240 kW of electricity and excess heat is transferred to the evaporator units which vaporise the blackwater produced from the Fig. 3. Illustration of the moving grate at the olive oil facility. gasifier bottom 2.2 Biomass Feeding System Units draft (FD) fan. As the solid feedstock is transformed into gas, it conducts the remaining material to The main feedstock hopper and the screw conveyor move through the reactor under a gravitational are shown in Figure 2. The main biomass hopper, effect. The char residues formed during the process which has a volume of 3 m3 at the top of the are automatically discharged into the char box by reactor, is packed with the pelleted feedstock. A intermittently rotating the screw conveyor. The screw conveyor intermittently feeds pelleted olive produced gas leaves the reactor at a temperature pomace into the reactor at the upper part of the range of 250°C to 350°C. The produced gas is gasifier. A frequency converter can convert the then flared at the well-designed burner and fed needed amount of feedstock. The fuel flow out of to a thermal oil boiler to generate 1.77 MWh of the hopper interconnects with the entire reactor thermal energy. This thermal energy is transferred and the rotation speed of the drive motor. to the ORC turbine to generate 240 kW (15%) of A container with an elevator in the basement feeds electrical power. The excess 1.35 MWh useable the pelleted olive pomace to the main hopper. After thermal energy of waste heat from the system is this, the feedstock is fed into the main hopper where used in the blackwater evaporation unit. a screw conveyor feeds the biomass to the reactor. The whole system used in this study consists of The main hopper system with a screw conveyor not an updraft gasifier reactor, hot gas cyclone, syngas only prevents air leakage to the reactor but also burner, thermal oil heater and ORC turbine, ID stack avoids gas leakage from the gasifier. fan, FD air fans and a stack which is illustrated in Figure 2. 2.3 The Gas Analyser Data obtained in gasification system experiments include the flow rates of feedstock and produced The gas sampling port is located at the syngas gas, produced gas compositions, temperatures, exit point between the cyclone and the gas burner pressure throughout the operation line and of the system as shown in Figure 2. A portable electrical power generated in the ORC system. Vario Plus (MRU, Germany) syngas analyser is Every 15 s, a programmable logic controller (PLC) used to measure the volumetric fractions of the records all temperatures for air inlet, oxidation main product gas components. After attaining a zone, reduction zone, pyrolytic zone, drying region, steady state condition, the product gas analyser cyclone outlet and thermal oil boiler. The pressure is switched on. A heated probe is sucked into a drop is recorded at the top of the reactor and the small stream of the produced gas; then the gas is cyclone outlet. The air flow rate is measured after passed through a filter box filled with glass wool. the primary ID fan. The generated gas flow rate is The gas flows from bottles filled with water which calculated from the gas exit channel of the gasifier. act as a cooler; the cooled and clean product gases The produced gas exits the gasifier at around are analysed by the MRU syngas analyser. The

250°C to 350°C. From one exit located at the volumetric fractions of the gas components (H2, top of the gasifier, product gas passes through CO, CO2, O2, CH4, ethylene (C2H4) and ethane the cyclone and then the thermal oil boiler. The (C2H6)) are measured on a data acquisition for a produced gas exiting from the reactor includes definite period during 24 h of process operation. some fine particulate matter passing through During the plant operation, the composition of the the cyclone which is used as a dedusting unit to gas is analysed and data are collected for about separate these particles. The cyclone eliminates half an hour. most of the fine particles and dust from the hot gas produced. The produced gas channels and cyclone 2.4 Control System are well insulated to prevent tar condensation. Afterwards, the produced hot gas is transferred The gasification system is controlled by a PLC. The to the syngas burner and combusted in a well- entire control strategy used in that research aims to

124 © 2020 Johnson Matthey https://doi.org/10.1595/205651320X15746781209529 Johnson Matthey Technol. Rev., 2020, 64, (2) generate a continuous syngas flow for the thermal temperature for gasification. First, charcoal was oil boiler to produce thermal heat and transfer it ignited using a natural gas burner through the to the ORC turbine. To achieve these tasks, the ID ignition point. The optimum amount of air supplied suction fan is functioned at a constant rate after to the oxidation zone was regulated by FD and ID reaching steady state. The algorithmic system is fans located at the inlet of the gasifier reactor and equipped with automatic security controllers and at the stack respectively. can be operated remotely. The experimental conditions, energy and mass Temperatures are recorded with an analogue-to- balance data are presented in Figure 4 for digital (ATD) converter. Four thermocouples are fixed autothermal updraft air gasification. After layer to the internal refractory wall inside the gasifier, to embers in the gasifier were attained, feeding prevent probable issues with the flow of feedstock of the olive pomace pellets was started in the while it is consumed. The thermocouples are located gasifier. From the bottom of the gasifier, at around at corresponding positions along the vertical axis of 680 m3 h–1, the gasifying agent air at maximum the gasifier as shown in Figure 2. The temperature load was supplied to obtain the updraft effect in the of the gas generated in the reactor is calculated reactor. When the temperature of the gasification at the outlet channel of the gasifier. Three digital region reached between 300°C and 400°C, biomass manometers are used to calculate the pressure at pellets were fed at 8.35 kg min–1. During this period, different locations and to measure the pressure 500 kg h–1 feedstock was fed to the gasifier reactor changes over the system. These fixed locations are forming around 5 m bed height. This was provided the top of the gasifier and the channel between gasifier to reach the maximum, to keep the bed height reactor and dust gas cyclone. The heat exchanger is steady during the operation modes of the gasifier. used for the assessment of the waste heat remaining Air permitted to form 0.25 equivalence ratio was from the gas combusted in the thermal oil heater. The preserved throughout the updraft process. The hot air generated from the heat exchanger is used as gasifier temperature was stabilised by achieving a gasifying agent and combustion air in the thermal steady-state conditions; then gas sampling was oil heater. The flow rates of the gasifying agent, carried out to analyse the gas composition. The produced gas and combustion air used in the syngas temperature and gas composition were measured burner flow rates are calculated by flow meters. The during the gasification experiments until all of the values are recorded every five seconds. These digital material in the bed was gasified. indicators are connected to the PLC system and a supervisory control and data acquisition (SCADA) 2.6 Test Procedure and Power computer for data retrieval. Generation FD and ID fans are placed in the system. One of these supplies gasifying agent from the bottom The pelleted olive pomace was gasified with the of the reactor through the gasification system method described above. The method was repeated generating the updraft effect. The other is located several times to attain reliable results. For power near the stack for a suction effect into the system generation, the produced gas was passed through so that the produced gas is pulled over from the combustion chamber of the syngas burner, the gasifier, resulting in a slight pressure drop. which is placed at the top of the thermal oil boiler. Negative pressure is provided at the top of the Then, the burner increased the temperature of the reactor for safety reasons. The gasifying agent thermal oil, the heated fluid was transferred to the flow rate is controlled to keep the temperature of ORC turbine to generate electricity and excess heat the oxidation zone between 900°C and 1200°C. As was passed through the blackwater evaporation mentioned earlier, four thermocouples are located units to vaporise the blackwater. in the different reaction zones of the reactor to The operation of the gasification system could be measure the temperature. In addition, there are portioned into three parts as described below. nine thermocouples located at the cyclone gas For the initial application, an amount of charcoal is inlet and outlet channels, the syngas burner, the ignited by the natural gas burner from the oxidation thermal oil heater, the ORC turbine inlet and outlet, region to warm up the system. Pre-weighed olive the combustion air inlet channel and the stack. pomace biomass in the form of densified logs (pelleted) is charged through the main hopper from 2.5 Experimental Procedure the top of the reactor. The maximum bed height level of the gasifier is determined by a mixer; then, In the primary phase of the start-up, the gasifier the ID stack fan and FD fans are adjusted for the was ignited with charcoal to reach the desired updraft gasification process. The start-up period

Parameter Value Unit Parameter Value Unit Parameter Value Unit Biomass moisture 20 % w/w Biomass dry 0% feedrate 417 kg h–1 Gross power:Dry fuel ratio 575 kWe ton–1 Higher heating value 17.5 MJ kg–1 Biomass wet 40% feedrate 584 kg h–1 Annual operation 8000 h yr–1 Lower heating value 15.8 MJ kg–1 Gas:fuel ratio 2.5 Nm3 ORC gross electrical 15.01 % kg–1 efficiency Biomass feedrate 500 kg h–1 Char:fuel ratio 0.2 – ORC net electrical 23.89 % 20% moisture efficiency without WCC Hot syngas 85.77 % Char heating value 12.5 MJ kg–1 ORC net electrical 22.42 % efficiency efficiency with WCC Gasifier heat loss 1.00 % Syngas heating value 7.00 MJ Nm–3

Ambient air Biomass 310ºC input A Hot air to syngas Heat 170ºC 165ºC burner Filter Stack gas P GCV B J exchanger Exhaust gas Energy in C K N

Electrical power GCL 330ºC Turboden 350ºC 260ºC Q gasifier Hot syngas H Thermal oil ORC turbine Swirl generator Ambient 1200ºC heater LT-oil air D Energy I burner 950ºC HT-oil Electrical WCC dis. ther. heat Thermal efficiency: 15% 310ºC R 33.0ºC efficiency: 85% Energy O 20.0ºC 310ºC Turbine Heat loss E internal loss: Char F Excess air M 100.00% Syngas burner Energy G air L

Position Item Value Unit Position Item Value Unit Position Item Value Unit A Biomass 500 Kg h–1 I Energy 6760 MJ N Exhaust gas 5504 m3 h–1 input –1 3 –1 B GCV 15.8 MJ kg 1878 kWhn 3641 Nm h –1 C Energy 7881 MJ 2084 kWhg 4296 Kg h input 3 –1 2189 kWh J Hot air to 4755 m h O Energy, 1596 kWhn syngas burner nominal 3 –1 3 –1 D Ambient 679 Nm h 2390 Nm h Energy, gas 1772 kWhg air 801 Kg h–1 2820 Kg h–1 12.8 Kg s–1 0 MJ kg–1 K Ambient air 7492 m3 h–1 2.5 kJ kg–1 K–1

E Heat loss 79 MJ 3641 Nm3 h–1 P Stack gas 5442 m3 h–1 22 kWh 4296 Kg h–1 3641 Nm3 h–1 F Char 83 Kg h–1 L Syngas 4135 m3 h–1 4296 Kg h–1 burner air G Energy 1043 MJ 2078 Nm3 h–1 Q Electrical 240 kW power, gross 290 kWh 2452 Kg h–1 Electrical 221 kW power, net H Hot 2660 m3 h–1 M Excess air 620 m3 h–1 R WCC distributed 1356 kWh syngas thermal heat 1251 Nm3 h–1 312 Nm3 h–1 1476 Kg h–1 368 Kg h–1

Fig. 4. Energy and mass balance diagram for olive pomace gasification and ORC turbine system (Run 3). WCC = water cooling circuit; kWhn = kWh (nominal) by actual calculation based on the data collected; kWhg = kWh (gas) by calculated energy value from the syngas data comprises all operations needed until a steady the generated gas is ignited at the syngas burner. state whereby the gas quality for the thermal oil When the produced gas steadily burns in the boiler is reached. syngas burner and thermal oil reaches 290°C, The gasification system generally attains a steady then the ORC turbine is started up to generate state about an hour after the initial ignition. 240 kWh electrical power. The data collected Afterward, the temperature of the oxidation during the steady operation of the gasification region reaches between 900°C and 1200°C and system are temperature and pressure, fuel-flow

126 © 2020 Johnson Matthey https://doi.org/10.1595/205651320X15746781209529 Johnson Matthey Technol. Rev., 2020, 64, (2) rate, gas composition and char rate. Temperatures porosity in the reactor and as a result, tends to lead were measured with an ATD converter every 15 s to higher pressure loss in the gasifier. Gasification for oxidation zones, pyrolysis zone, drying zone, of small size feedstock could lead to high pressure gasifier gas outlet, cyclone outlet, thermal oil boiler, drop as well as excessive fine particle content in the thermal oil inlet and outlet and stack. Pressure produced gas. Also, inconvenient build-up issues data were collected at the gasifier gas outlet pipe, arise in the reduction region of the gasification bed cyclone outlet and thermal oil boiler outlet channel. with small size and low-density feedstock. The flow rate of the produced gas was measured by Conversely, larger particle size feedstock decreases carefully calibrated gas flow meters placed before the reactivity of the fuel and triggers bridging and the gas burner and cyclone outlet to measure the channelling obstacles that reduce the amount air flow from the inlet channels. of gas produced. Feedstock size homogeneity Lastly, the shutdown procedure refers to all also influences the operation performance of the operations needed to seal the gasification system reactor. The gasifier efficiency rises with increasing safely. Gas suction ID and FD fans are turned off; feedstock size homogeneity. For all these reasons, gasifier inlet valves, outlet and stack gas valves as shown in Figure 6, feedstock fuel is prepared are switched off in a systematic arrangement. The by pelleting to fractions of the preferred particle off-gas burner remains on as a secondary natural diameter (dp) (10 mm < dp < 12 mm) with a bulk gas burner until no more product gas is generated. density of 589 kg m–3. The water content in the feedstock also affects 3. Results and Discussion the quality of produced gas. Feedstock with lower water content produces better-quality product The experimental tests were performed in gas than that with a higher moisture content. The Marmarabirlik’s pilot gasification facility at Bursa, heating value of the produced gas can be influenced Turkey (Figure 5). The gasifier reactor was by the feedstock water content. Feedstock with designed and built to implement experimental tests high water content generates produced gas with of olive pomace gasification at high temperature low CV. Feedstock with higher than 30% water with air as the gasifying agent. content reduces the CV of the produced gas due to low heat transfer to the endothermic pyrolysis zone 3.1 Feedstock Characteristics reactions during the gasification process. More of the heat is absorbed by biomass to evaporate water The quality of syngas is affected by feedstock in the drying process. Thus, the heat required in characteristics (water content, particle size and the pyrolysis zone for reactions is insufficient. composition). Proper homogeneous feedstock size For this reason, biomass with high moisture is an essential factor in generating better quality content (>30%) must be dried during feedstock fuel gas. Compared with small size feedstock, larger preparation before the gasification process. Prior to sizes produce lower quality syngas. However, the gasification experimental tests, raw olive pomace feedstock which contains fine particles has low with an original moisture content of 60% by weight was dried and then pelleted during the feedstock preparation process at 105°C for 6 h. Proximate,

Fig. 5. Pilot-scale gasification system and ORC turbine at the Marmarabirlik intensive and miniaturised gasification facility in Bursa, Turkey Fig. 6. Pomace biomass from olive production

Table II Chemical and Physical Compositions (Ultimate, Proximate and GCV Analyses) of Olive Pomace and Woodchips Woodchips Olive pomace (oak) Bulk density, 589 250 kg m–3 Absolute 916 837 density, kg m–3 C, % 43.54 42.70 H, % 6.36 6.58 S, % 0.17 0.37 Fig. 7. Pelleted olive pomace feedstock N, % 1.73 0.45 O, % 44.65 47.77 Moisture, wt% 25.54 21.10 where the LHV of olive pomace is theoretically calculated as 17.48 MJ kg–1. Ash, wt% 3.55 2.13 Standard biomass feedstock for gasification has Volatile matter, 71.13 70.21 –1 wt% LHVs of around 15–17 MJ kg ; woody feedstock that has been the conventional fuel for gasification Fixed carbon, wt% 17.10 7.73 systems has HHV in the range of 17–21 MJ kg–1. GCV, MJ kg–1 17.65 17.47 The LHV that was calculated as 17.48 MJ kg–1 for olive pomace demonstrated that this feedstock is final analysis and gross CV (GCV) of olive pomace suitable for gasification in terms of CV equivalent sample results are compared with oak woodchips and to woodchips. presented in Table II. Proximate analysis supplies The feedstock moisture content greatly affects the composition of a substance in terms of FC, both the quality of the produced gas and the moisture, ash and VM as well as GCV. The ultimate operational parameters of the gasifier. Excessive analysis provides elemental compositions containing water in the feedstock drops the operational C, H, sulfur, N, O and moisture. Absolute and bulk temperature of the reactor and that leads to long densities of both olive pomace and woodchips are chain hydrocarbons in the form of heavy tars in the shown in Table II for comparison. produced gas leaving the reactor. The water content The GCV (also known as HHV) based on the of the feedstock specifies the type of gasifier design ultimate analysis was derived using the Institute that is used. Higher moisture contents of biomass of Gas Technology (IGT) method, as shown in feedstocks are accepted for updraft reactors. Equation (viii) (25): The absolute and bulk density of biomass is essential for process design, handling and storage. HHV (MJ kg–1) = 341C + 1323H + 68S – Biomass with lower bulk densities frequently causes 15.3A – 120(O + N) (viii) deficient current under the gravitational force that The HHV of olive pomace is theoretically calculated leads to insufficient gas CV and char burnouts in as 17.65 MJ kg–1 (IGT method formula). In wt%: the gasification region. However, biomass with C = carbon; N = nitrogen; H = hydrogen; O = higher bulk densities requires lesser reactor vessels oxygen; S = sulfur; A = ash and M = moisture for a definite refuelling time. The bulk density of content of olive pomace. olive pomace is higher than that of woodchips To generate thermochemical conversion systems (589 kg m–3) and the experiments verified that such as gasification reactors, determination of the there was minimum char burnout that appeared LHV rather than the HHV of fuel in the calculation in the reduction region. Due to minimised reactor is more effective. The water heat of vaporisation dimensions and the feeding charge capability of and the moisture content of the feedstock can be the gasifier, the feedstock is compressed in the overlooked as these do not contribute any CV to form of pellets. Figure 7 presents the pelleted the biomass. feedstock (10–12 mm diameter): olive pomace A method of relating HHV to LHV is shown in obtained from Marmarabirlik’s olive oil facility Equation (ix) (8): was used in this pilot-scale system. Commonly, pellets are produced by pressing the pomace LHV = HHV – 0.212H – 0.008O – 0.0245M (ix) under high pressures using standard compress

128 © 2020 Johnson Matthey https://doi.org/10.1595/205651320X15746781209529 Johnson Matthey Technol. Rev., 2020, 64, (2) equipment. Intensification of fuel could decrease conditions are given in Table III and graphical the space occupied by the feedstock in the reactor. results are illustrated in Figure 8. Measured Fuel intensification has some advantages such compositions show CO in the range of 23 ± 1%, as reduction of gasifier dimensions, convenience H2 7 ± 2%, CH4 3.5 ± 0.8%, CO2 10 ± 2% and of feedstock management and inhibiting dust the balance N2. During steady state operating exposure. Pellets that have uniform dimensions conditions, power generation at 240 kW was enable identical flow by gravitational force and continuously observed via the ORC turbine with homogeneous pellets create a uniform void field the pelleted olive pomace whose moisture content in the gasifier which avoids channelling throughout was around 25 wt%. Gas with a typical LHV of the gasification section. 5.0–6.5 MJ Nm–3 was generated in the reactor. The characteristics of the syngas composition are 3.2 Gasification Characteristics presented in Table III. Alternatively, the CV of the gas can be calculated Gasification characteristics of pelleted olive using Equation (x) (8): pomace obtained during three runs are presented CV gas = 1.055 (121XH + 119XCO + in Table III and compared with oak woodchips. 2 37.36XCH ) 24 (x) Table III also illustrates the different flow rates 4 that cause different characteristics of produced where XH2, XCO and XCH4 are the mole fractions gas. The equivalence ratio and air intake of the of the main combustible gases, H2, CO and CH4 gasifier are also shown. respectively. During Run 1, the gasifier reactor operated at the Figure 8 illustrates the composition of the syngas lowest quantity of gasifying agent. Therefore, the generated by the gasification reactor in Run 3. reaction slowed and feed consumption decreased. In During that run, the composition of produced gas Run 2, the gasifier operated at 300 kg –1h (half the was quite stable, so the ORC turbine operated capacity) and the quality of the gas slightly improved. smoothly and stably. The flow rate of the gas However, in Run 3, the gasifier operated at maximum produced by the gasifier in steady state operation capacity (500 kg h–1) and the syngas CV was highest. is between 270 Nm3 h–1 and 1251 Nm3 h–1. In Therefore, it is understood that the gasifier reached the operational run, the hot gas generated had its maximum efficiency at the highest load. an average LHV of 6.30 MJ Nm–3 and the gas was During the pilot-scale updraft reactor operations, subsequently used in a thermal oil boiler to run produced gas was taken by a probe and analysed the ORC turbine. The turbine is designed for the externally. The analysis results during steady state conversion of 1.36 MWh thermal energy input to

Table III Product Gas Composition at Different Feed Flows Rates for Olive Pomace and Oak Woodchips Pelleted olive pomace Woodchips Parameter Run 1 Run 2 Run 3 Sample run Fuel feeding rate, kg h–1 100 300 500 500 Air intake, kg h–1 146 410 679 708 Syngas rate, Nm3 h–1 270 752 1251 1305 Gas composition

H2, vol% 6.63 7.98 9.28 17.76 CO, vol% 20.51 21.26 24.68 14.27

N2, vol% 54.92 53.47 51.95 51.25

O2, vol% 0.73 0.59 0.28 0.32

CO2, vol% 14.16 12.58 9.42 13.54

CH4, vol% 2.46 3.54 3.95 2.16

C2H4, vol% 0.21 0.37 0.29 0.52

C2H6, vol% 0.38 0.21 0.15 0.18 CV, MJ Nm–3 5.19 5.93 6.67 5.81

10.0 70 O2 H2 CO2 CO CH4 N2 CV Gas calorific value, MJ Nm 60 7.5 50

CV 40 5.0

30 Gas volume, % v/v Gas volume, 20 2.5 –3

0 0 00:00 01:00 01:30 02:00 02:30 03:00 03:30 04:00 04:30 05:00 05:30 06:00 Time, h:min

Fig. 8. Produced gas composition and CV at different loads

240 kW electricity power output, which means 15% the grate operation and high pressure drops can efficiency of electricity generation. The gasifier was occur in the zone. Consequently, the feedstock operated with a turndown ratio of around 5:1 and characteristics, fuel preparation and sizing, syngas generation was stable enough to operate gasifier design and operation parameters are all the ORC turbine. Water was used for the turbine critical and interdependent factors and need to cooling system; the input temperature of the be carefully evaluated to avoid these problems. 50 m3 circulating water was 60°C and the output In this pilot-scale system, all these features were temperature was 90°C. The hot water obtained evaluated and the operation was terminated from the waste heat of the ORC turbine was used without any problems. within the facility for the blackwater evaporation unit. 3.3 Energy and Mass Balance The performed runs indicated that the particle Analyses and Results size and shape of the pelleted olive pomace significantly affect gasifier operation. Therefore, Determination and evaluation of the energy pelleted feedstock of size 12 mm × 50 mm was and mass balance of the system are essential selected and used in the reactor. Referring to the to reveal the energy production potential of the size of the gasifier, it is assumed that there is an autothermal updraft gasifier from pelleted olive upper limit for the particle size of 12 mm. This pomace feedstock. The calculation of the energy feedstock size is optimised for smooth movement and mass balance for the gasification system and to prevent bridge formation inside the constitutes a significant factor in establishing gasifier. the efficiency of conversion of feedstock to the The design and actuation system of the grate is product gas and the determination of energy important to discharge the char in the gasification production. The determination of the energy operation. The amount of char byproduct from and mass balance varies according to the olive pomace feedstock gasification was observed type and characterisation of the feedstock and to be quite low. Hence, it can be discharged from the differences between the thermodynamic the gasifier less often, without interfering with the equilibrium and reaction kinetics and the three- continuous production of high-quality syngas. A reaction equilibrium that is essential in the small amount of ash in olive pomace is beneficial gasification as specified in the introduction to forestall probable clinker agglomeration in the section. It may also be changed according to the gasifier owing to higher operation temperatures type and operation of the gasifier reactor. in the reactor. Clinker agglomeration could cause The energy and mass balance calculations on bridging and channelling problems on the grate the process need an assessment of the inputs to just below the oxidation region. This can block and outputs from the reactor. To verify the mass

130 © 2020 Johnson Matthey https://doi.org/10.1595/205651320X15746781209529 Johnson Matthey Technol. Rev., 2020, 64, (2) and energy balance outputs, the results obtained heat. After the gasification of the gas produced from the olive pomace analyses, the fixed bed in the boiler and the heated thermal oil in the updraft gasifier capacity, the thermal oil boiler and ORC turbine is transformed into electricity and the ORC turbine efficiency were determined and waste heat energy, thermal power can operate calculated. There are difficulties in getting 100% the blackwater evaporation system. closure and obtaining these data. Nevertheless, Blackwater produced in the production of olive oil the average energy balance closeness for three is an environmental problem. Work continues on experimental runs was detected to be 96%, the vaporisation of this blackwater using excess indicating a reasonable figure for the initial heat with an evaporation system. In this study, demonstration of olive waste gasification. The the remaining solid substance from blackwater schematic diagram shown in Figure 4 is the vaporisation will be used in the gasification system energy and mass balance of the pelleted olive as a feedstock by mixing with olive pomace pomace as biomass feedstock in the gasification biomass. In future studies, the produced steam process. According to the energy and mass balance will be converted to a superheated gasification diagram, 500 kg h–1 of olive pomace is used. It agent in the reactor. Thus, the produced thermal has 17.65 MJ kg–1 chemical energy according power can also be evaluated efficiently within the to the fuel characteristic analysis. Pelleted olive facility. pomace has 25% moisture. The net energy value –1 of the 500 kg h fuel fed to the reactor is 7881 MJ 3.4 Gasifier Temperature Profile (2189 kWh). As stated in the literature for the updraft gasifier, the air:fuel ratio is determined to Figure 9 shows the temperature profiles of the be approximately 1 kg of fuel to 1.6 kg air flow oxidation, reduction, pyrolysis and drying zones rate (1:1.6) (10). For the autothermal gasifier, 1% in the updraft gasifier observed during 8 h of heat loss can be estimated (79 MJ). Depending on continuous operation in the test runs. In general, the feedstock, gasifier output char is about 17% there is only a small dependence on the feed rate. of the fuel input. Thus, in this process, 89 kg h–1 However, the variation is more pronounced at the of biochar is produced, the equivalent heat is gasifier outlet, which could be due to variations in 1043 MJ (290 kWh). the aeration rates, especially at higher throughputs. In the updraft gasifier, the ratio by mass of the However, as the air to fuel ratio increases, the feedstock and produced gas after gasification zone temperatures increase. Because of very is approximately 1:2.5 and the volumetric flow high temperatures around the moving grate zone rate of the product gas is 1251 Nm3 h–1. When (>1000°C), some forms of clinker were observed the density of the syngas is about 1.18 kg over the grate during the clean-up. In the literature, Nm–3, the production of hot gas is 1475 kg h–1. studies seem to have reached a consensus about Assuming that the temperature of produced gas the temperature (>1100°C) in the oxidation zone is 350°C, the volumetric flow rate of syngas at of an updraft gasifier (8). Reasonable residence this temperature is calculated as 2660 m3 h–1. time is necessary to destroy the refractory If the heat losses are calculated, the energy of unsubstituted aromatics (tars) in the product gas, produced gas at 350°C is 1878 kWh (6760 MJ). without catalytic tar cracking. The hot product gas is transferred to the syngas Therefore, the optimum operating temperature burner when gas is combusted in the thermal oil should be adjusted for each different fuel heater; the boiler thermal energy is calculated used in the reactor by considering tar cracking as 1596 kWh with 10% heat loss. This thermal versus clinker formation. Obviously, ash fusion energy produced is transmitted to the ORC temperatures of the fuels are decisive in selecting turbine with thermal oil circulation; heat loss is operating temperatures of an updraft gasifier. not calculated because it has sufficient insulation. Clinker formation has a more significant impact Since the ORC turbine efficiency is 15%, the than tar formation, which can be easily treated by turbine generates 240 kW gross, 221 kW net improving the clean-up of the system. Although electrical power as 8% parasitic load is internally tar formation above 900°C is small, the benefit of consumed. The ORC turbine also produces reducing clinker is substantial for the operation of 1356 kWh of thermal energy in the form of waste the gasifier.

1400 Drying Pyrolysis Reduction Oxidation

1200

1000

800

600

Temperature, ºC Temperature, 400

200

00:00:0000:20:5400:42:2601:03:2501:23:4801:49:2302:11:2002:32:5202:52:3403:12:2003:32:1503:52:4104:20:2404:52:2005:02:4205:21:4205:48:3506:08:1206:32:5106:54:0207:44:02 Time, h:min:s

Fig. 9. Temperature profile of the gasifier zones 3.5 Thermal Oil Boiler and Organic 2 –1 586.375 – 2.2809 ( Kinematic viscosity (mm s ) =( e T(ºC) + 62.5 Rankine Cycle Turbine Operation

Results (xiv) The operation parameters of the process are –9094.51 ( Vapour pressure (kPa) = e ( T(ºC) + 340 + 17.6371 1600 kWh energy generated in the thermal boiler as a result of burning syngas is transferred to (xv) 60 m3 h–1 Therminol 66 (Eastman, USA) fluids. Therminol 66 is a high performance, highly According to these formulas, at 280°C physical stable synthetic heat transfer fluid. The chemical properties of the fluid are 824.6 kg –3m (density), composition of this fluid was carefully selected 0.097 W m–1 K–1 (thermal conductivity), to minimise the formation of low boilers and 2.492 kJ kg–1 K–1 (heat capacity), 0.56 mm2 s–1 eliminate the risk of insoluble high boiler formation (kinematic viscosity) and 19.46 kPa (absolute and fouling, provided proper attention is given to vapour pressure). system design and operation is within the maximum bulk (345°C) and film (375°C) temperatures. To 4. Conclusion calculate the physical properties of the fluid such as density, heat capacity, thermal conductivity, In this study, a pilot-scale gasification system based kinematic viscosity and vapour pressure, formulas on CHP plant, its operation and energy production are given below (Equations (xi)–(xv)) (26): efficiency were investigated. Dried and pelleted olive pomace biomass from olive oil production Density (kg m–3) = –0.614254 × T (°C) – facilities was used as fuel in the autothermal fixed 0.000321 × T2 (°C) + 1020.62 (xi) bed updraft gasifier. It was seen that the feedstock characteristics and the design of the updraft Heat capacity (kJ kg–1 K–1) = 0.003313 × T (°C) gasifier play a significant role in process operation. + 0.0000008970785 × T2 (°C) + Appropriate design of the gasifier diminishes 1.496005 (xii) drawbacks, thus improving the performance of the system. Olive pomace, which is used as feedstock, Thermal conductivity (W m–1 K–1) = –0.000033 needs to be dried and pelleted before being × T (°C) – 0.00000015 × T2 (°C) + gasified to prevent blockage and pressure drops in 0.118294 (xiii) the gasifier.

The thermal oil boiler has a thermal capacity of 9. V. Dornburg and A. P. C. Faaji, Biomass Bioenerg., 1.60 MWh. Most of the generated thermal energy 2001, 21, (2), 91 is transferred via thermal oil to the ORC turbine to 10. J. R. Bunt and F. B. Waanders, Fuel, 2008, 87, generate an electric output of 240 kW transferred (10–11), 1814 to the rural grid at the maximum load. The excess 11. J. F. Pérez, A. Melgar and P. N. Benjumea, Fuel, useable waste heat energy of 1.35 MWh from 2012, 96, 487 the gasification system is used in the blackwater 12. M. Dogru, M. R. Beltran, S. Mitra, A. Erdem evaporation unit. Therefore, it was proved that this and E. S. Park, ‘Updraft Gasification of Waste innovative, unique system is an important source and Produced Syngas Treatment’, in “Waste of renewable energy for rural areas. Also, this Management and Resource Efficiency”, ed. type of process offers the possibility of converting S. K. Ghosh, Springer Nature Singapore Pte Ltd, feedstock biomass into power and heat with high Singapore, 2019, pp. 741–752 efficiency. C conversion during gasification was 13. P. McKendry, Biores. Technol., 2002, 83, (1), 55 around 85%, which is thought to be reasonable. 14. S. Priyadarsan, K. Annamalai, J. M. Sweeten, S. Mukhtar and M. T. Holtzapple, Trans. ASAE, Acknowledgement 2004, 47, (5), 1689 15. N. C. Taupe, D. Lynch, R. Wnetrzak, M. Kwapinska, This study was financially supported by Marzey, W. Kwapinski and J. J. Leahy, Waste Manage., a subsidiary of Marmarabirlik, and the Scientific 2016, 50, 324 and Technological Research Council of Turkey – 16. J. H. Kihedu, R. Yoshiie and I. Naruse, Fuel TÜBITAK (project number: 3130469). Process. Technol., 2016, 141, (1), 93 17. M. Puig-Arnavat, J. C. Bruno and A. Coronas, Appl. References Energy, 2014, 114, 845 18. B. de Mena, D. Vera, F. Jurado and M. Ortega, Fuel 1. R. De Vos, P. van Breevoort, N. Höhne, T. Winkel Process. Technol., 2017, 156, 394 and C. Sachweh, “Assessing the EU 2030 Climate 19. F. A. Boyaghchi, M. Chavoshi and V. Sabeti, and Energy Targets – A Briefing Paper”, Project No. Energy, 2018, 145, 38 DESNL14683, ECOFYS, Utrecht, The Netherlands, 20. M. Uris, J. I. Linares and E. Arenas, Renew. 17th March, 2014, 23 pp Energy, 2014, 66, 707 2. R. P. Overend, ‘Biomass Energy Heat Provision 21. I. Oberbernger, Biomass Bioenergy, 1998, 14, for Cooking and Heating in Developing Countries’, (1), 33 in “Energy from Organic Materials (Biomass)”, ed. M. Kaltschmitt, Vol. 2, Springer Science and 22. D. Vera, F. Jurado, J. Carpio and S. Kamel, Energy, Business Media LLC, New York, USA, pp. 513–531 2018, 144, 41 3. International Olive Council, Madrid, Spain, 2018 23. F. Cotana, A. Messineo, A. Petrozzi, V. Coccia, G. Cavalaglio and A. Aquino, Sustainability, 2014, 4. D. Vera, F. Jurado and J. Carpio, Fuel Process. 6, (9), 5714 Technol., 2011, 92, (10), 1970 24. T. B. Reed and A. Das, “Handbook on Biomass 5. A. García-Maraver, L. C. Terron, A. Ramos-Ridao Downdraft Gasifier Engine Systems”, Report and M. Zamorano, Biosys. Eng., 2014, 118, 167 No. SERI/SP-271-3022 and DE88001135, US 6. A. V. Bridgwater, Fuel, 1995, 74, (5), 631 Department of Energy, Washington, DC, USA, 7. A. V. Bridgwater, J. M. Double and D. M. Earp, March, 1988, 148 pp “Technical and Market Assessment of Biomass 25. M. Dogru, A. Midilli and C. R. Howarth, Fuel Gasification in the UK”, Report No. ETSU–B–1167, Process. Technol., 2002, 75, (1), 55 UKAEA Atomic Energy Research Establishment, 26. ‘Therminol 66 Heat Transfer Fluid – Product Harwell, UK, 1st January, 1986, 140 pp Description’, Eastman Chemical Company, 8. M. Dogru, ‘Fixed-Bed Gasification of Biomass’, PhD Kingsport, USA: https://www.therminol.com/ Thesis, University of Newcastle upon Tyne, United product/71093438?pn=Therminol-66-Heat- Kingdom, 2000, 342 pp Transfer-Fluid (Accessed on 7th January, 2020)

The Authors

Murat Dogru received his Doctorate degree (1996–2000) from the Chemical and Process Engineering department at Newcastle University, UK. He researched and lectured for eight years as a member of academic staff at Newcastle University from 2000 until 2008. He then worked as an associate consultant in the Adapt Low Carbon Group at the University of East Anglia, UK, between 2008 and 2012 and in 2013 he obtained his associate professorship in Environmental Engineering at Gebze Technical University in Istanbul, Turkey. Dogru is also Executive R&D Director at Beltran Technologies, Inc, New York, USA for the industrial development of gasification systems using a variety of biomass and wastes in order to develop renewable energy generation systems for local governments and industries. He has published more than 100 international publications in journals and international symposiums which all have direct relevance to renewable energy generation technologies, in particular biomass and waste gasification.

Ahmet Erdem received his Master’s degree (2013–2015) in Environmental Engineering from Gebze Technical University, Turkey. He has been a member of academic staff at Gebze Technical University since 2015. He worked as a commissioning engineer at a waste gasification research and development project in South Korea in 2014. The project was sponsored by the Korean Environmental Ministry and managed by Beltran Korea as subcontractor. He is in the last period of his doctoral study and his PhD thesis is on activated carbon production from biochar produced from gasification.

www.technology.matthey.com

In the Lab Targeting Industry-Compatible Synthesis of Two-Dimensional Materials Johnson Matthey Technology Review features new laboratory research

Niall McEvoy’s research is primarily focused on the synthesis and characterisation of nanomaterials, The Researcher particularly two-dimensional (2D) materials, and their subsequent assessment for use in a wide array of applications. A key aspect of this work involves developing and refining industry-relevant synthesis protocols for emerging 2D materials. One potentially industry-compatible way to produce these materials is using vapour-phase methodologies, such as chemical vapour deposition (CVD). Identifying 2D materials that can be synthesised at relatively low temperatures is vital if these materials are to be considered for real-world applications. Vapour-phase-grown 2D materials are of interest for diverse fields, in areas such as • Name: Dr Niall McEvoy electronics, optoelectronics, telecommunications, • Position: Science Foundation Ireland Funded sensing of analytes, detection and measurement Principal Investigator of strain and catalysis. The innovative potential • Department: AMBER and School of Chemistry of these materials has led to considerable • University: Trinity College Dublin, The interest and investment from private enterprise, University of Dublin particularly in the information and communication • Address: College Green, Dublin 2 technology sector. • Postcode: D02 PN40 McEvoy leads the Architecture and Synthesis of • Country: Ireland Integrated Nanostructures (ASIN) group at Trinity • Email: [email protected] College Dublin, Ireland. He is a funded investigator in the Advanced Materials and BioEngineering Research Centre (AMBER), also at Trinity College Dublin. He has co-authored over 100 peer- reviewed articles in the area of nanomaterials. His About the Research group benefits from an extensive research network involving active collaborations with research groups Since the isolation of graphene in 2004, research in Ireland, the UK, China, Germany, Italy, Austria, has unveiled the ever more impressive and diverse Switzerland and Denmark. He was the recipient of a properties of 2D materials, prompting them Science Foundation Ireland Technology Innovation to be linked with use in an increasing array of Development Award in 2015 and a Starting applications. While the properties of 2D materials Investigator Research Grant in 2016. are certainly revolutionary, and the associated

(a) (b) 0.05 Gas inlet Ar:H2, 9:1 0.04 0.03 0

R/R 0.02 D – 0.01 To pump 0 Zone 1 Zone 2 PtSe2 Pt substrates @ 400°C Se @ 220°C –0.01 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 e, %

Fig. 1. (a) Schematic for synthesis of PtSe2 thin films; (b) piezoresistive response of PtSe2/polyimide under flexure. Inset: photograph of PtSe2 grown directly on polyimide (5)

physics and chemistry indeed very exciting, the McEvoy and coworkers developed a simple, but hype surrounding the field should to some extent robust, vapour-phase process for the growth of thin be tempered by practical considerations of how films of PtSe2 (Figure 1(a)). The relatively low best they should be fabricated and subsequently growth temperatures involved (~400ºC) mean that processed. Many of the experimental reports on the material could potentially be integrated with their properties have used materials prepared by back-end-of-line processing in the semiconductor mechanical exfoliation, a laborious, serendipitous industry (3). and inherently unscalable production technique. The PtSe2 films grown in this manner have Efforts to improve the scalability of 2D materials shown very promising results in laboratory-based production broadly fall into two approaches: prototype devices. Like other 2D materials, PtSe2 liquid-phase exfoliation, a top-down method; possesses a near ideal surface-area to volume and vapour-phase growth, a bottom-up method. ratio which is in part responsible for its impressive Enormous progress has been made in these performance in gas-sensing devices (4). The fields in recent years. The scalable production relatively low growth temperature means that of 2D material dispersions by shear exfoliation PtSe2 can be grown directly on flexible polymer was reported by Professor Coleman’s group at substrates (5). These polymer/PtSe2 films show Trinity College Dublin (1). On the vapour-phase a piezoresistive effect (Figure 1(b)), i.e. when growth front, recent reports from researchers in they are bent the resistivity changes, suggesting

Interuniversity Microelectronics Centre (IMEC), potential use as gauges to monitor strain. PtSe2 Belgium have demonstrated wafer-scale growth films grown in the ASIN laboratory have also of the 2D material tungsten disulfide (WS2) in a shown promise for use in photodetectors (6), semiconductor fabrication setting (2). transistors (7) and pressure sensors (8). Much of the research undertaken by the ASIN Other ongoing projects in the ASIN group are group is centred on developing sensible and scalable focused on CVD synthesis of 2D material vapour-phase growth approaches for the synthesis heterostructures, synthesis and electrochemical of 2D materials. Particular focus has been placed on applications of transition metal ditellurides, developing growth recipes for less-commonly studied tailored functionalisation of 2D materials, resistive 2D materials, for instance those whose bulk form switching in 2D materials and scanning-probe is not naturally abundant. A recent example of the studies of defects in 2D materials. group’s research efforts is the vapour-phase growth of platinum diselenide (PtSe2). PtSe2 can be found Acknowledgements in nature in the form of the mineral sudovikovite but this is quite rare. In its 2D form PtSe2 benefits Niall McEvoy thanks all members of the ASIN group, from a high charge-carrier mobility, good stability the wonderful staff at AMBER and the School of in ambient conditions, a thickness-dependent band Chemistry, Trinity College Dublin, as well as his structure and promising electrocatalytic behaviour. external collaborators.

References

1. K. R. Paton, E. Varrla, C. Backes, R. J. Smith, 4. C. Yim, K. Lee, N. McEvoy, M. O’Brien, S. Riazimehr, U. Khan, A. O’Neill, C. Boland, M. Lotya, N. C. Berner, C. P. Cullen, J. Kotakoski, J. C. Meyer, O. M. Istrate, P. King, T. Higgins, S. Barwich, P. May, M. C. Lemme and G. S. Duesberg, ACS Nano, P. Puczkarski, I. Ahmed, M. Moebius, H. Pettersson, 2016, 10, (10), 9550 E. Long, J. Coelho, S. E. O’Brien, E. K. McGuire, 5. C. S. Boland, C. Ó Coileáin, S. Wagner, B. Mendoza Sanchez, G. S. Duesberg, N. McEvoy, J. B. McManus, C. P. Cullen, M. C. Lemme, T. J. Pennycook, C. Downing, A. Crossley, G. S. Duesberg and N. McEvoy, 2D Mater., 2019, V. Nicolosi and J. N. Coleman, Nature Mater., 6, (4), 045029 2014, 13, (6), 624 6. C. Yim, N. McEvoy, S. Riazimehr, D. S. Schneider, 2. T. Schram, Q. Smets, B. Groven, M. Heyne, F. Gity, S. Monaghan, P. K. Hurley, M. C. Lemme E. Kunnen, A. Thiam, K. Devriendt, A. Delabie, and G. S. Duesberg, Nano Lett., 2018, 18, (3), D. Lin, M. Lux, D. Chiappe, I. Asselberghs, S. Brus, 1794 C. Huyghebaert, S. Sayan, A. Juncker, M. Caymax 7. L. Ansari, S. Monaghan, N. McEvoy, C. Ó Coileáin, and I. Radu, ‘WS2 Transistors on 300 mm Wafers C. P. Cullen, J. Lin, R. Siris, T. Stimpel-Lindner, with BEOL Compatibility’, 47th European Solid- K. F. Burke, G. Mirabelli, R. Duffy, E. Caruso, State Device Research Conference, Leuven, R. E. Nagle, G. S. Duesberg, P. K. Hurley and Belgium, 11th–14th September, 2017, IEEE, F. Gity, npj 2D Mater. Appl., 2019, 3, 33 Piscataway, USA, pp. 212–215 8. S. Wagner, C. Yim, N. McEvoy, S. Kataria, 3. C. Yim, V. Passi, M. C. Lemme, G. S. Duesberg, V. Yokaribas, A. Kuc, S. Pindl, C.-P. Fritzen, C. Ó Coileáin, E. Pallecchi, D. Fadil and N. McEvoy, T. Heine, G. S. Duesberg and M. C. Lemme, Nano njp 2D Mater. Appl., 2018, 2, 5 Lett., 2018, 18, (6), 3738

www.technology.matthey.com

Plasma Catalysis: A Review of the Interdisciplinary Challenges Faced Realising the potential of plasma catalysis on a commercial scale

Peter Hinde*, Vladimir Demidyuk, that are charged or excited and free electrons. Alkis Gkelios, Carl Tipton Depending on the energy of the plasma it can be Johnson Matthey, PO Box 1, Chilton Office, fully ionised and have a bulk temperature of tens of Belasis Avenue, Billingham, TS23 1LB, UK thousands of kelvin (or in the case of nuclear fusion millions of kelvin) or it can be in a low temperature *Email: [email protected] non-equilibrium state where only a small portion of the gas phase is energised. At the lower end of the energy spectrum a non-thermal or cold plasma The work presented here introduces the topic of will have a bulk temperature a few tens of degrees plasma catalysis through selected work in scientific above ambient and yet still have some exceedingly literature and commercial applications, as well as high energy species present. Even though the identifying some of the key challenges faced when majority of the species in the plasma are not in attempting to utilise non-thermal atmospheric equilibrium, there can exist some partial equilibria plasma catalysis across multidisciplinary boundaries among species with similar kinetic temperatures including those of physics, chemistry and electrical existing at localised sites (1). engineering. Plasma can be generated by different This energy can be used to initiate chemical methods at many energy levels and can initiate reactions in the gas phase and on the surface of chemical reactions; the main challenges are to solids. The term plasma catalysis can be used to selectively initiate desirable reactions either within a describe both the use of plasma to initiate a reaction process stream or at the surface of a material. The directly and the use of plasma in combination with material, which may have intrinsic catalytic properties, a catalytic material. The catalyst can be positioned the nature of the process gas and the geometry of the after a plasma zone as post plasma catalysis (PPC) reactor will influence the products formed. Previous or within the plasma zone as in-plasma catalysis work has shown that the mechanism for plasma- (IPC). The PPC configuration allows the longer- initiated reactions can be different to that occurring lived excited species and the products from the from more traditional thermally stimulated reactions, plasma reaction to interact with the catalyst and which opens up possibilities of using different catalytic the IPC configuration allows a greater opportunity materials to optimise the reaction rate and product for the catalytic material to influence the nature speciation. In addition, the influence of a plasma at of the chemistry by directly interacting with the the surface of a material and the effects that can be plasma excitation as well as the products from the introduced will be discussed. plasma-initiated reactions. With the introduction of a plasma to a catalytic Introduction reactor the standard models of the chemistry and the traditional understanding of the mechanisms One area where physics and chemistry come by which reactions take place start to become less together across the disciplines is in the field of relevant due to the non-equilibrium concentration plasma catalysis. A plasma can be described as a of excited species such as free atoms, electrons ‘soup’ of species including molecules and atoms and radicals (2).

