Mechanochemistry and Organic Synthesis: From Mystical to Practical
By Joel Andersen and James Mack
Read this full article and more in the journal of Green Chemistry.
The allure of mechanochemistry is nearing a tipping point. With the rapid growth of articles (see Fig. 1) about it, it is becoming clear that this once obscure discipline is becoming increasingly mainstream. There are many reasons for this, and any researcher taking-up this rapidly evolving methodology can benefit from all of them. Various researchers are interested in exploring and pioneering the uncharted territory of reactivity associated with unique, solvent-free conditions. Some are interested in improved laboratory safety as well as simpler, cheaper reaction procedures that allow for more efficient use of research time resulting in increased productivity. While others are interested in the enhancement of mechanochemistry on fields such as catalysis, where the tandem of mechanochemistry and catalysis has been especially powerful. All these benefits are on top of the widely-touted environmental benefits of choosing mechanochemistry for greening the reaction component of a complete chemical process (e.g., reaction, work-up, purification). These motivations will twine together tightly and symbiotically over the coming years. For example, the success of those devoted to understanding the fundamentals of the field have enhanced and streamlined the transition from conventional, solution-based chemistry to mechanochemical, solvent-free conditions. In parallel, as this activation barrier to mechanochemistry shrinks, more researchers and practitioners will join the field for functional applications, inevitably contributing their own unique lessons of the solvent-free (or solvent-reduced) world of chemistry.
There are several impactful reviews regarding mechanochemistry as it relates to non-organic sub-disciplines of chemistry. Previous reviews regarding organic chemistry in the mechanochemical environment, discussed later, are in need of a companion review that addresses current paradigms and thought processes of the field. The intention of this tutorial is to address that need: Connecting the dots between the data. This is beneficial because much of the experimentation done on mechanochemistry has been empirically driven, but in aggregate there are now large, unifying themes that can be observed and discussed. To serve as a helpful introduction to the field and to the mentality of a mechanochemist, this article has a few objectives. First, to walk through the modern process of setting up typical mechanochemical organic reactions. Second, to outline the hardware required for a researcher to incorporate mechanochemistry into their synthetic arsenal. Third, to remove some of the mystique of mechanochemistry and jump-start the adoption of a mechanochemical mindset for the uninitiated chemist. The goal here is to provide a very high-level, broad perspective one can adopt when predicting and interpreting the results of mechanochemical reactions. We hope the final objective results naturally from the first three: To spark thoughts and questions in the reader’s mind about either the fundamentals of mechanochemistry or how the methodology might simplify or solve pesky chemical challenges that inevitably plague research from time to time. Scheme 1 summarizes many of the important concepts discussed in this paper and also serves as a convenient comparison of how some mechanochemical conditions may compare to conventional conditions.
Mechanochemistry in practice
Preparing a typical mechanochemical reaction
A typical mechanochemical reaction is easy to prepare, usually more so than the solution equivalent. Reactants and reagents are weighed and added to a vial/milling jar (often stainless steel) in a manner no different than if one were using a round-bottom flask. Depending on the type and size of mill, some number of stainless steel balls are then added. Temperature control of the vial may be provided by a Variac or alternative method (see “Energetics”). At this point the vial can be closed. With no solvent, there is no need for reflux condensers. Furthermore, since solvent is often water’s Trojan horse, the reaction environment is often dry enough for a successful reaction even when dry solvents would conventionally be mandatory. This can be a major benefit for many labs, as potentially dangerous solvent stills and/or expensive solvent purification systems can be unattractive. With respect to the vial material, certain reactions can benefit from prudent choice of vial material. For example, using a copper vial (instead of stainless steel) for a copper catalysed azide–alkyne cycloaddition removes the need to add or recover copper in another form. Several such systems have been designed, and the reader can find more information regarding this topic in a recent review. One vision for mechanochemistry involves kits containing vials coated with various immobilized chiral catalysts for enantioselective reactions without any need for a recovery step. Refocusing on the reaction at hand, the vial can be added to the mill and turned on, enabling chemistry at the flip of a switch.
While running, the mill’s physical container serves as a convenient safety measure, isolating the chemist from the reaction. Reaction time may be readily controlled via built-in timers.
