Mechanochemistry : A versatile synthesis strategy for new materials

Mechanochemistry deals with reactions induced by the input of mechanical energy – for example by impacts within a vibratory ball mill. The technique has a long history with significant contributions from Ostwald, Carey Lea and, notably, Faraday. Mechanochemistry has subsequently seen application in a variety of areas of materials science including mechanical alloying in metallurgy, the synthesis of complex organic molecules and, more recently, the discovery and development of new solid forms of active pharmaceutical ingredients. This paper overviews the broad areas of application of mechanochemistry, some key features which make it a particularly attractive approach to materials synthesis and some mechanistic aspects highlighted within the literature. A significant part, however, will focus on recent applications in the area of pharmaceuticals and its important role in exploring the rich variety of solid forms available for small, drug-like, molecules.

Please note that technical editing may introduce minor changes to the text and/or graphics, which may alter content. The journal's standard Terms & Conditions and the Ethical guidelines still apply. In no event shall the Royal Society of Chemistry be held responsible for any errors or omissions in this Accepted Manuscript or any consequences arising from the use of any information it contains.

Introduction
Mechanochemistry is a long-established method for the synthesis of solids and molecules. Its chemical importance was recorded by Faraday, who in 1820 demonstrated the reduction of AgCl to Ag by grinding in a mortar and pestle a mixture of AgCl and Zn. (Takacs, 2007) Significant progress followed, especially through the work of Carey Lea (1823 -1897), (Takacs, 2003) and the position of mechanochemistry within the framework of chemistry as a whole was established by Ostwald who noted the equivalency of electrochemistry, thermochemistry, photochemistry and mechanochemistry. (Boldyrev, 2006, Baláž, 2008, Takacs, 2013 Long held as an efficient process for the synthesis of metallic alloys, a recent extensive review by James et al (James et al., 2012) has highlighted several of the emergent areas now being actively pursued with regard to the application of mechanical energy in driving various transformationsindeed a working definition from that review is "mechanochemistry refers to reactions, normally of solids, induced by the impact of mechanical energy such as grinding in ball mills". A formal definition 2. Areas of application and development.
The areas in which mechanochemistry has been applied with success are varied and include: catalysts, (Ralphs et al., 2013) nanoparticles, (Urakaev and Boldyrev, 2005, Smolyakov et al., 2008, Baláž et al., 2013 organic synthesis, (Etter et al., 1989, Mikhailenko et al., 2004, Trotzki et al., 2008, Bruckmann et al., 2008, Štrukil et al., 2012, Stojaković et al., 2012, Wang, 2013 MOFs, (Friščić, 2012) oxides, (Šepelák et al., 2013) biomaterials, (Hick et al., 2010, Kleine et al., 2013 supramolecular chemistry, (Braga et al., 2006, Belenguer et al., 2011, Delori et al., 2012 process engineering, (Burmeister and Kwade, 2013) dyes and pigments, (Bučar et al., 2013b) fluorophores, (Yan et al., 2011) exfoliation of graphene nanosheets (Shang et al., 2012) and modification of fullerenes (see Figure 1). (Zhu et al., 2013) Numerous further examples are given in the review by James et al (James et al., 2012) and several other reviews on the application of mechanochemistry are available. (Blagden et al., 2008, Baláž and Dutková, 2009, Guo et al., 2009, Black et al., 2011, Boldyreva, 2011, Nasser and Mingelgrin, 2012, Friščić, 2012, Boldyreva, 2013, Majano et al., 2014 In particular, the contributions of V. V. Boldyrev and his colleagues to both experimental and theoretical aspects of mechanochemistry have been important to the subject. (Boldyrev, 1995, Boldyrev and Tkáčová, 2000, Boldyrev, 2004, Boldyrev, 2006 From a synthetic chemistry viewpoint an important observation is that substances can react by a significantly different pathway in the solid state than they would in solution leading to new products or to much higher selectivities. (Toda et al., 1987, Hollingsworth et al., 1991 This ability to generate a much broader product-spectrum, as we will demonstrate later, is also particularly important in areas other than chemical transformations e.g. in crystal polymorph conversions -and it is to be noted that the addition of very small amounts of liquid may have a significant impact on the outcome of a solid-solid reaction, even though quantities are much less than a conventional solvent-based reaction might require -see later. A key area of active research remains the understanding of the processes by which mechanically induced reactions occur. ) Given the variety of materials of interest, ranging from metals and ceramics to pharmaceuticals, the specific details of how a process occurs will likely be varied and complex. (Boldyrev, 2002, Boldyrev, 2006, Michalchuk et al., 2013 Also to be distinguished is the time-scale over which an event is studied. During the few milliseconds following the impact of two solids at high velocity, significant local consequences will ensue. Hot spots, magma-plasma regions, rapid induction of defects (point, linear and planar) and then propagation of these defects through the crystal are some of the concepts which have evolved. The distribution of imparted energy will also be different in those solids with high rigidity and few pathways for energy dissipation (so that the