Plasma Generation Current, 100 mA div Within the universe around us there are many examples of plasma (3) including the sun, stars, –1 auroras, lightning, welding arcs and fluorescent lighting tubes. A plasma can be generated in different ways, but all require energy to be applied, either as heat in the case of a thermal ionisation or –1 through the generation of an intense electric field 1 kV div Voltage, via the use of electrodes, radio frequency (rf) or microwave (MW).

Electrical excitation is one of the most feasible ways Time, 50 ms div–1 for producing well controlled plasma discharges at industrial scale. This type of excitation is controlled Fig. 1. Typical signals for sinusoidal voltage, by three main parameters: (a) the applied current and time voltage amplitude, (b) the applied frequency and (c) the waveform shape. The combination of the aforementioned parameters defines different operating regimes (4). The main types are: Current, 6 A div

• Direct current (DC) – earliest power supplies –1 where constant high voltage in tens of kilovolts are applied between the electrodes for creating a sustainable discharge between fixed anode and cathode electrodes. A resistor may be used –1 to limit the current and high voltage cables are Voltage, 3 kV div Voltage, used for power delivery to the electrodes • Sinusoidal – high voltage (0–40 kVp) continuous power sources in the frequency range of 50 Hz to 150 kHz are mostly used in dielectric barrier Time, 10 ms div–1 discharges (DBD). Those sources are easier to Fig. 2. Typical pulsed sources signal for current, manufacture and operate in a wide frequency voltage and time and power range • rf – continuous sinusoidal sources with hundreds of volts operating at 13.56 MHz. For as the introduction of an applied magnetic field, a optimal operation a matching network is used vortex flow or by reducing the pressure within the to restrict the reflected power system. • MW – continuous sinusoidal source operating In scientific and industrial applications, average normally at 2.45 GHz. Rectangular waveguides power delivered in the plasma discharge is or coaxial cables are used for delivering the critical. Accurately measuring and monitoring the energy to the load power consumption in plasma discharges is not a • Pulsed – fast pulses with well-defined pulse rise straightforward task. Average power consumption time, duration and voltage created by switching is calculated through the measurement of voltage a DC high voltage power supply. The fast and current waveforms. Although the voltage transition times and control capabilities allow signals are monitored with high accuracy through operation at higher power per pulse. capacitive voltage dividers, current waveforms as From the regimes described above, sinusoidal shown in Figure 1 and Figure 2 present significant and pulsed power sources present the advantage measuring difficulties. Current spikes with very of producing non-thermal plasma without the short (nanosecond) duration and high amplitude need for noble gasses and low pressure. Moreover, and frequency are imposed over the sinusoidal low the use of dielectric barrier material protects the frequency and amplitude signal. Monitoring those electrodes from erosion in chemical processes high dynamic range waveforms can be erroneous and limits the power needed for initiating and and special care needs to be taken in the method of maintaining plasma discharge. acquiring those signals. In the literature (5), three It is also possible to influence the nature and district ways are described for measuring directly stability of the plasma through external forces such the current or the charge in plasma.

• Shunt resistor method – a known resistor is Gas inlet inserted between the reactor and the ground electrode. The voltage drop across the resistor is logged and is converted to current using Outer electrode Ohms law • Monitor capacitor method – an integrating capacitor with known value is inserted between Generated plasma the plasma reactor and the ground electrode. Plotting its waveform versus the voltage Gas generates a Lissajous curve in which the outlet average power can be calculated Inner electrode • Rogowski coil – an inductive coil is used to measure the current through the ground Fig. 3. Schematic of a typical continuous flow electrode of the reactor. cylindrical plasma tube Although the above measurements have comparable accuracy in scaled plasma reactors, in physics disciplines. A schematic of a simple plasma industrial environments it is preferable that in situ tube is shown in Figure 3. Gas is passed through non-invasive techniques are used. Rogowski coils the cell, where a high voltage is applied across a have an intrinsically safe way of operation given central cylindrical electrode and a cylindrical outer that they are galvanically isolated from the main electrode. The generation of the plasma in the circuit without compromising the accuracy. cell leads to a shift in chemical make-up across Accurate power measurement and delivery the cell, hence the potential to use such cells in in combination with optimised mechanical and chemical processing. Even prior to the excitation electrical design can lead to improvements in of any plasma, the gas flow through the cell has energy efficiency. This is a critical parameter the potential to become a complex fluid mechanics that must be considered, especially when scaling problem. Once a plasma is generated the equations up plasma systems. It is unavoidable that some of magnetohydrodynamics become applicable. This part of the initial energy is lost in the electrical apparently innocuous point massively increases the transformation to high voltage and in the plasma experimental phase space that needs to be controlled reaction as heat. By carefully selecting the electric and understood. A general introduction into the field and reactor characteristics those losses can topic of magnetohydrodynamics can be gained from be minimised. Modelling can significantly enhance reading “An Introduction to Magnetohydrodynamics” the understanding of the plasma transitions in by P. A. Davidson (9). The bulk physical continuous short time and space frameworks (6). The next control parameters in a traditional reactor vessel improvement step involves the synergistic effects can be broadly listed as temperatures, pressures of plasma catalyst interaction. By introducing a and flow rates. In a plasma cell we have in addition catalyst in the plasma region or next to it, different voltages, currents and frequencies which must works have shown considerable improvement in be controlled and monitored. This opens huge conversion and in total efficiency (7). All these opportunities in chemical processing as significant optimisations allow energy consumption of the parts of this phase space remain largely unmapped. plasma system to be decreased and make plasma One of the oldest applications of plasma is in technologies feasible from the view point of economic fluorescent lighting. This fact means that there has and life cycle assessment. For example, a recent been a significant amount of study into the behaviour paper by Rooij et al. (8) shows that the combination of plasma cells where there is no flow. The graph of plasma with renewable energy sources is an shown in Figure 4 is a schematic illustration of the economical method even for such an expensive characteristic (current vs. voltage) curve of a typical process as carbon dioxide (CO2) reduction. The gas discharge in neon gas at a pressure of 1 torr, technoeconomic solution will of course be different between two planar electrodes separated by 50 cm. for each market application based upon comparison This figure has been recreated from the information to current scaled state of the art techniques. presented by Gallo (10). There are broadly speaking three classes of behaviour. A dark discharge region, a Mechanisms of Excitation glow region and an arc region. In the dark discharge region, a voltage lower than the breakdown voltage Even the simplest of plasma generation of the gas is applied to the tube. External radiation arrangements takes in aspects of a wide range of such as gamma photons and beta particles then

deodorisation and sanitisation. Commercially this 1200 is done using ultraviolet (UV) light at a wavelength of 185 nm, electrolytically or via a DBD plasma 1000 depicted in Figure 5. Siemens, Germany, were the

800 Dark discharge first to use plasma for an industrial application in 1857 to produce O3 (15). This plasma is generated Arc 600 as a large number of statistically distributed micro- discharges between electrodes where the potential Voltage, V Voltage, 400 is insufficiently high to create an arc. Diatomic oxygen is broken down through interaction with Glow 200 the electrons produced within these micro- discharges that have sufficient energy to split the

0 O2 double bond. These newly separated O atoms 10–20 10–15 10–10 10–5 100 105 then combine with other diatomic O2 molecules to Current, A produce O3. The micro-discharges are individually only present for a few nanoseconds each and Fig. 4. Transitions occurring in a fixed glow tube the number of micro-discharges generated is dependent upon the gap between the electrodes, trigger a Townsend cascade within the tube. This the humidity and pressure of the air, the properties behaviour is utilised in a number of simple nucleonic of the dielectric barrier and the characteristics of detection devices, notably Geiger counters (11). In the electrical supply (16). O3 production can be as this region the current is limited to remain low and high as 100 kg h–1 from a horizontal honeycomb the discharge is reliant on external excitation. In the reactor and is closely related to the specific voltage glow region, the current is no longer limited, and the and frequency applied, with typical voltages being accelerated electrons excite further electrons and 7–30 kV and frequencies between 50–1000s Hz. the voltage then drops. The gas emits light causing a glow, an effect utilised in fluorescent lighting. If Acetylene Production the current is high enough the neutral gas in the tube becomes heated and arcs start to form which Thermal plasma has been used to produce acetylene is the final region. (C2H2) since the 1940s in the Huels process. The graph shown in Figure 4 illustrates that The original Huels plant used the low-boiling the behaviour of plasmas is complex and many components of the motor fuel industry as raw transitions occur in a cell where all of the standard material; however, a wide range of hydrocarbons chemical processing variables are fixed i.e. including natural gas were shown to be suitable as pressure is constant, temperature is fixed and the process feed stocks. The equilibrium formation of flow rate is zero. In this example there was no time C2H2 is characterised by the requirement of very dependence to the applied voltage and current. high temperatures, around 3000°C. Therefore, the reaction gas should be rapidly cooled by liquid Practical Industrial Examples of water spray injection downstream of the plasma reaction zone to avoid formation of solid carbon Plasma Application that is a thermodynamically preferred product There are an increasing number of references in the between about 1000°C and 2500°C. This fast scientific literature giving examples of laboratory quenching prevents decomposition of the C2H2 scale plasma catalysis (12, 13) or industrial formed in the plasma. Because the formation of processes being investigated at semi-industrial C2H2 from methane (CH4) is strongly endothermic, level (14). However, there are currently no known relatively large amounts of energy are required. In large-scale industrial applications that combine industry the best energy performance was shown plasma with a catalytic material. The following two by the DuPont process (a modification of the Huels examples show the scaled production of chemicals reaction) with the specific energy consumption –1 using plasma excitation. 8.8 kWh kg of C2H2 (17).

More recently different methods of C2H2 formation Ozone Production in plasma were compared (18) and it was found that pulsed spark discharges gave the highest

Plasma is used to generate ozone (O3) industrially C2H2 yield (54%) with 69% of CH4 conversion for applications including cleaning, disinfection, in a pure CH4 system. It was suggested that the

Dielectric layers Fig. 5. O3 is formed by High voltage passing oxygen through an electrical discharge Oxygen Ozone that is diffused over an area using a dielectric to create a corona discharge

Ground

main disadvantage of the plasma method of C2H2 These plasma and catalyst interactions can be production was the fact that excited species reacted thought of as occurring between one or more with formed C2H2 and decomposed it to undesirable excited states and can be representative of byproducts. As a result, the energy costs for CH4 both the surface and the gas being excited. An conversion and C2H2 formation increased with electrically induced surface potential of a material

CH4 conversion percentage and were found to is equally defined as a catalyst as a material that be best in pulsed spark discharges (highest CH4 satisfies more traditional thermal catalysis ideals of conversions 18–69%). A Korean group reported adsorption and reaction, such as the use of glass –1 the energy cost in its systems as 9 kWh kg C2H2. beads for CH4 conversion (22). They calculated their theoretical minimum energy In addition to the above interactions, further –1 requirement as 4.03 kWh kg C2H2 based on thought should be given to the interdependency heat of reaction (19). In this publication, it of the plasma formed on the properties of the was also found that hydrogen in the plasma catalytic material present, and vice versa where counterintuitively increased the selectivity to C2H2 the plasma will have an impact on the properties of in the process, from a respectable 70% to well the catalyst surface, including even impacting upon above 90%. The paper claims that it is possible the physical morphology of the material. These to decrease the specific energy consumption to synergistic interactions have been proven to offer a –1 6 kWh kg C2H2 in the low-temperature arc with different mechanism for chemical reaction when the the application of argon as the recycling reactant. excitation comes from electrical plasma rather than It is close to the theoretical limit of 4.03 kWh kg–1 thermal means (23). A study of plasma activated of C2H2, which is 50% of the energy required for catalytic (palladium/aluminium oxide (Pd/ Al2O3)) the DuPont process. These numbers show that CH4 oxidation, conducted in a synchrotron there is space to improve energy efficiency of the beamline, concluded that the Pd nanoparticles are industrial production of C2H2 by plasma methods. heated within the plasma but the temperature of the nanoparticles remains lower than that required Plasma Catalysis to initiate the thermal CH4 oxidation reaction. Thus, an alternative reaction mechanism with a lower It is possible to introduce a catalyst after a plasma activation barrier must be taking place (24). zone and transform some of the still excited There have been many studies looking at the species present in the gas phase into products at application of plasma catalysis for chemical the surface of the catalyst, known as PPC (20). synthesis. In particular, small molecules that are Plasma can also generate activated species at difficult to activate using more traditional thermal a surface as well as in the gas phase. There are methods, such as CO2 and CH4 lend themselves many terms used to describe the use of a catalyst towards activation using plasma techniques (13). and a plasma together, including plasma catalysis, A special issue of the journal Catalysts titled plasma enhanced catalysis, plasma assisted ‘Plasma Catalysis’ was recently published including catalysis, plasma driven catalysis (21); this IPC papers covering the application of plasma catalysis uses the excitation of both the catalyst surface for CO2 splitting (25), ammonia (NH3) synthesis and the gas phase reactants to effect chemistry (26) and CH4 reforming (27). through reactions between: A key challenge for plasma catalysis is to design • Excited gas phase species interacting with a a reactor that is suitable to house a plasma and non-excited catalyst surface a catalyst that has low backpressure, but retain • Excited surface species interacting with non- good catalyst and gas-plasma interaction similar excited gas phase reactants to the ceramic monolith widely used in automotive • Excited gas phase interactions with excited emission treatment. Uytdenhouwen et al. (1) surface species. identify power, pressure and gap size in a reactor

142 © 2020 Johnson Matthey https://doi.org/10.1595/205651320X15759961130711 Johnson Matthey Technol. Rev., 2020, 64, (2) as key process parameters for utilisation in design covalent bonds due to the bombardment of the of plasma reactor and then go on to discuss their surface with high energy particles effect on CO2 disassociation in a DBD microreactor. • The partially destroyed polymers can react Mizuno (28) describes multiple approaches with other similarly separated units and form to tackle this problem in his review: including crosslinks, thus extending polymeric chains and (a) micro-discharge plasma with the process gas mechanically stabilising the surface flowing through a catalyst coated metal plate with • The surface can become functionalised by very narrow (micron) gaps to improve the catalyst- including a functional molecule in the plasma gas interaction, (b) a metal mesh DC powered gas. electrode in front of a ceramic monolith and a Examples of surface functionalisation include packed bed alternating current (AC) electrode at the the incorporation of hydroxyl groups (OH) from rear of the ceramic monolith in order to introduce humidity present in the plasma or N fixation surface streamers along the monolith channels using an NH3 plasma. The introduction of polar and (c) a sliding three-electrode DBD system groups such as hydroxyls allows for a significant combining a negative AC electrode, a DC electrode improvement in the wettability of polymers and and a ground to create more homogeneous and therefore has significant advantages for the widely dispersed surface plasma (29). printing and adhesives markets. This is a complex Plasma catalysis has been studied for automotive area and large bodies of work have been produced emission control in both the PPC and IPC documenting the effects of different types of plasma configurations. Some of the first studies were on different polymeric materials and summarised conducted using a packed bed for nitrogen oxides elsewhere (34, 36). (NOx) control (30, 31) and proposed a two- Increasingly plasma has been used for vapour stage process whereby the nitric oxide (NO) was deposition (VD) processes applying a uniform thin oxidised to nitrogen dioxide (NO2) which was film coating to a material (37). These films are then subsequently selectively reduced over the usually within the nanometre thickness range and catalyst by the hydrocarbons present. This system used for modification of optical, chemical, electronic, has also been proposed as a pre-particulate filter physical and decorative properties of the materials. plasma reactor to attain additional particulate The methods of plasma application for physical VD matter oxidation benefits from the increased include sputtering, ion plating and cathodic arc

NO2 generated by the plasma (32). A successful deposition. Sputter deposition involves deposition demonstration of a combined plasma and catalytic onto a substrate of a molecule previously vaporised system has also taken place for CH4 removal from from a target. The target is vaporised through the dual fuel engines at low temperatures (33). mechanism of momentum transfer from gaseous ions accelerated from a plasma. The plasma ion Plasma Surface Treatment plating process uses the material vaporised from a target (by whichever method is suitable) and If the plasma can generate excited species at a bombards the depositing film with molecules surface, then it follows that it should be possible produced from a reactive gas plasma as the film is to change the surface by plasma treatment deposited in order to change the properties of the of a material. One area where plasma surface depositing film. Arc VD occurs when an electrode treatment has garnered significant interest is in is vaporised through the application of high current the treatment of plastics and polymers. Plasma across a biased cell with the vaporised molecules treatment of plastics and polymers can have a being accelerated towards, and deposited on, the significant effect on the chemical and physical polarised substrate. properties of the materials, these changes occur These treatment and deposition techniques rapidly, often within seconds (34), through the require an understanding of physics and electrical following proposed mechanisms (35): engineering to generate, measure and optimise a • Etching and stripping surface material: plasma plasma in order to effect the chemical change on, which reacts with the surface to clean it of or within, a surface. contaminants, for example an oxygen plasma, or in more extreme cases such as with the Plasma for Catalyst Preparation inclusion of tetrafluoromethane (FreonTM (Chemours, USA)) to etch the surface itself. It has been reported that using plasma as a This happens through breaking of the polymeric preparative technique can improve catalyst

O2. They identified the two major reactions taking place as the dissociation of template molecules

by active species such as electrons or excited O2

and oxidation by excited O2 or O3 molecules (40). When directly comparing template removal from ZSM-5 zeolite by thermal and plasma techniques Liu et al. found that the rate of removal was approximately eight times higher using the DBD plasma method (41).

Plasma for Catalyst Modification

Using a DBD system AZO Materials, UK, reported Fig. 6. A catalytic DBD plasma reactor used for differences in the temperature and in the intensity of the demonstration of CH4 removal from an engine the peaks resulting from temperature programmed exhaust reduction of magnesium oxide (MgO) supported Ni catalysts compared to non-plasma treated dispersion, increase metal-support interactions and materials (42). The differences were attributed to change metal particle morphology, which in turn Ni particle morphology and dispersion. can lead to improved catalytic activity and stability Zhu et al. also treated supported Ni catalysts with (38). An example of improved activity from plasma DBD plasma and found an increase in the catalytic preparation is the supported nickel catalysts used activity and stability for the partial oxidation of for steam reforming; using a DBD plasma reactor CH4 (43). The scanning electron microscopy (SEM) to decompose the precursors such as nickel nitrate, images support a case for enhanced dispersion and for catalyst preparation it is possible to increase increased interaction between the 10% Ni and the the proportion of the (111) Ni facets which show Al2O3 support. As well as a measurable increase enhanced performance and coke resistance. in catalytic activity (3–5%) they also report a Another example of plasma used for catalyst reduction in the formation of C around the Ni. preparation is the use of DBD plasma instead of This is consistent with a change in the Ni particle the standard thermal calcination for silicon dioxide morphology towards having more (111) facets, as

(SiO2) supported cobalt materials for Fischer Tropsch observed by others during the plasma preparation synthesis. Li et al. found that plasma prepared of catalysts (38). materials had enhanced activity and a greater yield of heavy hydrocarbons when compared to Plasma for Material Regeneration the thermally calcined materials. This performance was attributed to the measured increase in Co The literature related to the plasma application for dispersion, smaller Co(II,III) oxide (Co3O4) cluster material regeneration is limited. The current state size and more even Co distribution. A byproduct of and perspectives of plasma applications for catalyst the plasma preparation is the claim that this route regeneration was discussed in a recent review can be a ‘greener’ method of preparing materials: (44). Plasma regeneration was successfully applied using a low temperature electron reduction instead for the reduction of oxidised catalysts and removal of using H2 as a reductant removes the need for of poisons and C deposits. The largest advantage dealing with H2 in the process (39). of plasma is that it allows catalyst regeneration to Another advantage of the low bulk temperature be performed at temperatures lower than those plasma treatment as a preparative technique in of typical thermal regeneration. The supply of comparison to a standard thermal treatment is gaseous reactive species and alteration of the the ability to remove precursors without inducing surface structure to a more energetic state were the detrimental changes that are associated with identified as prerequisites of successful low- the temperatures normally required to oxidise the temperature regeneration and it was also shown precursor molecules. An example of where this that plasma can supply heat in a more cost- is useful is in the preparation of zeolite materials effective way than conventional thermal treatment. through plasma template removal. Liu et al. have The energetic species produced in non-thermal conducted the removal of zeolite templates at plasma can initiate diverse reactions and open up around 125°C using a DBD plasma technique with or enhance reaction pathways other than those

144 © 2020 Johnson Matthey https://doi.org/10.1595/205651320X15759961130711 Johnson Matthey Technol. Rev., 2020, 64, (2) expected for equilibrium chemistry. As thermal nanosecond (5–100 ns) high voltage switches regeneration can result in catalyst sintering and with high frequency capabilities are to be thus a reduced number of active sites, plasma is a commercially available. Other important viable alternative to thermal treatment. components in order to improve electrical and The advantage of plasma regeneration was shown chemical efficiency are high dielectric strength in the recent work that has been undertaken at materials with different dielectric constants (6) the University of Central Lancashire, UK, studying and reduction of electromagnetic interference deactivated coked zeolite regenerated with the (EMI) presented in such fast-rising waves application of different techniques including • Physics: determining the appropriate signals thermal, MW plasma and DBD discharge plasma to measure to obtain information that (45). This work showed that plasma not only can be analysed in new ways. Analytical removes C from the deactivated catalyst but techniques borrowed from nuclear physics increases the activity of the catalyst significantly. may be appropriate, for example pulse Toluene disproportionation was used as the probe height spectrometry and more familiar optical reaction in this study. Unlike thermally regenerated spectrometry techniques can and should be used catalysts the material regenerated by plasma to characterise the species that are generated shows improved catalyst performance and the in real time. The outstanding challenge is to activity of the regenerated catalyst is even higher map out the outcomes and determine what they than that of the virgin material. Characterisation mean in the very large potential experimental methods including pyridine and collidine infrared space that presents itself in these systems studies, NH3 temperature programmed desorption • Engineering: scaling up a plasma reactor and solid state nuclear magnetic resonance were that requires both an interelectrode gap and used to explain the changes in catalytic activity. sufficient gas-plasma-surface interaction to Results showed MW plasma regeneration extended take advantage of the plasma and catalyst the catalytic life of zeolite due to the destruction synergies, while simultaneously having a low of Brönsted acid sites caused by dealumination, pressure drop without loss of crystal structure. In the toluene • Chemistry: knowledge around the catalogue of disproportionation reaction, this reduces the catalysts for thermally activated heterogenous amount of cracking which occurs, subsequently reactions is not fully valid for plasma activated leading to less coke deposition and therefore an catalytic processes and a relevant body extended catalytic life. of knowledge does not currently exist. An additional challenge is to address the limited Cross Disciplinary Challenges penetration depth of plasma into structures (46) and to develop methods for treating coated A key challenge for the field of plasma catalysis is to components including monolithic low pressure assemble teams with the relevant complementary drop structures skills in electronics, physics, engineering and • Data analysis and modelling: large amounts chemistry together to gain an understanding of data generated from multiple variables of the system in order to produce the desired increases the complexity of the experimental technological progress. A subset of this challenge design and interpretation. This therefore makes is the communication between the different the dataset an obvious candidate for supervised disciplines; involving not just the different language machine learning with input expertise from used, but also the models derived to express the each technical discipline. concepts and understanding of processes which are often not ideally accessible to other branches Concluding Remarks of science and engineering. As well as the broader topic of reducing barriers It is necessary to understand physics to be able to for communication, there are key challenges correctly use electronic engineering to generate the remaining within each of the disciplines required, requisite plasma to react with the designed catalyst including but not limited to: in order to affect the desired chemistry. Whether this • Electronic engineering: nanosecond pulsed chemistry change is in the gas phase, at the surface plasma sources tend to provide the most of a catalyst or within the surface, there are clearly energetic waveforms for plasma systems. a large number of challenges to be faced when For that reason, it is critical that solid state considering the large experimental space brought

145 © 2020 Johnson Matthey https://doi.org/10.1595/205651320X15759961130711 Johnson Matthey Technol. Rev., 2020, 64, (2) by the additional variables from non-chemistry 10. C. F. Gallo, IEEE Trans. Ind. Appl., 1975, IA–11, fields. The general challenges for plasma catalysis (6), 739 can be expressed in broad terms of, for instance, 11. G. F. Knoll, “Radiation Detection and Measurement”, increased efficiency, targeted catalysts or improved 3rd Edn., John Wiley and Sons Inc, New York, scaling of reactors; however within each of these USA, 2000, 802 pp targets underlying incremental improvements are 12. “Plasma Catalysis: Fundamentals and Applications”, required from additional scientific and engineering eds. X. Tu, J. C. Whitehead and T. Nozaki, Springer disciplines to achieve this. Series on Atomic, Optical, and Plasma Physics, Therefore, a key challenge for the field of plasma Vol. 106, Springer Nature Switzerland AG, Cham, Switzerland, 2019, 348 pp catalysis is to assemble teams with the relevant knowledge and skills in their own area of expertise 13. H. Puliyalil, D. Lašič Jurković, V. D. B. C. Dasireddy that can work together and communicate ideas and B. Likozar, RSC Adv., 2018, 8, (48), 27481 to initiate and progress the technology. Within 14. ‘Adaptable Reactors for Resource- and Energy- Johnson Matthey there are teams with expertise Efficient Methane Valorisation’, Spire 2030, across all the relevant disciples who can overcome ADREM Project: https://www.spire2030.eu/adrem (Accessed on 3rd February 2020) these barriers and work successfully together to realise the potential of plasma catalysis on a 15. U. Kogelschatz, Plasma Chem. Plasma Process., commercial scale. 2003, 23, (1), 1 16. U. Kogelschatz, B. Eliasson and M. Hirth, Ozone: Sci. Eng., 1988, 10, (4), 367 Acknowledgments 17. R. P. Anderson, J. R. Fincke and C. E. Taylor, Fuel, 2002, 81, (7), 909 Oli Beckett, Tracerco, UK, is thanked for drawing schematics. 18. X.-S. Li, A.-M. Zhu, K.-J. Wang, Y. Xu and Z.- M. Song, Catal. Today, 2004, 98, (4), 617 19. H. Kang, D. H. Lee, K. Kim, S. Jo, S. Pyun, Y. Song References and S. Yu, Fuel Process. Technol., 2016, 148, 209 20. “Plasma Chemistry and Catalysis in Gases and 1. Y. Uytdenhouwen, K. M. Bal, I. Michielsen, Liquids”, eds. V. I. Parvulescu, M. Magureanu and E. C. Neyts, V. Meynen, P. Cool and A. Bogaerts, P. Lukes, Wiley-VCH Verlag GmbH and Co KGaA, Chem. Eng. J., 2019, 372, 1253 Weinheim, Germany, 2012, 401 pp 2. P. Mehta, P. Barboun, F. A. Herrera, J. Kim, 21. J. C. Whitehead, J. Phys. D: Appl. Phys., 2016, P. Rumbach, D. B. Go, J. C. Hicks and 49, (24), 243001 W. F. Schneider, Nature Catal., 2018, 1, (4), 269 22. D. Lašič Jurković, H. Puliyalil, A. Pohar and 3. A. Fridman, “Plasma Chemistry”, Cambridge B. Likozar, Int. J. Energy Res., 2019, 43, (14), University Press, New York, USA, 2008, 978 pp 8085 4. H. Conrads and M. Schmidt, Plasma Sources Sci. 23. E. C. Neyts, K. Ostrikov, M. K. Sunkara and Technol., 2000, 9, (4), 441 A. Bogaerts, Chem. Rev., 2015, 115, (24), 13408 5. D. Ashpis, M. Laun and E. Griebeler, ‘Progress 24. E. K. Gibson, C. E. Stere, B. Curran-McAteer, Toward Accurate Measurements of Power W. Jones, G. Cibin, D. Gianolio, A. Goguet, Consumption of DBD Plasma Actuators’, 50th P. P. Wells, C. R. A. Catlow, P. Collier, P. Hinde and AIAA Aerospace Sciences Meeting Including C. Hardacre, Angew. Chem. Int. Ed., 2017, 56, the New Horizons Forum Aerospace Exposition, (32), 9351 Nashville, Tennessee, USA, 9th–12th January, 2012, American Institute of Aeronautics and 25. A. Zhou, D. Chen, C. Ma, F. Yu and B. Dai, Astronautics, Reston, USA Catalysts, 2018, 8, (7), 256 6. A. Bogaerts, Q.-Z. Zhang, Y.-R. Zhang, K. Van Laer 26. J. R. Shah, J. M. Harrison and M. L. Carreon, and W. Wang, Catal. Today, 2019, 337, 3 Catalysts, 2018, 8, (10), 437 7. V. Hessel, A. Anastasopoulou, Q. Wang, G. Kolb 27. I. Michielsen, Y. Uytdenhouwen, A. Bogaerts and and J. Lang, Catal. Today, 2013, 211, 9 V. Meynen, Catalysts, 2019, 9, (1), 51 8. G. J. van Rooij, H. N. Akse, W. A. Bongers and 28. A. Mizuno, Int. J. Plasma Environ. Sci. Technol., M. C. M. van de Sanden, Plasma Phys. Control. 2009, 3, (1), 1 Fusion, 2018, 60, (1), 014019 29. C. Louste, G. Artana, E. Moreau and G. Touchard, 9. P. A. Davidson, “An Introduction to J. Electrostat., 2005, 63, (6–10), 615 Magnetohydrodynamics”, Cambridge Texts in 30. J. Hoard, P. Laing, M. L. Balmer and R. Tonkyn, SAE Applied Mathematics, Cambridge University Press, Technical Paper 2000-01-2895, SAE International, Cambridge, UK, 2001, 431 pp Warrendale, USA, 16th October, 2000

31. R. Tonkyn, S. Barlow, M. L. Balmer, T. Orlando, 38. C. Liu, M. Li, J. Wang, X. Zhou, Q. Guo, J. Yan and J. H. Hoard and D. Goulette, SAE Technical Paper Y. Li, Chinese J. Catal., 2016, 37, (3), 340 971716, SAE International, Warrendale, USA, 1st 39. Y. Jiang, T. Fu, J. Lü and Z. Li, J. Energy Chem., May, 1997 2013, 22, (3), 506 32. B. M. Penetrante, R. M. Brusasco, B. T. Merritt, 40. Z. Wang, Y. Zhang, E. C. Neyts, X. Cao, X. Zhang, W. J. Pitz and G. E. Vogtlin, SAE Technical Paper B. W.-L. Jang and C. Liu, ACS Catal., 2018, 8, (3), 1999-01-3637, SAE International, Warrendale, 2093 USA, 25th October, 1999 41. Y. Liu, Y. Pan, P. Kuai and C. Liu, Catal. Lett., 2010, 33. ‘Plasma Removal of Methane from Natural Gas 135, (3–4), 241 Dual-Fuel Engines (PROMENADE)’, UK Research 42. Hiden Analytical, ‘Using a Dielectric Barrier and Innovation, Swindon, UK: https://gtr.ukri. Discharge for Plasma Modification of Catalysts’, org/projects?ref=102661 (Accessed on 30th AZO Materials, Manchester, UK, 28th March, 2018, September 2019) 10 pp 34. “Plasma Surface Modification of Polymers: 43. D. Cheng, X. Zhu, Y. Ben, F. He, L. Cui and C. Liu, Relevance to Adhesion”, eds. M. Strobel, Catal. Today, 2006, 115, (1–4), 205 C. S. Lyons and K. L. Mittal, VSP BV, Zeist, The 44. D. H. Lee, Y.-H. Song, K.-T. Kim, S. Jo and H. Kang, Netherlands, 1994, 290 pp Catal. Today, 2019, 337, 15 35. S. L. Kaplan and P. W. Rose, Int. J. Adhes. Adhes., 45. A. Halman, ‘Investigation of the Effects of Non 1991, 11, (2), 109 Thermal Plasma and Microwaves on Mordenite’, 36. S.L. Kaplan and P. W. Rose, Plastics Eng., 1988, PhD Thesis, Centre for Materials Science, School 44, (5), 77 of Forensic and Applied Sciences, University of 37. D. M. Mattox, “Handbook of Physical Vapor Central Lancashire, Preston, UK, 2020, submitted Deposition (PVD) Processing”, 2nd Edn., Elsevier 46. Y.-R. Zhang, K. Van Laer, E. C. Neyts and A. Bogaerts, Inc, Burlington, USA, 2010, 792 pp Appl. Catal. B: Environ., 2016, 185, 56

The Authors

Peter Hinde obtained his PhD Alkis Gkelios gained his Masters in photocatalysis from The in Electrical Engineering from University of Bradford, UK, and the University of Patras, Greece has been working for Johnson in 2014. He has been working Matthey at their central research on high voltage electronics and and development facility plasma generation systems. For supporting multiple global the last three years, he has been business units since 2004. working for Johnson Matthey. He currently leads Johnson His interests are focused on Matthey’s plasma catalysis efficient plasma production and research group. control for industrial process.

Vladimir Demidyuk received his Carl Tipton gained his PhD in MSc and PhD degrees in Physical Nonlinear Physics from the Chemistry from Moscow State University of Manchester, UK, University, Russia. For the past in 2003. Since then, he worked 15 years he has been working in as a Development Physicist at academia and in industry in the Tracerco, UK. He is currently UK. He joined Johnson Matthey, a Measurement Engineer at Sonning Common, UK, in 2015. Johnson Matthey, Chilton, UK, His research interests include where he works to optimise the application of plasma Johnson Matthey industrial technologies for industrial processes. purposes.

www.technology.matthey.com

Nanosurfaces 2019 Commercialised techniques involving plasma, surface coatings and graphene

Reviewed by Alistair Kean* polymerisation of surface molecules to provide Johnson Matthey, Blounts Court, Sonning a very thin hydrophobic conformal coating. This Common, Reading, RG4 9NH, UK is suitable for mass production with good quality control. If a direct current (DC) plasma is used, the Sara Coles molecular structure is disrupted whereas pulsed Johnson Matthey, Gate 2, Orchard Road, DC retains coherent molecular structure and the Royston, Hertfordshire, SG8 5HE, UK best liquid repellence. Photoelectron spectroscopy analysis was used to analyse bond structure. The *Email: [email protected] original product coating was splash proof but not completely waterproof, so a new coating was developed which is thicker (~100 nm) and has Introduction increased crosslinking. Thickness alone does not determine the level of water repellence (standards Nanosurfaces 2019 took place on 15th October from IPX2 to IPX8), the process parameters affect 2019 at the Institute of Materials, Minerals & porosity and film defects (1, 2). The coating was Mining (IOM3), London, UK. This event included further improved by ensuring all product pieces are commercialised techniques featuring plasma, placed close to a plasma electrode. The coating was surface coatings and graphene. Applications claimed to be durable and is tested with methods ranged from anticorrosion and waterproofing of including the adhesive tape test, thermal, chemical electronics to battery materials and biomedical and abrasion. It is economic for high-value items. applications. Scalability, reproducibility, economics The maximum temperature for the coating is and sustainability were considered by many of the 200ºC, determined by the maximum temperature speakers. for the items to be coated. ‘Enhancement in the Reactive Deposition Process Plasma Applications Using Remote Plasma Spluttering’ by James Dutson, Plasma Quest Ltd, UK, presented the The Keynote ‘Water Protection Technologies for use of a remote plasma to generate ions whose Smartphones and Beyond’ was presented by path when accelerated is curved by magnets and Angeliki Elina Siokou, P2i Ltd, UK. P2i originated focussed onto the sputter target. Key advantages as a spinoff from Durham University, UK. The of this technology are: high target usage, control original product was an oil repellent coating for over energy of sputter ions, low temperature at textiles created in 2004. They then developed the substrate. They have had good success with a hydrophobic coating for electronics. Initial aluminium doped zinc oxide as an indium tin applications were in hearing aids then, around oxide (ITO) replacement. Substrate heating is not 2011, the company moved into water protection required to control the phase or crystallinity of for smartphones. They are now coating half a deposited material as this can be done via sputter billion smartphones per year. Customers include ions energy control. Deposition rates are typically Samsung, Apple, Bose, Huawei and ASUS. Around 1–6 μm h–1, and 1 μm h–1 for the ZnO:Al materials. 18% of smartphones are thought to be damaged Plasma Quest is working with CSEM, Neuchâtel, by water ingress every year. P2i’s technology uses Switzerland, on photovoltaics and transistors. The a radio frequency (rf) plasma attachment and process can be carried out at ambient temperature

148 © 2020 Johnson Matthey https://doi.org/10.1595/205651320X15814173241108 Johnson Matthey Technol. Rev., 2020, 64, (2) so is suitable for coating plastics or sensitive Superior adhesion and solvent-free scalability were substrates. Reactive deposition is possible, i.e. claimed. The speaker claimed that the technique to incorporate an oxide or nitride from a metal can use two source virtual cathode deposition target. The main driver is replacement of ITO (VCD) to design materials in minutes, not hours. for transparent conductive coatings and there is a Horizon 2020 European Union (EU) project: Nanoparticles and Graphene INREP, Towards Indium Free TCOs. The company is currently scaling up to large surface areas ≥50 cm2. ‘Manufacture and Applications of Nanoclusters’ was The process does not require a clean room. presented by Richard Palmer, Swansea University, ‘Molecular Plasma Technology’ was presented UK. He set up Grove Nanomaterials to commercialise by Britta Kleinsorge, Molecular Plasma Group, matrix assembly cluster source (MACS) technology Luxembourg. This company offers atmospheric (3) for the solvent-free synthesis of nanoparticles plasma (50 kHz) processes for surface modification. (NP) for catalysis, sensors and electrodes. Atomic It is lower temperature compared to corona like nanoclusters can be created to mimic enzymes arc plasma processes. Organic precursors could with formation of grooves and pockets. Size be functionalised and adhere to a surface at low selected cluster beam deposition (CBD) of gold or temperature. Some interesting work on biological molybdenum sulfide, which can be nickel or cobalt non-specific binding and antibody immobilisation doped to replace platinum in water splitting, was was carried out with KU Leuven, Belgium, in which carried out. The present plan is to scale up the the process time was reduced from 24–72 h to 10 s. research system to be able to produce 1 g h–1 and Superhydrophobic barriers have been developed the technology has the potential to be scaled to for anticorrosion applications, waterproofing and mass production (tonne scale). adhesion. They presented a case study of a process ‘Silica Nanoparticles for Super-Hydrophobic Ice- they developed for a German automotive original Repellant Coatings’ was presented by Simon Haas, equipment manufacturer that could be automated Promethean Particles Ltd, UK. Promethean Particles with reduced chemical usage. Ltd is a spinout from the University of Nottingham, ‘A Radically New Antimicrobial Nanosurface UK. Using a continuous hydrothermal process, Formed by Plasma Processing’ was presented copper, silver, ZnO, barium titanate and silica NP by Alistair Kean, NikaWorks Ltd, Watlington, UK. can be produced as dispersion in liquid with no There is a need for more environmentally friendly dry powders for safer handling. The liquids can coatings, for example crisp packets which currently be used in further processes and agglomeration is use 30–40 nm Al on plastic film. Tungsten carbide prevented. The dispersions are highly stable and (WC) physical vapour deposition (PVD) coatings the process is scalable. A production plant has for surgical scissors need to be durable for been built which can produce 1000 tonnes per decontamination cycles. The BeBionic prosthetic year based on dry weight equivalent. Conductive hand includes a titanium alloy coating on one inks and printed electronics are currently being surface. Three dimensional (3D) nanomaterials developed with partners. The speaker presented an (metamaterials) are inspired by nature and Innovate UK funded programme, ICEMART, looking nanosurfaces can have surface areas of the order at SiO2 NP to prevent icing of plane wings. The of ~1000 m2 g–1. SOLAMON was a 7th Framework Welding Institute, Great Abington, UK, was part of Programme for ‘large’ NP 20 nm, 30 nm or 40 nm this project and provided a process to functionalise scale. Gencoa Ltd is a PVD company in Liverpool, the SiO2 particles to render them hydrophobic. UK, which is developing antimicrobial coatings with This technology is currently being commercialised NikaWorks. via Sharc® Matter, an Opus Materials Technology ‘Virtual Cathode Deposition’ was presented by Company, UK. Dmitry Yarmolich, Plasma App Ltd, UK. Plasma App ‘Graphene Enhanced Products’ was presented is based at Harwell Oxford Science and Innovation by Thanuja Galhena, Versarien plc, UK. Versarien Campus, UK. The speaker described a thin film acquired Cambridge Graphene Ltd, UK, in 2017 battery project with the University of Cambridge, and supplies proprietary grades of two-dimensional UK, which can deposit carbon as a combination of materials. There are two main products: NaneneTM graphite and graphene immediately followed by the (graphene) and HexoteneTM (boron nitride). Both lithium cobalt oxide (LCO) cathode. It is a platform can be manufactured at a 3 tonnes per annum technology that can deposit a thick or thin film of scale. The company has certification from the almost anything from nanometre scale to 50 µm. Graphene Council, New Bern, USA and the National

137 165 Fig. 1. SIMS images 123 140 of 3D silicon/tantalum 109 132 test object taken using

95 115 the CAMECA SIMS

82 99 instruments. Image courtesy of Mike Petty, 68 82 Loughborough Surface 54 66 Analysis Ltd, UK 41 45

27 33

13 16

0 0

137 137

123 123

109 109

95 95

82 82

68 68

54 54

41 41

27 27

13 13

0 0

Physical Laboratory (NPL), Teddington, UK, ‘Photocatalytic Antimicrobial Surfaces’ by European Registration, Evaluation, Authorisation Jeremy Ramsden, University of Buckingham, UK. and Restriction of Chemicals (REACH) registration Ramsden discussed the antimicrobial application in the EU and is applying for similar certifications of photocatalytic powders coated via spray from in China. Potential applications are composites, titanium dioxide powders in a water solution. fire retardants, barrier coatings, energy storage, He described the problem of hospital acquired filtration and heat dissipation. A fairly unusual infections, especially resistant infections such product application was a railway arch supporting as methicillin-resistant Staphylococcus aureus sensors, claiming that lighter weight reduces (MRSA), and presented studies of deaths and transport, installation time and personnel costs. disability-adjusted life year (DALY) statistics in Graphene based inks known as GRAPHINKS® have Europe, USA and elsewhere dating from 1930s to been developed (4). Another example is a graphene the present. There are many vectors of microbial coated diaphragm for some earphones – apparently transfer in and around hospital environments, bass is clearer. FLEXIBAT is an Innovate UK funded these are surprisingly little studied and there is project. The company is working on flexible much debate about the relative importance of batteries with Zinergy UK Ltd, Cambridge, UK, air, walls, floors, shoes or wheels and ceilings in making graphene coated electrodes for corrosion transmission of pathogens. In particular there are resistance. There are also supercapacitors using few or no controlled studies on hospital cleaning graphene plus metal oxide NP. The speaker noted efficacy. Hand hygiene compliance seems to that the cost of graphene has reduced considerably, have peaked at ~40%. He studies non-sacrificial making it cost effective for many applications. catalytic coatings, i.e. TiO2 which can kill microbes ‘Enhanced Surface-Analysis Capability’ was even under ambient lighting with continuous presented by Mike Petty, Loughborough Surface effectiveness. Estimated 10 colony forming units Analysis Ltd, UK. Petty focussed on the new (CFU) m–2 s–1 of bacteria arriving and 6000 s–1 secondary ion mass spectrometry (SIMS) system oxidising equivalents forming. He carried out a trial (Cameca, France) which can do quite sophisticated in which a coating was applied after a hospital deep analysis mainly on semiconductor materials. clean on the high-touch surfaces, for example bed Data cubes are formed so one can go back and rails, table and tubing. The surfaces were tested analyse specific coordinates within a scanned cube using Agar pads (selective for MRSA and general).

(Figure 1). This technique is sensitive to isotope The coating used a commercial sol of TiO2 NP in identification. a sol-gel process to form a hard coating within

20–30 mins. The coating is invisible (1 µm) and (a) (b) can be applied to mirrors and glass. Sampling was carried out for three months, the coating was durable in this timeframe. Evidence suggested the coating resulted in lower microbial growth (Figure 2) (5). Microbial resistance is unlikely as the microbial cell has no mechanism for defence against peroxide. The half-life of bacteria was ~6 h.

10 mm 10 mm Conclusions

(c) (d) Martin Kemp, Chairman of the IOM3 Nanomaterials Conference organising committee, remarked how many of the presentations featured plasma techniques and was excited by the many examples of commercialised techniques. There will be another meeting in early 2020.