At the conclusion of a reaction, there are a number of methods for removing the product. The most readily apparent approach is to use solvent. Note that although the process is no longer ‘solvent-free,’ at this stage (when the reaction is over) it is often the case that there is greater liberty in solvent choice. GlaxoSmithKline’s solvent selection guide can help keep this process, as well as any further needed purification, as green as possible. Depending on the application, it may also be appropriate to simply scrape out the material.
Preparing a lab for mechanochemistry
There are several pieces of hardware that must be invested in for any lab doing mechanochemistry. Generally, the most expensive piece of equipment is the ball mill itself. There are several types of mills, but three of the most important for chemistry are described in Fig. 2. The two types most commonly used are shaker mills (also known as mixer or vibratory mills) and planetary mills. These mills are typically in the range of $3000–$7000 US dollars, which is similar to the cost of a rotary evaporator. The two mills differ in the motion that causes the mixing. Shaker mills move somewhat like a paint shaker, whereas planetary mills use a rotating drum (similar to a top-loaded washing machine). Beyond the movement, another key difference between the mills is the milling media. Milling media are objects that facilitate the mixing and grinding. They may be inert (e.g., stainless steel or zirconia balls), or sources of a catalyst (e.g., copper balls). Shaker mills often use a single ball during operation, whereas a planetary mill may contain hundreds of balls. This can lead to wildly different operating temperatures inside the vial when comparing the two systems. Indeed, shaker mills commonly have temperatures ranging from 30 °C to 75 °C depending on the design, whereas planetary mills can reach several hundred degrees. Often, procedures for planetary mills may incorporate rest intervals to let the reactor cool down after even a brief (<20 minutes) reaction period. For bench-top work from milligram to gram scale, a mixer mill will be convenient for most organic syntheses. These mills have shown the greatest versatility in terms of customization. For example, custom vials can easily be used, and as discussed later they are readily interfaced with reaction-monitoring hardware. They also allow customizations for precise control of temperature over a wide range. In our observations, an unmodified SPEX 8000 M Mixer mill consumes 280 W, as measured by a Kill-A-Watt® Electricity Monitor. In comparison, our laboratory computer and monitor consume 100 W at idle. This is balanced in many cases by significant rate enhancements typically observed in a milled process (see “Energetics”). If one wishes to work on a larger scale, a planetary mill should be considered. If industrial scale is required, twin screw extrusion (TSE) is recommended. The Stuart James group demonstrated multi kilogram per hour utility via TSE for Knoevenagel, imine formation, aldol, and Michael addition condensations while providing a product that generally did not need further purification. TSE will be discussed in more detail in subsequent paragraphs. It is important to once again note that customizations will likely need to be applied to mills of all types, and the community would benefit greatly by those interested in taking on these engineering challenges.
Beyond the mill itself, there are a number of other components that are common in milling laboratories. Depending on the type of mill that is purchased, one may opt to use custom vials. These may be homemade in a machine shop or purchased from commercial sources. One start-up is offering vials made of different materials (acrylic, teflon, stainless steel, zirconia, and tungsten carbide) of different sizes. Often, the chemist elects to match the milling media or balls to the vial material. As mentioned earlier, some metal-catalyzed reactions can use the vial/milling media itself to serve as the catalyst, so copper vials or nickel vials may be beneficial as well.
Recent advancements and current limitations of mechanochemical reactors and reactions
The last several years has produced some truly outstanding work that has advanced the utility of mechanochemical reactors. Much of the motivation for this has come from the desire to monitor reactions in situ. In situ monitoring is important because pausing the milling process to obtain samples for measurement can interfere with the process itself. To this end, the first major advancement came from Friščić and co-workers when they paired the mill with a synchrotron to use X-rays to monitor crystallographic changes during a milling process. Not long after, Užarević et al. demonstrated the incorporation of Raman spectroscopy. In quick succession, a follow-up paper detailed the benefits of combining these two. There are now commercially available mills that come complete with Raman-monitoring capabilities. These are also configured for monitoring by synchrotron X-rays, but the user is expected to provide their own X-ray source.