effects at the specific point of impact will be great) than it is in those systems where structural movement (e.g. slippage of crystal planes) may allow the energy (and its consequences) to be distributed over a much larger distance. (Boldyrev, 2006) During this early stage, the making and breaking of chemical bonds would seem possible. Following the initial shortterm consequences, there will also be the longer term events associated with the final energy dissipation and relaxation of the system. The slow crystallisation of an intermediate amorphous phase may be such an example (see Figures 2 and 3). (Shakhtshneider, 1997, Nguyen et al., 2007) It is possible, then, to envisage that an individual particle, for example within a milled pharmaceutical material, may be completely amorphous initially, but over time may change to consist of domains of crystalline solid embedded within an amorphous shell (or more likely vice versa where, for example, humidity from the surrounding atmosphere results in each amorphous particle crystallising from the outside). This model might suggest that a description of a product as 25% amorphous does not mean that 25% of all the particles are amorphous and 75% are not -but that each individual particle within the product consists of domains of crystalline and amorphous in a 3:1 ratio.
A key factor for potential commercial exploitation in many instances would seem to be difficulties associated with scale-up. The progression from gram-scale, lab-based synthesis to kilogram-scale or higher outputs is an issue that has been addressed in many industrial applications 3 and more recently in the area of pharmaceuticals, where twin extrusion approaches are employed ( Figure  4). (Medina et al., 2010, Daurio et al., 2011 Furthermore, it is noteworthy that frequently the generation of a new material (e.g. in the case of molecular crystals) at the gram level can allow, through subsequent seeding of a supersaturated solution, production via a conventional solventbased crystallisation route. (Trask et al., 2005c) It is also necessary to recognise the need for methods of structurally characterising the products of mechanochemical processes. When applied to chemical conversions which may occur, molecular identity can be established by conventional analytical methods (e.g. solution NMR and mass spectrometry). For solid-solid transformations, where the identity of crystalline intermediates or final states is required, major improvements in crystal structure determination from powder X-ray diffraction data (frequently coupled with solid-state NMR) has been pivotal. (Cheung et al., 2003, Karki et al., 2007, Salager et al., 2010, Harris, 2012 In addition, the structural analysis of individual crystallites that result from mechanochemical processes is possible through utilisation of the electron diffraction facility of the transmission electron microscope. (Kolb et al., 2010, Eddleston et al., 2013b, Eddleston et al., 2013a 3. Pharmaceutical materials Within various areas of process engineering (including food science, chemicals and pharmaceuticals) particle size reduction is an important step. (Qiu et al., 2009, Burmeister andKwade, 2013) Approaches to size reduction include compression (e.g. rollers) or impact (e.g. impact mills and ball mills). All methods involve transfer of varying amounts of energy to the particle(s), and as a result other consequences can accompany the size reduction including the introduction of defects (such as dislocations, point defects and stacking faults) (Hüttenrauch et al., 1985) and in particular amorphisation (Trasi and Byrn, 2012) or polymorphic transformations. (Bruni et al., 2012) The work of Descamps and others with regard to pharmaceutical materials has clearly demonstrated the critical role of the glass transition temperature, T g , in determining whether or not a persistent amorphous phase will be generated. (Desprez and Descamps, 2006, Willart et al., 2006 The combination of increased milling intensity and the absence of any plasticisers, which might lower T g (either as liquid or vapour, e.g. water, from the surrounding atmosphere), will encourage the formation of an amorphous product ( Figure 5). The distinction between amorphous and crystalline products is particularly important for pharmaceutical applications given the inherent increased transient solubility (and hence improved bioavailability) associated with amorphous compared to crystalline solid forms. (Hancock and Parks, 2000, Murdande et al., 2010, Newman et al., 2012, Trasi and Byrn, 2012 Frequently, the amorphisation of a material can be a desirable outcome, especially as a means of increasing the transient solubility of the material or of eliminating compression issues with the crystalline equivalent. (Qiu et al., 2009) This attribute is frequently used in the pharmaceutical industry, and the complex nature of the resulting solids has been extensively studied (especially with regard to the instability of the amorphous form with respect to the crystalline state). (Hancock and Parks, 2000) Polymorphic transformations from the stable to a metastable form (again with increased apparent solubility) can also be a useful consequence of the application of mechanical energy ( Figure  6). (Pudipeddi and Serajuddin, 2005, Lin et al., 2006, Wildfong et al., 2007, Li et al., 2007 Jet-milling, a particle-size reduction method frequently used within the pharmaceutical industry where particleparticle collisions occur within a carrier gas (typically air), has, for example, been found to induce polymorph interconversions along with the size reductions.