10 mm 10 mm References 1. S. R. Coulson, D. Evans, T. Hellwig, F. Hopper, (e) (f) N. Poulter, A. Siokou and C. Telford, P2i Ltd, ‘Method for Forming a Coating on an Electronic or Electrical Device’, World Patent Appl. 2-16/198,855 2. S. R. Coulson, D. Evans, A. Siokou and C. Telford, P2i Ltd, ‘Coatings’, World Patent Appl. 2016/198,857 3. P. R. Ellis, C. M. Brown, P. T. Bishop, J. Yin, K. Cooke, 10 mm 10 mm W. D. Terry, J. Liu, F. Yinc and R. E. Palmer, Faraday Discuss., 2016, 188, 39 Fig. 2. Stained Staphyloccocus epidermidis 4. P. G. Karagiannidis, S. A. Hodge, L. Lombardi, within a biofilm (confocal laser scanning F. Tomarchio, N. Decorde, S. Milana, I. Goykhman, microscopy). Green fluorescence: live bacteria; red Y. Su, S. V. Mesite, D. N. Johnstone, R. K. Leary, fluorescence: membrane-compromised bacteria. P. A. Midgley, N. M. Pugno, F. Torrisi and (a) 1.5 h exposure to UVA only; (b) 1.5 h exposure to photocatalytic treatment; (c) 3 h exposure A. C. Ferrari, ACS Nano, 2017, 11, (3), 2742 to UVA only; (d) 3 h exposure to photocatalytic 5. P. S. M. Dunlop, C. P. Sheeran, J. A. Byrne, treatment; (e) 3 h exposure to TiO2 in the dark; M. A. S. McMahon, M. A. Boyle and K. G. McGuigan, (f) 3 h no treatment control (no TiO2, no UVA J. Photochem. Photobiol. A: Chem., 2010, 216, exposure). Reprinted from (5) with permission (2–3), 303 from Elsevier

The Reviewers

Alistair Kean is a New Business Sara Coles is Editor of Johnson Development Consultant at Johnson Matthey Technology Review. She Matthey, Sonning Common, UK. In obtained her Master’s degree 2013 he set up NikaWorks Ltd in Chemistry from Lancaster with the aim of commercialising University, UK, and has previous nanotechnology. He is a visiting experience in pharmaceuticals professor at Manchester Metropolitan research. She is interested in all University, UK and in 2019 he accepted areas of science and technology an appointment at University of the relevant to Johnson Matthey’s vision Highlands & Islands (UHI), UK, as of enabling cleaner air, improved professor of medical nanotechnology. health and the efficient use of natural resources.

www.technology.matthey.com

Observing Solvent Dynamics in Porous Carbons by Nuclear Magnetic Resonance Elucidating molecular-level dynamics of in-pore and ex-pore species

Luca Cervini is expected due to the shorter in-pore–ex-pore Department of Chemistry, Lancaster University, path length and faster diffusion in large pores. Our Lancaster, LA1 4YB, UK results also show that in-pore–ex-pore exchange of apolar solvents is slower than water, suggesting Nathan Barrow that the hydrophobic chemistry of the carbon Johnson Matthey, Blounts Court Road, Sonning surface plays a role in the diffusion kinetics, and Common, Reading, RG4 9NH, UK that increased viscosity also reduces the exchange kinetics. Our results also suggest the importance of John Griffin* other parameters, such as molecular diameter and Department of Chemistry, Lancaster University, solvent packing in micropores. Lancaster, LA1 4YB, UK; Materials Science Institute, Lancaster University, Lancaster, LA1 Introduction 4YB, UK Understanding the performance of activated carbons *Email: [email protected] in their applications as energy storage materials or catalyst supports requires a description of the behaviour of adsorbates, including solvents, gases, The adsorption and diffusion of species in activated organic molecules or ions. NMR allows adsorption of carbons is fundamental to many processes in species in porous carbons to be studied due to the catalysis and energy storage. Nuclear magnetic nucleus-independent chemical shift (NICS) arising resonance (NMR) gives an insight into the from the aromaticity of the pore walls (1, 2). The molecular-level mechanisms of these phenomena NICS allows adsorbed species to be distinguished thanks to the unique magnetic shielding properties from ex-pore species in the bulk solution external to of the porous carbon structure, which allows the carbon particles and has been shown to depend adsorbed (in-pore) species to be distinguished on, among other parameters, the distance between from those in the bulk (ex-pore). In this work we the adsorbate and the pore walls. In recent years, investigate exchange dynamics between ex- pore several studies have exploited this to use NMR as and in-pore solvent species in microporous a probe of the pore structure of a range of porous carbons using a combination of one-dimensional carbons. Borchardt et al. showed that the NICS (1D) and two-dimensional (2D) NMR experiments. of adsorbed organic electrolyte species varied in We systematically compare the effects of four accordance with the pore size of titanium carbide- variables: particle size, porosity, solvent polarity derived carbons (CDCs) which have very well- and solvent viscosity to build up a picture of how defined porosity (3). This was supported in work these factors influence the exchange kinetics. We by Forse et al. who also showed that ion adsorption show that exchange rates are greater in smaller in CDCs is significantly reduced when the average and more highly activated carbon particles, which pore size is smaller than the solvated ion size (4).

In a subsequent study, density functional theory 90º 90º 90º (DFT) was used to show that for model carbon slit pores, the NICS depends upon both the pore width t1 tmix and the size and curvature of carbon fragments making up the pore walls (5). Xing et al. used DFT calculations to derive a relationship between the magnitude of the NICS and the pore size assuming a slit pore geometry, which was found to agree Fig. 1. Pulse sequence of the EXSY experiment. For well with experimental measurements of NICS for each tmix chosen, the sequence is repeated with aqueous species adsorbed on poly‑ether-ether- increasing t1 values ketone (PEEK) derived carbons (PDCs) (6). Another important phenomenon that can be studied by NMR is the dynamics of the often highly precess at their characteristic frequencies and a mobile adsorbate and solvent species. Diffusion signal is detected. After a Fourier transformation coefficients of adsorbed species can be determined of both time dimensions is performed, a 2D experimentally via NMR using pulsed field gradient spectrum is obtained. Magnetisation that has not (PFG) techniques (7). A number of PFG NMR studies been exchanged will have precessed at the same have shown that diffusion of species confined characteristic frequency during both times, giving in carbon micropores is reduced significantly rise to a peak on the diagonal of the spectrum. compared to bulk solution. Furtado et al. observed Magnetisation that has been exchanged between a broad distribution of local diffusivities for a in-pore and ex-pore locations will have precessed carbon with a bimodal pore size distribution which at one frequency initially and another during was interpreted in terms of restricted internal detection. A peak will appear off-diagonal in the diffusion between pores of different sizes (8). PFG spectrum and at the direct-dimension chemical NMR measurements on ethylene carbonate and shift of the spin where the molecule finally resided. dimethyl carbonate mixtures by Alam and Osborn These two cases give rise to diagonal peaks and Popp showed that diffusion coefficients for species cross-peaks, respectively. Further information can adsorbed in carbon micropores were reduced by be extracted by performing the experiment with up to a factor of five compared to bulk solution (9). varying mixing times. Integrating the cross-peak Forse et al. used PFG NMR to study microporous area and plotting as a function of mixing time carbon supercapacitor electrodes and observed yields a build-up curve, from which exchange rate significant reductions in the diffusion coefficients constants can be extracted (12, 13). of adsorbed species, although acetonitrile solvent The EXSY approach was applied by Alam species were found to diffuse faster than the larger and Osborn Popp who showed that the in-pore electrolyte ions (10). environment is inhomogeneously broadened due to While PFG NMR provides significant insight into species occupying a range of pore environments, the dynamics of species confined within the porous between which exchange takes place on the carbon network, further information regarding millisecond timescale (9). Griffin et al. showed the dynamic exchange of species between pore for a commercial porous carbon saturated with an environments and also between the in-pore and organic electrolyte that in-pore–ex-pore exchange ex‑pore environments can be obtained from of the anionic species also takes place on the exchange spectroscopy (EXSY) measurements (11). millisecond timescale, although did not fit well In this experiment, for which the pulse sequence is to a single exchange process (14). Fulik et al. shown in Figure 1, the first pulse and subsequent subsequently showed that build up curves in EXSY delay allows the single-quantum magnetisation to data for commercial activated carbon saturated precess at a characteristic frequency. The second with organic electrolyte can be interpreted in pulse converts this coherence into a population terms of two processes with different rates, i.e. state, which is maintained for a duration called a slow process attributed to diffusion from the the mixing time, tmix. For zero mixing time it is centre of the particle to the surface, and a much expected that there will be no diffusion and for faster process attributed to effective exchange long mixing times it is expected that the solvent between ex-pore and in-pore (15). In addition, it molecules will reach an equilibrium between the was shown that exchange dynamics can also affect different pore locations. After the mixing time a the observed NICS and lineshape in 1D spectra, third pulse is applied and the spins once again whereby the onset of in-pore–ex-pore exchange

(a) (b) Ex-pore Pure water 2.4 1.8*PV 1.2*PV 1.8 Mesopores In-pore 1.0*PV Micropores 0.8*PV 1.6 NICS 0.6*PV –1 0.4*PV 1.2

0.2*PV ml g Dry 0.8

0.4

Variation of adsorbed volume, of adsorbed volume, Variation 0 5 0 –5 0.7 1.2 1.7 2.2 2.7 3.2

d 1H, ppm Pore size, nm Fig. 2. (a) 1H NMR spectra of a typical PDC wetted with increasing amounts of water shown in terms of multiples of the total pore volume (PV); (b) corresponding N2 gas sorption pattern

upon saturation of the micrometre-sized particles in-pore components coming from the regions of leads to a reduction of the observed NICS (4). In fast averaged NICS can, as shown by Merlet et these typical cases, the diffusion path between the al., be broadened by slow intra-particle exchange ex-pore and in-pore environment is much shorter, (17). The size of these regions of averaged NICS allowing faster exchange of species between can be estimated to be around 1 µm in diameter, ex-pore and in-pore which leads to partial or assuming that the in-pore diffusion coefficient is complete exchange averaging of the in-pore and the same as ex-pore, and that a 1 kHz frequency is ex-pore resonances. enough for the fast regime. From ex-pore to in-pore However in our previous work (16) we showed that however, the diffusion path was much longer in our the NICS did not change upon saturation for 100 µm particles. It arises that there must be regions with PDC particles soaked with water. An example of fast in‑pore–ex-pore exchange and more isolated this observation is shown in Figure 2(a). This was regions. attributed to the relatively large particle size used In this work, we systematically investigate the in the experiments, meaning that in-pore–ex-pore effects of carbon particle size, porosity and solvent exchange does not take place on the timescale of properties on in-pore–ex-pore and in-pore– in‑pore the NMR experiment. Furthermore, nitrogen gas exchange dynamics as viewed by NMR spectroscopy. sorption analyses of PDCs revealed the presence We chose PDCs as our model system due to the of at least three different pore widths of 0.8 nm, ease of synthesis and tunability of the porosity 1.2 nm and 2.2 nm, (Figure 2(b)) whereas only whereby the pore volume (PV) and average pore a single in-pore resonance was observed. This was size vary approximately linearly with burn-off rationalised in terms of fast exchange between (BO) and activation time (18), in addition to these micro- and mesopores within the pore network, materials giving rise to generally strong NICS, leading to motional averaging of the in-pore facilitating structural characterisation and analyses resonance. However, the single in-pore resonance of adsorbate behaviour (19–21). We first provide a was found to be subject to inhomogeneous review of the effects of two-site exchange on 1D broadening, as evidenced by a purely diagonal NMR spectra with specific reference to a model for in‑pore diagonal peak in 2D EXSY spectra at short a carbon particle saturated with solvent species. mixing time. This means that within a particle, there We then discuss the experimental results in three are regions of different averaged NICS, which can sections: first, we compare PDC particles of ~80 µm be due to a variation of aromaticity as well as local diameter with ~15–20 µm particles obtained from average pore size. The latter depends mostly on the the same sample, to confirm that our samples show inhomogeneity of activation throughout the carbon minimal exchange broadening due to a longer path particles. This was experimentally minimised between the two environments. Second, we compare by activating smaller particles. The symmetric PDC samples that were steam-activated with 20%

BO and 54% BO to see the extent at which diffusion molecules between regions giving rise to different coefficients in pores of different sizes influence NICS affects the NMR spectrum. The perturbation the exchange rate constants. Finally, we compare depends on the average time each molecule remains PDC samples saturated with different solvents, in each environment. Considering a nucleus able to namely water, hexane and cyclohexane. These explore two environments, several cases can be solvents were chosen because of their comparable distinguished: the slow exchange regime where both or different polarity and viscosity, two parameters peaks may be broadened but without being shifted, which are expected to influence the diffusion of the the intermediate regime where both peaks have solvent molecules in the pores. The pore filling is merged into a single very broad peak and the fast assessed in unsaturated samples to determine the exchange regime where the single peak narrows. To accessibility of the pore network. Ex-pore–in-pore understand these observations, we can turn to the exchange rate constants were determined using underlying principles of the Fourier transform. exchange experiments and were then used to better In the fast exchange regime, the resulting understand how various exchange regimes perturb peak is located at the average of the chemical 1D NMR spectra. shifts weighted by the residence time in each environment. To illustrate this, various NMR Exchange Averaging in One- signals were simulated using cosine waves and subsequently Fourier transformed by fitting Dimensional Nuclear Magnetic cosine functions of variable frequencies, whereby Resonance Spectra the NMR spectrum is obtained as the integral of Averaging of NMR signals is a common phenomenon the product between the original and the fitted observed when individual nuclei explore several functions (full details of the simulation are magnetic environments. In the case of a solvent given in Supplementary Information (SI)). This adsorbed in porous carbons, the diffusion of solvent simulation is a simplified case of fast exchange

(a) (b) 1.0 20 30 Hz 1:0 0.5 15

50 Hz 0 32 Hz 10:1 10 30 Hz Angle, rad –0.5 5 40 Hz 1:1 Intensity, arbitrary units arbitrary Intensity, –1.0 0 5 10 0 5 10 Acquisition time, arbitrary units Acquisition time, arbitrary units (c) Intensity, arbitrary units arbitrary Intensity,

0 25 30 35 40 45 50 55 60 Frequency, Hz Fig. 3. (a) Portion of simulated cosine waves oscillating at 30 Hz (blue) or 50 Hz (red) for various periods of time; (b) angle evolution as a function of time; and (c) corresponding Fourier transformed spectra

155 © 2020 Johnson Matthey https://doi.org/10.1595/205651320X15747624015789 Johnson Matthey Technol. Rev., 2020, 64, (2) where motional broadening is not accounted for. Its purpose is to show how the averaged chemical Ex-pore shift depends on the relative dwell times. Figure 3(a) shows the simulated NMR signals; Exchange-averaged the blue component oscillates at a frequency of exch 30 Hz and the red components oscillate at 50 Hz. Vex The ratio of the time spent oscillating at either exch frequency is given: an average of approximately In-pore Vin 32 Hz is obtained when adopting 30 Hz for 0.1 ms and 50 Hz for 0.01 ms, and an average of 40 Hz Ca rbo is obtained when oscillating at both frequencies n particle for the same time. For better visibility, the Macro/mesopore angle in radians travelled as a function of time is plotted in Figure 3(b). The spectra obtained after Fourier transformation, Figure 3(c), show how the apparent frequency of the exchange- Fig. 4. Scheme of an activated carbon particle averaged peak depends on the residence time in (black continuous line, delimiting the ex-pore/in- pore boundary) saturated with solvent, blue for each environment. When the residence times of adsorbed and green for bulk. The orange arrows the nucleus in both environments are identical, represent fast exchange of molecules between the we call it a symmetric exchange. This is found for two environments. The colour gradient represents example in an acid-base mixture when pH = pKa, the probability of a solvent molecule exchange where by definition both acid and conjugated base during data acquisition. Molecules between the are equimolar. However, asymmetric exchange is two dashed lines may exchange fast enough to appear at the exchange-averaged chemical shift, more relevant to the present study due to the exch exch with Vin and Vex standing for the volume of large population difference between the in-pore such exchanging solvent initially in-pore or ex-pore and ex-pore environments, and also between the respectively volume of the different connected pores. Different chemical shifts can be attributed to different regions inside and around a porous carbon each environment. However in practice, when particle. Figure 4 illustrates a carbon particle adsorption and desorption are at equilibrium, the (black continuous line) saturated with a solvent. solvent molecules exchange by pairs because the exch Molecules in the green region adopt the ex-pore volumes of exchanging adsorbed solvent Vin exch chemical shift, and molecules in the blue region an and exchanging free solvent Vex are constant. in-pore chemical shift. The deeper the colour, the The diffusion coefficient in the pores is smaller more likely the molecule is to change environment than in the bulk, so for the in-pore molecules to within a certain time. The dashed lines represent exchange in the fast regime they must reside exch the boundaries between molecules undergoing closer to the interface. Therefore, Vin is smaller exch fast and slow exchange: within the dashed lines, than Vex . The residence time of the exchanging the in-pore–ex-pore exchange (orange arrows) solvent molecules can now be correlated to exch exch is frequent enough for the chemical shift to be Vex and Vin and the in-pore/ex-pore ratio of averaged. Note that the ex-pore dashed line is at a diffusion coefficients. Experimentally it is possible exch exch constant distance from the particle surface because to calculate Vex and Vin from the position and the ex-pore diffusion coefficient is constant. The volume of the broad ex-pore peak. in-pore dashed line however is distorted due to inhomogeneities of in-pore diffusion coefficients Results and wraps around big mesopores penetrating deep into the particles. In summary we expect two Particle Size Effects non-exchanging peaks corresponding to ex-pore solvent far from the particle and in-pore solvent To directly observe the impact of particle size on in the core of the particle, and a broadened peak, 1D NMR spectra, two activated PDC samples were corresponding to solvent molecules undergoing reduced from approximately 100 µm particle size to fast exchange. approximately 15–20 µm by sieving (see Methods The position of the exchange-averaged peak in SI). The samples were named xBO_y, where x depends on the time the molecules spend in is the percentage of BO and y the median particle

therefore also partially homogeneously broadened (a) by exchange-averaging; the longer the residence Ex-pore In-pore time in the pores, the broader and the more shifted BO = 54% the broad component is, i.e. the bigger the ratio NICS before saturation D = 80 µm exch exch 50 Vin /Vex as per Figure 4. In big particles, the NICS after saturation broad component amounts to 30% of the ex‑pore peak with a FWHM = 0.30 ppm and is located within 0.10 ppm of bulk water chemical shift. On the other hand, with small particles the broad component represents 82% of the total ex-pore water with a (b) Narrow component much bigger FWHM = 0.70 ppm and is shifted by BO = 54% 0.20 ppm relative to bulk water. This indicates a exch exch D50 = 21 µm slight increase of V /V when the particles Broad component in ex are reduced; in other words, a bigger proportion of ex-pore water is able to experience the in-pore environment for a longer time. To explain the broadening of the ex-pore peak, we must adopt the point of view of water molecules 6 4 2 0 –2 –4 that spend the majority of their time in the ex‑pore d 1H, ppm environment, where we believe the packing of Fig. 5. 1H NMR spectra of: (a) 54BO_80 injected the particles plays an important role. A simple with 0.2*PV (red) and 2.3*PV (blue) of water; model consisting of spherical particles (Figure S3 and (b) 54BO_21 injected with 0.3*PV (red) and in the SI) allows exchange rates to be calculated 2.3*PV (blue) of water. The dashed lines show the maxima of the in-pore peaks in function of the particle diameter (Figure S4). With 80 µm particles (similar to 54BO_80) we find exchange rates of 2.5 Hz, and of 36.7 Hz with size D50, measured by dynamic light scattering. 54BO_21 particles (see SI for more details on the The samples were in the first instance wetted using calculations). The NICS being around 3000 Hz, a microsyringe with a defined volume of deionised the exchange regime would be slow in both cases, water less than the PV, then with a volume greater which is consistent with our observations, and than PV, to observe the 1H NMR spectrum before increases by a factor of 15 when reducing the and after saturation. We have previously shown particle size from 80 µm to 21 µm. that this method allows us to compare the NICS Examples of 2D EXSY NMR spectra are shown in averaged over the whole pore network without, Figures 6(a) and 6(b) corresponding to sample and then including, perturbations related to 54BO_80 and 54BO_21 saturated with water. The in‑pore–ex-pore exchange. This is because the mixing time (tmix) was 20 ms for both spectra, in‑pore peak before sample saturation corresponds and it can be seen that the cross-peaks are more to water located in completely filled particles that pronounced with small particles, giving a first are not yet surrounded by water. Figures 5(a) indication that the exchange is faster. Figures 6(c) and 5(b) show the spectra of sample 54BO_80 and 6(d) show the build-up curves of the intensity and 54BO_21, respectively. The in-pore peak ratio of cross-peaks over diagonal peaks. Visually, shifts by approximately 0.2 ppm upon saturation, we can see that the curve for big particles reaches regardless of the particle size. This means that in the maximum after long tmix intervals, whereas for this range of particle size, diffusion of water out small particles the build-up is complete within 0.1 s. of the pores has a negligible impact on the NICS In principle, the build-up of the ratio of cross- for most of the adsorbed water and the width of and diagonal peak intensity (Icross/Idiag) in EXSY the in-pore peak is solely due to the distribution spectra can be described by a single dependence of average NICS. However, exchange averaging on tanh(ktmix), where k is the exchange rate has a measurable impact on the ex-pore peaks. In constant. However, Fulik et al., have shown that both samples they were fitted with a narrow and better agreement is observed if two processes with a broad component, the intensity and full width at different rates are assumed, i.e. a slow process half maximum (FWHM) of which vary. The broad attributed to diffusion from the centre of the particle component is assigned to ex-pore water having to the surface and a much faster process attributed experienced the pores for a period of time and is to effective exchange between ex-pore and in-pore

(a) (b)

–4

–2 d

0 H, ppm

Diagonal Diagonal 2

6 4 2 0 –2 –4 6 4 2 0 –2 –4 d 1H, ppm d 1H, ppm

0.8 0.8

0.6 0.6 diag diag I I / / cross cross Exp I I 0.4 0.4 Exp Fit Fit 0.2 0.2

0 0.2 0.4 0.6 0.8 1.0 0 0.2 0.4 0.6 0.8 1.0 Mixing time, s Mixing time, s

Fig. 6. 2D exchange spectra for: (a) 54BO_80; and (b) 54BO_21 recorded with tmix = 20 m, with (c) and (d) corresponding build-up curves of the ratio of cross- and diagonal peak intensities (Icross/Idiag) as a function of mixing time

(15). However, over the course of our experiments, where I0 is an additional constant introduced it was observed that the ex-pore resonance reduced to account for t1 noise giving rise to spurious in intensity due to evaporation from the NMR rotor off-diagonal low-intensity signal at zero mixing leading to a global reduction in Icross/ Idiag for EXSY time. The fits were optimised by minimising the spectra recorded at the end of the series with root mean squared deviation (RMSD) between long mixing times. Although the exact kinetics of calculated and experimental points, to converged the solvent evaporation are complex and were values around 10–2. The errors on the data points not studied in detail, we found that this could be were calculated from the signal-to-noise ratio accounted for with sufficient accuracy through of each peak for a selection of 2D spectra and the incorporation of an exponential term with a propagated to Icross/Idiag, and were found to be characteristic decay constant Tevap (see SI). Rate smaller than 0.02%. The best fit for 54BO_80 was constants were therefore extracted from the EXSY obtained with k1 = 57 Hz and k2 = 4 Hz, and for data using Equation (i): 54BO_21 with k1 = 597 Hz and k2 = 50 Hz. The

carbon particles with D50 = 21 µm showed around IIcrossd/ iagm=+()Aktanh( 12*)tBix tanh(kt*)mix − (i) 10 times faster in-pore–ex-pore diffusion versus the tTmixe/ vap + + **()Ce DI0

D50 = 80 µm particles. This is close to the factor of 15 which we obtained from our simple calculations based on spherical particles (see Figure S4) and shows the impact of the particle size on the rate of the exchange processes, with larger particles significantly reducing the exchange kinetics between the in-pore and ex-pore environments. Therefore, particle size is an important factor to take into account when comparing in-pore–ex-pore exchange phenomena in porous carbon samples. 8 6 4 2 0 –2 –4 –6 Regarding the in-pore resonances, the shape and d 1H, ppm position in the NMR spectrum is largely unaffected Fig. 7. 1H NMR spectra of YP50 wetted with 0.5*PV by in-pore–ex-pore exchange despite the difference (red) and 5.3*PV (blue) of water. The dashed lines in particle size. We can therefore assume that the show the maxima of the in-pore peaks diffusion path out of the particle from any point of the pore network, apart from the very surface, is simply too long and exchange-averaging of the in‑pore a relatively flat valley between the two broad peaks: environment is in the slow regime. This would mean even a narrow particle size distribution in the few- that if the particle size is decreased further, at some micrometre range gives a very broad distribution point the in-pore peak should also become affected of exchange rates, therefore generating exchange- by faster diffusion of in-pore water into the ex‑pore averaged peaks at a continuum of shifts. environment. To test this hypothesis, 6 µm-sized One point worth addressing is whether the YP50 particles were wetted with deionised water. inhomogeneous broadening of the in-pore peak is YP50 is a commercial activated carbon, and responsible for broadening the exchange-averaged similar to our PDCs with respect to composition, ex-pore peak, which can give an idea of the reliability of pore size distribution and average pore size (see the exchange rate constants obtained. The spectrum Figure S13). The 1H NMR spectra before and after of Figure 7 was successfully reproduced in Express saturation are shown in Figure 7. As expected software (22) in the case of a homogeneous as due to the small particle size, an ex-pore peak well as a inhomogeneous broadening of the broad with sharp and broad components is observed. ex‑pore peak, so the simulation does not allow it to Relative to 54BO_21, the broad component is be determined whether the exchange-averaged peak shifted five times more, consistent with smaller is homogeneously or inhomogeneously broadened. inter-particle voids that facilitate adsorption of Refer to SI for further detail. However in 2D EXSY ex-pore water in close proximity. The in-pore spectra, for any tmix, the ex-pore peak shows peak shifts from a NICS of 7.3 ppm to 7.0 ppm symmetrical off-diagonal broadening meaning the upon saturation, which is similar to 54BO_21, broadening is homogeneous. This suggests that however the intensity decreases significantly, and one component of the in-pore peak is much more we notice that the peak exhibits a tail towards the exposed to the ex-pore environment than the ex-pore peak. This means that after saturation, a other components. In consequence the values of significant proportion of adsorbed water is able to exchange rate constants may be more reliable quickly diffuse into the ex-pore environment and than if the broadening were inhomogeneous. therefore does not appear at the purely in-pore This result is consistent with a radial distribution chemical shift, but rather intermediate between of NICS in the particle, in agreement with our the ex-pore and the in-pore chemical shifts. The previous observations where the centre of particles exchange rate constants were estimated to be of diameter greater than 100 µm is poorly affected k1 = 1117 Hz and k2 = 165 Hz. Assuming that k1 by steam activation. The particles employed here relates to the ex‑pore–in-pore exchange process were smaller than 100 µm but it seems that a small that we calculate as a function of the particle gradient of activation still remains. It is possible that size (Figure S4), we find that 4 µm particles give the valley between ex-pore and in-pore is the result a similar exchange rate. This is good agreement of in-pore–ex-pore exchange broadening, in the considering that 20% of YP50 particles are smaller fast or slow regime, of the other components of than 2.5 µm (Figure S14), especially because for the in‑pore peak, which are located further and particles smaller than 10 µm the exchange rate further away from the surface, and therefore have exch increases sharply. This explains why YP50 presents access to smaller and smaller volumes Vex .

In summary, from these observations it could be deduced that particles smaller than 10 µm offer (a) a diffusion path short enough for water to diffuse out of the pores at a rate that affects the in-pore peak as well. However, given that the diffusion coefficient of adsorbed water depends on the pore size, we cannot compare 54BO_21 and YP50 because they have different average pore sizes. Using the equation provided previously (6), these can be estimated using the NICS before saturation. We obtain for 54BO_21 and YP50, 1.17 nm and (b) 1.10 nm respectively. Therefore, it is necessary to investigate the effect of average pore size, which can be tuned by controlling the BO, on the diffusion of water and the appearance of the spectra.

The Effect of Burn-off

The effect of BO, and thus average pore size, on 6 4 2 0 –2 –4 –6 1 the dynamics of water in PDCs can be described d H, ppm by comparing 54BO samples (high BO) with 20BO Fig. 8. 1H NMR spectra of: (a) 20BO_83 injected samples (low BO). Figure 8(a) shows samples with 0.7*PV (red) and 3.6*PV (blue) of water; and (b) 20BO_15 injected with 0.2*PV (red) and 20BO_83 and Figure 8(b) shows 20BO_15 3.6*PV (blue) of water. The dashed lines show the injected with water. The NICS of samples 20BO maxima of one component of the in-pore peaks are 8.4–9.0 ppm, giving an average pore size less than 1.0 nm based on the previous equation (6), which is smaller than sample 54BO, in agreement appeared much less affected by diffusion. However, with the linear dependence of the average pore one must keep in mind that the exchange regime size on the BO (18, 19, 23). The in-pore peaks depends not only on the actual rates, but also on the appear to contain two components as is sometimes chemical shift difference of the two environments observed for samples with low BO. Possibly, the involved. The NICS of samples 20BO is higher than degree of activation was not homogeneous from in samples 54BO, so even identical exchange rates the surface to the centre of the particles. In sample still situate 20BO in a slower regime than 54BO.

20BO_83, the in-pore peak is constant regardless For the small particles 20BO_15, k1 = 162 Hz of the injected volume (shift < 0.10 ppm). Half the and k2 = 6 Hz, which are three times as fast when ex-pore peak was fitted with a sharp component compared to 20BO_83. However, the calculated of FWHM < 0.1 ppm, and half with a broad factor was 31, which is a clear discrepancy. Given component of FWHM = 0.16 ppm with a difference that sample 20BO_15 contains particles smaller in chemical shift < 0.10 ppm. This suggests in the than sample 54BO_21, we can attribute the first instance that the in-pore–ex-pore exchange discrepancy to a BO effect and not to a particle process is in a slower regime than in 54BO_80, size effect. Besides the average pore sizes, other where the broad component was broader and the structural parameters that could perhaps vary with in-pore peaks were shifted slightly, in agreement BO are oxygen content and tortuosity due to the with the study mentioned earlier (24). With small smaller amount of mesopores. These observations particles 20BO_15, 90% of the ex-pore peak is also raise the question of homogeneity of the broad, FWHM = 0.35 ppm but within 0.10 ppm of diffusion coefficients within the particle: it is likely the bulk water, suggesting again limited exchange that at low BOs a bigger gradient of activation exch with negligible Vin . within the particle is present. The exchange rate constants provide a more Overall, these results show that a smaller average quantitative description of the impact of average pore size hinders exchange as opposed to a smaller pore size. For the big particles 20BO_83, k1 = 54 Hz particle size, which promotes it. Since YP50 has a and k2 = 4 Hz, which are comparable to the values smaller average pore size than 54BO_21 but higher for 54BO_80 (57 Hz and 4 Hz). This was perhaps exchange rate constants, we can safely deduce not expected from the 1D spectra, where 20BO_83 that in YP50, the particles are truly small enough

160 © 2020 Johnson Matthey https://doi.org/10.1595/205651320X15747624015789 Johnson Matthey Technol. Rev., 2020, 64, (2) to allow for in-pore–ex-pore exchange to affect the in-pore peak. These conclusions are in agreement with simulations (25), and also an experimental study focusing on diffusion measurements in hydrophobic slit pores by neutron-scattering (23). The material contained approximately 95% carbon with 5% oxygen atoms on the surface of the pores and average pore sizes of 1.2 nm and 1.8 nm, which is similar to our PDCs. It was found that the diffusion coefficients of water in the 1.2 nm 4 2 0 –2 –4 –6 –8 and 1.8 nm pores are 40% and 30% smaller than d 1H, ppm that of bulk water, respectively. Furthermore, Fig. 9. 1H NMR spectra of YP50 injected with on the very surface of pore, water diffuses at 0.3*PV (red) and 2.9*PV (blue) of cyclohexane. 0.035 × 10–5 cm2 s–1 and 0.014 × 10–5 cm2 s–1 The dashed lines show the maxima of the in-pore respectively, which is two orders of magnitude peaks slower than in the bulk. Counterintuitively, the value on the surface of the bigger pores was found to be nearly half that in the smaller pores, of YP50 wetted with cyclohexane. As expected which was attributed to the promotion of slightly with such high exchange rate constants, we see a faster concerted motion in extreme confinement, narrow and a broad ex-pore peak, and an in-pore although the centre of the pore followed an overall peak that decreases and shifts upon saturation, slower regime. much like in Figure 7 with water in YP50. With water and cyclohexane, the in-pore peaks are The Effect of Solvent Properties similar; they shift by 0.30 ppm and are broadened by a factor two upon saturation. Interestingly, we Many applications of porous carbons (such as note that the NICS of cyclohexane (and hexane) is electric double-layer capacitors) commonly employ smaller than water by 0.2–0.3 ppm, which means electrolytes in organic solvents as an alternative that the apolar solvents are on average located to aqueous electrolytes. To extend the knowledge in slightly bigger pores than water. The diffusion presented here to such systems, it is desirable to be being faster in big pores, this is in contrast with the able to predict how the NICS may be affected. The slower exchange kinetics measured, showing that viscosity and the polarity are the two parameters solvent parameters are prevalent over pore size that will be considered here as tools to predict the effects. The broad ex-pore peak is narrower and diffusion regime adopted by any solvent. less shifted in cyclohexane (FWHM = 0.64 ppm and The effect of polarity can be estimated by shift = 0.27 ppm) than in water (FWHM = 1.69 ppm comparing water and cyclohexane which have the and shift = 0.98 ppm), consistent with smaller same viscosity, but different dipole moments. YP50 exchange rate constants. is chosen for this comparison because the exchange The effect of viscosity can be observed by comparing rates are high enough to significantly impact the cyclohexane (0.89 cP) and hexane (0.30 cP) which NMR spectra. The exchange rate constants were have the same very low polarity. Figure 10 shows measured to be 1117 Hz and 165 Hz for water, the 1H NMR spectrum of YP50 wetted with hexane. and 257 Hz and 23 Hz for cyclohexane, which Unlike water and cyclohexane, hexane shows two means that cyclohexane diffuses roughly four ex-pore peaks, which are simply due to the two times slower. On one hand, this could be related visible proton environments on the molecule; to increased van der Waals interactions with the 1.28 ppm are the methylene (CH2) and 0.88 ppm hydrophobic surface of the pores, as was also found are the terminal methyl (CH3) protons. The in‑pore to be the case in a xerogel using the same solvents peak before saturation can be fitted with two broad (24). On the other hand, the bulk self-diffusion peaks 0.30 ppm apart, which is similar to the coefficient of cyclohexane is smaller than for water difference between the CH2 and CH3 chemical shifts (1.42 × 10–5 cm2 s–1 vs. 2.3 × 10–5 cm2 s–1) (26), (refer to SI for details on the fits). This suggests which could also contribute. At this stage it is that all protons of adsorbed hexane molecules unclear whether other parameters such as the roughly experience the same NICS, therefore the different molecular sizes of water and cyclohexane molecule is either tumbling isotropically or aligned play a role. Figure 9 shows the 1H NMR spectrum with the pore wall. The exchange rate constants

hexane and also of cyclohexane in the other PDC samples, which goes against the trends. Another interesting observation was that there were two in-pore peaks in 20BO before saturation and both peaks shifted to the right upon saturation, meaning exchange-averaging takes place within different pores but not with ex-pore, while in 54BO there was only one in-pore peak (see Figures S5 and S6). We believe these observations are all 4 2 0 –2 –4 –6 –8 related and hint towards the ability of cyclohexane d 1H, ppm to form organised structures in slit-like micropores, Fig. 10. 1H NMR spectra of YP50 injected with which affects its diffusion coefficient. The diffusion 0.3*PV (red) and 3.0*PV (blue) of hexane. The coefficient of confined cyclohexane was shown by dashed lines show the maxima of the in-pore Fomin et al. to drop to zero in slit pores smaller peaks than 2.1 nm, and its density to increase two-fold from 2.6 nm to 1.2 nm pores (27). Another study showed that cyclohexane forms a monolayer in are k1 = 784 Hz and k2 = 104 Hz for hexane in 0.8 nm pores and a bilayer in 1.0 nm pores with YP50, which is approximately three times the rates a denser hexagonal packing structure resembling seen for cyclohexane. Therefore in this case, the the solid phase (28). Similar behaviour has also exchange rate was roughly proportional to the been observed for propylene carbonate which viscosity, as hexane is three times less viscous was observed to form an ordered structure upon than cyclohexane. nanoconfinement (29). In a disordered structure More interestingly, the broad ex-pore peak is the packing is expected to be less efficient, shifted by 0.61 ppm relative to the average between nonetheless the diffusion coefficient may still be the two narrow peaks, and the FWHM = 1.30 ppm. significantly lower. The exchange rate constants These values are between the values for determined take into account the diffusion of exch cyclohexane and water. This is consistent with the species in all types of pores located in Vin . shift and width being proportional to the exchange When cyclohexane forms an immobile structure in rate constants as they are even higher for water the smallest pores, only the diffusion coefficient in in YP50. Overall, pronounced solvent exchange the biggest pores where it is still liquid contributes affects the ex-pore and in-pore peaks and the to the exchange rate constant, explaining why valley between the two peaks, but the ex-pore peak in 20BO_15 the exchange rates are much higher seems to be the most reliable indicator to estimate than in the other PDC samples. The amount of at a glance perturbations due to exchange effects. mesopore is smaller than in 54BO and YP50, so the Further discussion is required about parameters long-range diffusion within the particles is probably that could not be assessed with this set of significantly slower. In that regard, the gradient of experiments, for example the kinetic diameter NICS will be less well averaged, and even more so of the solvent molecules. This factor was not with slow diffusing solvents like cyclohexane. The considered separately because it was first assumed 3.3 ppm NICS of the small peak gives a pore size to be reflected in the viscosity parameter, although of 2.2 nm and the 7.7 ppm NICS of the main peak in pores of similar size, viscosity and diameter may 1.0 nm. The small in-pore peak is therefore likely have independent contributions to the exchange to come from few isolated mesopores while the kinetics. In addition, to identify the impact of this other from micropores and mesopores in contact. factor alone, two solvents with similar viscosity and polarity but different molecular sizes should Conclusions be compared. The kinetic diameter of water and cyclohexane perhaps contributes to the exchange The series of exchange experiments in this work rate difference. provide a quantitative description of the in-pore– The behaviour of cyclohexane was peculiar in the ex-pore exchange rate constants of water, hexane case of sample 20BO_83 and 20BO_15. In the big and cyclohexane adsorbed in PDC samples. This particles, exchange was too slow to be observed, allowed us to distinguish the contributions of and by decreasing the particle size, the exchange particle size, porosity and solvent properties. became visible but the rate was higher than that of We showed that exchange rate constants are

162 © 2020 Johnson Matthey https://doi.org/10.1595/205651320X15747624015789 Johnson Matthey Technol. Rev., 2020, 64, (2) higher in small particles and increase with BO. 6. Y.-Z. Xing, Z.-X. Luo, A. Kleinhammes and Y. Wu, Carbon, 2014, 77, 1132 We also showed that the exchange rate constants are solvent dependent and increased in the 7. S. Hwang and J. Kärger,Magn. Reson. Imaging, 2019, 56, 3 order cyclohexane < hexane < water, which was attributed to differences in viscosity and stronger 8. F. Furtado, P. Galvosas, M. Gonçalves, van der Waals interactions between apolar solvents F.-D. Kopinke, S. Naumov, F. Rodríguez-Reinoso, and the pore walls. However some discrepancies U. Roland, R. Valiullin and J. Kärger, Micro. Meso. Mater., 2011, 141, (1–3), 184 were noted in low activated samples saturated with cyclohexane and assigned to its unusual packing in 9. T. M. Alam and T. M. Osborn Popp, Chem. Phys. micropores. More work is necessary to understand Lett., 2016, 658, 51 the contribution of other solvent properties such 10. A. C. Forse, J. M. Griffin, C. Merlet, J. Carretero- as the kinetic diameter. Based on these findings, Gonzalez, A.-R. O. Raji, N. M. Trease and C. P. Grey, we were able to rationalise the width and shift of Nature Energy, 2017, 2, (3), 16216 the ex-pore and in-pore peaks in 1H NMR spectra. 11. S. Macura, Y. Huang, D. Suter and R. R. Ernst, The appearance of the ex-pore peak in particular J. Magn. Reson., 1981, 43, (2), 259 provides a quick and reliable estimate of the extent 12. M. H. Levitt, “Spin Dynamics – Basics of Nuclear of exchange-averaging. For studying adsorption Magnetic Resonance”, John Wiley and Sons Ltd, phenomena in porous carbons by NMR spectroscopy, Chichester, UK, 2001, 686 pp we recommend the use of 50–100 µm activated 13. A. D. Bain, Prog. Nucl. Magn. Reson. Spectrosc., carbon particles as well as small, viscous and apolar 2003, 43, (3–4), 63 solvents to minimise exchange and to observe the 14. J. M. Griffin, A. C. Forse, H. Wang, N. M. Trease, ‘true’ NICS. P.-L. Taberna, P. Simon and C. P. Grey, Faraday Experimental methods and further details are Discuss., 2014, 176, 49 given in SI. The research data supporting this 15. N. Fulik, F. Hippauf, D. Leistenschneider, S. Paasch, publication can be accessed at Lancaster University S. Kaskel, E. Brunner and L. Borchardt, Energy Research Depository (30). Storage Mater., 2018, 12, 183 16. L. Cervini, O. D. Lynes, G. R. Akien, A. Kerridge, Acknowledgements N. S. Barrow and J. M. Griffin, Energy Storage Mater., 2019, 21, 335 We acknowledge Engineering and Physical Sciences 17. C. Merlet, A. C. Forse, J. M. Griffin, D. Frenkel Research Council (EPSRC) and Johnson Matthey and C. P. Grey, J. Chem. Phys., 2015, 142, (9), Plc for the provision of an iCASE studentship. We 094701 also thank Lancaster University and the European Regional Development Fund (ERDF) for the 18. I. P. P. Cansado, F. A. M. M. Gonçalves, P. J. M. Carrott and M. M. L. Ribeiro Carrott, provision of characterisation facilities under the Carbon, 2007, 45, (12), 2454 Collaborative Technology Access Program (cTAP). 19. R. J. Anderson, T. P. McNicholas, A. Kleinhammes, A. Wang, J. Liu and Y. Wu, J. Am. Chem. Soc., 2010, 132, (25), 8618 References 20. Z.-X. Luo, Y.-Z. Xing, Y.-C. Ling, A. Kleinhammes 1. R. K. Harris, T. V. Thompson, P. R. Norman and and Y. Wu, Nature Commun., 2015, 6, 6358 C. Pottage, J. Chem. Soc. Faraday Trans., 1996, 21. H. Wang, A. C. Forse, J. M. Griffin, N. M. Trease, 92, (14), 2615 L. Trognko, P.-L. Taberna, P. Simon and C. P. Grey, 2. P. von Ragué Schleyer, C. Maerker, A. Dransfeld, J. Am. Chem. Soc., 2013, 135, (50), 18968 H. Jiao and N. J. R. van Eikema Hommes, J. Am. 22. R. L. Vold and G. L. Hoatson, J. Magn. Reson., Chem. Soc., 1996, 118, (26), 6317 2009, 198, (1), 57 3. L. Borchardt, M. Oschatz, S. Paasch, S. Kaskel and 23. S. O. Diallo, Phys. Rev. E, 2015, 92, (1), 012312 E. Brunner, Phys. Chem. Chem. Phys., 2013, 15, (36), 15177 24. C. Cadar and I. Ardelean, Magn. Reson. Chem., 2019, 57, (10), 829 4. A. C. Forse, J. M. Griffin, H. Wang, N. M. Trease, V. Presser, Y. Gogotsi, P. Simon and C. P. Grey, 25. I. N. Tsimpanogiannis, O. A. Moultos, Phys. Chem. Chem. Phys., 2013, 15, (20), 7722 L. F. M. Franco, M. B. de M. Spera, M. Erdős and 5. A. C. Forse, J. M. Griffin, V. Presser, Y. Gogotsi and I. G. Economou, Mol. Simul., 2019, 45, (4–5), 425 C. P. Grey, J. Phys. Chem. C, 2014, 118, (14), 26. M. Holz, S. R. Heil and A. Sacco, Phys. Chem. 7508 Chem. Phys., 2000, 2, (20), 4740

27. Y. D. Fomin, V. N. Ryzhov and E. N. Tsiok, J. Chem. Y. Gogotsi, P. Simon and K. Kaneko, J. Phys. Phys., 2015, 143, (18), 184702 Chem. C, 2013, 117, (11), 5752

28. T. Shimoyama, K. Tashima and M. Ruike, Coll. 30. J. Griffin, ‘Observing Solvent Dynamics in Porous Surf. A: Physicochem. Eng. Asp., 2017, 533, 255 Carbons by Nuclear Magnetic Resonance’, Dataset, 29. M. Fukano, T. Fujimori, J. Ségalini, E. Iwama, Research Directory, Lancaster University, UK, P.-L. Taberna, T. Iiyama, T. Ohba, H. Kanoh, 2019

The Authors

Luca Cervini graduated with a BSc from the Department of Chemistry at the University of Geneva, Switzerland, in 2012. In 2014 he obtained a MSc jointly with the École Polytechnique Fédérale de Lausanne (EPFL), Switzerland, after a thesis on organic photochromic solar cells. In 2016 he started an industrial Cooperative Awards in Science & Technology (CASE) PhD in chemistry at Lancaster University, UK, supervised by John Griffin and Nathan Barrow where he has been studying dynamics in porous materials, specifically aqueous electrolytes in activated carbons using mostly NMR spectroscopy and gas sorption analysis.

Nathan Barrow is currently a Principal Scientist in the Advanced Characterisation department at Johnson Matthey, Sonning Common, UK. He graduated with an MPhys in 2006 from the University of Warwick, UK, where he remained to gain a PhD in solid- state NMR. In 2010 Barrow was a Knowledge Transfer Partnership associate between the University of Warwick and Johnson Matthey, helping to install and run a solid-state NMR service. His current research focuses on applying advanced characterisation to materials such as porous carbons, zeolites, alumina, glasses and battery materials.

John Griffin is a lecturer in Materials Chemistry at Lancaster University. He obtained his PhD from the University of Warwick in 2008 before moving to the University of St Andrews, UK, to carry out postdoctoral research in the group of Professor Sharon Ashbrook. In 2012 he moved to the University of Cambridge, UK, to join the group of Professor Clare Grey FRS. In 2015 he took up his current position where his current research interests concern the development and application of solid-state NMR methodologies for the study of energy conversion and storage materials.

www.technology.matthey.com

Insights into Automotive Particulate Filters using Magnetic Resonance Imaging Understanding filter drying in the manufacturing process and the effect of particulate matter on filter operation and fluid dynamics

J. D. Cooper, N. P. Ramskill, matter (PM) to understand the effect of PM on the A. J. Sederman, L. F. Gladden filter flow profiles and porous wall permeability as Department of Chemical Engineering and soot is deposited. Biotechnology, University of Cambridge, West Cambridge Site, Philippa Fawcett Drive, Introduction Cambridge, CB3 0AS, UK Particulate filters, comprising so-called wall-flow A. Tsolakis filter substrates, are of increasing importance in School of Mechanical Engineering, University of reducing pollutant emissions from vehicles to the Birmingham, Edgbaston, Birmingham, B15 2TT, levels required by legislation. Early legislation UK addressed the emissions of carbon monoxide, hydrocarbons and nitrogen oxides and the E. H. Stitt removal of these pollutants was achieved using Johnson Matthey, PO Box 1, Belasis Avenue, FTM catalysts. Later, when PM emissions from Billingham, Cleveland, TS23 1LB, UK diesel vehicles came under scrutiny, wall-flow filter devices were added to the emissions control A. P. E. York* system: often these were uncatalysed extruded Johnson Matthey, Blounts Court, Sonning cordierite or silicon carbide filter monoliths, Common, Reading, RG4 9NH, UK though sometimes a catalyst was incorporated on the filter to widen their operating window. *Email: [email protected] Wall-flow filters differ in their operation from FTMs, since adjacent channels are alternately blocked meaning exhaust gas must pass through Understanding the manufacture and operation of the porous monolith wall to flow from inlet to automotive emissions control particulate filters outlet; in this way PM is deposited on the inlet is important in the optimised design of these channel walls. Previous PM legislation was based emissions control systems. Here we show how on particulate mass emissions, however, more magnetic resonance imaging (MRI) can be used recently the legislation, such as in Euro 6 (1), has to understand the drying process, which is part of turned to address particulate number emissions. the manufacture of catalysed particulate filters. Due to this change in emphasis both diesel and Comparison between a wall-flow particulate filter gasoline vehicles now require filter systems to substrate and a flow-through monolith (FTM) achieve compliance. In gasoline applications, a has been performed, with MRI giving spatial GPF is typically a wall-flow filter, similar to that information on the drying process. We have also used for diesel, with a catalyst coating applied: used MRI to study the fluid dynamics of a gasoline by combining the catalyst and filter devices, particulate filter (GPF). Inlet and outlet channel multifunctional emissions control systems gas velocities have been measured for a clean have also been developed resulting in reduced GPF and two GPF samples loaded with particulate packaging volume.

In the manufacture of catalysed filter devices, affected by the gas flow fields and will impact the first a catalyst coating is applied to the monolith efficacy and safety of the regeneration process. substrate in the form of a slurry. The subsequent Large thermal gradients within the filter can be coated monolith is then dried and finally calcined formed if the soot distribution is non-uniform to fix the catalyst coating. The drying process that may damage the filter. While modelling and is both energy intensive and known to influence macroscopic measurements of these effects have the final metal distribution within the catalyst been performed by many authors, there is a lack throughout the catalysed filter and therefore also of experimental work focusing on the relationship influences filter performance on a vehicle. Indeed, between the filter structure, the gas transport and it has previously been shown (2, 3) that non- the perturbations of these by the loaded soot. ideal drying in FTMs can result in the macroscopic As with drying, the range of techniques available redistribution of the catalyst. In the work of to non-invasively measure flow in opaque filter Vergunst et al. (3) and Wahlberg et al. (4) it was systems is very limited, and most studies have observed that the metal phase, when not bound used models (7–10). Magnetic resonance (MR) can to the catalyst support, will migrate to the surface provide spatial information on the effect of the PM where evaporation occurs. This results in varying on the filter operation. degrees of inhomogeneous catalyst distribution MR techniques have gained prominence in depending on the method of drying used. Finally, chemical engineering research as they provide a enrichment of the catalyst in the washcoat layer of non-invasive method of studying the chemistry ceramic monoliths during drying has been observed and dynamics of a range of opaque systems (11). using MRI (5). Traditionally, drying is studied using They are particularly useful for the study of porous gravimetric methods, humidity and temperature media such as catalysts (12–16), construction measurements from which the drying kinetics can materials (17–19) and pharmaceuticals (20, 21). be determined (6). Although these techniques are Such techniques have also been used to provide well established in both industry and research, information on the drying mechanism in a they are somewhat limited in that they are only range of other applications such as detergent able to provide macroscopic measurements and powders (22, 23), dehydration and preparation of the process itself must be treated as a ‘black foodstuffs (24–26), evaporation from contaminated box’. Spatially resolved information has most surfaces (27–29) and fired-clay brick at elevated often been obtained by sectioning the sample temperature (30). Applications of MR to drying and and then weighing the individual components. To sorption have been covered in the review by Koptyug gain a greater understanding of the intrinsic water (31). While MR has previously been used to study migration characteristics during drying, in the first drying and active component distribution in FTMs part of this work we use MRI to image the time- by Koptyug and coworkers (5, 15), no studies have resolved water distribution during the drying. The looked at the drying process in particulate filters to data for the filter are compared with the same data date. This study investigates how the structure of acquired during drying of the related FTM. the filter substrate influences the water migration The second part of this work uses MRI methods behaviour during drying. MR techniques are used to measure the gas velocity in the channels of a to characterise water migration behaviour in a wall- clean and particulate loaded filter. During the flow filter and FTM substrate under identical drying operation of an automobile, PM-laden exhaust conditions. While MRI of liquids and their transport gas passes through the particulate filter and the is common, imaging studies of nuclear magnetic particulate or soot is deposited inside. However, resonance (NMR) active gas flows are relatively the micro- and macroscopic distribution of this few. This is mainly due to challenges associated soot deposition impacts the subsequent filtration with the low signal-to-noise ratio (SNR) presented behaviour, pressure drop, regeneration behaviour to the experimentalist. While hyperpolarised gases and ultimately the useful lifetime of the filter. (for example, xenon-129) offer a significant boost to Hence, a complete understanding of the filtration the SNR, they are costly and unsuitable for studying process is needed in order to optimise the function porous materials. The first application of imaging of particulate filters. The regeneration behaviour thermally polarised gases was demonstrated of particulate filters is coupled with the gas fluid by Koptyug et al. (32–34) who acquired two- dynamics. Both heat transfer and the mass transfer dimensional (2D) velocity images of hydrocarbon of oxidative species (oxygen for active regeneration gases at atmospheric pressure flowing through a and nitrogen dioxide for passive regeneration) are cylindrical pipe and alumina monoliths of different

166 © 2020 Johnson Matthey https://doi.org/10.1595/205651320X15754757907469 Johnson Matthey Technol. Rev., 2020, 64, (2) channel geometries. In the monolith studies (32), where γ is the gyromagnetic ratio which is from the spatially resolved profiles of the axial characteristic of the nuclei under observation. component of the velocity vector on the individual Considering the system in the rotating frame at channel scale, information pertaining to shear rates reference frequency, ω0, spatial dependence of and entry lengths were obtained which enabled the precession frequency is achieved by applying useful insights into the mass transfer between a spatially-dependent magnetic field gradient G( ) the bulk gas flow and the porous channel walls in and can be expressed as follows (Equation (ii)): the monolith to be made. Codd and Altobelli (35) ω(r) = γG∙r (ii) have also shown the application of using thermally polarised gases as a probe of the structure of where ω(r) is the precession frequency at position porous materials. Ramskill et al. (36) performed vector r. The precessing nuclei in a volume element an early study using sulfur hexafluoride gas to induce a voltage in the receiver coil and the image flow profiles inside a clean emissions control complex NMR signal, S(t), is detected in the time filter. They implemented compressed sensing (CS) domain. The time-dependent signal is then Fourier techniques, which enable spectra and images to transformed into the frequency domain where it is be reconstructed with sufficient accuracy from represented by the spin-density function, ρ(r). The relatively few data points and allow a reduction in Fourier conjugate relationship between the time data acquisition times (37–41). Here we show a and frequency domain of the NMR signal is shown continuation of this study with more commercially below in Equations (iii) and (iv): relevant samples and with particulate loading. S(t) = ∫∫∫ρ(r)exp[iγG∙rt]dr (iii) This study highlights two ways in which MRI can be used to gain insight into automotive particulate ρ(r) = ∫∫∫S(t)exp[–iγG∙rt]dt (iv) filters. First a comparison of water migration during drying within a wall-flow particulate filter is From this, a spin-density map, i.e. an image, is reported and compared with the analogous process obtained by taking the modulus of the complex occurring within a FTM. The mechanism of water function, ρ(r). In the case of the drying experiments migration during drying will influence catalyst the spin-density map provides a spatially-resolved distribution in the manufactured filter and valuable measurement of the water content within the insights are gained by being able to image this monolith sample. process in more than one dimension. Second, MRI is Magnetic field gradients can be used to make used to image gas fluid dynamics within a GPF both the NMR signal sensitive to nuclei displacements in the clean state and following two stages of soot in addition to position. This is achieved by first loading. From these flow profiles, permeabilities as encoding the nuclei positions by applying a given a function of space and time are predicted. Using magnetic field gradient (G) for time (δ), then MRI to gain a greater understanding of automotive decoding by applying the same magnetic field particulate filters could eventually lead to improved gradient in the opposing direction after an evolution catalyst properties, filtration efficiency and more time (Δ). Any static nuclei will be unchanged but efficient and improved manufacturing. any moving nuclei will create a phase shift in the NMR signal, ΔΦ proportional to their displacement Principles of Magnetic Resonance Δr (Equation (v)): Imaging ΔΦ = γδΔ(G∙Δr) (v)

This section provides the reader with a brief Hence the gas velocity can be measured by introduction to the principles of MRI but the consideration of the signal phase at each spatial interested reader is directed to the texts by location. Callaghan (42) and Haacke (43) and review articles by Mantle and Sederman (44) and Caprihan and Materials and Methods Fukushima (45) for further detail. When NMR active nuclei (such as 1H or 19F) are Comparison of Drying in a placed in an external magnetic field B( 0), the nuclei Particulate Filter and a Flow Through will precess at a characteristic frequency known as Monolith the Larmor frequency (ω0) as given by Equation (i): A cordierite wall-flow particulate filter and a

ω0 = γB0 (i) cordierite FTM were used. These are typical of

Table I Properties of the Diesel Particulate and outlet of the drying cell were recorded at 10 s Filter and FTM Substratesa intervals over the course of the process using a Wall-flow FTM Humidiprobe (Pico Technology, UK). Temperature Material cordierite and RH measurements were recorded with an Length (L), mm 75 accuracy of ±0.5°C and ±2%, respectively. These measurements allow the total uptake of moisture Core diameter, mm 26 by the air to be determined and thus the drying Channel hydraulic 1 diameter, mm rate can be calculated through conservation of the total water mass (36, 46). Substrate porosity (ε), % 48±4 24±2 Mean pore size, µm 13.8±7.8 2.8±1.1 Effect of Soot Loading on Gas Fluid Water content (mc), g 6.1±0.1 2.8±0.1 aThe mean pore size and porosity were determined by Dynamics in a Gasoline Particulate mercury porosimetry. Filter

A cordierite GPF sample was prepared in the laboratory for this study. The cordierite substrate 26 mm (55% porosity) was coated with a Pd/Rh alumina G three-way catalyst typical of commercial catalysts used for GPF applications. The properties of the sample are shown in Table II, with the porosity E 70 mm and pore size of the catalyst coated filter listed for 120 mm H the front, middle and rear of the sample. Samples were soot loaded using a 2 l, four-cylinder D F gasoline direct injection (GDI) turbocharged engine, and then subsequently removed and y transferred to a different sample holder for the B A C MRI flow experiments. Such engines are typical of z current passenger automobiles. The engine was Fig. 1. Schematic of the experimental setup for run at 2100 rpm, producing a torque of 60 Nm. the drying experiments. A = compressed air line, Three 25 mm diameter cores were bored from the B = pressure regulator, C = rotameter, D = air centre of the GPF, each acting as a sample for the distributor plate, E = wall-flow filter or FTM soot loading. The filter sample to be loaded was substrate sample, F = imaging region, G = MRI held downstream inside the exhaust manifold. The spectrometer and H = magnetic field gradients manifold was surrounded by a furnace, allowing the filter to be held at different temperatures. substrates used commercially and the relevant A temperature of 300°C was chosen as properties of the two samples are listed in Table I. representative of real-world gasoline exhausts; the Porosity and mean pore size measurements were temperature was measured using a thermocouple made using an AutoPore IV system mercury porosimeter (Micromeritics Instrument Corporation, Table II Properties of the GPF Samplea USA). Drying of pure water from the respective Bare filter Catalyst substrates has been investigated under identical coated filter conditions i.e. with air at 20 l min–1 ± 2 l min–1 and Material cordierite temperature of 19.5°C ± 0.5°C. To saturate the samples, the substrates were Length (L), mm 145 immersed in deionised water for two minutes and Core diameter, 25 mm for soot loading, mm 6 mm for MRI then shaken to remove any water blocking the Channel hydraulic channels. A schematic of the experimental set up 1 diameter, mm used is shown in Figure 1. The filter substrate was Substrate 28.1, 25.9, ® 58 held within a PERSPEX cell (Perspex International porosity (ε), % 29.2 Ltd, UK) above an air distributor plate used to Mean pore size, 21 19, 16, 18 produce a uniform flow of air over the cross- µm sectional area of the filter. The relative humidity aPorosity and pore size are given for the front, middle and (RH) and temperature of the air flow at the inlet rear of the sample.

3.5 s and eight scans for signal averaging were Protocol II 14 used, resulting in a total acquisition time of Protocol III 0.5 min per spectrum 12 • 2D images were acquired over the course of drying using the rapid acquisition with relaxation 10 enhancement (RARE) pulse sequence (47). , kPa P 8 Images were acquired in the yz plane with a Δ slice width of 10 mm, an in-plane field-of-view 6 (FOV) of 80 mm × 30 mm and a data matrix symbol size of 32 × 32, giving an in-plane pixel 4 resolution of 2.5 mm px–1 × 0.94 mm px–1 in the 0 20 40 60 read (z) and phase (y) directions respectively. A Time, min RARE factor of four with eight scans were used, Fig. 2. Pressure drop measurements for the GPF allowing acquisition of a full image in 3.5 min sample subject to particulate loading Protocols II • One-dimensional (1D) profiles in the axial and III (z) direction were acquired using a spin-echo profiling sequence that integrates the spin inserted into the manifold upstream of the filter. density along the x and y directions (42). A FOV Two pressure transducers were placed either of 80 mm in the z direction and a matrix size of side of the filter sample, allowing measurement 128 points was used, giving a spatial resolution of the pressure drop during the loading process. of 0.625 mm px–1. Eight scans were used for Measurements were made at 180 ms intervals signal averaging, giving a total acquisition time at both transducers. The pressure readings were of 0.5 min. The echo time between excitation subtracted and averaged over 2 min intervals to and acquisition of the NMR signal was 10 ms, give the transient pressure drop. Three particulate giving a maximum error of 3% for the relaxation loading protocols were used: times present in the system. • Protocol I: no soot loading The MRI method used to measure gas velocity • Protocol II: normal running of the engine for in the filter samples is described fully in

50 min Ramskill et al. (36). In the present study SF6 has • Protocol III: Protocol II followed by 10 min of been chosen as the NMR active gas to be used for accelerated soot loading achieved by delaying velocity imaging due to its favourable MR properties the fuel injection by a crank-shaft angle of in comparison with other potential candidates 50 degrees. such as the hydrocarbon gases (48). Eleven The backpressure profiles recorded for Protocols II images were acquired along the length of the GPF and III are shown in Figure 2. samples, each with a slice width of 6 mm. An SF6 gas pressure of 5.0 barg ± 0.1 barg and mass Magnetic Resonance Characterisation flow rate of 16 g min−1 was used for each sample. For the drying experiments, the MR experiments Axial velocity profiles were acquired for the GPF were performed using a 2 Tesla (85 MHz for 1H) samples after all three soot loading protocols. The horizontal bore magnet controlled by an AV mean volume flow for each sample agreed with the spectrometer (Bruker Corporation, USA). An value calculated from the mass flow rate to within 85 mm radio frequency (rf) coil tuned to a frequency 8.5%. The through-wall velocities were calculated of 85.1 MHz was used for excitation and signal for each based on the gas mass balance. Velocity detection and spatial resolution was achieved with profiles inside the inlet channels were extracted magnetic field gradients with a maximum strength from the MR velocity images through the mid-point of 10.7 gauss cm–1. Three MR techniques were of the channels parallel to the filter wall. used as follows. • NMR spectroscopy was used to provide a Results and Discussion quantitative measurement of the bulk water content through calibration with gravimetric Comparison of Drying in a measurements. Due to the short deadtime Particulate Filter and a Flow Through between rf excitation and detection, negligible Monolith relaxation weighting is associated with the spectra and hence they are directly proportional 2D images in the yz plane of the wall-flow filter and to the spin density of water. A recycle time of FTM have been acquired over the course of drying as

(a) (b) (c) (d) (e) (f) High Signal intensity, arbitrary unit 75 mm

Low

y A B C Fig. 3. 2D images over the course of the drying of a wall-flow filter at: (a) 3.5 min; (b) 10.5 min; (c) 17.5 min; (d) 24.5 min; (e) 30.5 min and (f) 36.5 min. The signal intensity has been normalised relative to the maximum signal intensity in the filter substrate. Air flow is from the bottom with a volumetric flow rate of 20 l min–1. These images were acquired with a FOV of 80 mm × 30 mm in the zy plane corresponding to a spatial resolution of 2.5 mm × 0.94 mm

(a) (b) (c) (d) (e) (f) High Signal intensity, arbitrary unit 75 mm

Low

y A B C Fig. 4. 2D qualitative images over the course of the drying of a FTM at: (a) 3.5 min; (b) 7 min; (c) 10.5 min; (d) 14 min: (e) 17 min and (f) 20.5 min. The signal intensity has been normalised relative to the maximum signal intensity in the filter material. Air flow is from the bottom with a volumetric flow rate of 20 l –1min . These images were acquired with a FOV of 80 mm × 30 mm in the zy plane corresponding to a spatial resolution of 2.5 mm × 0.94 mm shown in Figure 3 and Figure 4 respectively. The the wall-flow filter substrate, the behaviour of the initial water distribution along the two substrates water saturation at the three positions across the has been determined by integrating the signal monolith as a function of time show very similar intensity in the first image of each sequence behaviour. This result confirms that the 1D profiles (Figure 3(a) and Figure 4(a)) in the y-direction. in the axial (z) direction are sufficient to be able These data are shown in Figure 5. A uniform to study the drying mechanism in the wall-flow wetting of the channels of both substrates is seen filter and FTM and thereby allow the process to be with an area of relatively high signal intensity studied at a higher temporal resolution than would corresponding to the higher moisture content be permitted using 2D MRI. contained in the plugs of the wall-flow filter Figures 7 and 8 show the quantitative drying substrate. Figure 6 shows the moisture content curves and rate of drying for the filter and FTM, at the three radial positions (marked as A, B and C respectively. The water content data shown in on Figures 3 and 4) as a function of drying time. Figure 7(a) and Figure 8(a) are quantitative and Apart from the longer drying time associated with obtained directly from the traditional measurements

(a) (b)

2.0 1.0

1.5 0.8

0.6 1.0 0.4 0.5 0.2 Signal intensity, arbitrary unit arbitrary intensity, Signal Signal intensity, arbitrary unit arbitrary intensity, Signal

0 20 40 60 80 0 20 40 60 80 z, mm z, mm Fig. 5. 1D profiles of signal intensity in: (a) the wall-flow filter obtained though numerical integration of the 2D images in the radial (y) direction; (b) FTM obtained though numerical integration of the 2D images in the radial (y) direction

(a) (b) A 1.0 1.0 B C 0.8 0.8

0.6 0.6

0.4 0.4

0.2 0.2 Relative signal intensity Relative Relative signal intensity Relative

0 10 20 30 40 50 0 10 20 30 Time, min Time, min

Fig. 6. Relative saturation taken at three radial (y) positions (A, B and C) from the 2D images (Figure 3(a) and Figure 4(a)) plotted over the course of drying for: (a) the wall-flow filter; and (b) FTM. Drying appears to be uniform in the radial direction

(RH and temperature measurement) and integration of the NMR signal. As the sensitivity of the NMR of the signal from NMR spectroscopy; the error signal depends on the population difference of bars were calculated based on the instrument nuclear spin energy levels, as described by the sensitivities and the standard deviation of repeat Boltzmann distribution, a reduction in temperature measurements respectively. The rate of drying causes an increase in the observed NMR signal. data shown in Figure 7(b) and Figure 8(b) are During the induction period, a temperature obtained from the time derivative of the data in drop of up to 6 K is observed due to the heat of Figure 7(a) and Figure 8(a). It is seen that for evaporation, resulting in an increase of up to 2% in both the wall-flow filter and FTM the trends in drying the observed NMR signal and hence a reduction in behaviour appear similar; in particular, a slow the calculated drying rate. While this is negligible falling rate period followed by a faster falling rate for the modest temperature changes during most as drying proceeds. The only significant difference of the drying process, the temperature drop during is during the induction period of both samples; the the induction period is sharp and the increase in NMR data shows an increasing rate whereas the NMR signal decreases the measured drying rate by traditional measurements show a decreasing rate. up to 20%. Thus, from a quantification standpoint This is attributed to the temperature dependence we have successfully been able to validate the MRI

(a) (b) 7 4 Relative humidity/ temperature 6 Magnetic resonance 3 × 5 –1

4 2 3

2 1 Moisture content, g 1 of drying, mg s Rate 0 0

0 10 20 30 40 50 0 10 20 30 40 50

Time, min Time, min Fig. 7. (a) Drying; and (b) rate of drying curves for the wall-flow filter as determined from the RH/temperature and MR measurements. An air flow rate of 20 l min–1 was used

(a) (b) 4 3 Relative humidity/ temperature Magnetic resonance 3 –1

2 2

1 1 Moisture content, g Rate of drying, mg s Rate

0 0 0 10 20 30 0 10 20 30 Time, min Time, min

Fig. 8. (a) Drying; and (b) rate of drying curves for the FTM as determined from the RH/temperature and MR measurements. An air flow rate of 20 l min–1 was used

using simple humidity measurements; however, the drying front develops and the front of the filter the MRI is able to give spatial information, as will will dry more quickly than the middle and back be shown. sections. Between 20 min and 35 min, the middle For the wall-flow filter and FTM respectively, and back sections of the filter continue to dry at Figures 9 and 10 show the time evolution of the the same rate until the drying front reaches the 1D profiles during the drying process, with 15 mm middle section and begins to dry more quickly than slices extracted from the data to show the average the back section. Finally, the remaining water is water saturation time evolution at 20 mm, 40 mm removed as the drying front moves through to the and 60 mm from the front of the substrates. In the back of the filter and is completely dry after 50 min. case of the filter substrate (Figure 9), it can be In contrast, as is seen in Figure 10, for the FTM seen that drying proceeds uniformly in the axial (z) the drying front propagates through the substrate direction up to a critical point at which a developing from the very beginning of the drying process. drying front is present until the filter is dry. In the The most striking result from these studies is that initial stages, the rate of drying at the three axial whilst the spatially-unresolved data (Figures 7 positions is the same until ~20 min, after which and 8) might suggest similar drying characteristics

(a) (b) High 80 20 mm 1.0 40 mm Signal intensity 60 mm 60 0.8

0.6 40 , mm z 0.4

20 Low

Relative signal intensity Relative 0.2

0 10 20 30 40 50 60 0 10 20 30 40 50 60 Time, min Time, min Fig. 9. (a) Time series of 1D axial (z) profiles over the course of the drying of a wall-flow filter; (b) average saturation over three 15 mm slices centred at 20 mm, 40 mm and 60 mm along the length of the sample. Air flow is from the bottom with a flow rate of 20 l min–1. FOV in the z direction was 80 mm and data matrix with 128 points thus corresponding to a spatial resolution of 0.625 mm px–1. The development of the drying front at ~20 min can be clearly seen

(a) (b) High 80 20 mm 1.0 40 mm Signal intensity 60 mm 60 0.8

0.6 40 , mm z 0.4

20 Low

Relative signal intensity Relative 0.2

0 10 20 30 40 0 10 20 30 40 Time, min Time, min Fig. 10. (a) Time series of 1D axial (z) profiles over the course of the drying of a FTM; (b) average saturation over three 15 mm slices at 20 mm, 40 mm and 60 mm along the length of the sample. Air flow is from the bottom with a flow rate of 20 l min–1. The FOV in the z direction was 80 mm and data matrix with 128 points thus corresponding to a spatial resolution of 0.625 mm px–1. It is evident that a drying front is present from the start of drying between the wall-flow filter and FTM substrates, filters, since water migration and transport could the spatially-resolved measurements obtained result in a final catalyst location and properties that from 1D and 2D MR measurements (Figures 6, 9 are different from those expected. and 10) reveal significant differences. These data In order to reconcile the predictions of the therefore provide important information on the bulk and spatially resolved measurements, the microscopic contributions to the drying process structure and operation of each sample must be which must be reflected in any computational considered. FTMs operate through the axial flow model. Furthermore, the differences seen imply of gas through the channels of the structure, that care must be taken when drying particulate with no flow through the substrate itself. Thus,

173 © 2020 Johnson Matthey https://doi.org/10.1595/205651320X15754757907469 Johnson Matthey Technol. Rev., 2020, 64, (2) drying occurs through evaporation at the interface propagation of a drying front in the filter material of the channels and the substrate, with water will create similar issues as have been observed in migrating from the saturated pores to the surface FTMs by Vergunst et al. (3). However, the effect of through capillary flow. Hence the rate of drying the uniform drying regime is uncertain due to the is determined by the humidity gradient in the limited number of drying studies on particulate channels. Air entering the front of the FTM has the filters. Spatially uniform drying methods such as lowest humidity, and so the rate of evaporation is microwave drying (50) produce more homogeneous greatest at the front and decreases along the length distributions of the active phase in the monolith, but of the filter. As water is removed from the front, as with static and mobile air drying, the mobility of the humidity gradient shifts down the channels the catalyst and the solvent leads to non-uniform and the drying front propagates until it reaches distributions. Limiting or ceasing this mobility, the end of the filter as seen in Figure 10(a), either through fast drying protocols, deposition- causing the slowly decreasing drying rate seen precipitation (51) or freeze-drying (3), can improve between 5 min and 18 min in Figure 8(b). Once the homogeneity of catalyst distribution at the most of the moisture has been removed from expense of economic viability. each axial position, the residual is controlled by diffusive mass transport in the porous medium Effect of Soot Loading on Gas Fluid and is slower, resulting in the quickly decreasing Dynamics in a Gasoline Particulate drying rate after 18 min in Figure 8(b). This is Filter consistent with the findings of Koptyug et al. (3) for a segmented monolith, suggesting that axial The velocity profiles measured for the catalyst capillary flow is not significant due to the larger coated wall-flow filter are shown in Figure 11 for pore sizes present in this study. (a) axial and (b) through-wall velocities. Under no In particulate filters, alternate channels are soot loading (Protocol I) the axial velocity profile blocked meaning that gas is forced to flow through resembles those seen previously (36), with the the porous substrate in order to exit the filter, characteristic U-shaped form of the through-wall adding extra complexity to the drying process. velocity profile. After Protocol II of soot loading, The air first removes water in the largest pores in the axial profile has only changed slightly, with the substrate, so the air is humidified in the walls a more linear change in channel velocities and and not the channels, so there is no humidity the cross-over of the inlet and outlet velocities gradient along the length of the sample. These occurring further forward in the filter. A more pores form the paths of least resistance between uniform through-wall velocity is seen, although channels, allowing air to flow through the filter there is still a parabola-like section in the filter walls and become fully saturated. For this air flow centre. After Protocol III, the axial velocity profile rate, the transverse velocity through the walls is shows a linear decrease and increase in the inlet expected to be reasonably uniform (36, 49), and and outlet channel velocity respectively. The so the drying rate in this regime is highly uniform corresponding through-wall velocity profile is across the filter. From Figure 9, this takes around much more uniform. Velocity profiles were then 20 min, at which point the air becomes less than extracted from the MRI measurements for the fully saturated and a humidity gradient can form, central inlet channel of the sample, and show the starting the drying front that propagates through gas velocity radially across the cross-section of the wall-flow filter between 20 min and 50min the single channel to show the evolution of the in Figure 9(a). Figure 7(b) shows no change radial flow profile at different axial positions with at 20 min, indicating that despite the change in increasing soot load (Figure 12): axial positions of drying mechanism for the wall-flow filter, the rate- z/L = 0.14, 0.33, 0.52, 0.70 and 0.87, referred to limiting process has not changed and is a similar as P1, P2, P3, P4 and P5 respectively. The profiles rate to the FTM shown in Figure 8(b). The rest of at P1 to P4 are close together for the soot-free the moisture is removed via a similar mechanism sample (Protocol I) but become more spaced out to that in FTMs; the transition from evaporation as the soot loading increases. Some profiles show limited to diffusion limited drying can be seen in step-like features towards the channel edge, for Figure 9(b) for each axial position. example P5 in Protocol I, P3 in Protocol II and P1 This additional mechanism in the drying of wall- and P3 in Protocol III. At the highest soot loading flow substrates may affect the distribution of catalyst (Protocol III), the shape of the flow profiles at P4 in the monolith following the drying process. The and P5 are narrower than the expected paraboloid.

(a) Protocol I Protocol II Protocol III

Inlet 50 50 Outlet 30 40 40 –1 –1 –1 30 30 20 , cm s , cm s , cm s

z 20 z 20 z v v v 10 10 10

0 0.5 1 0 0.5 1 0 0.5 1 z/L z/L z/L (b) 0.30 0.30 0.30 Through- wall 0.25 0.25 0.25

–1 –1 –1 velocities 0.20 0.20 0.20 0.15 0.15 0.15 , mm s , mm s , mm s

xy 0.10 xy 0.10 xy 0.10 v v v 0.05 0.05 0.05

0 0.5 1 0 0.5 1 0 0.5 1 z/L z/L z/L

Fig. 11. MRI measurements (markers) of: (a) the inlet and outlet channel velocities; (b) through-wall velocities for the GPF sample with Protocol I, Protocol II and Protocol III

(a) (b) (c) 60 50 60 50 40 40 –1 –1 –1 40 30 30 , cm s , cm s , cm s

z z 20 z v v v 20 20 10 10

0 0 0 –0.5 0 0.5 –0.5 0 0.5 –0.5 0 0.5 x, cm x, cm x, cm

Axial positions (z/L): 0.14 0.33 0.52 0.70 0.87

Fig. 12. Axial velocity flow profiles for the GPF sample after loading: (a) Protocol I; (b) Protocol II; and (c) Protocol III. Profiles are shown at axial positions ofz/L = 0.14, 0.33, 0.52, 0.70 and 0.87. Lines are only shown as a guide

Using a numerical 1D model (7) and the MRI profile for the clean filter is observed, butthe data, it is possible to extract information regarding permeability is much more uniform after loading the axial properties of the filter wall in the form of Protocol II and especially after Protocol III. The a permeability from Darcy’s law. The permeability results here show that as the GPF operates, regions calculated is an average value for the combined of higher through-wall velocity will have a greater effect of the filter wall and the particulate. However, loading of soot due to the correspondingly high mass because the through-wall velocity is known from flow of PM; this leads to the ‘self-correction’ effect the MRI measurements, it is possible to calculate predicted by Bensaid et al. (52). The axial velocity a spatial permeability in the axial dimension (z) of profiles inside the inlet channels (Figure 12) the filter. The permeability profiles are shown in show changes at the filter rear with increasing Figure 13. Some variability in the permeability soot load. The profiles become ‘narrower’ with

(a) (b) (c)

1.2 1.2 1.2

1.0 1.0 1.0 2 2 2 0.8 0.8 0.8 , m , m , m 12 0.6 12 0.6 12 0.6

× 10 0.4 × 10 0.4 × 10 0.4 k k k 0.2 0.2 0.2

0 0 0 0 0.5 1 0 0.5 1 0 0.5 1 z/L z/L z/L Fig. 13. Simulated permeability profiles for the GPF sample for loading: (a) Protocol I; (b) Protocol II; and (c) Protocol III. Lines are only shown as a guide lower velocity towards the wall. This is similar to and substrate properties. The application of MRI the profiles observed by Yorket al. (53) also using to study filters used in vehicle emissions control MRI for high soot loadings in a diesel particulate has provided new insight into their manufacturing filter and may be consistent with the development and operation. The greater understanding of of a soot cake layer in these regions. These the drying process could ultimately result in an regions also correspond to the largest reduction improved and more efficient drying process, while in wall permeability (Figure 13). Two limitations great understanding of their operation can lead to of the MRI method are that it cannot quantify the improved final product performance, for example soot loading and it cannot differentiate between higher filtration efficiency. different diameters of soot particle. However, other techniques such as gravimetric analysis or microscopy may allow these to be related to the Acknowledgements MRI results in the future. The authors would like to thank the Engineering and Physical Sciences Research Council (EPSRC) Conclusions for Cooperative Awards in Science and Technology (CASE) to N. P. Ramskill and J. D. Cooper. We A range of MR and traditional techniques, namely would also like to thank Johnson Matthey, UK, for RH and temperature measurement, have been permission to publish. applied to study wall-flow filter substrates used for PM vehicle emissions control. Drying of the filter material has been compared with an FTM using References 2D RARE imaging to investigate the effect of the structure on the drying kinetics. Since little deviation 1. Commission Regulation (EU) 459/2012, Official in the radial drying profile was seen the problem J. Eur. Union, 2012, 55, (L142), 16 was reduced to 1D in the axial (z) direction. This 2. T. A. Nijhuis, A. E. W. Beers, T. Vergunst, I. Hoek, allowed increased temporal resolution that revealed F. Kapteijn and J. A. Moulijn, Catal. Rev. Sci. Eng., different drying mechanisms are associated with 2001, 43, (4), 345 the wall-flow and FTM substrates, attributable to 3. T. Vergunst, F. Kapteijn and J. A. Moulijn, Appl. differences in the physical structures of the two Catal. A: Gen., 2001, 213, (2), 179 autocatalyst substrates. MRI and velocimetry has 4. A. Wahlberg, L. J. Pettersson, K. Bruce, also been used to investigate the effect of PM on M. Andersson and K. Jansson, Appl. Catal. B: filter channel fluid dynamics. In the gasoline system Environ., 1999, 23, (4), 271 studied, any inhomogeneities in the filter wall 5. Z. R. Ismagilov, S. A. Yashnik, A. A. Matveev, permeability are ‘self-corrected’ by the particulate I. V. Koptyug and J. A. Moulijn, Catal. Today, loading over the initial hour of filter operation. 2005, 105, (3–4), 484 MRI is a method that can indirectly visualise 6. “Handbook of Industrial Drying”, 3rd Edn., ed. soot location in the filter, while also providing A. S. Mujumdar, Taylor and Francis Group LLC, important information on the flow characteristics 2007, 1280 pp

7. E. J. Bissett, Chem. Eng. Sci., 1984, 39, (7–8), 28. N. C. Reis, R. F. Griffiths, M. D. Mantle and 1233 L. F. Gladden, Int. J. Heat Mass Trans., 2003, 46, 8. A. G. Konstandopoulos and J. H. Johnson, SAE (7), 1279 Technical Paper 890405, SAE International, 29. N. C. Reis, R. F. Griffiths, M. D. Mantle, L. F. Gladden Warrendale, USA, 1st February, 1989 and J. M. Santos, Int. J. Heat Mass Trans., 2006, 9. E. J. Bissett, M. Kostoglou and 49, (5–6), 951 A. G. Konstandopoulos, Chem. Eng. Sci., 2012, 30. G. H. A. van der Heijden, H. P. Huinink, L. Pel 84, 255 and K. Kopinga, Chem. Eng. Sci., 2009, 64, (12), 10. A. P. E. York, T. C. Watling, M. Ahmadinejad, 3010 D. Bergeal, P. R. Phillips and D. Swallow, SAE Int. 31. I. V. Koptyug, Prog. Nucl. Magn. Reson. Spectrosc., J. Fuels Lubr., 2009, 2, (1), 578 2012, 65, 1 11. L. F. Gladden, Am. Inst. Chem. Eng. J., 2004, 49, 32. I. V. Koptyug, S. Altobelli, E. Fukushima, (1), 2 A. V. Matveev and R. Z. Sagdeev, J. Magn. Reson., 12. M. P. Hollewand and L. F. Gladden, Magn. Reson. 2000, 147, (1), 36 Imaging, 1994, 12, (2), 291 33. I. V. Koptyug, L. Yu Ilyina, A. V. Matveev, 13. L. Yu Khitrina, I. V. Koptyug, N. A. Pakhomov, R. Z. Sagdeev, V. N. Parmon and S. A. Altobelli, R. Z. Sagdeev and V. N. Parmon, J. Phys. Chem. Catal. Today, 2001, 69, (1–4), 385 B, 2000, 104, (9), 1966 34. I. V. Koptyug, A. V. Matveev and S. A. Altobelli, 14. I. V. Koptyug, S. I. Kabanikhin, K. T. Iskakov, Appl. Magn. Reson., 2002, 22, (2), 187 V. B. Fenelonov, L. Yu Khitrina, R. Z. Sagdeev and 35. S. L. Codd and S. A. Altobelli, J. Magn. Reson., V. N. Parmon, Chem. Eng. Sci., 2000, 55, (9), 2003, 163, (1), 16 1559 36. N. P. Ramskill, A. P. E. York, A. J. Sederman and 15. I. V. Koptyug, L. Yu Khitrina, V. N. Parmon and L. F. Gladden, Chem. Eng. Sci., 2017, 158, 490 R. Z. Sagdeev, Magn. Reson. Imaging, 2001, 19, 37. M. Lustig, D. Donoho and J. M. Pauly, Magn. (3–4), 531 Reson. Med., 2007, 58, (6), 1182 16. A. A. Lysova, I. V. Koptyug, R. Z. Sagdeev, 38. M. Lustig, ‘Sparse MRI’, PhD thesis, Department V. N. Parmon, J. A. Bergwerff and B. M. Weckhuysen, of Electrical Engineering, Stanford University, J. Am. Chem. Soc., 2005, 127, (34), 11916 Stanford, USA, 2008, 131 pp 17. R. Enjilela, P. F. J. Cano-Barrita, A. Komar, 39. D. J. Holland, M. J. Bostock, L. F. Gladden and A. J. Boyd and B. J. Balcom, Mater. Struct., 2017, D. Nietlispach, Angew. Chem. Int. Ed., 2011, 50, 50, (2), 151 (29), 6548 18. G. Guillot, A. Trokiner, L. Darrasse and H. Saint- 40. Y. Wu, C. D’Agostino, D. J. Holland and Jalmes, J. Phys. D.: Appl. Phys., 1989, 22, (11), L. F. Gladden, Chem. Commun., 2014, 50, (91), 1646 14137 19. K. M. Song, J. Mitchell, H. Jaffel and L. F. Gladden, 41. D. J. Holland and L. F. Gladden, Angew. Chem. Int. J. Mater. Sci., 2010, 45, (19), 5282 Ed., 2014, 53, (49), 13330 20. M. D. Mantle, Int. J. Pharm., 2011, 417, (1–2), 42. P. T. Callaghan, “Principles of Nuclear Magnetic 173 Resonance Microscopy”, Oxford University Press, 21. Q. Zhang, L. Gladden, P. Avalle and M. Mantle, Oxford, UK, 1993, 492 pp J. Control. Release, 2011, 156, (3), 345 43. E. M. Haacke, R. W. Brown, M. R. Thompson and 22. J. D. Griffith, A. E. Bayly and M. L. Johns, Chem. R. Venkatesan, “Magnetic Resonance Imaging Eng. Sci., 2008, 63, (13), 3449 – Physical Principles and Sequence Design”, 23. J. D. Griffith, A. E. Bayly and M. L. Johns, J. Colloid John Wiley and Sons Inc, New York, USA, 1999, Interface Sci., 2007, 315, (1), 223 914 pp 24. G. Adiletta, G. Iannone, P. Russo, G. Patimo, 44. M. D. Mantle and A. J. Sederman, Prog. Nucl. S. De Pasquale and M. Di Matteo, Int. J. Food Sci. Magn. Reson. Spectrosc., 2003, 43, (1–2), 3 Technol., 2014, 49, (12), 2602 45. A. Caprihan and E. Fukushima, Phys. Rep., 1990, 25. L. Manzocco, M. Anese, S. Marzona, N. Innocente, 198, (4), 195 C. Lagazio and M. C. Nicoli, Food Chem., 2013, 46. “Perry’s Chemical Engineers’ Handbook”, 8th Edn., 141, (3), 2246 eds. R. H. Perry and D. W. Green, McGraw-Hill 26. T. Tsuruta, H. Tanigawa and H. Sashi, Dry. Education, New York, USA, 2008 Technol., 2015, 33, (15–16), 1830 47. J. Hennig, A. Nauerth and H. Friedburg, Magn. 27. M. D. Mantle, N. C. Reis, R. F. Griffiths and Reson. Med., 1986, 3, (6), 823 L. F. Gladden, Magn. Reson. Imaging, 2003, 21, 48. M. Schejbal, M. Marek, M. Kubíček and P. Kočí, (3–4), 293 Chem. Eng. J., 2009, 154, (1–3), 219

49. M. Yu, D. Luss and V. Balakotaiah, Chem. Eng. J., 52. S. Bensaid, D. L. Marchisio and D. Fino, Chem. 2013, 226, 68 Eng. Sci., 2010, 65, (1), 357 50. G. Bond, R. B. Moyes, S. D. Pollington and 53. A. P. E. York, T. C. Watling, N. P. Ramskill, D. A. Whan, Stud. Surf. Sci. Catal., 1993, 75, L. F. Gladden, A. J. Sederman, A. Tsolakis, 1805 J. M. Herreros and I. Lefort, SAE Technical Paper 51. K. P. de Jong, Stud. Surf. Sci. Catal., 1991, 2015-01-2009, SAE International, Warrendale, 63, 19 USA, 1st September, 2015

The Authors

Jonathan Cooper has a Master’s degree in chemistry and recently completed his PhD at the Department of Chemical Engineering, University of Cambridge. His research focused on applying MRI methods to study the gas hydrodynamics in particulate filter systems. He now teaches chemistry at a school in London.

Nicholas Ramskill holds an MEng in Chemical Engineering from the University of Leeds, UK, and a PhD in Chemical Engineering from the University of Cambridge, UK. Nicholas completed his PhD at the Magnetic Resonance Research Centre in collaboration with industrial partner Johnson Matthey where his research focused on MRI studies of diesel particulate filters (DPFs). Subsequently, Nicholas was a Postdoctoral Research Associate in Cambridge where he worked in collaboration with Royal Dutch Shell, The Netherlands, on the development and application of MRI techniques to characterise enhanced oil recovery processes in the laboratory at representative reservoir conditions.

Andy Sederman is a Reader in Magnetic Resonance in Engineering at the University of Cambridge where he also gained his PhD in 1998. His research focus is on developing magnetic resonance techniques for application to engineering and materials. He has worked extensively in the area of velocity and transport measurement and methods to increase the imaging speed to be able to investigate transient systems, both by fast data acquisition and by utilising innovative reconstruction methods allied to data under-sampling. Areas of application for these methods have focused on single and multiphase flows, fluid flow in porous media and reaction and hydrodynamics in multiphase reactors.

Lynn Gladden is Shell Professor of Chemical Engineering at the University of Cambridge in the Department of Chemical Engineering and Biotechnology. She is recognised for her work on advancing magnetic resonance imaging techniques, originally developed for use in the medical environment, and using them in engineering research to gain a greater understanding of the physical and chemical phenomena that determine the performance of chemical processes and their resulting products.

Professor Athanasios Tsolakis has academic and industrial expertise in the field of low carbon energy carriers, environmental catalysts, combustion and pollutant control technologies. He works at the forefront of basic and translational research to improve fuel efficiency and reduce the environmental impact of the transportation and power generation sectors. Prior to his academic appointment at the University of Birmingham in 2005 he worked as a research scientist at Johnson Matthey in the design and characterisation of environmental catalysts for modern aftertreatment systems.