With grinding experiments, an approach for directing the polymorphic outcome through addition of a small amount of liquid was described by Trask et al for the cases of anthranilic acid and succinic acid (see Figure 7). (Trask et al., 2005b) For anthranilic acid, interconversion between the three known polymorphic forms occurred depending upon the choice of the added liquid. (Madusanka et al., In Preparation) In the case of succinic acid, while the β-polymorph appeared stable to neat grinding, addition of a few drops of a low polarity liquid (e.g. hexane) caused conversion to the αpolymorph. More polar liquids (e.g. water, acetonitrile or methanol) appeared not to induce the transformation. The actual mechanism of the solid-solid transformation is an area of ongoing interest, especially where the change appears to be crystal-crystal in nature. (Halasz, 2010) An additional reason for interest in mechanochemical processes comes from the fact that the majority of drugs (80-90%) are formulated as tablets, which are seen as a reliable and general way of delivering medicines, and that the tableting process itself requires the application of significant applied pressures (see Figure 8). (Shakhtshneider, 1997, Brittain, 2002, Boldyreva et al., 2006, Oswald et al., 2009, Moore et al., 2009, Buckner et al., 2010, Mazel et al., 2011, Roopwani and Buckner, 2011, Christensen et al., 2012. Importantly, tablets contain other components (excipients), which are added to enhance performance e.g. lubricants, dissolving agents and so on. When pressure is applied during compression, there is a risk that surface-surface reactions between the components and drug, such as that between dicalcium phosphate dihydrate and aspirin, will lead to enhanced instability. (Cassidy et al., 2012) More recent interest has focussed on the formation of cocrystals (see Figure 9) by milling or grinding. (Braga and Grepioni, 2004, Trask and Jones, 2005, Braga et al., 2013 One of the earliest examples of a cocrystal obtained by grinding is that of quinhydrone reported in the work of Wohler published in 1844. (Wöhler, 1844) The advantage for this system compared to solution crystallization was shown by Patil et al, (Patil et al., 1984) who demonstrated that simple grinding of mixtures was significantly superior to attempted solution growth, where self oxidation-reduction lowered the efficiency of complex formation. Subsequently, numerous examples of host-complex formation by solid-solid mixing were reported by Toda and co-workers. (Toda et al., 1987) Work by Etter and colleagues also demonstrated how solid-solid grinding could be used to produce hydrogen bonded cocrystals of adenine and thymine derivatives, where a marked selectivity amongst various base pairs was observed. (Etter et al., 1993) Hollingsworth et al (Hollingsworth et al., 1991) used grinding as a means of creating cocrystals of dinitriles and urea, as part of a wide study of inclusion crystals, in cases where solution growth was not possible. Pedireddi et al demonstrated that in some cases a third component is necessary in order for a cocrystallisation reaction to take place (Pedireddi et al., 1996) and Kuroda and her colleagues studied the influence of the addition of a third component on the optical properties of the resulting cocrystals. (Kuroda et al., 2002) Frequently, the solid state properties of a pure drug can present processing or application problems e.g. particle flow, thermal stability, stability to moisture or ability to be compacted, (Qiu et al., 2009) and, increasingly, poor solubility is an issue. (Blagden et al., 2007, Bethune et al., 2011 The challenge with a badly behaving drug, therefore, is to be able to control the properties of the molecule in the solid state without recourse to chemical modification by covalent chemistry. The use of multicomponent systems, especially pharmaceutical cocrystals, and more recently ionic cocrystals, (Braga et al., 2010, Braga et al., 2011b has been demonstrated to be an effective strategy in such instances. (Almarsson and Zaworotko, 2004, Stahly, 2007, Shan and Zaworotko, 2008, Meanwell, 2008, Stahly, 2009, Friščić and Jones, 2010, Qiao et al., 2011, Brittain, 2013, Thakuria et al., 2013, Steed, 2013 In terms of property improvement with pharmaceutical cocrystals, various