E. Hugh Stitt is a Scientific Consultant at Johnson Matthey, Chilton, UK. He is a Visiting Professor at the University of Birmingham; Fellow of the Institution of Chemical Engineers and a Fellow of the Royal Academy of Engineering. He has 30 years of industrial research experience across a variety of themes related to catalytic reaction engineering and catalyst manufacture with over 100 refereed publications.

Andrew York is a Research Manager responsible for the Emissions Control Reaction Engineering and Modelling group. He joined Johnson Matthey, Sonning Common, in 2000, and has had a variety of roles, including in gasoline and diesel catalyst research, and leading many collaborations with academia on various engineering and catalysis related projects.

www.technology.matthey.com

Manufacturing and Characterisation of Robot Assisted Microplasma Multilayer Coating of Titanium Implants Biocompatible coatings for medical implants with improved density and crystallinity

D. Alontseva* the optimised composition and structure of the School of Engineering, D. Serikbayev East titanium/hydroxyapatite (HA) multilayer coatings. Kazakhstan State Technical University, 69 It is desirable that the Ti coated lower layer offer Protozanov Street, Ust-Kamenogorsk, 070004, a dense layer to provide the implant with suitable Kazakhstan structural integrity and the Ti porous layer and HA top layer present biocompatible layers which E. Ghassemieh** are suitable for implant and tissue integration. The Wolfson School of Mechanical, Electrical Scanning electron microscopy (SEM), transmission and Manufacturing Engineering, Loughborough electron microscopy (TEM) and X-ray diffraction University, Loughborough, Leicestershire, LE11 (XRD) were used to analyse the structure of the 3TU, UK coatings. The new robot assisted MPS technique resulting from this research provides a promising S. Voinarovych, O. Kyslytsia, Y. solution for medical implant technology. Polovetskyi E.O.Paton Electric Welding Institute of NAS of 1. Introduction Ukraine, 11 Kazymyr Malevich Street, Kyiv, 03150, Ukraine Manufacturing technology for medical implants undergoes constant improvement in order to N. Prokhorenkova, A. Kadyroldina accelerate the patient’s recovery and increase the D. Serikbayev East Kazakhstan State service life of the implant. Following this trend, Technical University, 69 Protozanov Street, special attention is paid to applying biocompatible Ust-Kamenogorsk, 070004, Kazakhstan coatings on the surface of implants (1–7). There is a huge clinical need for advanced biomaterials Email: *[email protected]; with enhanced functionality to improve the quality **[email protected] of life of patients and reduce the burden of health care for the world’s ageing population. In recent years, Ti and HA have been widely used in medical This study focuses on new technologies for the devices due to their favourable biocompatibility production of medical implants using a combination (1–8). of robotics and microplasma coatings. This involves The research presented here offers robot assisted robot assisted microplasma spraying (MPS) of MPS of Ti wires and HA powder onto Ti substrates a multilayer surface structure on a biomedical (9–12). Currently, MPS technology is highly implant. The robot motion design provides specialised with few research groups working a consistent and customised plasma coating on its development. Yet it has shown promising operation. Based on the analytical model results, results in delivery of high quality and well-designed certain spraying modes were chosen to form coatings (1–3). Among existing plasma spraying

180 © 2020 Johnson Matthey https://doi.org/10.1595/205651320X15737283268284 Johnson Matthey Technol. Rev., 2020, 64, (2) processes, MPS is particularly characterised by phosphate (ACP) has a higher rate of dissolution, low plasma power (up to 4 kW), small diameter which reduces the recovery time of the patient, but of the spray deposition spot on the surface (up to at the same time some reduction of the reliability 15 mm) and the possibility of forming a laminar of fixation of the endoprosthesis in the bone is also jet. This laminar jet can be up to 150 mm in possible. Thus, increased crystallinity provides length. It heats the refractory material in a stream reliable fixation of the implant in the bone (2, 3). of argon plasma and provides low heat input into It is assumed that to increase the biocompatibility the substrate (2). The process provides deposition of the implants and rapid accretion with the bone, of Ti or HA on small size parts and components, the implant surface should be covered with a including those with complex geometry. This is biocompatible coating with an extensive surface normally unachievable with any other method. morphology, with recommended pore sizes in The MPS generally provides a micro-rough surface the coating from 20 µm to 200 µm, and closed and a higher degree of porosity (~20%) that in porosity of at least 30% (2–4, 16, 17). At the same the case of biocompatible coatings facilitates bony time, the coating must be firmly connected to the tissues ingrowth; in most cases the bond strength implant, without transversal porosity, so that the of MPS coatings with substrates is acceptable (2). implant material does not interact directly with However, there are still a number of challenges the human body. Therefore in this research, the remaining to be addressed. The most important proposed composition and thickness of the coating issue is the formation of coatings with specified is designed to meet these requirements. In order structure and properties. to characterise the coated layer structures, SEM, Currently, endoprosthesis medical practice widely TEM and XRD are used. Optimum modes of plasma uses metal implants coated with HA (1–8, 13, 14). HA spraying are chosen on the basis of the results of is the calcium phosphate mineral Ca10(PO4)6(OH)2 structural characterisation. of the apatite group, which is chemically similar The aim of this study is to select the modes of to the apatite of the host bone and is a source of MPS of both Ti wire and HA powder to obtain a Ca and P for the bone-HA interface (3, 4, 6, 8). sufficiently thick (up to 300 μm) multilayer HA coatings improve osseointegration and can Ti/HA coating. The dense Ti sublayer is supposed significantly reduce the duration of implantation of to provide good adhesion to the substrate and the endoprosthesis. It provides a reliable connection the porous Ti middle layer and HA top layer can with the bone and increases the reliability of the accelerate the bone ingrowth. Another objective is implant (1–8, 13, 14). In the case of thermal to clarify the relationship between various plasma spraying of HA powder, the chemical composition spray process parameters and the resultant of the final HA coating is dependent on the thermal coating structure. This is achieved through the decomposition occurring during spraying. The development of process models that relate process high temperatures experienced by HA powder parameters to various coating structures. particles in the plasma spraying process lead to the dihydroxylation and decomposition of the particles. 2. Materials and Methods At temperatures above 1050°C HA decomposes to tricalcium phosphate (β-TCP, Ca3(PO4)2) and The HA was synthesised in the laboratories of tetracalcium phosphate (TTCP, Ca4(PO4)2O), and East Kazakhstan State Technical University. above 1120°C β-TCP is converted to α-tricalcium The process of synthesis of HA powder with the phosphate (α-TCP, Ca3(PO4)2) (13, 14). Thus, ratio Ca:P of 1.65 by a chemical precipitation the resultant coating phase composition depends method was described in our previous paper on the thermal history of the powder particles. (10). The purity of HA powder was 99.5%, which The higher the plasma jet temperature and the met the purity requirement (not less than 95%) longer the exposure of the particles to plasma, set out in the ASTM International (USA) standard the greater the degree of phase transformation. ASTM F1185- 03(2014) (18). The purity of the According to the British Standards Institution synthesised HA is determined using the XRD standard specification BS ISO 13779-2:2000 (15), results which are further explained below. Before the maximum allowable content of non-HA phases spraying, the powder was dried at a temperature in a HA coating is 5%, and the minimum allowable of 120°C for 6 h, in order to avoid clogging during percentage of crystallinity is 50%. The degree of powder feeding. HA powder was used as a sprayed crystallinity of the HA coating affects the process coating material. The particles had irregular shape of osseointegration (16). Amorphous calcium with smooth edges (Figure 1).

(a) (b) Fig. 1. SEM images of HA particles indicating particle size 80.57 µm 48.23 µm

40.45 µm 48.73 µm

500 µm 50 µm

Table I Chemical Composition of Ti-Based Materials Materials Reference wt% of element grade composition Al V O C N H Fe Si Ti Grade 5 ELI (19), (20) 5.50–6.75 3.50–4.50 0.13–0.20 0.08 0.05 0.015 0.25–0.40 – base alloy VT1-00 commercially (21) 0.3 – 0.12 0.05 0.03 0.003 0.15 0.08 base pure Ti

The shapes of the particles are particularly on an industrial robot arm (RS010L, Kawasaki Heavy designed with smooth edges to make sure that Industries, Japan). It is able to move horizontally they do not stick together. Moreover the HA along a computed trajectory at set speed. The particles should not cling to each other in order thickness of the coatings was varied from 80 μm to provide the necessary flowability of the powder. to 300 μm by changing the MPS parameters. The The flowability is essential in gradual supply of the speed of linear movement of the plasmatron along powder to the plasma jet. the substrate was chosen to be 50 mm s–1. The According to previous studies (9, 10), the particle choice of speed of the plasmatron was based on size of the HA powder for MPS should be in the preliminary estimates of the temperature of the range of 40 μm to 90 μm. Before MPS, the dried substrate when exposed to a plasma jet (12). and milled HA powder was sieved through mesh This was to make sure the temperatures remain diameters 40 µm and 90 µm to obtain only the well below the melting temperature. Ar served as powder fraction within the desired range. The a plasma-forming and transporting gas for MPS; flow rate of the HA powder is in the range of additional heating of the substrate was not carried 120 s 50 g–1. out. Using a robotic arm allows precise spraying of Samples of medical Ti alloy of Grade 5 extra coatings with a uniform speed of movement of the low interstitials (ELI) (Table I) were used as plasmatron along the surface of the implant, as well substrates for MPS. For deposition of Ti coatings, as moving the plasmatron along a predetermined wires of VT1‑00 commercially pure Ti with path. a diameter of 0.3 mm were used (Table I). Before MPS, the surfaces of the samples were The samples of medical Ti alloy of Grade 5 ELI degreased with acetone and subjected to ultrasonic (Table I) were cut with thicknesses of 3 mm from cleaning. To ensure proper adhesion of the coatings, rods with a diameter of 50 mm and 30 mm on it was important to pre-treat the surfaces of the CTX 510 ecoline computer numerical control (CNC) substrates to increase their roughness. For surface machine (DMG MORI AG, Germany). Plates of size activation, gas abrasive surface treatment was 15 mm × 15 mm × 2 mm were cut from large carried out on a Contracor ECO abrasive blasting sheet of Grade 5 ELI alloy. machine (Comprag Group GmbH, Germany) using MPS of the HA powders and Ti wires was carried normal grade A14 electrocorundum. The chemical out by MPN-004 microplasmatron (produced reactivity of the substrate’s surface rapidly falls due by E.O.Paton Institute of Electric Welding, Kiev, to oxidation and the adsorption of chemical gases Ukraine) (22). The microplasmatron was mounted from the atmosphere. Therefore it is important

182 © 2020 Johnson Matthey https://doi.org/10.1595/205651320X15737283268284 Johnson Matthey Technol. Rev., 2020, 64, (2) that the time interval between the gas abrasive diffuse background in this region. Diffraction treatment and the coating on the surface does not scans of the HA powder and coatings were carried exceed 2 h. Before coating, samples were stored in out in accordance with ASTM F2024- 10(2016) a tightly closed container. (26). The purity of HA powder and HA coatings To evaluate the porosity of biocompatible was evaluated by calculating the areas of all coatings, the images obtained by the scanning non‑HA peaks found in the diffraction pattern. electron microscope JSM-6390LV (JEOL Ltd, Japan) The impurity area was determined by calculating were processed using MicroCapture (MustCam, the area in the region where the highest impurity Hong Kong) and ATLAS.ti (ATLAS.ti Scientific phase peaks were present. The impurity peaks Software Development GmbH) computer-aided that would be expected to be present in HA programs. The measurements were carried out powders and HA coatings were those of TTCP, on the polished cross-section of the coatings α-TCP and β-TCP. The Rietveld method was used according to ASTM E2109-01(2014) standard (23). to quantitatively determine the percentage of The surface roughness of the substrates and the various phases of impurity in HA coatings. X’Pert as-sprayed coatings was measured in accordance HighScore Plus software was used to calculate with ISO 4287:1997 (24) using a MarSurf PS 10 the impurity. Three purity measurements were mobile roughness measuring instrument (Mahr, carried out for each of the XRD patterns. Germany). Four measurements were taken for Electron diffraction patterns of samples of HA each sample and the average was determined. The powder were obtained by TEM on JEM-2100 (JEOL). adhesion strength of coatings to the substrates The structural-phase composition data obtained was measured in tension using a AG-X universal using TEM were compared with the data obtained testing machine (Shimadzu, Japan) in a static using XRD analysis. TEM sample preparation experiment according to ASTM C633-13(2017) techniques are described in detail in a previous standard (25). publication (11). The crystallinity and the structure-phase The particles of the sprayed Ti wire after compositions of HA powders and plasma sprayed collision with the substrate were studied using HA coatings were measured by X’Pert PRO splat tests (2). MPS of Ti wire onto the plates diffractometer (PANalytical, The Netherlands). of polished Ti alloy was performed in the plane The interpretation of the XRD patterns was carried perpendicular to the axis of the plasma jet. out using Rietveld method and powder diffraction Speed of the linear movement of plasma jet was data from the database of the International Centre set at 50 mm s–1. As a result, single particles of for Diffraction Data (ICDD, USA, 2003), the ASTM the sprayed material (splats) were fixed on the card file and X’Pert HighScore Plus software substrate and deformed upon contact with the (Malvern Panalytical, UK). The percentage of substrate surface. The splats’ visual analysis was crystallinity of the HA powder was calculated carried out by SEM; the splats were classified using the area of crystalline peaks in the region according to their appearance and their spraying of 15° to 45° 2θ and the area of the amorphous modes (Table II).

Тable II MPS Deposition Parameters of Ti Wires, the Porosity and the Arithmetic Ra of Sprayed Coatings

Parameters/settings Porosity of Spraying sprayed Ti Ra, μm mode –1 I, A Vpg, slpm H, mm Vw, m min coatings, % 1 25 3.7 120 4.3 6.2 ± 0.74 12.0 ± 0.97 2 25 3.7 40 3.0 13.8 ± 1.90 45.6 ± 4.40 3 25 2.3 120 3.0 12.0 ± 1.50 44.4 ± 3.85 4 25 2.3 40 4,3 5.7 ± 0.45 12.0 ± 1.13 5 15 3.7 120 3.0 9.6 ± 0.35 31.5 ± 3.10 6 15 3.7 40 4.3 10.6 ± 1.32 34.8 ± 3.23 7 15 2.3 120 4.3 8.7 ± 2.32 30.2 ± 3.87 8 15 2.3 40 3.0 31.0 ± 3.87 >50

3. Results and Discussion Examination of the Ti particles splats obtained in Modes 1, 4 and 8 (Table II) showed that the The effect of parameters of MPS such as electric samples are completely melted and have formed a arc current (I, A), plasma gas flow rate in disk (Figures 2(a) and 2(b)). The thicker splats standard litre per minute (Vpg, slpm), spraying are formed in Mode 8 (Table II, Figure 2(c)). – 1 distance (H, mm) and wire flow rate (Vw, m min ) The beginning of the process of solidification of the –1 or powder consumption (Ppowder, g min ) on the particles upon impact with the substrate is shown surface morphology, porosity and structural‑phase in Figure 2(c). The increase in the thickness of the transformations of the coatings have been splats in Mode 8 (Table II) is due to the minimum studied. The coating experiments for MPS were velocity of the particles during their interaction accomplished in a two level fractional factorial with the substrate. design (24–1). The experimental conditions in As can be seen in Figure 2, the splats obtained fractional factorial designs have been selected to in different modes have similar area but different provide balanced design (27). The maximum and thickness. The increase in the thickness of the minimum values of the parameters for feasible splats in Mode 8 (Table II) is due to the minimum processing of high-quality coatings were chosen velocity of the particles during their interaction empirically. with the substrate. The minimum particle MPS of Ti wires on Ti alloy gas-abrasive treated velocity in Mode 8 is caused by the large size of substrates was carried out in various modes as the sprayed particles, low gas flow rate and low shown in Table II. The key characteristics of spraying distance. This also leads to a decrease the resultant coatings such as their porosity and in the speed of the sprayed particles. Particles roughness were examined. The data in Table II of Ti wire melted by a plasma jet move towards and Table III represent the averaged values for the substrate during plasma spraying. On the one three experimental runs. hand, the size of these particles depends on the

Table III The Dependency of the СTE, Phase Composition and Crystallinity of HA Coating on the Spraying Parameters –1 Mode I, A Vpg, slpm H, mm Ppowder, g min СTE, % СTE, % estimated HAcryst. Aph β-TCP 1 45 2.0 160 1.2 54 58 92 5 3 2 45 2.0 80 0.4 64 71 98 0 2 3 45 1.0 160 0.4 89 89 93 2 5 4 45 1.0 80 1.2 69 69 96 0 4 5 35 2.0 160 0.4 29 29 93 4 3 6 35 2.0 80 1.2 48 48 94 3 3 7 35 1.0 160 1.2 40 47 88 7 5 8 35 1.0 80 0.4 56 60 98 0 2 9 40 1.5 120 0.8 60 59 90 6 4

(a) Mode 1 (b) Mode 4 (c) Mode 8

500 µm 500 µm 500 µm

Fig. 2. SEM images of Ti splats sprayed in: (a) Mode 1; (b) Mode 4; and (c) Mode 8 (according to Table II)

184 © 2020 Johnson Matthey https://doi.org/10.1595/205651320X15737283268284 Johnson Matthey Technol. Rev., 2020, 64, (2) set spraying parameters. On the other hand, both using Mode 1 and Mode 4 (Table II), the remaining the set spraying parameters and the size of the modes provide a high surface roughness (above particles affect the speed and the degree of heating 30 μm). The maximum size of open pores (up to of the particles in the plasma jet. Before interacting 300 µm) is observed on the surface of coatings with the substrate, the particles of molten metal obtained in Mode 8 (Figure 3). can either heat up, being in the high-temperature The analysis of the cross-sections of the Ti coatings zone of the plasma jet (the initial section of the showed that the coatings sprayed in Mode 8 have the plasma), or, more likely, cool down by going to the highest average porosity of 31.0%, while Mode 4 low-temperature zone (the end of the plasma jet). makes it possible to obtain dense coatings with an The size of the sprayed particles and the degree average porosity of 5.7% (Figure 3 and Table II). of their melting in the plasma jet can be varied The tensile strength test established the average by spraying parameters. A porous coating with a adhesion strength of the coating with a thickness of high surface roughness can be achieved by large 100 µm sprayed in Mode 4 (Table II) to be 38.7 MPa. particles moving with low speed, as in Mode 8. This meets the requirements of ISO 13179‑1:2014 A dense coating with a relatively low surface (29). According to ISO 13179‑1:2014 (29), the roughness can be obtained by high speed small average static tensile strength of a Ti coating and completely melted particles, as in Mode 1 or in should be more than 22 MPa. Mode 4 (Table II). As can be seen from the results presented The appearance and structure of the splats varies in Table II, the arithmetic Ra of the coatings depending on the interactions of the sprayed surface correlates with the porosity of the sprayed particles with the substrate. The interaction is coatings. The condition of the presence of large, generally defined by the velocity of the particles on incompletely molten particles with a low speed in impact and also their degree of melting. It is possible the plasma jet leads to a high surface roughness to correlate the temperature and velocity of the and increased porosity of the coatings. Thus, the particles before the collision to the substrate with roughness of the coating is affected by the size the resultant structure of the coating. The profile of the particles involved in the formation of the and cracks on the surface of the splat correlate with coating. The main purpose of plasma spraying of the stress state of the coating (2, 28). the Ti layers at the initial stage of the coating is The substrate average roughness (Ra) after gas to form a porous surface with high roughness in abrasive treatment was Ra = 7.0 ± 0.35 μm. The order to spray fully molten HA particles onto it. It analysis of the surface morphology of Ti coatings was previously shown (30) that partially melted showed the possibility of obtaining dense coatings HA particles were not able to form dense coatings. with relatively low roughness (Ra is about 12.0 μm) This leads to poor adhesion of the HA coating to

Mode 4 Mode 8 Fig. 3. SEM images of surfaces and cross-sections of Ti coatings sprayed in Mode 4 and Mode 8 (according to Table II) Surface

500 µm 500 µm

Cross section

200 µm 200 µm

the substrate. A desirable HA coating shows good (HAcryst)) and the coating transfer efficiency (СTE). adhesion and high surface roughness. The surface In powder coating, transfer efficiency is the ratio of roughness of the HA coating affects osteoblast cell the quantity of powder deposited on the part to the attachment and thus accelerates bone growth into quantity of powder directed at the part. Transfer the implant. Whereas fibroblasts and epithelial efficiency is provided as a percentage, with 100% cells prefer smoother surfaces, osteoblasts attach being most desirable. An experiment was conducted and proliferate better on rough surfaces (31, 32). to determine the impact of process parameters

A relatively thin HA layer (not more than 100 μm) such as I, Vpg, H and Ppowder on the СTE. sprayed on a porous surface with a high roughness The application of well-known methods of (above 50 μm) can ensure the reliable adhesion fractional factorial design (27) and the design of HA coating to the surface, while maintaining of experiment method (33) for the analysis of high surface roughness and porosity. XRD analysis plasma spraying of HA powder are described confirmed that the phase composition of the initial elsewhere (34, 35) and in our previous paper (10).

HA powder was fully crystalline Ca10(PO4)6(OH)2. A factorial experimental design to investigate the The TEM images (Figures 4(a) and 4(b)) of relationship between plasma spray parameters the HA powder particle and the corresponding and the microstructure of HA coatings was first microelectron diffraction pattern (Figure 4(c)) are used by Dyshlovenko et al. (35). Three responses shown in Figure 4. The results of TEM analysis were examined (35). This included the fraction are in good agreement with the results of XRD of HA, the fraction of decomposition phases and analysis (Figure 4(d)). X-ray phase analysis the amorphous content of the coatings. In this showed that the main phase (99.5%) is HA with study, the next three responses were examined: the hexagonal crystal system P63/m. The electron the fraction of HA, the amorphous content of the diffraction pattern corresponds to the hexagonal coatings and CTE. The first two responses were phase of HA with unit cell parameters a = 0.94 nm, chosen to ensure purity and crystallinity of the HA c = 0.68 nm. coating. CTE was selected in order to increase the The plasma spraying of HA powder was carried efficiency of the plasma spraying process. Moreover out using nine different modesTable ( III). The it allows interpretation by a linear regression model key criteria were the phase composition, the degree which could quite easily and reliably be measured of crystallinity (the proportion of the amorphous in experimental runs. The linear regression model phase (Aph), the proportion of the crystalline phase was chosen to estimate CTE. The coefficients in the

(a) (b) (c) 111

311 331 010 110 100 331 131 111 200 nm 100 nm (d) HA arbitrary unit arbitrary Relative intensity, intensity, Relative

10 20 30 40 50 60 Degrees 2θ, °

Fig. 4. (a, b) TEM images of the HA powder particles; (b, c) the corresponding indexed microelectron diffraction pattern; and (d) the XRD patterns of HA powder

186 © 2020 Johnson Matthey https://doi.org/10.1595/205651320X15737283268284 Johnson Matthey Technol. Rev., 2020, 64, (2) regression Equation (i) were calculated by assigning The modes of application of multilayer coatings the corresponding units of measure: 2.575 A–1; of Ti/HA were selected on the basis of the analysis –0.246 slpm–1; –0.203 mm–1; 4.06 min g–1; of the tests of MPS of Ti wire and HA powders –0.825. with the measurement of thickness and porosity of the deposited layers. The porosity and surface СTE, % = 2.575 I – 0.246 Vpg – 0.203 H roughness of Ti coatings and the high CTE value, + 4.06 Ppowder – 0.825 (i) purity and crystallinity of the HA coating indicate The comparison of calculated and experimental the optimum coating composition and MPS modes. results indicates good agreement (Table III). The coating thickness and the expected adhesion Therefore, Equation (i) can be used for preliminary to the substrate are other parameters indicating estimation of CTE when selecting MPS modes. A the quality of the coating. The first relatively thin more complex regression model that takes into (up to 80 µm) and dense layer of Ti wire coating is account the mutual influence of factors should applied in Mode 4 (Table II), then another layer be applied to determine the dependency of other of Ti wire up to 100 µm thick is sprayed in Mode 8 values presented in Table II and Table III on (Table II) to form a porous coating with a rough the spraying parameters. This requires further surface, then the top layer of HA powder coating investigation using the obtained experimental is applied with thickness up to 100 µm in Mode 3 data. (Table III). The microstructure of microplasma The results of XRD analysis presented in sprayed multilayer Ti/HА coatings under the above Table III show that the phase compositions of modes is shown in Figures 5(a) and 5(b). The all coatings comply with ISO 13779-2:2000 (15). XRD pattern for microplasma-sprayed HA coating However, Mode 3 provides the highest CTE. Thus, is presented in Figure 5(c). we consider Mode 3 to be the most cost-effective. The desired porosity was achieved in the Ti lower This mode allows a desired HA coating thickness layer (30 vol%) (Figure 5(b)). Pore sizes in both (about 100 µm) to be obtained in one pass of a the Ti middle layer and the HA top layer are in plasma jet. the range 20–50 μm (Figure 5(b)). HA coating

(a) (b)

100 µm 50 µm

(c)

Amorphous region HA

Relative intensity, arbitrary unit arbitrary intensity, Relative 10 20 30 40 50 60 Degrees 2θ, °

Fig. 5. SEM images of the multilayer coating with the Ti lower and middle layers and the HA upper layer: (a) the surface; (b) the cross-section; and (c) the XRD patterns of microplasma-sprayed HA coating

187 © 2020 Johnson Matthey https://doi.org/10.1595/205651320X15737283268284 Johnson Matthey Technol. Rev., 2020, 64, (2) porosity is about 20% (Figure 5(b)). It should shown promising directions for further development be noted that for biocompatible coatings, open of the research presented here. Cizek et al. (36) have porosity is essential: that is the egress of the pore reviewed the patents concerning thermal spraying on the surface of the coating, where the bone for biomedical applications for the period 2005 to grows. Therefore, measuring the diameters of the 2018. They have also reported recent research pore craters on the surface of the coating is an and development trends in this field. Among the appropriate way to indicate the surface morphology materials recommended for bio-applications, they and profile. In our experiment, the maximum pore have mentioned Ta. It can be noted that MPS of Ta diameter on the surface of the HA coating was wire onto Ti alloys using the technology presented about 150 μm (Figure 5(a)). in our paper is highly feasible. Our trials with Ta It was established by XRD that the mode specified have indicated great potential. This could open for HA powder provides the required structure- the potential to apply the developed technology phase composition in the HA coating: 93% by for other materials. Fotovvati et al. (37) have weight of the HAcryst, 5% by weight of β-TCP phase, compared the results of obtaining biocompatible and 2% by weight of the Aph. The coating purity coatings by cold and thermal spraying in favour was determined using the procedure outlined in of thermal spraying. Fousova et al. (38) have Materials and Methods section above. The highest shown the benefits of using thermal plasma peaks of HA and β phases for HA coatings were spray to prepare bulk Ti for bone enlargements. located at 32° 2θ and 31° 2θ respectively. The However, despite the advantages and relative cost purity of coatings was found to be 95.1%. This effectiveness of thermal plasma spraying, its use shows that the purity meets the 95% purity for the manufacture of medical implants has not requirements of ISO 13779-2:2000 (15). The yet become widespread. This is mainly due to the measurement error was 0.06. high temperatures of the bulk resulting from the

The loci of Aph have been found on the XRD thermal spraying process. MPS avoids the issue of patterns between 18° and 38° 2θ (Figure 5(c)). overheating. It allows coatings to be obtained from All the diffraction patterns in the range of 37.3° 2θ materials with a high melting point, such as Ti and were thoroughly investigated, but even weak Ta, by a microplasma jet while introducing a very peaks of calcium oxide (CaO) were not found small thermal impact into the substrate. (Figure 5(с)). It confirms that no harmful СаО The use of robotic MPS could be considered compound is formed through MPS coating of HA promising for the production of patient specific powder. implants. Three-dimensional scanning and rapid For a multilayer coating, the porosity and adhesion prototyping technologies facilitate the manufacture of the top HA layer depends on the characteristics of specifically designed complex geometry of the lower Ti layers such as the roughness and implants and robot assisted plasma coating is open porosity of the middle Ti layer. To determine used for coating. This is more advantageous for the dependency of the porosity and adhesion of the production of small endoprostheses with the HA coating on the parameters of MPS, further biocompatible coatings, such as vertebral cages research is needed. and dental implants (39, 40). This study proves that it is possible to obtain coatings from biocompatible materials with the 4. Conclusion desired level of porosity and satisfactory adhesion to the substrate using MPS. A robot assisted MPS of It has been established that the main parameters coatings from biocompatible materials of Ti and HA controlling the porosity of microplasma sprayed onto Ti implants has been implemented. Also the coatings are I and Vpg. MPS parameters for the composition and modes of microplasma deposition formation of porous coatings of Ti wire and HA of multilayer coatings for Ti implants have been powders with rough surfaces have been determined identified. The next stage of the research includes and are reported here. The advantages of applying the study of the biocompatibility of microplasma- SEM, TEM and XRD to analyse the structure of sprayed coatings (in vitro tests) and MPS of sprayed Ti and HA coatings to substantiate the different materials such as tantalum and zirconium. choice of plasma spraying modes of the coatings Among the number of works that have were demonstrated. It is also proven that by using demonstrated the advantages of thermal spraying the appropriate MPS process parameters, a layer of biocompatible coatings for use in medical of HA with a high degree of crystallinity (93%) can applications, three recent papers (36–38) have be obtained, controlled by changing the deposition

188 © 2020 Johnson Matthey https://doi.org/10.1595/205651320X15737283268284 Johnson Matthey Technol. Rev., 2020, 64, (2) mode. The small size of the spraying spot (up to 10. D. L. Alontseva, M. B. Abilev, A. M. Zhilkashinova, S. G. Voinarovych, O. N. Kyslytsia, E. Ghassemieh, 8 mm) provides a significant reduction inpowder P A. Russakova and L. Łatka, Adv. Mater. Sci., 2018, when depositing on implants of small size compared 18, (3), 79 to conventional plasma spraying. 11. D. Alontseva, E. Ghassemieh and A. Dzhes, Acta The composition and modes of microplasma Phys. Pol. A, 2019, 135, (5), 1113 deposition of multilayer coatings for Ti implants, 12. D. L. Alontseva, A. L. Krasavin, D. M. Nurekenov including a dense Ti sublayer, porous Ti middle and Ye. T. Zhanuzakov, ‘Mathematical Modeling of layer and HA top layer have been established. The Temperature Fields in Two-Layer Heat Absorbers total thickness of such coatings is about 300 µm. for the Development of Robotic Technology for The porous middle coating layer has a porosity Microplasma Spraying of Biocompatible Coatings’, of 30% and a pore size varying from 20 µm to Computational and Information Technologies in 50 µm. Moreover the upper layer of HA indicates Science, Engineering and Education (CITech), Ust- a thickness up to 100 µm with a 95% level of HA Kamenogorsk, Kazakhstan, 25th–28th September, phases and 93% crystallinity. The results of this 2018, Springer Nature Switzerland AG, Cham, research are of significance for a wide range of Switzerland, 2019, pp. 11–22 researchers developing plasma spray technologies 13. T. M. Sridhar, U. Kamachi Mudali and M. Subbaiyan, for biocompatible coatings manufacture. Corros. Sci., 2003, 45, (10), 2337 14. S. Vahabzadeh, M. Roy, A. Bandyopadhyay and Acknowledgments S. Bose, Acta Biomater., 2015, 17, 47 15. ‘Implants for Surgery: Hydroxyapatite: Coatings The study has been conducted with the financial of Hydroxyapatite’, BS ISO 13779-2:2000, support of the Science Committee of the Ministry of British Standards Institution, London, UK, 15th Education and Science of the Republic of Kazakhstan November, 2000, 12 pp by the project AP05130525 “The intelligent robotic 16. L. M. R. de Vasconcellos, D. O. Leite, F. N. de system for plasma processing and cutting of Oliveira, Y. R. Carvalho and C. A. A. Cairo, Braz. large‑size products of complex shape”. Oral Res., 2010, 24, (4), 399 17. F. Matassi, A. Botti, L. Sirleo, C. Carulli and M. Innocenti, Clin. Cases Miner. Bone Metab., References 2013, 10, (2), 111

1. E. J. Tobin, Adv. Drug Deliv. Rev., 2017, 112, 88 18. ‘Standard Specification for Composition of Hydroxylapatite for Surgical Implants’, ASTM 2. S. V. Dorozhkin, Mater. Sci. Eng.: C, 2015, 55, F1185-03(2014), ASTM International, West 272 Conshohocken, USA, 2014, 3 pp 3. “Bone-Implant Interface in Orthopedic Surgery: 19. ‘Implants for surgery — Metallic materials — Part 3: Basic Science to Clinical Applications”, ed. Wrought titanium 6-aluminium 4-vanadium alloy’, T. Karachalios, Springer-Verlag, London, UK, ISO 5832-3:2016, International Organization for 2014, 342 pp Standardization, Geneva, Switzerland, October, 4. W. Jing, M. Zhang, L. Jin, J. Zhao, Q. Gao, M. Ren 2016, 7 pp and Q. Fan, Int. J. Surg., 2015, 24, (A), 51 20. ‘Standard Specification for Wrought Titanium- 5. S. B. Goodman, Z. Yao, M. Keeney and F. Yang, 6Aluminum-4Vanadium ELI (Extra Low Interstitial) Biomaterials, 2013, 34, (13), 3174 Alloy for Surgical Implant Applications (UNS 6. R. B. Heimann, Open Biomed. Eng. J., 2015, 9, R56401)’, ASTM F136-13, ASTM International, (Suppl. 1-M1), 25 West Conshohocken, USA, 2013, 5 pp 7. S. Sakka, J. Bouaziz and F. B. Ayed, ‘Mechanical 21. ‘Wrought Titanium and Titanium Alloys. Properties of Biomaterials Based on Calcium Grades’, GOST 19807-91, Euro-Asian Council Phosphates and Bioinert Oxides for Applications in for Standardization, Metrology and Certification Biomedicine’, in “Advances in Biomaterials Science (EASC), Russia, 17th July, 1991, 6 pp and Biomedical Applications”, eds. R. Pignatello, 22. K. Yushenko, Y. Borisov, S. Voynarovych and Ch. 2, IntechOpen Ltd, London, UK, 2013, pp. O. Fomakin, International Association Interm, 23–50 ‘Plasmatron for Spraying of Coatings’, World 8. C. J. Wilcock, P. Gentile, P. V. Hatton and Patent Appl. 2004/010,747 C. A. Miller, J. Vis. Exp., 2017, (120), e55343 23. ‘Standard Test Methods for Determining Area 9. D. Alontseva, Y. Borisov, S. Voinarovych, О. Percentage Porosity in Thermal Sprayed Coatings’, Kyslytsia, T. Kolesnikova, N. Prokhorenkova and ASTM E2109-01(2014), ASTM International, West A. Kadyroldina, Prz. Elektrotech, 2018, 7, 94 Conshohocken, USA, 2014, 8 pp

24. ‘Geometrical Product Specifications (GPS) 31. B. D. Boyan, T. W. Hummert, D. D. Dean and – Surface Texture: Profile Method – Terms, Z. Schwartz, Biomaterials, 1996, 17, (2), 137 Definitions and Surface Texture Parameters’, 32. A. Boyde, A. Corsi, R. Quarto, R. Cancedda and ISO 4287:1997, International Organization for P. Bianco, Bone, 1999, 24, (6), 579 Standardization, Geneva, Switzerland, April, 1997, 25 pp 33. R. K. Roy, “Design of Experiments using the Taguchi Approach: 16 Steps to Product and 25. ‘Standard Test Method for Adhesion or Cohesion Process Improvement”, John Wiley and Sons Inc, Strength of Thermal Spray Coatings’, ASTM New York, USA, 2001, 538 pp C633-13(2017), ASTM International, West Conshohocken, USA, 2017, 8 pp 34. C. Pierlot, L. Pawlowski, M. Bigan and P. Chagnon, 26. ‘Standard Practice for X-ray Diffraction Surf. Coatings Technol., 2008, 202, (18), 4483 Determination of Phase Content of Plasma-Sprayed 35. S. Dyshlovenko, L. Pawlowski, P. Roussel, Hydroxyapatite Coatings’, ASTM F2024-10(2016), D. Murano and A. Le Maguer, Surf. Coatings ASTM International, West Conshohocken, USA, Technol., 2006, 200, (12–13), 3845 2016, 4 pp 36. J. Cizek and J. Matejicek, J. Therm. Spray Technol., 27. D. C. Montgomery, G. C. Runger and N. R. Hubele, 2018, 27, (8), 1251 “Engineering Statistics”, 2nd Edn., John Wiley and 37. B. Fotovvati, N. Namdari and Sons Inc, Hoboken, USA, 2001 A. Dehghanghadikolaei, J. Manuf. Mater. Process., 28. M. Mutter, G. Mauer, R. Mücke, O. Guillon and 2019, 3, (1), 28 R. Vaßen, Surf. Coatings Technol., 2017, 318, 157 38. M. Fousova, D. Vojtech, E. Jablonska, J. Fojt and 29. ‘Implants for surgery – Plasma-sprayed unalloyed J. Lipov, Materials, 2017, 10, (9), 987 titanium coatings on metallic surgical implants – Part 1: General requirements’, ISO 13179-1:2014, 39. P. Honigmann, N. Sharma, B. Okolo, U. Popp, International Organization for Standardization, B. Msallem and F. M. Thieringer, BioMed Res. Int., Geneva, Switzerland, June, 2014, 5 pp 2018, 4520636 30. K. A. Gross and M. Babovic, Biomaterials, 2002, 40. W. Jamróz, J. Szafraniec, M. Kurek and 23, (24), 4731 R. Jachowicz, Pharm. Res., 2018, 35, (9), 176

The Authors

Darya Alontseva completed her PhD in Physics at East Kazakhstan State University in 2002. She completed her postdoctoral studies at Altai State Technical University, Russia, in 2013 and received the degree of Doctor of Sciences in Physics and Mathematics. In 2016 she was awarded the academic title of Full Professor of Physics. She has 19 years of research experience in developing new material and processes and management of funded scientific projects. She is a lead researcher in her research area: physics of condensed state, material science and surface engineering.

Elaheh Ghassemieh has 25 years of research experience in the areas of advanced manufacturing including additive manufacturing, development of novel materials especially composites using range of experimental and multiscale numerical methods. After completing her PhD in simulation of micromechanics of composite materials, she worked at several universities in the UK including The University of Sheffield, Queen’s University Belfast and Loughborough University. She has obtained a number of research funding grants from UK and international research councils and has also secured and managed many industrially funded projects where she has transferred novel research outcomes to relevant industrial users in the aerospace, automotive or biomedical fields.

Sergey Voinarovych completed his postgraduate study at E.O.Paton Electric Welding Institute, Ukraine, in 2008 and received a PhD degree in the specialty “Welding and related processes and technologies”. Since 2010 he has been working as a Senior Researcher at E.O.Paton Electric Welding Institute. He has 20 years of research experience in developing new processes, materials and equipment in the area of thermal spray coatings. He has designed MPS equipment and technology for forming biocompatible coatings on parts of endoprostheses. Currently he is researching new biocompatible materials and coatings.

Oleksandr Kyslytsia completed his postgraduate study at E.O.Paton Electric Welding Institute in 2010 and received PhD degree in the specialty “Welding and related processes and technologies”. Since 2011 he has been working as a Senior Researcher at E.O.Paton Electric Welding Institute. For over 20 years he has been developing new processes, materials and equipment for producing coatings by gas thermal spraying methods. He developed equipment and technology for MPS from wire materials to obtain biocompatible coatings on parts of endoprostheses. Currently he is researching new biocompatible materials and coatings.

Yuri Polovetskyi graduated from the Chernihiv Technological Institute, Ukraine, in 1999 with a degree in Welding Technology and Equipment and entered the doctoral program at the E.O.Paton Electric Welding Institute. After completing his doctorate in 2002 and to the present, Polovetskyi is a senior researcher of the department of physical and chemical research of materials. His area of scientific expertise is the structural characteristics, chemical composition and mechanical properties of welded joints and plasma coatings for various purposes.

Nadezhda Prokhorenkova received her PhD in Technical Physics in 2014. She has nine years of research experience in developing new materials and processes. She is an accomplished researcher in her research area: material science. Her area of scientific expertise is X-ray analysis of coating structures. Currently Prokhorenkova is an associate professor at the School of Engineering at D. Serikbayev East Kazakhstan State Technical University.

Albina Kadyroldina received her BS and MS degrees from D. Serikbayev East Kazakhstan State Technical University. She is currently pursuing a PhD with the School of Engineering, D. Serikbayev East Kazakhstan State Technical University. Her research interests include automation, control and mathematical modelling.

www.technology.matthey.com

EuropaCat 2019 Catalysis without borders

Reviewed by Andrew Richardson*, examples of work studying the control of molecular

Katie Smart O2 reactivity including maximising the energy

Johnson Matthey, PO Box 1, Chilton Office, efficiency of O2 reduction to water in fuel cells and Belasis Avenue, Billingham, TS23 1LB, UK achieving selective oxidation of organic molecules without overoxidation to carbon dioxide or other *Email: [email protected] undesirable byproducts (1–3). A high-impact plenary was presented by Professor 1. Introduction Ib Chorkendorff from the Technical University of Denmark. The talk started with a few headline The 14th European Congress on Catalysis data to emphasise that sectors such as aviation, (EuropaCat 2019), themed ‘Catalysis without long haul transport and the chemical industry will Borders’, was held on the 18th–23rd August 2019 always have to rely on the same chemical building at the Eurogress conference centre in Aachen, blocks we use today rather than full electrification. Germany. The conference hosted over 1500 The speaker predicted that the chemical demands participants from academia and industry across the of the future should be met by ‘solar’ fuels in the world, with around 400 lectures and 800 posters form of photovoltaics providing the energy for presented throughout the week. There were six electrochemistry. Several areas of focus within parallel sessions; this review is a small selection of the research team were summarised including a the talks attended by the reviewers. collaboration with Haldor Topsøe, Denmark, in electrified steam reforming (4), green hydrogen 2. Plenary Lectures generation (5) and electrochemical hydrogenation of CO2 (6). Plenary lectures of an hour were intended to Professor Valentin Parmon of the Boreskov showcase the significant breadth and depth of Institute of Catalysis SB RAS in Novosibirsk, Russia, scientific knowledge acquired by established presented an overview of some non-traditional academics over the course of their careers. approaches to achieve various endothermic Summaries of key themes presented in the lectures thermocatalytic transformations. An example was are provided below. the application of nuclear radiation to supply heat Two plenary lectures were delivered on the for steam reforming in an integrated chemical- subject of using molecular oxygen as a reagent, nuclear reactor. A steam reforming catalyst was one by Professor Karen Goldberg of the University developed for the system; a porous uranium oxide of Pennsylvania, USA and another by Professor support with nickel on the surface and titanium Shannon Stahl from the University of Wisconsin- dioxide to protect against nucleotides (7). Solar Madison, USA. Professor Goldberg opened by energy was also considered to provide heating for highlighting how molecular O2 represents the steam reforming, temperatures >1000 K can be ideal oxidant for chemical transformations, with achieved by concentrating solar radiation (8). More the ‘jackpot’ being to achieve benign and facile recent work in the Parmon group focused on the routes to chemicals with atmospheric O2, avoiding use of microwaves to supply energy to reactions

O2 separation from air. Professor Stahl provided such as H2 evolution from alkanes though the cost

Method

Surface organometallic chemistry: a molecular approach

(a) Key steps of surface organometallic chemistry (SOMC)

Step 1: Step 2: grafting Y Step 3: support preparation tailored molecular M O O Ox precursors post-treatment Xx–1 Support H H L M Isolated Δ – vacuum n Y = OH, O or vacant site O O OH O OH O O sites for – H2O LnMXx Support Support Support M = early TM –HX

Metal nanoparticle OH for M = late TM or O coinage metal Support

(b) Well-defined catalysts (c) Supported nanoparticles (d) Isolated metal hydrides (e) Low-coordinate sites via SOMC/TMP

R Hx O (HO)y x M L x M M OH OH O O O O O O OH O O O O O O O

Support Support Support Support

Fig. 1. Molecular processes investigated by the Copéret group reproduced with kind permission from Professor Copéret (12, 13)

of the electrical energy required is prohibitive for and acknowledges the ‘pioneering’ work of Denis large scale application (9). Ballard, Imperial Chemical Industries (ICI), UK. A range of fundamental structure-activity relationships were described in the lecture of 3. Industrial Forum Professor Christophe Copéret from ETH Zürich, Switzerland. Work in the group is focused on The industrial forum sessions of the conference the understanding and control of chemistry are to present work undertaken by industrial on surfaces, with the ultimate goal to generate corporations or of direct industrial application. isolated metal sites with defined chemical Over the course of the conference there were 21 environment to elucidate structure-activity talks in the industrial forum by both academics and relationships. The approach in the team has speakers from commercial organisations. led to highly active and selective single-site catalysts that out-performed their homogeneous 3.1 Growth of Chemicals counterparts, but that also provided useful information to understand industrial catalysts. Joe Scheper from ExxonMobil Chemical Europe Inc, Several examples were provided in the lecture Belgium, opened the industrial forum with a talk including details of research into chromium focussed on the projected growth in global chemical polymerisation catalysts to generate a Cr(II) demand, set to outpace gross domestic product species directly rather than the need to reduce (GDP) growth. Attention was drawn to <15% of a Cr(VI) in situ (10). An example was also provided barrel of oil being converted to chemicals pre-1990, for copper-particle catalysed methanol synthesis versus 25–40% at present, and projected to be from methane, investigating the effect of the 40–80% post 2020. Specific examples were given support; copper on alumina displayed higher of technology to enable this growth, including the activity than copper on silica (11). Figure 1 ExxonMobil EMTAMSM process for selective para- summarises work undertaken in the Copéret team xylene production from toluene and methanol.

electrifies, oil capacity will be liberated for use in +H2 chemicals and a 4% annual growth in chemicals + C8 2 could be expected. Some data was also shared

looking at CO2 cost versus product value, with formaldehyde and monoethylene glycol coming Scheme I out as the most expensive (i.e. relatively cheap

chemical product made with high CO2 emissions). Para-xylene is smaller than the meta- or ortho- variants, so can be targeted for selective transport 3.2 European Federation of Catalysis through specifically sized zeolite pores. Avoiding Societies Applied Catalysis Award reaction on the outer surfaces of zeolites minimises byproduct formation and catalyst attrition resistance The European Federation of Catalysis Societies was highlighted as critical for this fluid bed (EFCATS) award for 2019 was awarded to Professor regenerative process. Some future developments Glenn Sunley of BP Plc, UK, and was accompanied required to sustain the necessary global growth in by a talk titled ‘Adventures in C1 Chemistry’. Glenn chemicals were presented. Processes to take plastic talked through some of his career highlights, waste and turn this into a chemical feedstock as well including work done on the BP CativaTM process as algae-based routes to biofuels with 50,000 barrels for acetic acid production and the elucidation of (bbl) of petroleum products being projected to be the reaction mechanism in collaboration with the produced from algae in the not too distant future. University of Sheffield, UK (15). Work was also Andrei Parvulescu from BASF SE, Germany, then presented about a collaboration with University of went on to discuss growing demand for highly California, Berkeley, USA and the California Institute branched C7 and C8 alkanes owing to their superior of Technology, USA, investigating the conversion fuel burning efficiency. C4 alkene dimerisation or and mechanistic features of methanol homologation alkylation of isobutane with propene or butenes to triptane over an indium iodide catalyst (16, 17). were presented as routes (Scheme I). Special mention was also given to some recent Solid acid catalysts have been proposed for both Fischer-Tropsch technology development as a of these processes but suffer from relatively poor result of the long-term collaboration between BP stability and selectivity. One successfully evaluated and Johnson Matthey, UK, described as ‘stunning catalyst for alkylation was a zeolite beta made via engineering’. The first license for this technology seed-assisted synthesis without organic structure has already been sold and will enable Fulcrum directing agent, providing a high density of acid sites BioEnergy Inc, USA, to convert 175,000 tonnes of and easy ability to be modified by dealumination. household rubbish into 11 million gallons of jet fuel The catalyst was shown to have superior activity each year (18, 19). and selectivity compared to conventional zeolite beta and control of product mix was demonstrated 3.3 SunCarbon – Creating a through manipulation of reaction temperature and New Value Chain from Forest to olefin concentration (14). Refineries Jean-Paul Lange from Shell Global Solutions International BV, The Netherlands, continued the Christian Hulteberg from SunCarbon (both a theme of growth in chemicals, discussing how process and Swedish based company) described as populations become wealthier demand for their newly launched process for the transformation petrochemicals will rise. Fossil fuels would be able of lignin into vehicle fuel. This process is dual- to meet this increase in demand, as they are still function, removing lignin from a typical pulp mill abundant and will have lower demand in future as cycle allowing debottlenecking of the pulp mill the automotive sector decarbonises. Shell analysis evaporator and also providing the mill with an predicts that the petrochemical industry will be additional value stream. Lignin is extracted by affected by limited carrying capacity of the earth membrane filtration and then homogenised in a and societal calls for a more circular economy. homogeneously catalysed process. The resultant Therefore, focus must be placed on transforming product is then purified and mixed with vacuum the linear petrochemicals industry into a circular one gas oil to make it liquid and pumpable, before being built on recycled plastics. There was also agreement shipped to a refinery. The adjacent pulp mill is between Shell and the earlier discussions led by capable to take care of any and all waste products ExxonMobil, with Shell predicting that as transport of the SunCarbon process. Once at the refinery,

194 © 2020 Johnson Matthey https://doi.org/10.1595/205651320X15822789454902 Johnson Matthey Technol. Rev., 2020, 64, (2) the material undergoes standard hydrotreatment Shohei Tada from The University of Tokyo, Japan, and is transformed into vehicle fuel. The feedstock gave a talk on work done investigating catalysts was reported to contain up to 20% O, which would for the conversion of CO2 to methanol. Zirconia place significant additional H2 demand on the supported Cu catalysts were discussed and work was refinery hydrotreater. presented showing Cu species being metallic during reaction and methanol synthesis being performed at 4. Carbon Dioxide Utilisation Cu-ZrO2 interface sites (22). Tada went on to discuss the importance of finding a good support material to An ongoing theme of EuropaCat 2019 was the suppress methanol decomposition to CO, with the focus of many research groups on utilisation of methanol-support interaction being said to be key.

CO2. Many talks were presented on addressing Methanol weakly adsorbs on an amorphous surface various aspects of the Sabatier reaction for CO2 (a-ZrO2) and strongly adsorbs on a monoclinic hydrogenation, on the dry reforming of methane surface (m-ZrO2), with the stronger interaction

(DRM) for H2 production or for the direct synthesis yielding more unfavourable decomposition of of methanol from CO2. methanol to CO (23, 24). Matteo Monai from Utrecht University, The Netherlands, presented a short talk on work done 5. Conclusions tuning metal-support interactions on supported Ni catalysts for the Sabatier reaction hydrogenating EuropaCat 2019 was a large conference with wide-

CO2 to CH4 (Equation (i)): ranging themes delivered in multiple sessions over six days. It has not been possible to summarise CO2 + 4H2 CH4 + 2H2O (i) the conference in its entirety. Major topics Both a reaction→ mechanism with carbide and included: alternative energy inputs (plasma, formate intermediates are observed on Ni catalysts electrochemistry, light); ammonia synthesis; and both mechanisms were reported to be biomass valorisation; CO2 to chemicals; dry sensitive to particle size and local structure. Using reforming; Fischer-Tropsch catalysis; industrial strong metal-support interactions and a reducible forum; mechanistic insights; nitrogen oxides (NOx) support, it was possible to ‘embed’ Ni particles reduction; organometallic catalysis; and zeolite into the support material and suppress sintering. catalysis. More information can be found on the Enhanced C–C coupling was also observed on conference website. these materials when utilising the embedded metal particles, the mechanism for which was still subject References to investigation (20). An experimental programme investigating the 1. J. M. Keith, R. P. Muller, R. A. Kemp, K. I. Goldberg, mechanism for carbon formation over rhodium W. A. Goddard and J. Oxgaard, Inorg. Chem., on alumina catalysts in the DRM was shared by 2006, 45, (24), 9631 Gianluca Moroni from the Politecnico di Milano, Italy (Equation (ii)): 2. J. L. Look, D. D. Wick, J. M. Mayer and K. I. Goldberg, Inorg. Chem., 2009, 48, (4), 1356 CH + CO 2CO + 2H (ii) 4 2 2 3. M. C. Denney, N. A. Smythe, K. L. Cetto, R. A. Kemp and K. I. Goldberg, J. Am. Chem. Soc., A variety of → CH4:CO2 ratios were flowed over the catalyst at temperatures from 300–700°C 2006, 128, (8), 2508 in an operando-Raman annular reactor with 4. S. T. Wismann, J. S. Engbæk, S. B. Vendelbo, gas analysis by Micro GC Natural Gas Analyzer F. B. Bendixen, W. L. Eriksen, K. Aasberg-Petersen, (Agilent, USA). An adverse dependence between C. Frandsen, I. Chorkendorff and P. M. Mortensen, Science, 2019, 364, (6442), 756 CH4 concentration and activity was observed and a presented microkinetic model was said to predict 5. J. Kibsgaard and I. Chorkendorff, Nature Energy, experimentally observed activity with catalyst 2019, 4, (6), 430 surface area as an input. Catalyst activity was 6. S. Nitopi, E. Bertheussen, S. B. Scott, also observed to reduce with increased carbon X. Liu, A. K. Engstfeld, S. Horch, B. Seger, monoxide levels, and C deposition was said to begin I. E. L. Stephens, K. Chan, C. Hahn, J. K. Nørskov, on the Rh sites before migration and deposition on T. F. Jaramillo and I. Chorkendorff, Chem. Rev., adjacent sites on the support (21). 2019, 119, (12), 7610

7. Y. I. Aristov, Y. Y. Tanashev, S. I. Prokopiev, M. Payne, J. M. Pearson, M. J. Taylor, P. W. Vickers L. G. Gordeeva and V. N. Parmon, Int. J. Hydrogen and R. J. Watt, J. Am. Chem. Soc., 2004, 126, Energy, 1993, 18, (1), 45 (9), 2847 8. Y. I. Aristov, V. I. Fedoseev and V. N. Parmon, Int. 16. J. E. Bercaw, P. L. Diaconescu, R. H. Grubbs, J. Hydrogen Energy, 1997, 22, (9), 869 N. Hazari, R. D. Kay, J. A. Labinger, 9. S. Horikoshi, M. Kamata, T. Sumi and N. Serpone, P. Mehrkhodavandi, G. E. Morris, G. J. Sunley and Int. J. Hydrogen Energy, 2016, 41, (28), 12029 P. Vagner, Inorg. Chem., 2007, 46, (26), 11371 10. M. P. Conley, M. F. Delley, G. Siddiqi, G. Lapadula, 17. J. H. Ahn, B. Temel and E. Iglesia, Angew. Chemie S. Norsic, V. Monteil, O. V. Safonova and Int. Ed., 2009, 48, (21), 3814 C. Copéret, Angew. Chemie Int. Ed., 2014, 53, 18. ‘JM and BP License Waste-to-Fuels Technology to (7), 1872 Fulcrum BioEnergy’, Johnson Matthey, London, 11. J. Meyet, K. Searles, M. A. Newton, M. Wörle, A. P. UK, 25th September, 2018 van Bavel, A. D. Horton, J. A. van Bokhoven and 19. A. Coe and J. Paterson, Chem. Eng., 2019, C. Copéret, Angew. Chemie Int. Ed., 2019, 58, (937/938), 33 (29), 9841 20. C. Hernández Mejía, C. Vogt, B. M. Weckhuysen 12. C. Copéret, A. Comas-Vives, M. P. Conley, Deven and K. P. de Jong, Catal. Today, 2020, 343, 56 P. Estes, A. Fedorov, V. Mougel, H. Nagae, 21. A. Maghsoumi, A. Ravanelli, F. Consonni, F. Nanni, F. Núñez-Zarur and Pavel A. Zhizhko, Chem. Rev., A. Lucotti, M. Tommasini, A. Donazzi and 2016, 116, (2), 323 M. Maestri, React. Chem. Eng., 2017, 2, (6), 908 13. C. Copéret, Acc. Chem. Res., 2019, 52, (6), 1697 22. K. Larmier, S. Tada, A. Comas-Vives and 14. B. Yilmaz, U. Müller, M. Feyen, S. Maurer, H. Zhang, C. Copéret, J. Phys. Chem. Lett., 2016, 7, (16), X. Meng, F.-S. Xiao, X. Bao, W. Zhang, H. Imai, 3259 T. Yokoi, T. Tatsumi, H. Gies, T. De Baerdemaeker 23. S. Tada, A. Katagiri, K. Kiyota, T. Honma, H. Kamei, and D. De Vos, Catal. Sci. Technol., 2013, 3, (10), A. Nariyuki, S. Uchida and S. Satokawa, J. Phys. 2580 Chem. C, 2018, 122, (10), 5430 15. A. Haynes, P. M. Maitlis, G. E. Morris, G. J. Sunley, 24. S. Tada, S. Kayamori, T. Honma, H. Kamei, H. Adams, P. W. Badger, C. M. Bowers, D. B. Cook, A. Nariyuki, K. Kon, T. Toyao, K. Shimizu and P. I. P. Elliott, T. Ghaffar, H. Green, T. R. Griffin, S. Satokawa, ACS Catal., 2018, 8, (9), 7809

The Reviewers

Andrew Richardson graduated from the University of Newcastle Upon Tyne, UK with an MChem in Chemistry. Andrew joined Johnson Matthey in 2012 and has worked across a number of catalysis and absorbent research areas. He is currently a Principal Researcher at Johnson Matthey’s Chilton site in Billingham, UK.

Katie Smart has a PhD in Chemistry from the University of Paul Sabatier, France, where she studied at the Laboratoire de Chimie de Coordination in Toulouse. Katie has worked at Johnson Matthey since 2013 and is currently Technical Development Manager for the high-temperature shift research and development team at the Chilton site in Billingham.

www.technology.matthey.com

“Solid-State NMR in Zeolite Catalysis” By Jun Xu, Qiang Wang, Shenhui Li and Feng Deng (Wuhan Institute of Physics and Mathematics, China), Lecture Notes in Chemistry, Vol. 103, Springer Singapore, 2019, 260 pages, ISBN: 978-981-13-6965-0, £109.99, €139.09, US$159.99

Reviewed by Shingo Watanabe Principles of Nuclear Magnetic Nanoramic Laboratories, 21 DryDock Ave, Suite Resonance 820E, Boston, MA 02210, USA In Chapter 1, the authors provide a snapshot of Email: [email protected] NMR theory with the current state of methodology development, including strategies to enhance NMR sensitivity. This chapter gives sufficient background Introduction to understand the subsequent chapters without “Solid-State NMR in Zeolite Catalysis” was written overwhelming. by four professors from the Wuhan Institute of Physics and Mathematics, Chinese Academy of Solid-State Nuclear Magnetic Sciences, who have tremendous expertise in the Resonance Studies of Zeolites and fields of solid-state nuclear magnetic resonance Related Materials (solid-state NMR) and heterogeneous catalysis. This book is Volume 103 in the series ‘Lecture Highlights and lowlights of solid-state NMR for Notes in Chemistry’ published by Springer. It can be material characterisation are addressed at the organised into three sections: (a) principles of NMR beginning of Chapter 2. The chapter goes on to theory (Chapter 1); (b) solid-state NMR studies of explain the local framework structure characterised zeolites and related materials (Chapters 2 and 3); by different chemical environments in zeolites. and (c) surface chemistry and reaction chemistry NMR spectroscopy provides detailed information over zeolites (Chapters 4–6). Each chapter offers on synthesis mechanisms in the solid and liquid a sufficient introduction to move smoothly into phases, which may be complementary to X-ray the main content. The book guides readers to diffraction and electron microscopy methods that familiarise themselves with solid‑state NMR are often used to gain structural information. and gives sufficient practical examples, detail The application of NMR spectroscopy in zeolite and illustration. However, visually supported crystallisation, in situ and ex situ methods is explanations by the authors were included only also introduced to investigate the kinetics of after Chapter 1. Adding to this could help deepen crystallisation for the convenience of a general the understanding of pre-graduate readers. A few audience. As an example, the structural changes abbreviations were not explained, which could in the aluminosilicate gels of zeolite faujasite confuse readers. Advantages and disadvantages (FAU) were confirmed in addition to the changes of solid-state NMR for each purpose are addressed in silicon:aluminium ratios during the ageing and in comparison with other techniques in each crystallising processes. In addition to zeolite chapter. The discussion in this book further applies analysis, xenon-129 chemical shifts were used to materials for heterogeneous catalysts, glasses, to examine the pore geometry and chemical ceramics, superconductors, lithium-ion batteries surroundings. This also fits the discussion of the and biological systems. guest-host interaction study. Additional visual aids

197 © 2020 Johnson Matthey https://doi.org/10.1595/205651320X1582541146190 Johnson Matthey Technol. Rev., 2020, 64, (2) could improve understanding in this chapter. The reading and reminded the reviewer of the ability basics of characterisation of Brønsted acid sites may of solid-state NMR to assess the mechanisms of also be complemented by a review by Hunger (1). intermolecular interactions, chemical reactions and transport phenomena. Additional visual supporting Surface Chemistry and Reaction materials should be added to deepen the readers’ understanding of the authors’ explanations, Chemistry Over Zeolites particularly in Chapter 4. NMR spectroscopy has the advantage of providing detailed information on local bonding “Solid-State and solid- state interactions. Investigations on NMR in Zeolite host- guest interactions between organic molecules, Catalysis” such as aromatics to surfactants and inorganic moieties of zeolites, are well summarised in Chapter 4 with sufficient examples and explanation of local structures and dynamic behaviours between adsorbed molecules and active sites. Although several informative figures were cited from references, additional visual supports to explain the host-guest interaction would further elucidate this interaction. The content of Chapter 4 reintroduced and extended this reviewer’s understanding of Li‑ion battery and Li-ion ultracapacitor materials (2). Chapters 5 and 6 elucidate interfacial chemistry by solid-state NMR approaches. Several representative reactions are given including in situ studies with a focus on how surface and interfacial phenomena significantly influence the catalytic performances (activity and selectivity) of heterogeneous catalysts. Many essential aspects of interfacial chemistry can References be emphasised on the chemical and electrochemical 1. M. Hunger, Catal. Rev.: Sci. Eng., 1997, 39, (4), reactions at interfaces. The following readings 345 may bring additional value (3, 4), and this is also 2. C. P. Grey and N. Dupré, Chem. Rev., 2004, 104, applicable for an understanding of Li-ion battery (10), 4493 and Li-ion ultracapacitor materials (5). 3. H. Koller and M. Weiß, ‘Solid State NMR of Porous Materials’, in “Solid State NMR”, ed. J. C. C. Chan, Conclusions Topics in Current Chemistry, Vol. 306, Springer- Verlag, Berlin, Heidelberg, Germany, 2012, pp. The reviewer would like to recommend this book 189-227 for anyone who would like to quickly grasp the 4. W. Zhang, S. Xu, X. Han and X. Bao, Chem. Soc. capability of solid-state NMR as well as researchers Rev., 2012, 41, (1), 192 working on nanoporous materials, nanocrystals, 5. N. Leifer, M. C. Smart, G. K. S. Prakash, L. nanomaterials, Li-ion batteries and capacitors with Gonzalez, L. Sanchez, K. A. Smith, P. Bhalla, C. P. surface and interfacial phenomena in addition to Grey and S. G. Greenbaum, J. Electrochem. Soc., zeolites and zeolite related materials. This is exciting 2011, 158, (5), A471

The Reviewer

Shingo Watanabe received his PhD from the Pennsylvania State University, USA, in the field of fuel, surface and materials chemistry. He worked for Johnson Matthey, USA, for eight years in heterogeneous catalysis and purification in petrochemicals, oleochemicals, biorenewable chemicals and gas and liquid purification. Currently, he is working for a start-up company, Nanoramic Laboratories, USA, as a business development director for the Japan market in the areas of Li-ion batteries, ultra- and supercapacitors and thermal interface materials.

www.technology.matthey.com

Johnson Matthey Highlights A selection of recent publications by Johnson Matthey R&D staff and collaborators

Activity of Molybdenum and Tungsten Oxycarbides Correlating Physicochemical Properties of in Hydrodenitrogenation of Carbazole Leading to Commercial Membranes with CO2 Absorption Isomerization Secondary Reaction of Bicyclohexyl. Performance in Gas-Liquid Membrane Contactor Results Using Bicyclohexyl as Feedstock Y. Xu, C. Malde and R. Wang, J. Membr. Sci. Res., M. Lewandowski, A. Szymańska-Kolasa, C. Sayag 2020, 6, (1), 30 and G. Djéga-Mariadassou, Appl. Catal. B: Environ., Polypropylene (PP) and polyvinylidene fluoride 2020, 261, 118239 (PVDF) hollow fibre membranes were applied to Bulk W and Mo oxycarbides were synthesised investigate the impact of their physicochemical and characterised by a range of physicochemical properties on long-term CO2 absorption techniques. The catalytic behaviour of the performance in a gas-liquid membrane oxycarbides was tested for the carbazole contactor (GLMC). CO2 transport was hindered hydrodenitrogenation (HDN) and the bicyclohexyl by membrane wetting and fouling when water (BCH) isomerisation reactions. High isomerisation was used as an absorbent, which resulted in catalytic activity was demonstrated for both continuous flux during long-term operation. oxycarbides. The addition of 50 ppm of sulfur When monoethanolamine (MEA) was used as increased this activity further during the HDN an absorbent, both PP and PVDF membranes demonstrated dramatic flux decline. The pore of carbazole. W2C demonstrated the highest catalytic activity for both pure BCH isomerisation size, stability, morphology and hydrophobicity and secondary BCH isomerisation during the of commercial membranes were shown to be HDN of carbazole. Ethylbicyclo[4.4.0]decane (an affected by MEA over long-term operations. isomerisation product) and n-hexylcyclohexane (a Therefore, the authors propose selection criterion ring opening product) were the main BCH isomers of microporous membranes. detected for both oxycarbides. The presence of Implications of the Molybdenum Coordination these products, in significant quantities, indicated Environment in MFI Zeolites on Methane that the bulk oxycarbides are bifunctional catalysts Dehydroaromatisation Performance with both acid and metallic sites. M. Agote-Arán, R. E. Fletcher, M. Briceno, Solid State NMR Service Across the World A. B. Kroner, I. V. Sazanovich, B. Slater, M. E. Rivas, A. W. J. Smith, P. Collier, I. Lezcano-González and N. S. Barrow and P. Jonsen, Solid State Nucl. A. M. Beale, ChemCatChem, 2020, 12, (1), 294 Magn. Reson., 2020, 105, 101626 In situ XAS and DFT were used to compare the 24 global SSNMR laboratories were interviewed on structure and activity of Mo/Silicalite-1 (Si:Al = ∞) the telephone and face-to-face in early 2019. Data to Mo/H-ZSM-5 (Si:Al = 15). Calcination in was collected related to service throughput, staff, Mo/ Silicalite-1 was shown to disperse the MoO3 barriers and equipment. The hardware profile precursor into tetrahedral Mo-oxo species, in close observed in this study agreed with a previous proximity to the microporous framework. Both report published in 2013 which primarily looked catalysts were shown to be active for methane at SSNMR in the UK. This study highlights that dehydroaromatisation (MDA). Mo/Silicalite-1 a lack of knowledge about SSNMR capabilities is demonstrated faster sintering of the Mo species, the biggest barrier. By conducting this research, which led to rapid deactivation. This contributed to a strong benchmark has been set. SSNMR the accumulation of carbon deposits on the zeolite laboratories can therefore identify their barriers outer surface. Thus, this study highlights the and implement changes to maximise the use of importance of framework Al for the stabilisation SSNMR within their own research. of active Mo species, when under MDA conditions.

Strength and Fragmentation Behaviour of Flow and Forced Convection Heat Transfer Complex-Shaped Catalyst Pellets: A Numerical and Characteristics of Developed Laminar Flow in the Experimental Study Octahedral Channels of Octo-Square Asymmetric A. Farsi, J. Xiang, J. P. Latham, M. Carlsson, Particulate Filters E. H. Stitt and M. Marigo, Chem. Eng. Sci., 2020, T. C. Watling, Res. Eng., 2020, 5, 100086 213, 115409 Octahedral channels are present in octo-square The relationship between catalyst support shape asymmetrical diesel and gasoline particulate filters. and final strength and fragmentation behaviour In this study, the flow and forced convection heat was investigated. Pellet crushing behaviours were transfer characteristics of these channels were examined by performing uniaxial compression investigated. Least squares and point matching tests on solid and four-holed discs (Figure 1). The methods were employed to determine the finite-discrete element method (FDEM) was used temperature and velocity fields, with the least to provide a detailed analysis of the fragmentation squares method being the most effective. The evolution of pellets. The FDEM method also revealed viscous loss coefficient and the product of the a primary failure undetected in laboratory tests. channel perimeter and heat transfer coefficient Concentration of compressive stress in the pellet decreased as the channel cross section became core led to a greater fraction of fines and the choking closer to a regular octahedron, whereas the effect of these fines contributed to pressure drops. friction factor and Nusselt number were shown to These drops had a greater impact on fixed-bed increase. When comparing an octahedral channel reactors made with solid cylindrical catalysts than and a square channel of the same width, there was to those made with four-holed catalyst supports. minimal difference between the heat transfer and along-channel pressure drop. Structure and Ion Transport of Lithium-Rich Li1+xAlxTi2−x(PO4)3 with 0.3

Measuring Velocity and Turbulent Diffusivity in Wall‑Flow Filters Using Compressed Sensing Magnetic Resonance J. D. Cooper, A. P. E. York, A. J. Sederman and L. F. Gladden, Chem. Eng. J., 2019, 377, 119690

Recirculating flows and turbulent diffusivity in wall-flow particulate filters were observed for the first time using gas-phase compressed sensing magnetic resonance (MR) methods. Entrance and exit gas flow distributions were characterised Fig. 1. Reprinted from A. Farsi et al., Chem. Eng. using 2D MR velocity imaging. Fast 3D compressed Sci., 2020, 213, 115409, Copyright (2020), with sensing MRI was used to quantify turbulent permission from Elsevier diffusivity. Contrary to numerical predictions, the

200 © 2020 Johnson Matthey https://doi.org/10.1595/205651320X15832385550263 Johnson Matthey Technol. Rev., 2020, 64, (2) data collected demonstrated that different regions 100% perovskite phase was observed after 3 h of of turbulent diffusivity were present within the filter. milling, as derived from XRD analysis. However, For instance, at the entrance, two different regions results from XAS analysis demonstrated that of turbulent diffusivity were observed, both within significant structural alterations took place after the inlet channels. just 30 min. XAS also highlighted the presence of amorphous, oxygen deficient, content. Increased Synthesis and Characterization of LiFe1−xMnxPO4 (x = 0.25, 0.50, 0.75) Lithium Ion Battery Cathode oxygen deficiency at the surface led to the LaMnO3 Synthesized via a Melting Process catalyst displaying early onset production of N2 (in comparison to sol-gel synthesised LaMnO ). E. B. Fredj, S. Rousselot, L. Danis, T. Bibienne, 3 Therefore, mixed metal oxide catalysts with M. Gauthier, G. Liang and M. Dollé, J. Energy enhanced catalytic properties can be prepared via Storage, 2020, 27, 101116 mechanochemical routes. Melt synthesis, a low-cost and simple method, was used to synthesise electrochemically active Influence of Synthesis Conditions on the Structure of Nickel Nanoparticles and their Reactivity in LiFe Mn PO (x = 0.25, 0.50, 0.75) cathode 1− x x 4 Selective Asymmetric Hydrogenation materials for the first time. SEM, XRD and galvanostatic charge/discharge cycling were used R. Arrigo, S. Gallarati, M. E. Schuster, J. M. Seymour, D. Gianolio, I. da Silva, J. Callison H. Feng, to characterise the LiFe1−xMnxPO4 materials. Results were compared to those of solid‑state J. E. Proctor, P. Ferrer, F. Venturini, D. Grinter and synthesised materials. Overall, capacity retention, G. Held, ChemCatChem, 2020, 12, (5), 1491 rate capability and discharge capacity were similar A hot-injection colloidal route was used to for both materials. However, melt synthesised synthesise unsupported and silica-supported Ni LiFe0.25Mn0.75PO4 had a higher capacity than NPs. NP size and Ni electronic structure were solid‑state synthesised LiFe0.25Mn0.75PO4. Therefore, affected by changing equivalents of reducing melt synthesis could offer a viable alternative to and protective agents. (R,R)-Tartaric acid (TA) current synthetic techniques. was used to modify the NPs which were then investigated in the asymmetric hydrogenation of Understanding the Mechanochemical Synthesis of methyl acetoacetate to chiral methyl-3‑hydroxy the Perovskite LaMnO3 and its Catalytic Behaviour butyrate. A Ni metallic active surface was R. H. Blackmore, M. E. Rivas, T. E. Erden, identified where activity was shown to increase T. D. Tran, H. R. Marchbank, D. Ozkaya, with metallic domain size. Particle size had M. Briceno de Gutierrez, A. Wagland, P. Collier and no impact on selectivity for unsupported NPs. P. P. Wells, Dalton Trans., 2020, 49, (1), 232 Catalysts that contained positively charged Ni Lanthanum manganite was synthesised from species demonstrated very high (R)-selectivity. metal oxide powders (Mn2O3 and La2O3) via At long reaction times, TA modification of metallic mechanochemistry using a planetary ball mill. Ni NPs was shown to be unsatisfactory for the During the milling process, ‘time slices’ were taken. maintenance of high (R)-enantiomer selectivity.

www.technology.matthey.com

A Short Review on Properties and Applications of Zinc Oxide Based Thin Films and Devices ZnO as a promising material for applications in electronics, optoelectronics, biomedical and sensors

Sumit Vyas growing semiconductor industry. Nanotechnology, Department of Electronics and Communication in which the products contain very small particles Engineering, Thapar Institute of Engineering and demonstrate special properties, is one of the and Technology, Patiala-147001, India most recent and active areas of research. In this regard, thin-film technology plays an important Email: [email protected] role that allows deposition of very thin layers (from a few nanometres down to the angstrom level) of semiconductor material on a supporting substrate. Zinc oxide has emerged as an attractive material for The resulting material exhibits novel mechanical, various applications in electronics, optoelectronics, chemical, optical and electrical properties with the biomedical and sensing. The large excitonic reduction in size to the nanometre scale, which is binding energy of 60 meV at room temperature as the result of surface and quantum confinement compared to 25 meV of gallium nitride, an III-V effects. compound makes ZnO an efficient light emitter A thin film is defined as a very thin layer (10 nm in the ultraviolet (UV) spectral region and hence to 1–2 µm) of material deposited on a supporting favourable for optoelectronic applications. The material (substrate) by the controlled condensation high conductivity and transparency of ZnO makes of vapours, ions or molecules by a physical or it important for applications like transparent chemical process (1). This technology is known conducting oxides (TCO) and thin-film transistors as thin-film technology. Thin films are deposited (TFT). In this paper, the optoelectronic, electronic over a wide range of substrates (2–4). Thin films and other properties that make ZnO attractive for can be classified based on material into various a variety of applications are discussed. Various categories: for example metallic, dielectric, organic applications of ZnO thin film and its devices or semiconductor films. The material can be in such as light-emitting diodes (LED), UV sensors, monocrystalline, polycrystalline or amorphous biosensors, photodetectors and TFT that have forms. The properties of thin films are completely been described by various research groups are different from their bulk form. Materials in bulk presented. form have fixed properties whereas the properties of thin films and devices depend on the quality 1. Introduction of the surface rather than the bulk (5). Also, the properties of thin films can be modulated by The dependence of daily life on the products various techniques like doping, thickness variation of the semiconductor industry has resulted in or surface treatments. Multilayer thin films can enormous growth of this industry. Progress exhibit completely unknown properties. Thin-film demands the development of smaller and smaller technology also makes efficient use of raw material. devices with higher speed, flexibility, better The progressive development of thin-film performance and lower cost. This demand has technology has resulted in its extensive use in resulted in the development of new technologies fields of optics, electronics, aircraft, defence, and materials to meet the requirements of the space science and other industries. The categories

202 © 2020 Johnson Matthey https://doi.org/10.1595/205651320X15694993568524 Johnson Matthey Technol. Rev., 2020, 64, (2) in which thin-film technology finds applications photovoltaic devices, batteries, sensors, information are mechanical, chemical, thermal, electrical, storage, lighting and large-area electronics. Various magnetic, electronic, chemical, optical and materials like silicon, GaN, gallium arsenide and optoelectronic (2). The main applications of thin- oxide-based semiconductors (including ZnO) film technology primarily include optical coatings (7–16) have continued to receive considerable and semiconductor thin film devices. Various attention for fundamental as well as application- applications of thin-film technology are listed in oriented research. However, research interest in Table I. A thin film of materials can be deposited ZnO is enormously growing because of its excellent from the gas, vapour, liquid or solid phase. In optical, electrical, magnetic, piezoelectric, catalytic Figure 1, various thin film deposition methods are and gas-sensing properties that make it specifically classified and summarised (6). attractive for nanoelectronic, optoelectronic, With advances in nanotechnology and thin film nanophotonic and piezoelectric devices (17, 18). deposition techniques, significant interest has been Different nanostructures of ZnO including developed in recent years for the development of nanorods, nanowires, nanotubes and nanoribbons

Table I Applications of Thin Film Technology Category Typical applications Engineering and Protective layers and low friction coatings to reduce wear, corrosion and erosion; high- processing temperature corrosion; surface passivation; decorative coatings; catalytic coatings Photodetectors; liquid crystal display (LCD); TFT; optical memories; light amplification by Optoelectronics stimulated emission of radiation (laser); LED Integrated optics; antireflex and high reflecting coatings (laser mirrors, interference filters, Optics mirrors); beam splitter; thin film polariser Active thin film elements (diodes, transistor); passive thin film element (interconnects, Electronics resistors, condensers); charge coupled device (CCD); very large scale integrated circuits (VLSI) Superconducting quantum interference devices (SQUIDS); superconducting thin films; Cryotechnics switches; memories Sensors Gas sensor; biosensors

Thin film deposition methods

Vacuum based Solution based

Chemical vapour deposition Physical vapour deposition (CVD) (PVD)

Plasma enhanced CVD (PECVD) Sputtering Printing Low pressure CVD (LPCVD) Thermal evaporation Electrode deposition Metal organic CVD (MOCVD) Electron beam evaporation Spin coating Pulsed laser deposition Colloidal synthesis Molecular beam epitaxy Layer by layer synthesis Atomic layer deposition

Fig. 1. Classification of thin film deposition techniques

(19, 20) can be deposited on various substrates high-speed electronic devices and optoelectronic using conventional thin film deposition methods devices in the near-IR region. However, due to like radiofrequency (rf) sputtering, thermal the narrow bandgap of GaAs, it does not possess evaporation and sol-gel (11). With the availability the properties for optoelectronic devices in the of large single-crystal ZnO, epitaxial films with UV/blue spectral range. Optoelectronic devices in very few defects can be obtained hence very high the UV/blue spectral range are in great demand performance electronic and optoelectronic devices for commercial applications in astronomy, medical, can be fabricated. The processing temperature of healthcare, water treatment and the military. ZnO nanostructures is very low. Therefore, cheap The development of blue LED has resulted in substrates like glass and plastic can also be used the development of low-power white LED that is for fabricating ZnO-based devices. Moreover, the replacing incandescent and fluorescent lighting. electrical and optical properties of ZnO can be The blue LED has also resulted in the development easily tuned by post-deposition treatments like of blue-ray discs for storing high-definition video. annealing, surface treatments and doping with Therefore, wide bandgap semiconductors such as materials like aluminium, gallium, indium, tin GaN and ZnO have received considerable attention. and copper (21–25). It is an n-type transparent For semiconductor-based photonic devices such as material with a direct bandgap of 3.37 eV with UV/blue LED and laser diodes, wide bandgap group good electrical conductivity (26–28). Therefore, III-nitrides have been the focus of intensive research it can also be used for near-UV emission and due to their specific properties (9). However, detection, as a transparent conductor and as a research interest in ZnO is growing because of its channel material in TFT. large excitonic binding energy of 60 meV at room This paper presents the various important temperature as compared to 25 meV of GaN, which properties that make ZnO suitable for electronic makes ZnO an efficient light emitter in the UV and optoelectronic applications. Further, research spectral region. Also, the crystal growth technology into applications of ZnO thin films and its and processing of GaN is complex as compared devices including LED, biosensors, UV sensors, to that of ZnO thin films and crystals that make photodetectors and TFT given by various research it more attractive for optoelectronic devices in the groups are presented. UV/blue spectral range (29–31).

2. Relevant Semiconductor Materials 3. Relevant Semiconductor Materials for Optoelectronic Applications for Thin-Film Transistors

Based on the bandgap, semiconductor materials At present hydrogenated amorphous-Si (a-Si:H) can be divided into two categories: narrow bandgap and polycrystalline-Si (poly-Si) are commercially and wide bandgap materials. They can be further used for large area display TFT and high-speed, classified as indirect bandgap and direct bandgap high-resolution displays respectively. However, materials. Narrow bandgap materials with a direct a-Si TFT has a low field-effect mobility value bandgap are desired for optoelectronic devices in that makes it unacceptable for high-resolution the visible/infrared (IR) region whereas materials displays with faster switching speeds. The field- with a wide and direct bandgap are desired for effect mobility of poly-Si TFT is very high but it optoelectronic devices in the UV/blue region. It is requires a very high-temperature crystallisation well known that Si dominates the semiconductor and is a very time-consuming process. As a result, industry due to its exceptional material properties the cost and time of production both increase. The and compatibility with conventional processing. high processing temperature restricts the use of However, the indirect bandgap of Si greatly limits substrates like glass and plastic. Poly-Si TFT suffer its application in optoelectronic devices. Therefore, from non-uniform electrical properties due to its

GaAs, a direct bandgap material (Eg = 1.43 eV) with polycrystalline nature that makes it unsuitable very high electron mobility (>8500 cm2 V–1 s–1) for large-area displays. Also, Si is sensitive to and related III-V compounds like indium gallium light because of its low bandgap, therefore its arsenide and aluminium gallium arsenide are characteristics degrade on exposure to visible used for fabricating optoelectronic devices like light. Hence, shielding is required that limits the LED, lasers and other very high-speed electronic resolution of the display (11). Considering all devices (7, 12). GaAs and its related materials these limitations there is a constant search for have many advantages and are suitable for very new materials and ZnO seems promising. ZnO thin

204 © 2020 Johnson Matthey https://doi.org/10.1595/205651320X15694993568524 Johnson Matthey Technol. Rev., 2020, 64, (2) film has high field-effect mobility, is insensitive to surrounded by four O atoms and vice versa. ZnO visible light and has a low processing temperature exists in three crystal structures i.e., wurtzite, zinc (18). Further, the quality of its film and devices can blende and rock salt. At ambient conditions, ZnO be very easily enhanced by doping with materials exists in wurtzite form (11). A stable zinc blende like In, Ga or Al. phase can be achieved by growing ZnO on a cubic substrate (32–34). The rock salt structure can be 4. Properties of Zinc Oxide obtained by applying very high pressure to the wurtzite structure (35). For the wurtzite structure, Some of the physical properties of ZnO that make it the lattice parameters a and b are equal and in attractive for electronic and optoelectronic devices the range 3.2475–3.5201 Å and c is in the range are summarised in Table II and are discussed one 5.2042–5.2075 Å. The bond between Zn and O by one in the following sections. in the crystal lattice possesses very strong ionic character. Therefore, ZnO is classed as being 4.1 Crystal Structure and Lattice between an ionic and covalent compound (11). Constant 4.2 Electronic Band Structure In the crystal lattice, zinc and oxygen are arranged in tetrahedral geometry with each Zn atom ZnO is a direct bandgap material. Figure 2 shows the band structure of ZnO. It can be observed that in the Brillouin zone at k = 0, the lowest of the Table II Basic Properties of Zinc Oxide conduction band and topmost of the valence band (31) lies at the same point. The electron configuration Parameters Value of Zn is 1s2 2s2 2p6 3s2 3p6 and O is 1s2 2s2 3p4. In Bandgap 3.4 eV (direct bandgap) a ZnO crystal, the bottom of the conduction band –3 Density 5.606 g cm is due to occupied 2p states of O2– and the top of Wurtzite, rock salt and the valence band is due to the empty 4s states Crystal structures zinc blende of Zn2+. The valence band further splits into three Stable phase at 300 K Wurtzite subvalence bands under the influence of spin that Amorphous white or can be seen in Figure 2 (36). Appearance yellowish white powder Melting point 1975ºC 4.3 Defects in Zinc Oxide Odour Odourless Nature of oxide Amphoteric oxide ZnO exhibits n-type properties due to intrinsic defects. The defects arise because of deviation Lattice constants at a0: 0.32495 nm 300 K from stoichiometry. Major defects present in ZnO c0: 0.52069 nm Relative dielectric 8.66 constant

Refractive index 2.0041 E –1 Solubility in water 0.16 mg 100 ml 2+ Conduction band Zn 4s Intrinsic carrier 1016 to 1020 cm–3 concentration Breakdown voltage 5.0 × 106 V cm–1 Electron effective 0.24 m mass 0 Exciton binding 60 meV energy Electron Hall mobility k 200 cm2 V–1 s–1 at 300 K Valence band 2– Hole Hall mobility at O 2p 5–50 cm2 V–1 s–1 300 K A Ionicity 62% B Max p-type doping Intrinsic carrier C ~1017 cm–3; max n-type concentration doping ~1020 cm–3 Fig. 2. Electronic band structure of ZnO

Electrically pumped lasing from ZnO nanowires Conduction band (CB) 60 meV has also been achieved by some research groups (45–47).

Zni 0.22eV

Eg = 3.4eV 4.5 Electrical Properties 2.28eV Oi V o 0.9eV The conductivity of a thin film mainly depends on VZn carrier concentration and mobility. The relation 0.3eV Valence band (VB) between conductivity, mobility and carrier concentration is given by Equation (i): Fig. 3. Defects level and luminescence associated σ np, = nqµ (i) with the defects level where n is the density of electron (hole) concentration in the conduction band (valence band), q is the –19 are oxygen vacancies (VO) and zinc interstitials charge on the electron (1.6 × 10 ) and μ is the

(Zni). However, which one of the defects dominates mobility of charge carriers. ZnO exhibits n-type is still unclear (20). Due to these major defects, characteristics due to the intrinsic defects (VO and

ZnO exhibits n-type characteristics. Figure 3 Zni). The carrier concentration and mobility highly shows the defects and energy levels associated depend on the level of defects. In 2011 Torricelli with it. In the Figure 3, Zn and O stand for zinc et al. (48) proposed a multi-trapping-and- and oxygen respectively and V, and i correspond release-transport mechanism for charge transport to vacancy and interstitial site respectively. Zni phenomena in disordered ZnO. According to and VO result in a donor level in the forbidden gap this model, the conductivity can be explained as whereas Zn vacancies create an acceptor level. Equations (ii) and (iii):

The VO creates deep level donor states while the T ° shallow level donor states are due to Zn . The T °  T  T i sin π   °  ° difficulty in achieving p-type conductivity is due  T  T  T (ii) σσ= ° nt to the compensation of acceptor atoms by deep  N π  T  t  level donors that are the result of VO (37). The   luminescence in green, blue and violet light regions is also attributed to these defects (38). Figure 3 (iii) σµ° = qNbb shows possible luminescence from ZnO due to the various defect levels. where μb is band mobility at infinite temperature,

Nb is total states per unit volume in the transport 4.4 Optical Properties band, To is the characteristic temperature that accounts for the energetic disorder, Nt is the total

For materials to be used in optical emitting number of trap states and nt is the charge-carrier devices, they should have direct bandgap and high concentration in the trap states in the disordered exciton energy. ZnO is a direct and wide bandgap ZnO. The authors assumed that the charge carriers semiconductor with high refractive index (2.008). nt in the trap state are much greater than that of Its bandgap is around 3.4 eV at room temperature. the carriers in the transport band n. Therefore, the

It has an exciton binding energy around 60 meV as total carrier concentration nT was approximated compared to 25 meV of GaN. Due to this, exciton as nT = n + nt ~ nt. The defects and hence the recombination is possible at room temperature carrier concentration and mobility in the ZnO and above. Therefore, ZnO is a stable light emitter highly depend on the deposition method and as compared to GaN. Because of the excitonic the growth conditions. The concentration and process, emission in the UV region (380 nm) is mobility of electrons in ZnO have been found in observed from ZnO. However, due to the intrinsic the range 1016~1017 cm−3 and 20~400 cm2 V–1 s–1 defects of lower energy states, emission of violet, respectively (11, 25, 49–51). blue and green light has also been observed (39–41). Therefore, ZnO is an efficient material for 4.6 Ohmic and Schottky Contact phosphor applications (42). Stimulated emission under optical pumping has also been observed For high performance electronic and optoelectronic from ZnO. This phenomenon may be due to devices, high-quality metallic contact on the ZnO excitonic-excitonic scattering or emission (43, 44). thin film is very important. The electrical properties

206 © 2020 Johnson Matthey https://doi.org/10.1595/205651320X15694993568524 Johnson Matthey Technol. Rev., 2020, 64, (2) of semiconductor devices are greatly affected by The transparency of GZO and AZO was found to be the contact used. The metallic contact on ZnO can greater than 90% equivalent to the transparency be Schottky barriers or ohmic depending on the of ITO (21, 61, 62). Therefore, it can be concluded difference between the work function of the metal that ZnO can be a good choice for TCO. and the electron affinity of ZnO. For a Schottky contact on ZnO thin film, metals with high work 5.2 Gas Sensors function are required. Platinum, palladium, tantalum and gold are high work function metals Gas sensors have many important applications that are generally used for making Schottky like environmental pollution control, fire detection, contact with ZnO film. Pd (ϕm = 5.12 eV) and Au as an alcohol breath analyser, industrial process

(ϕm = 5.1 eV) have been reported to form the controller or for detection of harmful gas leaks in most stable Schottky barrier contact on ZnO thin mines and other industries (63). Semiconducting films (52, 53). An ohmic contact plays an important oxide-based gas sensors are easy to fabricate, have role in the performance of devices like solar cells, low cost and their surfaces have good sensitivity to TFT, varistors and LED. A good ohmic contact on the adsorbed gases (64). For good sensitivity, the a semiconductor film is characterised by a linear film surface should have high grain density with current-voltage (I-V) relationship and negligible a porous surface (65). ZnO being physically and contact resistance. To create an ohmic contact chemically stable can be a good choice for thin film on ZnO, the work function of the metal should be gas sensors. Doping ZnO with suitable elements close to the electron affinity of ZnO (χ = 4.35 eV) in appropriate amounts increases the surface (54). Al, In and titanium have work function values density of grains and porosity thereby improving close to 4.28 eV, their resistivity is very low and the the sensing selectivity and response time of the contact resistance formed between these metals film (66). The sensitivity further improves at high and ZnO is also negligible. Therefore, these metals temperature. The conductivity of ZnO thin film can be a good choice for making ohmic contact surfaces will increase or decrease depending upon with ZnO films. the nature of reaction (oxidation or reduction) of the adsorbed oxygen on the surface of the ZnO 5. Applications of Zinc Oxide thin film and the gas under test (65). There are numerous reports of ZnO thin film gas sensors for 5.1 Transparent Conducting Oxides detecting species such as ammonia, ammonium, nitrogen dioxide, water, ozone, carbon monoxide, TCO are widely used as electrodes in optical and hydrogen, hydrogen sulfide and ethanol for various electronic devices like displays, solar cells, LED applications (17). Chou et al. (63) reported and organic light-emitting diodes (OLED) (55). At Al‑doped ZnO thin film by rf sputtering method present indium tin oxide (ITO) is used as a TCO with interdigitated Pt electrodes that can be used due to its excellent transparency and conductivity as a breath analyser for sensing ethanol. Kim et al. but its availability is limited and this makes it (67) reported a Sn-doped ZnO thin film gas sensor very costly (56). As a result, the cost of devices for NO2 detection with improved selectivity. Other incorporating ITO as electrodes is very high. reported works include Pd-doped ZnO gas sensors

ZnO is widely available, cheap and also has very for H2 detection by Al-zaidi et al. (68) and a ZnO good transparency in the visible region and good thin film gas sensor by rf sputtering for H2, NO2 conductivity. Therefore, it can be an alternative and hydrocarbon detection by Sadek et al. (69). choice as TCO. Highly crystalline transparent ZnO Balakrishnan et al. (70) reported the detection of film with good conductivity is easy to process NH3 gas by a p-type ZnO thin film. The p-type thin at low temperatures making it compatible with film was obtained by co-doping with aluminium plastic and glass substrates (55, 57). The electrical nitride and aluminium arsenide and then depositing conductivity of ZnO is not equivalent to ITO but with rf sputtering method. the conductivity of ZnO can be modified by doping it with elements like Al, In and Ga (58). Agura 5.3 Light-Emitting Diodes et al. (59) and Jun et al. (60) have reported the lowest resistivity of Al-doped and Ga-doped thin The large bandgap of ZnO and high exciton energy films respectively. The reported resistivity was makes it an ideal material for blue and UV LED. 8.1 × 10–5 Ω cm for Al‑doped ZnO (AZO) and ZnO is widely available and cheap, so it has an 7.7 × 10–5 Ω cm for Ga‑doped ZnO (GZO) thin film. advantage over GaN from the cost point of view.