examples are known including: enhancing stability against hydration of an anhydrate form (see Figure 10), (Trask et al., 2005a, Trask et al., 2006, Eddleston et al., 2014b, Eddleston et al., 2014a developing forms with improved compression properties (for tablet formation)(see Figures 11 and  12), Hou, 2008, Karki et al., 2009a) significantly improving the bioavailability of poorly soluble drugs and controlled release strategies e.g. for insulin. (Brader et al., 2002) Trask has reported on the potential intellectual property implications of pharmaceutical cocrystals, (Trask, 2007) and recent interest has also focussed on regulatory guidelines. (Aitipamula et al., 2012) In nonpharmaceutical applications, benefits of cocrystallisation have been shown for fluorescent materials (Yan et al., 2011, Frenette et al., 2013 and explosives. (Landenberger andMatzger, 2010, Shen et al., 2011) An early indication of the potential to form pharmaceutical cocrystals by grinding was provided by Caira et al in 1995 who used grinding to prepare a series of crystalline products containing sulfonamide (similar to those obtained from solution) in a facile manner. (Caira et al., 1995) They also reported detailed solid state kinetic data based on powder X-ray measurements and were able to show a distinct selectivity between reactants during grinding.
An ongoing challenge, however, is selecting possible coformers which might cocrystallise with the API rather than remain as a separate and distinct crystalline phase. Whilst various strategies have been developed to predict the likelihood of cocrystal formation, (Fábián, 2009, Musumeci et al., 2011, Springuel et al., 2012, Wood et al., 2014 experimental screening for viable coformers remains vital ( Figure 13). (Andre et al., 2009, Bysouth et al., 2011, Abourahma et al., 2012 While solution growth remains the ideal approach -especially with regard to scale-up in later stages of development -significant differences in solubility between the drug and coformer often presents difficulties. (Gagnière et al., 2009, Gagnière et al., 2012, Eddleston et al., 2013c, Eddleston et al., 2014b, Eddleston et al., 2014a Mechanochemistry has emerged as a powerful tool in the experimental screening area, and has the advantage of minimising the need for large volumes of solvent ( Figure 14). It has also emerged as a powerful way of searching for different polymorphic forms of cocrystals (Eddleston et al., 2013c, Madusanka et al., 2014 as well as for various stoichiometric compositions e.g. 1:1, 1:2 and 2:1. (Trask et al., 2005c) Furthermore, mechanochemistry has been used to identify chiral agents suitable for converting a racemate into two distinct diastereomeric cocrystal phases (Eddleston et al., 2012) as well as for coformer exchange. (Braga et al., 2011a) An extension to simple (i.e. neat or dry) grinding of mixtures was reported by Shan et al who added a few drops of liquid to the reaction mixture (liquid assisted grinding, LAG) (see Figure 15). (Shan et al., 2002) Initial attention focused on LAG as a means of accelerating cocrystallisation kinetics, although it was subsequently shown to provide a means of increasing product diversity. For example, the outcome of mechanochemical formation of cocrystals has been compared with that from using other approaches to cocrystallsation. Trask et al reported three possible outcomes in the case of cocrystals with caffeine: (i) solution growth and milling produced identical products, (ii) a different stoichiometric product could be obtained by milling compared to solution and (iii) different polymorphic forms were possible. (Trask et al., 2005c) Numerous recent publications have demonstrated that LAG is likely to be the most efficient way of screening for cocrystal formation. (Myz et al., 2009, Aakeröy et al., 2011, Fucke et al., 2012, Yamamoto et al., 2012, Lin et al., 2014 The key advantage of LAG over solution based approaches was interpreted by Davey, (Chiarella et al., 2007, Chadwick et al., 2009) Childs(Childs et al., 2008 and others ) on the basis of solution phase chemistry (Figure 16). Other developments, such as ion-and liquid-assisted grinding (ILAG) (Friščić et al., 2010b) and vapour digestion of solid mixtures, Grepioni, 2005, Braga et al., 2007) have also been reported.