The limiting factor in realising ZnO based LED was energy hence lasing is observed under moderate the lack of stable and reproducible p-type ZnO. pumping. Therefore, ZnO-based lasers have a low The alternative approach was that n-type ZnO threshold value (11). Ozgur et al. (83) in 2004 thin film was grown on other p-type materials like reported low threshold exciton-exciton scattering- Si, GaN, zinc telluride, copper(I) oxide and GaAs induced stimulated emission in rf-sputtered ZnO (11, 27). Various ZnO based heterojunction LED thin films. Random stimulated emission from a in the UV and visible ranges (red, blue, green or ZnO polycrystalline thin film was observed by Cao white) have been reported. Rogers et al. (71) and et al. (88). Gadallah et al. (89) in 2013 reported Alivov et al. (72) have reported n-ZnO/p-GaN surface and edge emission under optical pumping and n-ZnO and p-AlGaN LED in the UV range of from a ZnO thin film grown on a sapphire substrate 375 nm and 385 nm by pulsed laser deposition by PLD with the highest gain and lowest loss (at (PLD) and chemical vapour deposition (CVD) that period). Waveguide assisted random lasing processes, respectively. Yang et al. (73) and Alivov was also observed from an epitaxial ZnO thin film et al. (74) have reported n-ZnO/p-GaN and n-ZnO/ (90). Although there are various reports on lasing p‑GaN/Al2O3 blue LED by metalorganic chemical through ZnO, there are no reports on ZnO-based vapour deposition (MOCVD) and CVD processes, laser diode. The limitation in fabricating ZnO-based respectively. An n-ZnO/n‑MgZnO/n‑CdZnO/ laser diodes was that a stable p-type ZnO thin p‑MgZnO LED emitting red light has been reported film was not realisable. But now with the various by Ohashi et al. (75) by a MOCVD process. reports on p-type ZnO (78–81), it is expected that Chichibu et al. (76) fabricated greenish-white LED a ZnO-based laser diode will be available soon. using helicon-wave-excited plasma-sputtering from p-type copper gallium sulfide heterojunction 5.5 Biosensors diodes using n-type ZnO as an electron injector. They also reported IR‑LED (780 nm) by using a A biosensor is a transducer that detects a biological p-CuGaS2/n-ZnO-Al structure fabricated by the response and converts it into an equivalent helicon-wave-excited plasma-sputtering method electrical signal. Biosensors have many important (77). Earlier it was difficult to achieve p-type doping applications especially in the healthcare and food- in ZnO but, at present, several researchers have processing industries. They are used for chemical reported p-type ZnO and homojunction LED based and biological analysis. They are also used for on it. Wang et al. (78) reported p-ZnMgO/ZnO/ clinical analysis and environmental monitoring. n‑ZnMgO p-n junction LED. Tsukazaki et al. (79) Materials to be used for biosensors should be represented a p-i-n homojunction structure on a (0 biocompatible and non-toxic so that the biological

0 0 1) ScAlMgO4 substrate. The p-type conductivity activity of the element to be recognised is retained. was achieved by doping ZnO with nitrogen. Ryu et It should also present a high surface area to the al. (80) fabricated arsenic-doped p-type ZnO and element to be detected for better sensitivity. demonstrated (Zn, Be)ZnO/n-ZnO-based LED. Lim The large surface-to-volume ratio and electron et al. (81) fabricated p-ZnO/n-ZnO/sapphire LED and phonon confinement of nanomaterials make by rf sputtering method. them favourable for biosensors. ZnO due to its biocompatibility, non-toxicity and antibacterial 5.4 Laser properties is a good choice for biosensors. ZnO has a high isoelectric point (9.5) therefore elements For short-wavelength semiconductor laser diodes, with a low isoelectric point can also be immobilised wide bandgap materials are ideal (82). At present on it through electrostatic interaction. ZnO blue and UV lasers are based on GaN materials nanostructures as biosensors can be used to detect (83). Because of the large exciton binding energy species like hydrogen peroxide, urea, protein, of 60 meV as compared to 25 meV of GaN, ZnO glucose, human immunoglobulin G (IgG), DNA could be a promising material for UV and blue (phosphinothricin acetyltransferase (PAT) gene), laser applications. The lasing phenomenon in ZnO phenol, catechol or cholesterol (26, 91). occurs due to exciton-exciton scattering. Various researchers have observed stimulated emission 5.6 Photodetector from ZnO (84–87). Stimulated emission from the surface and edges of a ZnO thin film is observed A photodetector is a device that senses under optical pumping. ZnO has high excitonic electromagnetic waves. It converts the optical

208 © 2020 Johnson Matthey https://doi.org/10.1595/205651320X15694993568524 Johnson Matthey Technol. Rev., 2020, 64, (2) signal into an equivalent electrical signal. If for detecting UV rays. Mollow (93) in 1940 was the a light wave with energy greater than or equal first to observe the UV photoresponse of ZnO thin to the bandgap of the semiconductor falls on films. The 3.4 eV bandgap of ZnO makes it very it, then electron-hole pairs are generated. The sensitive to UV rays compared to visible and IR rays. charge pairs drift towards the anode and cathode The bandgap can, however, be tuned by doping respectively under the influence of the appropriate with materials like In, Al or magnesium to make a electric field resulting in the generation of current. detector for a specific wavelength. UV sensors have This current is called photocurrent, and it is a wide range of applications. They are used in space proportional to the intensity of light falling on the applications for communication and in the military semiconductor. Photodetectors can be classified as for missile warning and guiding systems. They can photoconductors and photodiodes. Photodiodes are be used for environmental monitoring as an ozone further classified as metal-semiconductor-metal layer monitor and for commercial purposes as (MSM) photodiodes, Schottky photodiodes, p-n a fire detector (92, 94). Materials to be used for homojunction photodiodes and p-n heterojunction space and military applications should be thermally, photodiodes. In a photoconductor, the conductivity mechanically and chemically stable and should have of the semiconductor changes under the influence high radiation resistance. ZnO is an ideal material of light. If light with energy greater than or equal with all these properties along with high gain and to the bandgap of the semiconductor falls on it, an high photoresponse. ZnO is almost transparent electron in the valence band absorbs energy and in IR and in the visible region hence ZnO-based jumps to the conduction band. The concentration of UV detectors exhibit less dark current and better electrons in the conduction band increases hence sensitivity to UV rays as compared to Si-based the conductivity also increases. Thus, the current UV detectors. Figures 4(a)–4(c) show some in the external circuit under biased conditions important structures of ZnO-based UV detectors. increases proportionally to the photocurrent. A photodiode works in reverse bias mode. In a 5.6.1 Photoconductor diode, a depletion region is formed at the junction of p and n regions and the width of this region Figure 4(a) shows the structure of a ZnO increases on increasing the reverse bias voltage. photoconductor. It is very simple to fabricate. The In this region, there are no free charge carriers ohmic contacts are patterned over a ZnO thin film but because of thermal energy, very few electron- layer. For ohmic contacts, Al, Ti and ITO can be used hole pairs may be generated. Under the influence (95–97). A ZnO based photoconductor exhibits high of an electric field in the depletion region, the internal gain. The disadvantage is that it exhibits a electrons and holes drift towards the n-region and the p-region respectively. This current has (a) very small magnitude and is known as leakage current or dark current (current in the absence of visible light). This current depends on the ambient temperature, reverse bias voltage and external series resistance. If the diode is exposed to a light wave of appropriate wavelength, more electron (b) Substrate and hole pairs generate, and more current flows Semiconductor in the external circuit. That current is the sum of Ohmic contact dark current and photocurrent. The photocurrent Schottky contact is proportional to the intensity of light falling on the diode. (c) For efficient detection of light, the photodetector Interdigitated should have some desirable features. It should contact be sensitive in the required spectral region with high responsivity, high quantum efficiency, fast response time and small noise equivalent power Fig. 4. ZnO based: (a) photoconductor; (b) MSM (NEP). It should have low noise current in the photodiode; (c) Schottky photodiode undesired spectral range (92). ZnO is mainly used

209 © 2020 Johnson Matthey https://doi.org/10.1595/205651320X15694993568524 Johnson Matthey Technol. Rev., 2020, 64, (2) very high dark current. The responsivity and linear Ru, Ag and Pd can be used for making Schottky dynamic range are also low (92). contacts on ZnO. In 1986 the first ZnO Schottky photodiode was reported by Fabricius et al. (102). 5.6.2 Metal-Semiconductor-Metal Au and manganese were used to form Schottky and ohmic contacts respectively. The observed Photodiode efficiency was not good. After that various efforts Figure 4(b) shows the structure of a ZnO based were made to improve efficiency. Most of them used MSM photodetector. The Schottky metal contacts Pd, Au and Pt due to the stability of the Schottky are patterned in an interdigitated form on ZnO contact with ZnO films and Al as the ohmic contact. thin film. The Schottky contact should have a large Recently Tang et al. (103) fabricated a graphene barrier height and should form a stable contact nanodots array (GNDA) with ZnO nanofilm spin- with ZnO. The larger the barrier height, the lower coated on it for UV photodetection. They found a will be the leakage current and better will be two-fold increase in external quantum efficiency the photocurrent to dark current contrast ratio. (9.32%) and responsivity (22.55 mA W−1) of However, at the same time, quantum efficiency and ZnO/GNDA for 20 nm and 30 nm sizes. As the size responsivity will decrease (92, 97, 98). The main of GNDA increased to 45 nm, the performance advantage of a MSM photodiode is that there is was comparatively poor. Su et al. (104) fabricated very low capacitance between the Schottky contact a high performance and self-powered beryllium and the thin film. Therefore, it has high speed. zinc oxide based dual-colour UV photodetector High work function metals like Pd, Au, Pt, nickel, through a one-step electron beam evaporation chromium, ruthenium or silver are preferred for of an asymmetric Ti/Au pair. The device exhibits making interdigitated Schottky contact on ZnO thin ultrafast response speed, with a rise time of films (27, 28, 92). The two interdigitated contacts ~35 μs and a decay time of ~880 μs and also are similar but creating dissimilarities in the two two cut-off response wavelengths at ~275 nm and contacts may result in a self-powered device. ~360 nm under zero bias, which correspond to Chen et al. (99) reported a self-powered ZnO MSM the UVA and UVC regions. Very high-performance photodetector with Au contact. One interdigitated UV detectors have so far been reported by groups contact had narrow Au fingers whereas others had like Somvanshi et al., Ali et al. and Oh et al. wide Au fingers. The observed responsivity was very (52, 105–108). high. The responsivity was reported to be highest, –1 at 20 nA W , when the asymmetric ratio was 20:1. 5.6.4 p-n Heterojunction Photodiode Very low-cost ZnO based MSM detectors have also been reported using graphite electrodes and ZnO based heterojunction photodiodes can paper substrate (100, 101). Gimenez et al. (100) be fabricated by depositing a ZnO thin film on in 2011 and Hasan et al. (101) in 2012 reported other p-type films or substrates like GaN, Si,

ZnO nanocrystals based MSM photodetectors with silicon carbide, nickel(II) oxide, ZnTe and Cu2O interdigitated graphite contacts drawn by pencil (11, 92, 109). Generally, the p-Si substrate is on paper. The interdigitated pattern was drawn used because of its low cost, easy availability and on paper by using an appropriate pencil and ZnO compatibility with Si-based complementary metal- nanocrystals were grown by a solution-based oxide-semiconductor (CMOS) technology. By using technique and then transferred to the paper. This a Si substrate, it is possible to integrate ZnO-based MSM detector was very easy to fabricate and very devices with Si-based CMOS technology (52). cheap with performance comparable to a MSM The problem with the n-ZnO/p-Si UV detector is photodetector with metal contacts. that ZnO is transparent to visible light whereas Si exhibits photocurrent in the visible region so 5.6.3 Schottky Photodiode it cannot be used in the presence of visible light. This problem can be solved by either insertion Figure 4(c) shows the structure of a ZnO based of an insulator layer between ZnO and Si (110) Schottky diode. It has well-patterned Schottky or coating the surface with nanoparticles (111). and ohmic contacts. It has many advantages over Zhang et al. (110) reported a n-ZnO/insulator- photoconductor and MSM photodiodes including MgO/p-Si visible-blind UV photodetector. A visible- low dark current, high contrast ratio, high speed blind n-ZnO/p-Si UV detector was also obtained and high quantum efficiency. As discussed earlier by Chen et al. (111) by coating the surface of ZnO various high work function metals like Pt, Ni, Cr, with silica nanoparticles. Hu et al. (112) reported

210 © 2020 Johnson Matthey https://doi.org/10.1595/205651320X15694993568524 Johnson Matthey Technol. Rev., 2020, 64, (2) a high-performance UV photodetector (nearly 104 formed by the accumulation of charges and in the at zero set bias under 370 nm (~0.85 mW cm−2)) MOSFET the channel is formed by inversion. Figures with high signal-to-noise ratio, high speed, high 5(a)–5(c) show the working principle of n-channel selectivity and high detectivity. Ouyang et al. TFT and Figures 5(d)–5(e) represent the output (113) reported a heterojunction photodetector in characteristics of n-channel TFT (119, 120). which a CdMoO4–ZnO composite film was prepared For n-channel TFT, a positive bias is applied by spin-coating CdMoO4 microplates on ZnO film. between the drain to source and gate to source The responsivity was 18-fold higher and the decay contact. The source contact is biased at 0 V. time was half compared to ZnO film by optimising Figures 5(d)–5(e) show the graphs ID vs. VG the amount of CdMoO4 microplates. Further, the and ID vs. VD respectively for an n-channel TFT. photocurrent was two-fold higher if Au nanoparticles In Figure 5(e), Region 1 is known as the linear are deposited to the CdMoO4–ZnO composite film. region. In this region VD<

Zhao et al. (114) fabricated a highly crystallised, ID is given by Equation (iv): self-powered solar-blind (200–280 nm)  2  ZnO–Ga2O3 core-shell heterostructure using a W VDS IDF=−µ EiCV()GS VVth DS −  (iv) one-step CVD method. The device exhibited a L  2  sharp cut-off wavelength at 266 nm, fast response speed and decay time and showed an ultrahigh where W is the width, L is the length of the channel, −1 responsivity (9.7 mA W ) at 251 nm with a high µFE is the mobility of electrons, Vth is the turn-on

UV:visible rejection ratio (R251 nm:R400 nm) of voltage, Ci is the capacitance of gate insulator per 2 6.9 × 10 under zero bias. The device was highly unit area, VGS is gate-to-source voltage and VDS is suitable in practical self-powered solar-blind drain-to-source voltage. Since VD<

can be observed that in the linear region ID varies 5.6.5 p-n Homojunction Diode linearly with VDS, Equation (v): W There are very few reports on ZnO-based p-n I =−µ CV()VV DFEi GthDS  (v) homojunction UV detectors due to difficulty in L achieving p-type ZnO thin films as discussed Region 2 is known as the saturation region. In earlier. But a few groups like Liu et al. (115), Moon this region (VG–Vth) >> VD and the drain current is et al. (116) and Chiu et al. (117) have succeeded given by Equation (v). It can be seen that the value in growing stable p-type ZnO thin films and of drain current in this region is constant and does have reported ZnO based p-n homojunction UV not vary with VDS, Equation (vi): detectors. ZnO was doped mainly with As, nitrogen W 2 I =−µ CV()V  (vi) or antimony to achieve p-type conductivity. DF2L Ei GS th 

5.7 Thin-Film Transistor Based on the position of the gate terminal and source/drain electrodes there can be four possible The TFT was first patented (in 1952) as a solid- TFT structures. Figures 6(a)–6(d) show the state amplifier (118). It is a three-terminal (source, possible structures of TFT: (a) staggered bottom- drain and gate) device similar to the metal-oxide- gate, (b) co-planar bottom-gate, (c) staggered semiconductor field-effect transistor (MOSFET) with top-gate and (d) co-planar top-gate (115–117, the same working principle. It has a substrate (for 121–123). In the coplanar structure, the source/ providing mechanical support to the structure), drain contacts and the gate contact are placed a dielectric layer, an active channel layer and on the same side of the semiconductor/oxide source/drain and gate contacts. Charge carriers interface. In staggered structures, the gate are injected through the source electrode at one electrode is placed on one side, and the source/ end and collected at the drain electrode at another drain contact is placed on the other side of the end. The gate electrode is to control the flow of semiconductor/oxide interface. The bottom gate charge between the source and drain terminal. structure is easy to fabricate, but the disadvantage The dielectric layer between the gate electrode and is that the channel layer is exposed directly to the the channel layer is to prevent the flow of charge atmosphere. Therefore, the performance of the carriers between them. The difference between bottom gate TFT is easily affected by the presence MOSFET and TFT is that the channel in the TFT is of light, gases and humidity. Passivation of the

(a) =0 (+) VS VD

< VG Vth (b) =0 (+) Semiconductor VS VD

Source/drain

Gate < < Vth VG VD Channel

(+) VG (d) (e) V – th G D VD VG Vth Off Linear Saturation region region D D I I

Vth Vth

VG VD

Fig. 5. Working principle and output characteristics of an n-channel TFT

channel layer is required to prevent exposure. in the commercial market thereby increasing the The passivated bottom gate TFT is more stable, research interest in the field of TFT. At present, reliable and gives a much better performance as active-matrix liquid-crystal display (AMLCD) compared to an unpassivated one (124). The top technology is based on a-Si:H. But it has many gate structure has the advantage that the active disadvantages. The field-effect mobility of a-Si is layer is covered by the gate oxide and the gate very low (~1 cm2 V–1 s–1). This makes it unsuitable contact, but an extra masking step is needed to for ultra-high-definition displays where very high fabricate it. The first TFT was based on cadmium switching rates are required. Poly-Si has a very selenide material. In 1962 Weimer et al. (125) high field-effect mobility (>50 2cm V–1 s–1) but the reported the first TFT in which he used CdSe as problem is that it requires a very high processing an active channel material. In 1973, Brody et al. temperature (>500ºC) and the crystallisation (126) demonstrated the use of TFT in the LCD. He process is very time-consuming. The high used a matrix of 120 × 120 CdSe TFT for switching temperature makes it incompatible with cheap glass of pixels. But the very high cost and issues like and plastic substrates and hence the cost of a poly- reliability, stability and the invention of low power Si TFT display is very high. Due to its polycrystalline CMOS technology limited the research interest in nature, it exhibits different characteristics across TFT at that time. In 1979 le Comber et al. (127) the film area. Therefore, poly-Si TFT is unsuitable reported the a-Si:H TFT. The channel layer was for large-area displays. The common problem with deposited using plasma-enhanced chemical vapour Si-based TFT is that they are sensitive to visible deposition (PECVD) and doping of hydrogen was light, so a shield in the form of an array is required done by a glow discharge technique. After that, it that blocks the backlight, therefore the resolution took ten years for TFT LCD to become attractive of the display is degraded. These limiting factors

(a) and are suitable for high-resolution displays. ZnO TFT with field effect mobility up to 50 2cm V–1 s–1 5 and Ion:Ioff ratio greater than 10 can be obtained. The performance of the ZnO TFT can be further improved by various techniques like the use of high-k dielectric, doping of ZnO and post-deposition (b) treatments.

5.8 Memristor Substrate Memristor is of interest to many research groups as it Semiconductor (c) finds important application in fields like non‑volatile Oxide memory, neural networks optoelectronics, radiation Source/drain sensors and neuromorphic systems. There are Gate some reports on ZnO based memristor devices. Patil et al. (135), Fauzi et al. (136), Barnes et al. (137), Santos et al. (138) and Le et al. (139) have (d) reported ZnO based memristor with low power and fast switching activity.

6. Conclusion Fig. 6. TFT structures: (a) staggered bottom-gate; (b) co-planar bottom-gate; (c) staggered top-gate; ZnO has emerged as an important semiconductor and (d) co-planar top-gate material because of its excellent electrical, optical, piezoelectric and gas sensing properties. Hence, it can turned attention toward other materials especially be used for near-UV emission or detection and as a to wide bandgap materials due to their insensitivity transparent electrode. It has a large excitonic binding to visible light (11). energy of 60 meV at room temperature as compared There are various reports on ZnO-based TFT to 25 meV of GaN, an III-V compound having limited dated from 1968. The insensitivity to visible prospects. This makes ZnO an efficient light emitter light, low processing temperature, deposition of in the UV spectral region and comparably favourable a highly crystalline thin film over a large area by for optoelectronic applications. The high conductivity conventional processes like sputtering and devices and transparency of ZnO are important for with very high field-effect mobility in the range of applications like transparent conducting oxides and 0.2 cm2 V–1 s–1 to 40 cm2 V–1 s–1 have made ZnO a TFT. ZnO is fast emerging as a future material for the very attractive channel material for TFT (128, 129). fabrication of low cost, high performance electronic The first ZnO TFT was reported by Boesen et al. in and optoelectronic devices including transparent 1968 (130). Numerous ZnO TFT have since been conductive films, solar cells, LED and TFT. However, reported with very high mobility as compared to there are certain challenges and limitations. First a-Si TFT. Due to the transparent nature of ZnO in is the realisation of stable p-type ZnO. It is very the visible region, it is possible to realise ZnO based difficult to achieve p-type conductivity hence the fully transparent TFT. In 2003 Hoffmanet al., Carcia fabrication of ZnO-based p-n junction devices and et al. and Masuda et al. (131–133) reported fully CMOS is not currently viable. The most important transparent ZnO TFT. In 2008 Hirao et al. (134) factor is the stability of electrical characteristics in demonstrated a 1.46 inch LCD with 61,600 pixels the presence of oxygen. ZnO reacts with the oxygen driven by bottom gate ZnO TFT arrays. in the environment. Due to this, the conductivity The main performance parameters for TFT are varies and the electrical properties change over turn-on voltage, drain current on-to-off ratio time. The variation of the electrical properties makes

(Ion:Ioff) and channel mobility. Turn-on-voltage is ZnO‑based devices unstable. The characteristics the minimum gate voltage required to turn on the of devices based on silicon technology are highly TFT. The lower the turn-on voltage, the lower the reproducible and stable under varying ambient requirement of biasing voltages leading to lower conditions, and the devices are highly reliable. For power consumption. TFT with high mobility and a commercialisation of ZnO based devices, it is very high Ion:Ioff ratio can work at a higher frequency important to resolve these issues.

Glossary

light amplification by stimulated Al2O3 alumina laser emission of radiation AlAs aluminium arsenide LCD liquid-crystal display AlGaAs aluminium gallium arsenide LED light-emitting diode AlGaN aluminium gallium nitride MgO magnesium oxide AlN aluminium nitride MgZnO magnesium zinc oxide active-matrix liquid-crystal AMLCD metalorganic chemical vapour display MOCVD deposition a-Si amorphous silicon metal-oxide-semiconductor field- MOSFET a-Si:H hydrogenated amorphous Si effect transistor AZO aluminium-doped zinc oxide MSM metal-semiconductor-metal BeZnO beryllium zinc oxide NEP noise equivalent power CCD charge coupled device OLED organic light-emitting diode

CdMoO4 cadmium molybdenum oxide PAT phosphinothricin acetyltransferase plasma-enhanced chemical vapour CdSe cadmium sellenide PECVD deposition CdZnO cadmium zinc oxide PLD pulsed laser deposition complementary metal-oxide- CMOS semiconductor poly-Si polycrystalline silicon

Cu2O copper(I) oxide rf radiofrequency scandium aluminium magnesium CuGaS2 copper gallium sulfide ScAlMgO 4 oxide CVD chemical vapour deposition SiC silicon carbide Ga2O3 gallium oxide superconducting quantum SQUIDS GaAs gallium arsenide interference devices GaN gallium nitride TCO transparent conducting oxides GNDA graphene nanodots array TFT thin-film transistors GZO gallium-doped zinc oxide UV ultraviolet IgG immunoglobulin G VLSI very large scale integrated circuits InGaAs indium gallium arsenide ZnMgO zinc magnesium oxide IR infrared ZnO zinc oxide ITO indium tin oxide ZnTe zinc telluride

References

1. K. L. Chopra, P. D. Paulson and V. Dutta, Prog. 5. Y. Zhou, Li Yang and Y. Huang, “Micro- and Photovoltaics Res. Appl., 2004, 12, (23), 69 Macromechanical Properties of Materials”, CRC 2. K. Wasa, M. Kitabatake and H. Adachi, Press, Boca Raton, USA, 2013, 620 pp ‘Deposition of Compound Thin Films’, in 6. K. Seshan, ‘Scaling and its Implications “Thin Film Materials Technology – Sputtering for the Integration and Design of Thin Film of Compound Materials”, Ch. 5, William and Processes’, in ”Handbook of Thin Film Andrew Inc, Norwich, New York, USA, 2004, Deposositions”, 3rd Edn., Ch. 2, Elsevier Inc, pp. 191–403 Waltham, USA, 2012, pp. 19–40 3. “Handbook of Thin Film Technology”, eds. 7. J. Zhu, Int. Org. Sci. Res. J. Eng., 2015, 5, (4), 13 L. I. Maissel and R. Glang, McGraw-Hill, New 8. “Silicon-Based Materials and Devices – Properties York, USA, 1970 and Devices”, ed. H. S. Nalwa, Vol. 2, Academic 4. H. J. Choi, W. Jang, B. C. Mohanty, Y. S. Jung, Press, San Diego, USA, 2001 A. Soon and Y. S. Cho, J. Phys. Chem. Lett., 9. J. Sheu, M. Lee, Y. Lu and K. Shu, IEEE J. Quantum 2018, 9, (20), 5934 Elect., 2008, 44, (12), 1211

10. H. Hosono, Thin Solid Films, 2007, 515, (15), 31. A. Janotti and C. G. Van de Walle, Rep. Prog. 6000 Phys., 2009, 72, (12), 126501 11. Ü. Özgür, Y. I. Alivov, C. Liu, A. Teke, 32. H. Morkoç and U. Özgür, “Zinc Oxide – M. A. Reshchikov, S. Doğan, V. Avrutin, S.-J. Cho Fundamentals, Materials and Device Technology”, and H. Morkoç, J. Appl. Phys., 2005, 98, (4), Wiley-VCH Verlag GmbH and Co KGaA, Weinheim, 041301 Germany, 2009, 477 pp 12. G. E. Patil, D. D. Kajale, D. N. Chavan, 33. S.-K. Kim, S.-Y. Jeong and C.-R. Cho, Appl. Phys. N. K. Pawar, P. T. Ahire, S. D. Shinde, Lett., 2003, 82, (4), 562 V. B. Gaikwad and G. H. Jain, Bull. Mater. Sci., 34. A. B. M. A. Ashrafi, A. Ueta, A. Avramescu, 2011, 34, (1), 1 H. Kumano, I. Suemune, Y.-W. Ok and 13. R. H. Bari, P. P. Patil, S. B. Patil and A. R. Bari, T.‑Y. Seong, Appl. Phys. Lett., 2000, 76, (5), Bull. Mater. Sci., 2013, 36, (6), 967 550 14. S.-S. Lin and D.-K. Wu, Ceram. Int., 2010, 36, 35. A. Segura, J. A. Sans, F. J. Manjón, A. Muñoz and (1), 87 M. J. Herrera-Cabrera, Appl. Phys. Lett., 2003, 15. Q. Zhou, Z. Ji, B. Hu, C. Chen, L. Zhao and 83, (2), 278 C. Wang, Mater. Lett., 2007, 61, (2), 531 36. C. F. Klingshirn, B. K. Meyer, A. Waag, 16. A. Walsh, J. L. F. Da Silva, S.-H. Wei, C. Körber, A. Hoffmann and J. Geurts, “Zinc Oxide – A. Klein, L. F. J. Piper, A. DeMasi, K. E. Smith, From Fundamental Properties Towards Novel G. Panaccione, P. Torelli, D. J. Payne, A. Bourlange Applications”, Springer-Verlag, Berlin, Germany, and R. G. Egdell, Phys. Rev. Lett., 2008, 100, 2010, 359 pp (16), 167402 37. A. Janotti and C. G. Van de Walle, Phys. Rev. B, 17. Ü. Özgür, D. Hofstetter and H. Morkoç, Proc. 2007, 76, (16), 165202 IEEE, 2010, 98, (7), 1255 38. N. Karak, P. K. Samanta and T. K. Kundu, Optik, 18. D. P. Norton, Y. W. Heo, M. P. Ivill, K. Ip, 2013, 124, (23), 6227 S. J. Pearton, M. F. Chisholm and T. Steiner, 39. J.-J. Wu and S.-C. Liu, Adv. Mater., 2002, 14, Mater. Today, 2004, 7, (6), 34 (3), 215 19. Z. L. Wang, Mater. Today, 2004, 7, (6), 26 40. P. K. Samanta, S. K. Patra and P. Roy Chaudhuri, 20. L. Schmidt-Mende and J. L. MacManus-Driscoll, Phys. E: Low-dimensional Syst. Nanostructures, Mater. Today, 2007, 10, (5), 40 2009, 41, (4), 664 21. S.-M. Park, T. Ikegami and K. Ebihara, Thin Solid 41. B. Lin, Z. Fu and Y. Jia, Appl. Phys. Lett., 2001, Films, 2006, 513, (1–2), 90 79, (7), 943 22. T. Tynell, H. Yamauchi, M. Karppinen, R. Okazaki 42. P. A. Rodnyi and I. V Khodyuk, Opt. Spectrosc., and I. Terasaki, J. Vac. Sci. Technol. A, 2013, 2011, 111, (5), 776 31, (1), 01A109 43. D. M. Bagnall, Y. F. Chen, Z. Zhu, T. Yao, 23. H. Gong, J. Q. Hu, J. H. Wang, C. H. Ong and S. Koyama, M. Y. Shen and T. Goto, Appl. Phys. F. R. Zhu, Sensors Actuators B: Chem., 2006, Lett., 1997, 70, (17), 2230 115, (1), 247 44. A. Ohtomo, M. Kawasaki, Y. Sakurai, Y. Yoshida, 24. S. Singh, R. Nunna, C. Periasamy and H. Koinuma, P. Yu, Z. K. Tang, G. K. L. Wong and P. Chakrabarti, Int. J. Contemp. Res. Eng. Tech., Y. Segawa, Mater. Sci. Eng.: B, 1998, 54, (1–2), 2011, 1, (1), 14 24 25. C. Periasamy and P. Chakrabarti, J. Electron. 45. S. Chu, G. Wang, W. Zhou, Y. Lin, L. Chernyak, Mater., 2011, 40, (3), 259 J. Zhao, J. Kong, L. Li, J. Ren and J. Liu, Nature Nanotechnol., 2011, 6, (8), 506 26. S. K. Arya, S. Saha, J. E. Ramirez-Vick, V. Gupta, S. Bhansali and S. P. Singh, Anal. Chim. Acta, 46. Y. Tian, X. Ma, L. Jin and D. Yang, Appl. Phys. 2012, 737, 1 Lett., 2010, 97, (25), 251115 27. H. Ohta and H. Hosono, Mater. Today, 2004, 7, 47. F. Gao, M. M. Morshed, S. B. Bashar, Y. Zheng, (6), 42 Y. Shi and J. Liu, ‘Electrically Pumped Random Lasing Based on Au-ZnO Nanowire Schottky 28. Y. Liu, Y. Li and H. Zeng, J. Nanomater., 2013, Junction’, Conference on Lasers and Electro- 196521 Optics, San Jose, USA, 10th–15th May 2015, 29. D. Basak, G. Amin, B. Mallik, G. K. Paul and Paper SM1F.7, The Optical Society, Washington, S. K. Sen, J. Cryst. Growth, 2003, 256, (1–2), DC, USA 73 48. F. Torricelli, J. R. Meijboom, E. Smits, 30. P. K. Nayak, J. Jang, C. Lee and Y. Hong, Appl. A. K. Tripathi, M. Ferroni, S. Federici, Phys. Lett., 2009, 95, (19), 193503 G. H. Gelinck, L. Colalongo, Z. M. Kovacs-Vajna,

D. de Leeuw and E. Cantatore, IEEE Trans. 71. D. J. Rogers, F. H. Teherani, A. Yasan, K. Minder, Electron Devices, 2011, 58, (8), 2610 P. Kung and M. Razeghi, Appl. Phys. Lett., 2006, 49. Y. Igasaki and H. Saito, J. Appl. Phys., 1991, 70, 88, (14), 141918 (7), 3613 72. Y. I. Alivov, E. V. Kalinina, A. E. Cherenkov, 50. S. P. Lau, H. Y. Yang, S. F. Yu, H. D. Li, D. C. Look, B. M. Ataev, A. K. Omaev, M. Tanemura, T. Okita, H. Hatano and H. H. Hng, M. V. Chukichev and D. M. Bagnall, Appl. Phys. Appl. Phys. Lett., 2005, 87, (1), 013104 Lett., 2003, 83, (23), 4719 51. A. Tsukazaki, A. Ohtomo, S. Yoshida, 73. T. P. Yang, H. C. Zhu, J. M. Bian, J. C. Sun, M. Kawasaki, C. H. Chia, T. Makino, Y. Segawa, X. Dong, B. L. Zhang, H. W. Liang, X. P. Li, T. Koida, S. F. Chichibu and H. Koinuma, Appl. Y. G. Cui and G. T. Du, Mater. Res. Bull., 2008, Phys. Lett., 2003, 83, (14), 2784 43, (12), 3614 52. D. Somvanshi and S. Jit, J. Nanoelectron. 74. Y. I. Alivov, J. E. Van Nostrand, D. C. Look, M. V Optoelectron., 2014, 9, (1), 21 Chukichev and B. M. Ataev, Appl. Phys. Lett., 2003, 83, (14), 2943 53. C. Periasamy and P. Chakrabarti, J. Nanoelectron. Optoelectron., 2010, 5, (1), 38 75. T. Ohashi, K. Yamamoto, A. Nakamura and 54. L. J. Brillson and Y. Lu, J. Appl. Phys., 2011, 109, J. Temmyo, Japan. J. Appl. Phys., 2008, 47, (12), 121301 (4S), 2961 55. H. Liu, V. Avrutin, N. Izyumskaya, Ü. Özgür and 76. S. F. Chichibu, T. Ohmori, N. Shibata, T. Koyama H. Morkoç, Superlattices Microstruct., 2010, 48, and T. Onuma, Appl. Phys. Lett., 2004, 85, (19), (5), 458 4403 56. S.-J. Kim, IEEE Photonics Technol. Lett., 2005, 77. S. F. Chichibu, T. Ohmori, N. Shibata, T. Koyama 17, (8), 1617 and T. Onuma, J. Phys. Chem. Solids, 2005, 66, 57. A. W. Ott and R. P. H. Chang, Mater. Chem. (11), 1868 Phys., 1999, 58, (2), 132 78. Y.-L. Wang, F. Ren, H. S. Kim, D. P. Norton and 58. T. Minami, Thin Solid Films, 2008, 516, (17), 5822 S. J. Pearton, IEEE J. Select. Topics Quantum Electron., 2008, 14, (4), 1048 59. H. Agura, A. Suzuki, T. Matsushita, T. Aoki and M. Okuda, Thin Solid Films, 2003, 445, (2), 263 79. A. Tsukazaki, A. Ohtomo, T. Onuma, 60. M.-C. Jun, S.-U. Park and J.-H. Koh, Nanoscale M. Ohtani, T. Makino, M. Sumiya, K. Ohtani, Res. Lett., 2012, 7, 639 S. F. Chichibu, S. Fuke, Y. Segawa, H. Ohno, H. Koinuma and M. Kawasaki, Nature Mater., 61. B.-Z. Dong, G.-J. Fang, J.-F. Wang, W.-J. Guan 2005, 4, (1), 42 and X.-Z. Zhao, J. Appl. Phys., 2007, 101, (3), 033713 80. Y. Ryu, T.-S. Lee, J. A. Lubguban, H. W. White, B.-J. Kim, Y.-S. Park and C.-J. Youn, Appl. Phys. 62. S. Shirakata, T. Sakemi, K. Awai and T. Yamamoto, Lett., 2006, 88, (24), 241108 Superlattices Microstruct., 2006, 39, (1–4), 218 63. S. M. Chou, L. G. Teoh, W. H. Lai, Y. H. Su and 81. J.-H. Lim, C.-K. Kang, K.-K. Kim, I.-K. Park, D.- M. H. Hon, Sensors, 2006, 6, (10), 1420 K. Hwang and S.-J. Park, Adv. Mater., 2006, 18, (20), 2720 64. H. Shokry Hassan, A. B. Kashyout, I. Morsi, A. A. A. Nasser and I. Ali, Beni-Suef Univ. J. Basic 82. Z. K. Tang, G. K. L. Wong, P. Yu, M. Kawasaki, Appl. Sci., 2014, 3, (3), 216 A. Ohtomo, H. Koinuma and Y. Segawa, Appl. Phys. Lett., 1998, 72, (25), 3270 65. S. Roy and S. Basu, Bull. Mater. Sci., 2002, 25, (6), 513 83. Ü. Özgür, A. Teke, C. Liu, S.-J. Cho, H. Morkoç 66. S. T. Shishiyanu, T. S. Shishiyanu and O. I. Lupan, and H. O. Everitt, Appl. Phys. Lett., 2004, 84, Sensors Actuators B: Chem., 2005, 107, (1), (17), 3223 379 84. H.-C. Chen, M.-J. Chen, M.-K. Wu, Y.-C. Cheng 67. P.-S. Cho, K.-W. Kim and J.-H. Lee, and F.-Y. Tsai, IEEE J. Select. Topics Quantum J. Electroceramics, 2006, 17, (2–4), 975 Electron., 2008, 14, (4), 1053 68. Q. Al-zaidi, A. Suhail and W. Al-azawi, Appl. 85. X. Q. Zhang, Z. K. Tang, M. Kawasaki, A. Ohtomo Phys. Res., 2011, 3, (1), 89 and H. Koinuma, J. Crystal Growth, 2003, 259, 69. A. Z. Sadek, S. Choopun, W. Wlodarski, (3), 286 S. J. Ippolito and K. Kalantar-zadeh, IEEE 86. Z. K. Tang, M. Kawasaki, A. Ohtomo, H. Koinuma Sensors J., 2007, 7, (6), 919 and Y. Segawa, J. Crystal Growth, 2006, 287, 70. L. N. Balakrishnan, S. Gowrishankar and (1), 169 N. Gopalakrishnan, IEEE Sensors J., 2013, 13, 87. L. Miao, S. Tanemura, H. Y. Yang and K. Yoshida, (6), 2055 J. Nanosci. Nanotechnol., 2011, 11, (10), 9326

88. H. Cao, Y. G. Zhao, H. C. Ong, S. T. Ho, J. Y. Dai, 108. G. M. Ali and P. Chakrabarti, J. Vac. Sci. Technol. J. Y. Wu and R. P. H. Chang, Appl. Phys. Lett., B, 2012, 30, (3), 031206 1998, 73, (25), 3656 109. F. Teng, K. Hu, W. Ouyang and X. Fang, Adv. 89. A.-S. Gadallah, K. Nomenyo, C. Couteau, Mater., 2018, 30, (35), 1706262 D. J. Rogers and G. Lérondel, Appl. Phys. Lett., 110. T. C. Zhang, Y. Guo, Z. X. Mei, C. Z. Gu and 2013, 102, (17), 171105 X. L. Du, Appl. Phys. Lett., 2009, 94, (11), 90. P.-H. Dupont, C. Couteau, D. J. Rogers, 113508 F. H. Téhérani and G. Lérondel, Appl. Phys. Lett., 111. C.-P. Chen, P.-H. Lin, L.-Y. Chen, M.-Y. Ke, 2010, 97, (26), 261109 Y.‑W. Cheng and J. Huang, Nanotechnology, 91. N. Batra, M. Tomar and V. Gupta, J. Appl. Phys., 2009, 20, (24), 245204 2012, 112, (11), 114701 112. K. Hu, F. Teng, L. Zheng, P. Yu, Z. Zhang, H. Chen 92. W. Ouyang, F. Teng, J.-H. He and X. Fang, Adv. and X. Fang, Laser Photon. Rev., 2017, 11, (1), Funct. Mater., 2019, 29, (9), 1807672 1600257 93. E. Mollow, Proceedings of the Photoconductivity 113. W. Ouyang, F. Teng, M. Jiang and X. Fang, Small, Conference, 4th–6th November, 1954, Atlantic 2017, 13, (39), 1702177 City, USA, ed. R. G. Breckenridge, Wiley, New 114. B. Zhao, F. Wang, H. Chen, L. Zheng, L. Su, York, USA, p. 509 D. Zhao and X. Fang, Adv. Funct. Mater., 2017, 94. G. M. Ali and P. Chakrabarti, IEEE Photonics J., 27, (17), 1700264 2010, 2, (5), 784 115. J. L. Liu, F. X. Xiu, L. J. Mandalapu and Z. Yang, 95. Q. A. Xu, J. W. Zhang, K. R. Ju, X. D. Yang and ‘P-Type ZnO by Sb Doping for PN-Junction X. Hou, J. Crystal Growth, 2006, 289, (1), 44 Photodetectors’, Integrated Optoelectronic 96. Z. Bi, X. Yang, J. Zhang, X. Bian, D. Wang, Devices, San Jose, USA, 21st–26th January, X. Zhang and X. Hou, J. Electron. Mater., 2009, 2006, “Zinc Oxide Materials and Devices”, eds. 38, (4), 609 F. H. Teherani and C. W. Litton, Vol. 6122, SPIE, Bellingham, USA 97. S. P. Chang, S. J. Chang, Y. Z. Chiou, C. Y. Lu, T. K. Lin, Y. C. Lin, C. F. Kuo and H. M. Chang, 116. T.-H. Moon, M.-C. Jeong, W. Lee and J.-M. Myoung, Sensors Actuators A: Phys., 2007, 140, (1), 60 Appl. Surf. Sci., 2005, 240, (1–4), 280 98. S. J. Young, L. W. Ji, S. J. Chang and X. L. Du, 117. H.-J. Chiu, T.-H. Chen, L.-W. Lai, C.-T. Lee, J. Electrochem. Soc., 2007, 154, (1), H26 J.‑D. Hong and D.-S. Liu, J. Nanomater., 2015, 284835 99. H.-Y. Chen, K.-W. Liu, X. Chen, Z.-Z. Zhang, M.- M. Fan, M.-M. Jiang, X.-H. Xie, H.-F. Zhao and 118. E. Fortunato, P. Barquinha, A. Pimentel, D.-Z. Shen, J. Mater. Chem. C, 2014, 2, (45), A. Gonçalves, A. Marques, L. Pereira and R. Martins, 9689 Thin Solid Films, 2005, 487, (1–2), 205 100. A. J. Gimenez, J. M. Yáñez-Limón and 119. K. Long, A. Z. Kattamis, I.-C. Cheng, H. Gleskova, J. M. Seminario, J. Phys. Chem. C, 2011, 115, S. Wagner and J. C. Sturm, IEEE Electron Dev. (1), 282 Lett., 2006, 27, (2), 111 101. K. ul Hasan, O. Nur and M. Willander, Appl. Phys. 120. K. A. Gupta, D. K. Anvekar and V. Venkateswarlu, Lett., 2012, 100, (21), 211104 Int. J. Model. Optim., 2013, 3, (3), 266 102. H. Fabricius, T. Skettrup and P. Bisgaard, Appl. 121. H. Q. Chiang, J. F. Wager, R. L. Hoffman, J. Jeong Optics, 1986, 25, (16), 2764 and D. A. Keszler, Appl. Phys. Lett., 2005, 86, (1), 013503 103. R. Tang, S. Han, F. Teng, K. Hu, Z. Zhang, M. Hu and X. Fang, Adv. Sci., 2018, 5, (1), 1700334 122. R. E. Presley, D. Hong, H. Q. Chiang, C. M. Hung, R. L. Hoffman and J. F. Wager, Solid‑State 104. L. Su, H. Chen, X. Xu and X. Fang, Laser Photon. Electron., 2006, 50, (3), 500 Rev., 2017, 11, (6), 1700222 123. S. Sze, “Physics of Semiconductor Devices”, 2nd 105. H. von Wenckstern, S. Müller, G. Biehne, Edn., John Wiley and Sons, Hoboken, USA, 1981, H. Hochmuth, M. Lorenz and M. Grundmann, 868 pp J. Electron. Mater., 2010, 39, (5), 559 106. D. C. Oh, T. Suzuki, T. Hanada, T. Yao, H. Makino 124. C.-L. Fan, M.-C. Shang, B.-J. Li, Y.-Z. Lin, S.- and H. J. Ko, J. Vac. Sci. Technol. B: Microelectron. J. Wang, W.-D. Lee and B.-R. Hung, Materials, Nanom. Struct., 2006, 24, (3), 1595 2015, 8, (4), 1704 107. H. Endo, M. Sugibuchi, K. Takahashi, S. Goto, 125. P. K. Welmer, Proc. IRC, 1962, 50, 1462 S. Sugimura, K. Hane and Y. Kashiwaba, Appl. 126. T. P. Brody, J. A. Asars and G. D. Dixon, IEEE Phys. Lett., 2007, 90, (12), 121906 Trans. Electron Devices, 1973, 20, (11), 995