Further insight has been given by the recent work of Bučar et al for the case of caffeine and benzoic acid. (Bučar et al., 2013a) No cocrystals for this pair of molecules were known despite a strong expectation based on studies of analogous systems (Heiden et al., 2012) and despite extensive experimental efforts to generate such a species over many years. A systematic study using a large range of cocrystallisation methods including neat grinding, liquid assisted grinding, melt crystallization and various solution methods was employed ( Figure 17). All failed to produce the elusive cocrystal even though theoretical work had clearly confirmed a cocrystal was thermodynamically feasible. During grinding experiments, however, the addition of a few seed crystals of an isostructural fluorobenzoic acid cocrystal readily generated the elusive cocrystal, with the inference being that, as in solution crystallisation, a high barrier to nucleation of a particular form can frequently exist.
Accompanying the recent developments in mechanochemistry has been the development of techniques to follow the course of solid-solid reactions using both ex situ and in situ approaches. (Nguyen et al., 2007, Cinčić et al., 2008, Tumanov et al., 2011, Du et al., 2013, Frenette et al., 2013 Ex situ analysis has frequently suggested that there may be distinct intermediates generated (Figure 18). (Cinčić et al., 2008, Karki et al., 2009b The application of synchrotron X-ray diffraction for in situ studies has indeed confirmed the presence of crystalline intermediates and a possible role of amorphous phases.  To better understand the processes occurring during milling, several recent efforts have concerned developing methods to follow in situ the course of the reaction. Noteworthy has been the use of synchrotron Xray diffraction. In a study of the milling of a mixture of carbamazepine and succinic acid, Halasz et al  showed that, in the absence of any added liquid, the outcome was the gradual disappearance of reflections of the two components and apparent amorphisation. A similar experiment, but now with the addition of 50 μL of acetonitrile showed that within 10 seconds of milling, reflections associated with a triclinic (cbz)(sac) cocrystal emerged. The reaction was complete within 4 minutes. Such rapidity was to be compared with other solid-solid grinding systems (e.g. the reaction of ZnO with imidazoles) where reaction timescales were much longer. Similar in situ studies were conducted for suberic acid and nicotinamide demonstrating that formation of a 2:1 cocrystal during grinding proceeds via a 1:1 cocrystal intermediate (see Figure 19). Friščić and his colleagues have also used in situ methods, based on the changes in fluorescence (Frenette et al., 2013) accompanying cocrystallisation, to monitor directly events within the grinding jar.

Concluding remarks
In the case of pharmaceutical solids, recent work using atomic force microscopy (AFM), terahertz spectroscopy and transmission electron microscopy (TEM) suggests that these techniques will be important for characterising mechanically treated samples and understanding changes which occur during such processing. For example, there are particle-particle effects -especially important during compression in a tablet containing various components. In the case of aspirin (acetylsalicylic acid) and dicalcium phosphate, when the excipient was attached to the tip of the AFM cantilever and different pressures and contact times under various humidity conditions there was evidence for enhanced reactivity induced by surface contact with excipient. (Cassidy et al., 2012) Terahertz spectroscopy has emerged as a sensitive method of monitoring changes in solid form during mechanical activation (see Figure 20), (Nguyen et al., 2007, Du et al., 2013 whereas TEM has been shown to be a powerful technique for characterising the size and shape of particulates generated mechanochemically, as well as determining their crystal form (see Figure 21). TEM can also be used to identify defects in pharmaceutical crystals (see Figure 22), and will enable a greater understanding of how defects introduced into crystals during mechanical processing influence material properties, for example a link between defects in theophylline crystals and crystal fracture has been established. (Eddleston et al., 2010) Mechanochemistry, as mentioned at the outset, provides many opportunities for developing environmentally-friendly processes (especially the avoidance of large volumes of solvent). It also provides routes to materials not readily obtained by other methods. It has a long history and no doubt will continue to demonstrate useful applications whilst at the same time generate considerable activity in rationalising the mechanistic aspects behind the many types of transformations observed.  . Schematic of the characteristics of polymorphs and multicomponent systems. Cocrystals represent a sub-set of the multicomponent group, frequently that characterised by the requirement that the components be neutral and solid at room temperature. Figure 10. Supramolecular synthon approach used in the design of cocrystals of caffeine with various dicarboxylic acids. This approach lead to a variety of solid forms of caffeine which could be tested for stability under conditions of high humidity. Mechanochemistry was the technique used to both screen for cocrystal formation and also prepare powders for humidity assessment.