127. P. G. le Comber, W. E. Spear and A. Ghaith, C. Li, H. Furuta, H. Hokari, M. Yoshida, H. Ishii Electron. Lett., 1979, 15, (6), 179 and M. Kakegawa, IEEE Trans. Electron Devices, 128. P. F. Carcia, R. S. McLean and M. H. Reilly, Appl. 2008, 55, (11), 3136 Phys. Lett., 2006, 88, (12), 123509 135. S. R. Patil, M. Y. Chougale, T. D. Rane, S. S. Khot, A. A. Patil, O. S. Bagal, S. D. Jadhav, A. D. Sheikh, 129. C. Brox-Nilsen, J. Jin, Y. Luo, P. Bao and S. Kim and T. D. Dongale, Electronics, 2018, 7, A. M. Song, IEEE Trans. Electron Devices, 2013, (12), 445 60, (10), 3424 136. F. B. Fauzi, M. H. Ani, S. H. Herman and 130. G. F. Boesen and J. E. Jacobs, Proc. IEEE, 1968, M. A. Mohamed, IOP Conf. Ser.: Mater. Sci. Eng., 56, (11), 2094 2018, 340, 12006 131. R. L. Hoffman, B. J. Norris and J. F. Wager, Appl. 137. B. K. Barnes, Sci. Rep., 2018, 8, 2184 LINK Phys. Lett., 2003, 82, (5), 733 https://doi.org/10.1038/s41598-018-20598-5 132. P. F. Carcia, R. S. McLean, M. H. Reilly and 138. Y. P. Santos, E. Valença, R. Machado and G. Nunes, Appl. Phys. Lett., 2003, 82, (7), 1117 M. A. Macêdo, Mater. Sci. Semicond. Process., 133. S. Masuda, K. Kitamura, Y. Okumura, S. Miyatake, 2018, 86, 43 H. Tabata and T. Kawai, J. Appl. Phys., 2003, 93, 139. V.-Q. Le, T.-H. Do, J. R. D. Retamal, P.-W. Shao, (3), 1624 Y.-H. Lai, W.-W. Wu, J.-H. He, Y.-L. Chueh and Y.- 134. T. Hirao, M. Furuta, T. Hiramatsu, T. Matsuda, H. Chu, Nano Energy, 2019, 56, 322

The Author

Sumit Vyas received his BE degree in Electronics and Communication Engineering from Shri Vaishnav Institute of Technology and Science, India and an MTech degree in Microelectronics and VLSI Design from Motilal Nehru National Institute of Technology Allahabad, India, in 2010 and 2012 respectively. He completed his PhD degree in 2016 from Motilal Nehru National Institute of Technology Allahabad. Currently he is working as an Assistant Professor in Thapar Institute of Technology, India. His current research interests include fabrication, characterisation and modelling of semiconducting oxide thin film based electronic and optoelectronic devices. He has published more than 17 papers in various international journals and conference proceedings.

www.technology.matthey.com

Advances in Cold Sintering Improving energy consumption and unlocking new potential in component manufacturing

Jessica Andrews*, Daniel Button and from volatility, melting, interaction and mismatch Ian M. Reaney in thermal expansion with the ceramic. This leads Functional Materials and Devices Group, to complications in the production of components, Department of Materials Science and which include warping, delamination and the high Engineering, The University of Sheffield, cost of inert noble metals such as platinum and Western Bank, Sheffield, S10 2TN, UK palladium. Figure 1 shows the compatibility of several materials at various sintering temperatures. *Email: [email protected] Reducing sintering temperatures is a critical strategy in the goal of decarbonising foundation industries. A number of methods have been Ceramics are traditionally sintered at high proposed to reduce sintering temperature, or temperatures (~80% melting temperature (Tm)). more precisely the energy for densification, these There are numerous incentives to reduce processing include the addition of sintering aids, the utilisation temperature: the reduction in processing energy; of Joule heating through processes such as spark integration of polymeric and non-noble metals; plasma sintering (SPS) and flash sintering and greater control of microstructure and final most recently cold sintering (2). There have also component geometries. ‘Cold sintering’ has been been some investigations into combinations of cold developed as a novel method of densification which sintering and SPS-flash sintering. uses a transient liquid phase, pressure and heat to Sintering aids are often utilised to reduce achieve dense ceramics. This review explores the conventional sintering temperatures and typically process of cold sintering and its potential to densify form a liquid phase flux through which mass is more various ceramic materials and components at rapidly transported than within the solid state. In low temperatures (<300°C), primarily describing electroceramics, lithium-doping of barium titanate recent results at The University of Sheffield, UK. has been shown to successfully reduce sintering temperatures: Kimura et al. demonstrated a 1. Introduction reduction from 1300°C to 1000°C (3) and Randall et al. to 750°C by the addition of 15 mol% lithium Sintering is a crucial stage in the manufacturing fluoride (4). Many other sintering aids have been of dense ceramic products from a green body. investigated, but none have been found to reduce

Archaeologists have dated some of the earliest BaTiO3 sintering temperatures to <900°C (2, 5), ceramic artefacts to 24,000 BCE (1) and sintering which is desirable for metal internal electrode was empirically developed over thousands of technologies in multilayer devices (6). years prior to the appearance of modern sintering Flash sintering is a field assisted sintering theories (1). Traditionally, sintering to create dense technique (FAST) for the consolidation of ceramics. products requires heat treatment up to 80% of Tm A green body is placed in contact with electrodes in to promote the transport of material to eliminate a furnace, an electric field is applied and the furnace pores. Such high temperatures are costly in terms heated until a specific temperature or field and the of energy and can be restrictive in the manufacture ‘flash’ phenomenon occurs (7). The densification of functional ceramic devices which often require occurs rapidly and significantly reduces the integration of metals and polymers that suffer sintering temperature, for example Downs et al.

Table I Comparison of Energy 1800°C Metal Consumption for Barium compatibility Titanate of Sintering High T W, Mo Techniques (2) Sintering method Energy consumption, –1 1200°C kJ g Solid state 2800 1000°C Liquid phase 2000 Low T Cu, Au, Ag 900°C Field assisted 1050

700°C Microwave 540 Fast firing 130 Ultra-low T Cu, Au, Ag, Al Cold sintering 30 400°C

300°C the environmental impact of manufacturing. The Most metals Cold some polymers energy required to sinter BaTiO3 conventionally is 2800 kJ g–1. A comparison of energy consumption for several sintering methods is shown in Table I. Compared with conventional sintering, an energy reduction of ~99% has been Fig. 1. Material compatibility with ceramic sintering reported for BaTiO cold sintered at 300°C (2), temperatures 3 thereby demonstrating its potential importance to foundation industries such as ceramics which densified cubic yttria-stabilised zirconia (8YSZ) at are required to decarbonise. 390–1000°C below conventional temperatures (8). There are a number of challenges to be overcome 2. Cold Sintering in flash sintering including the formation of hot spots in larger components due to the electric Cold sintering or the cold sintering process current concentration (7). (CSP) is a novel method of sintering ceramics SPS is a field assisted, high pressure method first introduced by Jantunen and coworkers (12) which is particularly useful for materials that are but developed further at Pennsylvania State difficult to fabricate using conventional technology University, USA, by Randall and colleagues. CSP such as bismuth telluride-based thermoelectrics utilises a liquid phase to aid rearrangement and (9). It typically utilises a graphite die filled with interdiffusion of the particles alongside pressure powder which is subjected to high pressure and and modest heat to dramatically reduce the temperature in the presence of a field. Not only is sintering temperature (Figure 2). The proposed the temperature to achieve densification reduced mechanisms of CSP densification are closely but also the sintering time (<1 min). Oxides related to those in liquid phase sintering (2). however, require re-oxidation post densification as During cold sintering, powdered material is mixed the graphite die reduces the compound, which to with a transient liquid in which it is partially some extent negates the benefits of SPS. soluble. The moistened powder is then placed Cool SPS exploits high pressure in a vacuum into a die and pressure (100–500 MPa) and heat to densify materials with low decomposition (<300°C) are applied to aid rearrangement of temperatures or unfavourable phase transitions the particles and the reprecipitation of the solid at <400°C (10, 11). Compounds such as material (13). The development of the process manganese(II) sulfate (96%), potassium has mainly been on a material by material basis, bis‑(carbonato)cuprate(II) (94–95%), sodium optimising parameters empirically. bis‑(carbonato)cuprate(II) (97–98%), ammonium Prior to research emerging from Randall et al. iron(III) diphosphate (95–98%) and zirconia at Pennsylvania State University, Yamasaki et al. (66– 80%) have been densifiedvia this method at described a combined process of hydrothermal 300–600 MPa and 300–400°C. reactions with isostatic pressing referred to as The highly energy intensive nature of hydrothermal hot pressing (HHP) to densify conventional ceramic sintering contributes heavily ceramics at <200°C in 1986 (14). The process to the cost of the materials and products and to was used to demonstrate the densification of a

(a) (b) (c)

Powder mixed with Dissolution of ions into Dried powder weighed transient solvent the transient solvent

(d) (e) (f)

Application of uniaxial Application of uniaxial Evaporation of transient pressure causing particle pressure and heat. Increased liquid and precipitaion of rearrangement dissolution material

Fig. 2. Schematic illustrating the different stages of the cold sintering process

range of materials including silicates, cements, at 540°C but via cold sintering it can be densified

BaTiO3, porous anatase and hydroxyapatite at 120°C with the addition of 2–10 wt% distilled ceramics (14–16). The process was applied to water under applied pressure (12, 17, 18). The bonding hydroxyapatite with metal and densifying properties of cold sintered LMO are comparable ceramics which decompose at low temperatures. with conventional samples but a slight increase Cold sintering has many similarities with HHP such in dielectric loss is thought to relate to residual as the requirement for a liquid phase to facilitate hydroxyl groups at the grain boundaries (12). mass transport and the occurrence of dissolution Materials related to LMO such as sodium and precipitation reactions. HHP was inspired by molybdate (NMO) can also be densified via cold natural geographical phenomenon and is now sintering. Whilst NMO is not hygroscopic, it is considered to be a subset of the broader definition highly soluble in water making it another ideal of CSP and the equipment for HHP (17). candidate for cold sintering. NMO is conventionally sintered at 610°C, whilst the material can be cold 2.1 Congruently Dissolving Materials sintered at 150°C with the addition of 5–10 wt% water and the application of 200 MPa of pressure. 2.1.1 Molybdates Wang et al. achieved relative densities of 87% after conventional sintering but 96% after cold Lithium molybdate (LMO) is a hygroscopic material sintering. The dielectric properties of NMO are which is congruently soluble in water and was one also comparable between conventional and cold of the first materials used to demonstrate cold sintered. An increase in permittivity (εr) is observed sintering (4, 13, 18). LMO is conventionally sintered (conventional: 11.6, CSP: 12.7), due to increased

221 © 2020 Johnson Matthey https://doi.org/10.1595/205651320X15814150061554 Johnson Matthey Technol. Rev., 2020, 64, (2) density whilst residual hydroxyl groups increased dielectric loss (19, 20). (a) (b) As LMO and NMO are readily densified via cold sintering, they have been used as a starting point to create many composites with other materials with more favourable properties, but which are harder to cold sinter such as barium hexaferrite, sodium bismuth molybdate and bismuth lithium vanadium molybdate (19, 21, 22). (c) (d)

2.1.2 Zinc Oxide

Zinc oxide has also gained interest as a material that can be densifiedvia cold sintering. ZnO is wide band-gap (3.4 eV) semiconductor traditionally 5 µm used in electronics such as varistors and requires temperatures in excess of 1100°C to sinter (23, Fig. 3. Fracture surface SEM micrographs of 24). The high temperatures for conventional cold sintered ZnO fabricated at The University sintering leads to grain coarsening and other effects of Sheffield at a range of temperatures and deleterious to the electrical properties and therefore pressures: (a) 120°C and 250 MPa; (b) 200°C and 250 MPa; (c) 300°C and 250 MPa; (d) methods of reducing sintering temperature have 300°C and 375 MPa. All micrographs are at same been widely investigated. Several studies have magnification successfully cold sintered ZnO at ≤300°C. Unlike molybdate compounds, ZnO has limited solubility in water and therefore an alternative transient producing a sample with a microstructure similar solvent such as aqueous acetic acid or zinc acetate to conventional sintering. (Zn-Ac) is utilised to achieve sufficient dissolution Kang et al. studied a large range of processing to promote densification (25, 26). conditions, including the use of various solvent Recent work at The University of Sheffield has chemicals and the effect of pressure, temperature, investigated the effects of pressure and temperature pH and ion concentration. Initial findings agreed on the cold sintering of ZnO. Samples of ZnO with Funahashi with pressure promoting necking were produced at temperatures 125–300°C and and temperature having the largest effect on grain pressures of 187–375 MPa with 25–30 wt% of 1 M growth (25, 26). However, it was proposed that acetic acid. Scanning electron microscopy (SEM) pressure has a threshold below which densities of cold sintered ZnO showed that temperature and are significantly affected. Beyond this threshold, pressure have a significant effect on grain growth densification becomes pressure independent. and morphology, helping to corroborate previous Other chemicals explored by Kang et al. (other work by Funahashi et al. and Kang et al. (25, 26). than acetic acid or Zn-Ac) include hydrochloric Funahashi studied cold sintering of ZnO using acid, sulfuric acid, zinc chloride and zinc sulfate a wide variety of pressures, temperatures and but only 70–75% densities were achieved, as well concentrations of acetic acid (0.1–17.5 M) and as unwanted cement and hydroxide phases (25). water for comparison. The presence of the acetic Densification was independent of pH due mainly acid was critical to achieve high density, with 1.0 M to the presence of Zn2+ and acetate ions. The the optimal concentration. Pressures of 387 MPa exchange of Zn2+ ions through solution enabled combined with 300°C were reported to produce by applied pressure is the largest contributor to the highest densities from the test pressures and densification (25). Secondary phases observed temperatures. The lower pressure of 77 MPa did by Kang et al. were also present, contradicting produce high densities however no neck-growth earlier suggestions that samples were single phase was observed (26), in agreement with work done (26). Raman analysis showed evidence of residual in The University of Sheffield. Pellets pressed at acetate or Zn-Ac (25) which was confirmed in our 250 MPa at 300°C showed high density (>98% studies. Figure 4 shows a comparison of ZnO theoretical) and grain growth but no necking, cold sintered at 125°C and 300°C compared to a Figure 3(c). When pressure was increased conventionally sintered sample produced at The to 374 MPa necking is observed (Figure 3(d)) University of Sheffield. At 125°C and 300°C, three

* Currently under investigation

C–C C–OC–H

* 300°C 375 MPa a.u. y,

* 125°C 375 MPa Normalised intensit *

1100°C Conventional

0 500 1000 1500 2000 2500 3000 3500 4000 Raman stokes shift, cm–1

Fig. 4. Raman spectra of ZnO comparing sintering conditions. Conventional sintering at 1100°C for 2 h, cold sintering performed at 125°C, 375 MPa and 300°C, 375 MPa with 30 wt% 1 M acetic acid

peaks are observed which are not present in the Table II Parameters Used by Gonzalez- –1 conventional sample. The peak at ~943 cm is Julian et al. (27) to Create typical of a C–C bond, ~1435 cm–1 a C–O bond CSP and CSP-SPS Samples –1 and finally at ~2930 cm a mode characteristic CSP- CSP of a C–H and the peak ~650 cm–1 is still under SPS investigation. These organic peaks are weaker in 1.6–3.2 1.6 H2O the 300°C sample, suggesting that some of the H O 1.6 H O Solvent content, wt% 2 2 acetate has been removed. Acetate decomposes at 3.2 H2O + + 0.5 0.5 Zn-Ac Zn-Ac ~225°C consistent with this hypothesis. Temperature ramp rate, Gonzalez-Julian et al. densified ZnO via cold 20 100 °C min–1 sintering and a combined cold sinter-FAST/SPS Holding temperature, °C 250 250 method, using Kelvin probe force microscopy 100, (KPFM) on resulting samples, to better understand Pressure, MPa 150, 300 125, 150 the mechanisms behind cold sintering (27). For both CSP and CSP‑SPS, powders were moistened with 1.6– 3.2 wt% distilled water or a 0.5% Zn–Ac solution. surface potentials are reduced indicating a lower Samples were then sintered according to parameters defect concentration, which the authors attribute to in Table II. The addition of Zn–Ac to the transient the observed grain growth. The increase in surface solvent was found to significantly reduce the onset potential indicates that the solvent phase not only temperature of densification from 90–130°C to promotes transport but also raises the sintering ~25°C at all pressures. This demonstrates the potential through the creation of activation energy crucial role of powder dissolution in densificationvia lowering defects, as OH– and H+ ions diffuse into cold sintering. the surface of the crystal structure (25–27). KPFM was used to analyse the surface potentials From impedance spectroscopy, ZnO densified of samples sintered via CSP-SPS at 150 MPa. using Ac-H2O had the highest conductivity with The addition of water increased surface potential significantly lower total activation energy than compared to the as-received powder, which implies other conditions. The bulk activation energy of an increase in defect concentration. A contrasting ZnO was found to be significantly reduced by effect is seen with the addition of Ac-H2O, where sintering with both water and acetate, whereas

the grain boundary (Ea) was found to be increased as reducing the particle size to nanoscale, thereby with H2O and decreased with Ac-H2O. This increasing the reactivity of the powder and altering lowering of activation energy is thought to be due liquid additive to include more complex acids, to the manufacture of highly defective diffusion alkalis or ionic solvents. pathways, which helps to encourage sintering at In cases where the powder dissolves incongruently, low temperatures. a method of hydrothermal assisted cold sintering Overall Gonzalez-Julian et al. theorised the liquid is utilised through reactive intermediate phases. phase has five main roles during the CSP in ZnO: The liquid utilised in cold sintering of incongruent (a) a better initial packing of the powder material materials is often a solution containing a deep due to interparticle friction; (b) dissolution of eutectic reaction precursor to form the desired Zn2+ and O2– ions from the powder surface; products at temperatures below that of a solid‑state (c) formation of defects in ZnO crystals due to process (29–32). + – H and OH diffusion; (d) formations of highly When particles of BaTiO3 are exposed to water, Ba defective diffusion pathways between grains and ions leach from the surface, leaving a titanium‑rich

(e) elimination of carbonates. These effects are layer (33). To cold sinter BaTiO3, Guo et al. thought to be further enhanced by the presence utilised nanoscale particles of BaTiO3 and a barium of the acetate phase by improving dissolution. This hydroxide on titania suspension in deionised water. paper indicates that the liquid phase has a more This prevents the dissolution of Ba during cold complex role than first suggested by initial studies; sintering and the Ba(OH)2 and TiO2 react to form better understanding of defect chemistry effects BaTiO3 during annealing at 700–900°C. is in understanding and improving the results Strontium titanate is conventionally sintered achieved from the CSP. at over 1400°C (34). Boston et al. developed a

To unite cold and flash sintering, Nieet al. studied method of cold sintering for SrTiO3 which utilised the effect of an aqueous transient liquid phase reactive intermediate phases (29). Nanoscale on flash sintered ZnO. An electrode green body SrTiO3 and TiO2 powders were mixed with 0.2 ml produced by uniaxial pressing was placed in a flash of a 1.5 M strontium chloride aqueous solution chamber and flowing wet argon + 5% hydrogen with 1.5 M equivalent of anatase nanoparticles. was introduced after 1 h. The conductivity of The mixture was pestle and mortared to produce the hydrated pellet increased by a factor of four a free-flowing powder which was then pressed at compared to the unhydrated form (3 × 10–7 S cm–1 750 MPa for 10 min at room temperature before to ~7 × 10–3 S cm–1). The presence of water was increasing to 180°C for 60 min. After cold sintering, found to trigger flash sintering at room temperature a 4 h heat treatment at 950°C was utilised to due to the higher conductivity, producing relative promote microreactions between SrCl2 and TiO2 densities of ~98%. The water was also proposed to intermediate phases, forming SrTiO3. Electrical assist with densification via mass transport due to testing of cold sintered SrTiO3 showed similar trends partial dissolution of the substrate (28). to conventionally sintered materials, however the relative permittivity values exhibited frequency 2.2 Non-Congruently Dissolving dependence. Particle size in the conventionally sintered samples is shown to affect the permittivity Materials and loss (Table III). As already discussed, Li, NMO and ZnO are relatively easy to cold sinter and coarse powders 2.3 Challenges for Cold Sintering can be densified with the addition of water or acetic acid as the transient solvent. To cold sinter Cold sintering is an exciting area for development a wider variety of materials, with a broader range in ceramic science. There are however a number of properties, several methods are employed, such of challenges which will need to be overcome to

Table III Relative Permittivity and Tan δ from 25–250°C for Cold vs. Conventional Sintered

SrTiO3 (29) Nanoscale Micron-scale Permittivity Tan δ Permittivity Tan δ Conventional 130–210 0–0.55 120–180 0–0.14 Cold 70–120 0–0.21 70–120 0–0.21

224 © 2020 Johnson Matthey https://doi.org/10.1595/205651320X15814150061554 Johnson Matthey Technol. Rev., 2020, 64, (2) improve the commercial prospects of this new formation. Even for more successful solvent phases technology. used to densify ZnO, such as acetic acid and Zn-Ac, small amounts of residual acetate phases have 2.3.1 Processing Parameters been detected and affect properties. When BaTiO3 interacts with distilled water, Ba is Most developments in cold sintering so far have leached from the material leading to an amorphous been on an empirical, material-specific, ‘trial Ti-rich surface layer. This preferential leaching and error’ basis. A greater understanding of the is overcome using a solution containing high mechanisms and how they relate to processing concentrations of Ba and Ti ions but amorphous parameters will allow a wider range of materials material forms which requires a further post CSP to be densified via cold sintering. There are crystallisation step at high temperatures. numerous processing parameters which can be altered to tailor the densification of material during 2.3.3 Nanoparticle Manufacture cold sintering, including composition of transient liquid phase, volume of transient liquid required, While the amount of energy required to densify pressure, temperature and powder particle size. material via cold sintering has shown to be The transient phase should allow for the congruent significantly reduced, the energy of nanopowder dissolution of the solid phase or react to form a production has not been routinely considered desirable composition upon heating during sintering when evaluating the total energy consumed. To or subsequent heat treatment. Therefore, it is produce nanopowders significant amounts of important to understand the dissolution behaviour energy or complex chemical reactions are often of the ceramic within the temperature range of required, transferring the energy consumption sintering. The amount of liquid used during cold and environmental costs to a different stage of the sintering is mostly quoted in weight percent of manufacturing process. the solid phase. This does not consider the effect of surface area, far greater for nano- as opposed 3. Applications of Cold Sintering in to micropowders. The purpose of pressure during Radiofrequency Technology cold sintering is the rearrangement of particles, but it also plays a more complex role in dissolution, 3.1 Microwave Dielectric Composites grain growth and activation of reactions due to inhomogeneous pressure distributions. Microwave (MW) dielectric materials show The temperatures used during cold sintering are strong interactions with electromagnetic waves, largely dependent on the evaporation point of the making them extremely important in modern solvent. Grain growth has also been observed in communications as resonators, filters and some materials cold sintered significantly above substrates (22, 35). The three selective parameters the solvent evaporation temperature but below of MW dielectric ceramics are high quality factor conventional sintering temperatures, this could be (Qf), near-zero temperature coefficient of resonant used to achieve specific grain sizes and structures. frequency (TCF) and high εr (19, 22, 35). With fifth generation (5G) network technologies 2.3.2 Residual Secondary Phases beginning to be utilised and installed in numerous countries, the material challenge is to develop In some cases, secondary phases can form during systems of very high resonant frequency and low cold sintering, due to reactions between the solid latency. Whilst fourth generation (4G) systems phase and transient solvent or residual solvent operate in the 2–8 GHz range, the operating range after sintering completion. Kang et al. used a of 5G systems will eventually be up to 30 GHz. For number of acids in the cold sintering of ZnO to these 5G systems, the dielectric loss of polymeric observe their effectiveness as solvent phases. substrates used in 4G is too high and other

For a ZnCl2 solution, significant amounts of zinc substrates must be investigated (36–39). oxychloride phases were observed, with similar Cold sintering has shown promise within this area results produced when sulfate and nitrate based at The University of Sheffield with the densification solvents were used. While the use of such strong of several known MW ceramics achieved at low solvents is an extreme example, it demonstrates temperatures. However, none of the early materials the importance of correct solvent choice for the such as LMO and NMO exhibited near zero TCF. sintering process to prevent secondary phase Consequently, Wang et al. (17, 19) has developed

225 © 2020 Johnson Matthey https://doi.org/10.1595/205651320X15814150061554 Johnson Matthey Technol. Rev., 2020, 64, (2) several, two component temperature stable MW Although the Qf values of these composites do ceramic composites via CSP. not compete with conventionally sintered ceramics

Na0.5Bi0.5MoO4-Li2MoO4 (NBMO-LMO) composite for resonator applications, their properties, ease samples were produced by mixing NBMO and LMO of integration and low energy consumption show powders with 5–10 wt% of deionised water and promise for a wide range of novel devices. pressing pellets 30 min at 150°C and 200 MPa. Sintered pellets were dried for 24 h at 120°C to 3.2 Graded-Index Lenses remove any residual moisture (22). The NBMO‑LMO ceramic composites in this study showed no Graded-index (GRIN) lenses (19, 43) are able to chemical reaction between the phases during convert a point electromagnetic source to a planar cold sintering and near zero TCF was achieved at wave and vice versa. They are normally used in

~20% LMO with εr = 17 and Qf = 8000 GHz (22) optics but CSP can be used to fabricate devices (Figure 5). from ceramics in MW applications. A MW GRIN lens

(Bi0.95Li0.05)(V0.9Mo0.1)O4-Na2Mo2O7 (BLVMO‑NMO) (19, 22) consists of concentric rings of material, composites were also sintered by combining the with decreasing εr towards the outer edge of the mixtures with 5–10 wt% of deionised water and structure, preferably reaching values close to hot pressing for 30 min at 150°C and 200 MPa. εr = 1. Dimensions and εr of the layers are tailored A post sintering drying step of 120°C for 24 h to ensure they have the same focal point (O) to was performed to remove residual moisture (19). convert the incident spherical wave to a plane Electrical and MW analysis of the BLVMO-NMO wave; a schematic GRIN lens design is shown in showed similar trends to the NBMO-LMO and near Figure 7 and a simulation (CST Microwave Studio, zero TCF was obtained at ~20% NMO with εr ~40 Dassault Systèmes, France) of a working GRIN lens and Qf = 4000, Figure 6. is shown in Figure 8.

Fig. 5. (a) Density; (b) permittivity; (c) Qf; and (d) TCF of NBMO-xLMO composite ceramics produced by cold sintering, comparing permittivity to conventional samples produced and measured by Zhou et al. (19). Adapted from (19) under Creative Commons Attribution 4.0 International (CC BY 4.0)

Fig. 6. (a) Density; (b) permittivity; (c) Qf; and (d) TCF of cold sintered BLVMO-NMO composite ceramics and conventionally sintered BLVMO and NMO. Adapted from (19) under Creative Commons Attribution 4.0 International (CC BY 4.0)

εr1 εr2 εr3 εr4 εr5 D O

εr1 F εr2

εr3

εr4

εr5 εr1>εr2>εr3>εr4>εr5

Fig. 7. Schematic of a GRIN lens design principal. Where D is the external diameter of the layer, θ is the angle from focal point (O) to middle of ring and F is the focal length. Reproduced with permission from (22). Further permissions related to this material should be directed to the American Chemical Society (ACS)

The simulated lens was illuminated with an compared to the case with no lens was between open-ended Ka-band waveguide (7.112 mm 4.6 dB and 8.5 dB. The simulated BLVMO-NMO × 3.556 mm). Across the whole frequency and NBMO-LMO lenses exhibited an aperture range the boresight directivity was increased efficiency ~70% at 26 GHz and ~78% at 34 GHz from 26 GHz to 40 GHz, the relative increase respectively (19, 22).

dB 10

–2

–6

–10

–14

–18

–22

–26

–30

–35

Fig. 8. Simulation of electric fields as a GRIN lens transforms spherical wave fronts into planar waves. Reproduced from (19) under Creative Commons Attribution 4.0 International (CC BY 4.0)

Due to the ability to control lateral dimensions composite. Composites were produced at 150°C during cold sintering, it is possible to co‑sinter under a uniaxial pressure of 200 MPa for 30 min, multiple layers of ceramics without the achieving high density for all compositions. From detrimental effects of differential shrinkage energy-dispersive X-ray spectroscopy (EDS) and divergent thermal expansion coefficients. mapping, the authors showed that two chemically Wang et al. co‑sintered three layers of ceramic discrete regions are present in the composites after in the LMO‑NBMO system (LMO, NMBO-10 wt% cold sintering indicating that no reaction occurs LMO and NMBO-50 wt% LMO) to create a between the two phases during sintering (45). macroscopic ceramic-ceramic composite (22). This For 5G antenna substrates, materials should have demonstrated the ability to utilise cold sintering low εr (<15), a near-zero TCF and high-quality to produce graded dielectrics and thus illustrated factor (45, 46). In previous studies, the dielectric proof of concept for the fabrication of a MW GRIN properties of composites approximately follow lens. Simulations were performed to understand known mixing laws and can therefore be tailored the potential efficiency of lenses composed of six to suit specific applications. A CTO-KMO composite –1 concentric rings of radially reducing εr, illuminated produced with 92 wt% KMO had TCF ~–4 ppm °C , with a Kα-band waveguide between 26–40 GHz. εr = 8.5 and Qf ~11,000 GHz. This composition Peak aperture efficiencies were 78% at 34 GHz for was then used to create a cold sintered MPA which the NBMO-LMO lens and 70% at 26 GHz for the operates at 2.51 GHz with a 62% radiation efficiency BLVMO-NMO lens. Demonstrating high conversion (Figure 9). The combination of high antenna rates between input and output of the lens. performance and low temperature densification demonstrate the potential for the direct fabrication 3.3 Microstrip Patch Antennas of antenna substrates onto PCBs (45).

Microstrip patch antennae (MPA) are a low-profile 3.4 Multilayer Ceramic Capacitors form of antenna that can be integrated effectively where space and weight restrictions apply and Multilayer ceramic capacitors (MLCCs) consist of maybe printed onto polymer-based printed circuit alternate layers of ceramic and metallic electrode boards (PCBs) for mobile device applications and over three trillion are produced every year (44, 45). (20). MLCCs are conventionally sintered at high At The University of Sheffield, Wang et al. made temperatures which presents several challenges, use of cold sintering to produce an MPA from a not least the electrode melting point and chemical calcium titanite-potassium molybdate (CTO-KMO) compatibility with the ceramic.

(a) (b) 0 0 30 0 –30

–10 60 –60 –5 –20

90 –30 –90 s11, dB –10

120 –120

E-plane H-plane –15 150 –150 1 1.5 2 2.5 3 3.5 4 180 Frequency, GHz

Fig. 9. (a) Efficiency; and (b) radiation pattern of a microstrip patch antenna (inset (a)) fabricated from CTO- KMO

20 10 0 –10 –20 –30 CC, %

T –40 –50 –60 –70 –80 –75 –50 –25 025 50 75 100 125 150 400 µm Temperature, °C

Fig. 10. Comparison of commonly used capacitor categories, according to TCC and temperature Fig. 11. Cold sintered multilayer capacitor with C0G characteristics fabricated at 150°C from BLVMO- 0.2NMO. Reproduced with permission from (20)

Capacitors are used in a wide variety of applications δ ≈ 0.001) and εr = 40 (19, 20). Using BLVMO-xNMO and conditions and they are characterised by the (x = 0.2) composites, researchers at The University temperature dependency of their properties. C0G of Sheffield have demonstrated the use of cold (or NP0) Class 1 dielectric materials do not show sintering to produce multilayer ceramic capacitors a significant variation in capacitance over a wide with comparable properties to conventional range of temperatures. Materials with positive and calcium zirconate C0G MLCCs manufactured at negative temperature coefficients of capacitance 1100°C (20). Laminated stacks were made from (TCC) can be mixed to create a temperature stable tape cast BLVMO-NMO with screen printed silver ceramic composite. Figure 10 compares the TCC electrodes. After binder burnout, at 180°C for 3 h, of several common Class 2 and C0G capacitors. stacks were exposed to water vapour in a sealed The TCC of BLVMO and NMO ceramics are beaker at 80°C. The moistened stacks were then approximately +81 ppm °C–1 and –99 ppm °C–1 cold sintered at 150°C under 100 MPa of pressure respectively. The combination of these materials for 30 min. The SEM image of the cross-section of creates temperature stable composites the cold sintered MLCC in Figure 11 shows the <10 ppm °C–1 with low dielectric loss (tan dielectric layers are well densified, well laminated

229 © 2020 Johnson Matthey https://doi.org/10.1595/205651320X15814150061554 Johnson Matthey Technol. Rev., 2020, 64, (2) and unwarped. The Ag electrodes also appear well 3. T. Kimura, Q. Dong, S. Yin, T. Hashimoto, A. Sasaki defined indicating no reaction at the metal-ceramic and T. Sato, J. Eur. Ceram. Soc., 2013, 33, (5), 1009 interface (20). The room temperature εr and loss at 1 MHz were found to be 39 and 0.01 respectively 4. C. A. Randall, S. F. Wang, D. Laubscher, and the TCC was within 0.013% up to 150°C. J. P. Dougherty and W. Huebner, J. Mater. Res., 1993, 8, (4), 871 5. H.-I. Hsiang, C.-S. Hsi, C.-C. Huang and S.-L. Fu, 4. Commercialisation: Challenges for J. Alloys Compd., 2008, 459, (1–2), 307 Scale-up and Complex Structures 6. “Sintering of Advanced Ceramics”, eds. C. A. Handwerker, J. E. Blendell and W. Kaisser, Some ceramic materials require high pressures to Ceramic Transactions, Vol. 7, American Ceramic achieve the desired densification. To scale up the CSP Society, Ohio, USA, 1990, 789 pp to produce larger components therefore, a significant 7. M. Biesuz and V. M. Sglavo, J. Eur. Ceram. Soc., increase in applied pressure will be required compared 2019, 39, (2–3), 115 to laboratory scale. Cold sintering is currently a batch process with a small number of samples produced. 8. J. A. Downs and V. M. Sglavo, J. Am. Ceram. Soc., 2013, 96, (5), 1342 For this technology to be adopted more widely, rates of manufacture should be increased. Currently only 9. M. Al. Minnath, ‘Metals and Alloys for Biomedical simple two-dimensional and pellet shapes have Applications’, in “Fundamental Biomaterials: Metals”, eds. P. Balakrishnan, M. S. Sreekala and been reported. Many potential applications for cold S. Thomas, Ch. 7, Elsevier Ltd, Duxford, UK, 2018, sintering ceramics will require more complex shapes, pp. 167–174 which could be achieved by creating near net-shape 10. T. Herisson de Beauvoir, A. Sangregorio, I. Cornu, die pieces, although this brings its own challenges C. Elissalde and M. Josse, J. Mater. Chem. C, of die filling and variations in pressure across the 2018, 6, (9), 2229 component during sintering. 11. C. Elissalde, U.-C. Chung, M. Josse, G. Goglio, M. R. Suchomel, J. Majimel, A. Weibel, F. Soubie, A. Flaureau, A. Fregeac and C. Estournès, Scripta 5. Conclusion Mater., 2019, 168, 134 12. H. Kähäri, M. Teirikangas, J. Juuti and H. Jantunen, Alongside the reduction in manufacturing energy J. Am. Ceram. Soc., 2014, 97, (11), 3378 consumption, lower sintering temperatures allow 13. J. Guo, H. Guo, A. L. Baker, M. T. Lanagan, densification of materials that would otherwise react, E. R. Kupp, G. L. Messing and C. A. Randall, decompose or volatilise. This new area of research Angew. Chemie, 2016, 128, (38), 11629 is gaining traction as more materials and prototype 14. N. Yamasaki, K. Yanagisawa, M. Nishioka and components are developed. Cold sintering has the S. Kanahara, J. Mater. Sci. Lett., 1986, 5, (3), potential to be a game changing technology in 355 ceramic processing and component manufacturing 15. K. Yanagisawa, K. Ioku and N. Yamasaki, J. Am. and work at The University of Sheffield has Ceram. Soc., 2005, 80, (5), 1303 demonstrated several potential avenues for cold 16. K. Hosoi, T. Hashida, H. Takahashi, N. Yamasaki sintering to create devices and composites which and T. Korenaga, J. Am. Ceram. Soc., 1996, 79, are not achievable through conventional sintering. (10), 2771 17. J.-P. Maria, X. Kang, R. D. Floyd, E. C. Dickey, H. Guo, J. Guo, A. Baker, S. Funihashi and Acknowledgements C. A. Randall, J. Mater. Res., 2017, 32, (17), 3205 The authors would like to acknowledge the support of the Johnson Matthey Strategic Fund and the 18. H. Kähäri, M. Teirikangas, J. Juuti and H. Jantunen, Engineering and Physical Science Research Council J. Am. Ceram. Soc., 2015, 98, (3), 687 (EPSRC) grants, EP/L017563/1 and EP/N010493/1. 19. D. Wang, S. Zhang, D. Zhou, K. Song, A. Feteira, Y. Vardaxoglou, W. Whittow, D. Cadman and I. M. Reaney, Materials, 2019, 12, (9), 1370 References 20. D. Wang, D. Zhou, K. Song, A. Feteira, C. A. Randall and I. M. Reaney, Adv. Electron. Mater., 2019, 5, 1. R. M. German, Powder Metall., 2013, 56, (2), 117 (7), 1900025 2. D. S. B. Heidary, M. Lanagan and C. A. Randall, 21. S. S. Faouri, A. Mostaed, J. S. Dean, D. Wang, J. Eur. Ceram. Soc., 2018, 38, (4), 1018 D. C. Sinclair, S. Zhang, W. G. Whittow,

Y. Vardaxoglou and I. M. Reaney, Acta Mater., 35. N. M. Alford, S. J. Penn, A. Templeton, X. Wang, 2019, 166, 202 J. C. Gallop, N. Klein, C. Zuccaro and P. Filhol, 22. D. Wang, D. Zhou, S. Zhang, Y. Vardaxoglou, ‘Microwave Dielectrics’, IEE Colloquium on Electro- W. G. Whittow, D. Cadman and I. M. Reaney, ACS Technical Ceramics – Processing, Properties and Sustain. Chem. Eng., 2018, 6, (2), 2438 Applications, The Institution of Engineering and Technology, London, UK, 14th November, 1997 23. T. K. Gupta, J. Am. Ceram. Soc., 1990, 73, (7), 1817 36. I. F. Akyildiz, S. Nie, S.-C. Lin and M. Chandrasekaran, Comput. Networks, 2016, 24. T. K. Roy, D. Bhowmick, D. Sanyal and 106, 17 A. Chakrabarti, Ceram. Int., 2008, 34, (1), 81 37. E. Hossain and M. Hasan, IEEE Instrum. Meas. 25. X. Kang, R. Floyd, S. Lowum, M. Cabral, E. Dickey Mag., 2015, 18, (3), 11 and J.P. Maria, J. Am. Ceram. Soc., 2019, 102, (8), 4459 38. G. Ancans, V. Bobrovs, A. Ancans and D. Kalibatiene, Procedia Comput. Sci., 2017, 104, 26. S. Funahashi, J. Guo, H. Guo, K. Wang, A. L. Baker, 509 K. Shiratsuyu and C. A. Randall, J. Am. Ceram. Soc., 2016, 100, (2), 546 39. P. Wang, Y. Li, L. Song and B. Vucetic, IEEE Commun. Mag., 2015, 53, (1), 168 27. J. Gonzalez-Julian, K. Neuhaus, M. Bernemann, J. Pereira da Silva, A. Laptev, M. Bram and 40. D. Zhou, C. A. Randall, L.-X. Pang, H. Wang, O. Guillon, Acta Mater., 2018, 144, 116 J. Guo, G.-Q. Zhang, Y. Wu, K.-T. Guo, L. Shui and X. Yao, Mater. Chem. Phys., 2011, 129, (3), 688 28. J. Nie, Y. Zhang, J. M. Chan, R. Huang and J. Luo, Scripta Mater., 2018, 142, 79 41. J. Guo, S. S. Berbano, H. Guo, A. L. Baker, M. T. Lanagan and C. A. Randall, Adv. Funct. 29. R. Boston, J. Guo, S. Funahashi, A. L. Baker, Mater., 2016, 26, (39), 7115 I. M. Reaney and C. A. Randall, RSC Adv., 2018, 8, (36), 20372 42. D. Zhou, C. A. Randall, H. Wang, L.-X. Pang and X. Yao, J. Am. Ceram. Soc., 2010, 93, (4), 1096 30. H. Guo, J. Guo, A. Baker and C. A. Randall, J. Am. 43. S. Zhang, R. K. Arya, S. Pandey, Y. Vardaxoglou, Ceram. Soc., 2016, 100, (2), 491 W. Whittow and R. Mittra, IET Microwaves 31. C. Vakifahmetoglu, J. F. Anger, V. Atakan, S. Quinn, Antennas Propag., 2016, 10, (13), 1411 S. Gupta, Q. Li, L. Tang and R. E. Riman, J. Am. 44. L. V. Blake and M. W. Long, “Antennas: Ceram. Soc., 2016, 99, (12), 3893 Fundamentals, Design, Measurement”, 3rd Edn., 32. J.-P. Ma, X.-M. Chen, W.-Q. Ouyang, J. Wang, H. Li SciTech Publishing, Raleigh, USA, 2009, 503 pp and J.-L. Fang, Ceram. Int., 2018, 44, (4), 4436 45. D. Wang, S. Zhang, G. Wang, Y. Vardaxoglou, W. Whittow, D. Cadman, D. Zhou, K. Song and 33. M. C. Blanco-Lopez, B. Rand and F. L. Riley, J. Eur. I. M. Reaney, Appl. Mater. Today, 2020, 18, Ceram. Soc., 1997, 17, (2–3), 281 100519 34. M. Bäurer, H. Kungl and M. J. Hoffmann, J. Am. 46. I. M. Reaney and D. Iddles, J. Am. Ceram. Soc., Ceram. Soc., 2009, 92, (3), 601 2006, 89, (7), 2063

The Authors

Jessica Andrews completed an MEng in Materials Science and Engineering in 2016. She is now a PhD student within the Functional Materials and Devices Group at The University of Sheffield. Supervised by Professor Reaney and sponsored by Johnson Matthey, her project is focused on developing methods of cold sintering for several ceramic materials and polymer-ceramic composites.

Daniel Button originally graduated at The University of Sheffield with an MEng in Materials Science and Engineering in 2017 and is currently a PhD student in the Functional Materials and Devices at the same university. He is working on a project sponsored by Johnson Matthey, investigating the cold sintering of ZnO and its possible use cases for industrial application.

Ian M. Reaney was appointed Professor in Ceramics in 2007 and in 2017 to the Dyson Chair in Ceramics at The University of Sheffield. Ian’s research interests are in development of MW and piezoelectric materials and functional glasses. He’s published more than 400 papers and given over 100 invited, keynote and plenary talks at international conferences. His h-index is 63 and his papers have received >16,000 citations. He won the Edward C. Henry award in 2002, best knowledge transfer partnership building on EPSRC funded research in 2009 and the Verulam Medal in 2017. He is European site director of the Centre for Dielectrics and Piezoelectrics.

232 © 2020 Johnson Matthey This page is intentionally blank www.technology.matthey.com Johnson Matthey Technology Review is Johnson Matthey’s international journal of research exploring science and technology in industrial applications www.technology.matthey.com Editorial team

Manager Dan Carter Editor Sara Coles Editorial Assistant Yasmin Stephens Senior Information OfficerElisabeth Riley

Johnson Matthey Technology Review Johnson Matthey Plc Orchard Road Royston SG8 5HE UK Tel +44 (0)1763 253 000 Email [email protected] www.technology.matthey.com