The above developments were not done with typical organic syntheses in mind. In that case, there are several considerations to make. First, it is likely that, from the above list of advancements, the Raman capabilities would prove most useful. However, mechanistic studies of organic reactions benefit greatly from NMR spectroscopy, and more developments will be required to bring mechanistic studies of mechanochemical organic reactions on par with conventional, solution-based reactions. Furthermore, as is addressed below in “Energetics,” the temperature of the reaction container is key for controlling chemical reactivity. Andersen et al. outlined a method for controlling vial temperature with high precision up to approximately 200 °C. Unfortunately, few methods have been outlined for decreasing temperature, and they are limited to only minor cooling of a few degrees below operating temperature or, alternatively, extreme cooling at −196 °C using liquid nitrogen, neither near the −78 °C routinely seen in conventional syntheses. More work on developing cold temperature conditions for vibratory mills and planetary mills remains.
Fortunately, twin screw extruders are capable of fine temperature control, which is important as TSE is the most readily apparent method for industrial scale up of mechanochemistry. In this type of extrusion, two interlocking screws convey material down a barrel as they are sheared and compressed. Over the 60 years since their inception, TS extruders have been optimized for energy consumption at an industrial scale. Although some optimization will be required to adjust for the properties of materials being used, TSE’s are well established in the food and plastics industries. Furthermore, TSE is a continuous process and not a batch process as is typically associated with mechanochemistry. The use of twin-screw extruders in mechanochemistry began with a pair of studies describing the formation of co-crystals of pharmaceutical significance. Since then, the Stuart James group has pioneered the path. In 2015, they reported the synthesis of metal–organic frameworks, and in 2016 they followed up with synthesis of deep eutectic solvents. It was their following paper, however, that demonstrated not only the practicality of TSE as a scale-up of mechanochemistry for organic synthesis, but an exceptionally convenient one, too. Crawford et al. investigated four different condensation reactions, as presented in Scheme 1(b). By balancing variables such as temperature, screw speed, and residence time they achieved 100% conversion in all cases. Moreover, of the four reactions, only one produced a by-product, which was present only in trace amounts. Of further note is the fact that all products could be fed directly into any desired receiving container as dry powders.
Clearly, mechanochemical reactors have made significant improvements in the last several years. Many of the remaining limitations exist not because they are insurmountable, but rather because the field is young and still maturing. As such, several of the advancements mentioned here are still commercially unavailable. Until such commercialization is realized, the development of a truly robust and ready-out-the-door mechanochemical reactor for organic synthesis will continue to require after-market customizations. Just as the commercialization of routine and specialized glassware expanded the accessibility (and productivity) of chemistry, so too will the commercialization of these in-house customizations.
That being said, there are some important current design limitations that have not been addressed even with in-house customizations. To begin, there is no mechanochemical method comparable to solution’s “dropwise addition.” Given the importance of this technique, it would be remarkably beneficial for someone to develop an equivalent of this for mechanochemistry. There is also no “TLC-equivalent” for organic reaction monitoring. One may take a sample for TLC, but it will require interrupting the reactor. When temperature control is in place, this would be very challenging. The development of an effective alternative would be highly valuable for the mechanochemistry community. Another limitation is the lack of knowledge regarding how to drive a reaction by exploiting Le Chatelier’s principle whether by instrumentation or some form of chemical sequestration (e.g., incorporating a drying agent such as P2O5). However, there may be non-conventional methods for forcing equilibrium far to one side that could offer a solution to this problem (see “Driving Forces”).
It is important to keep in mind that the end goal of mechanochemistry is not, per se, to duplicate solution reactions for getting from Reactants A and B to Product C. The goal is simply to obtain Product C. That may mean using new starting materials, new reactions, and/or new methods for controlling product distribution.
The mechanochemical mindset
The last half decade provided key advancements for the world’s understanding of how molecules in a milling environment obtain the energy required for chemical reactivity. Prior to those advancements, various theories were put forward to account for the peculiarity surrounding mechanochemical reactions, which were often successful at temperatures significantly lower than what may commonly be required for the same conventional, solution reaction. Perhaps the most dominant theory was the so-called “Hot Spot” theory. This theory suggested that the impact of the milling media with the wall could create a localized, short-lived hot spot of high-intensity energy that could be converted into chemically usable energy. That is to say that there is a direct link between the energy available for reaction and the nature of the impact. As it turns out, however, recent research has indicated that energetic transformations in the mill may be more straightforward and more easily tuned than this Hot Spot Theory would dictate.
The current understanding in the community is readily understood by the Arrhenius equation. Several studies targeted low-activation barrier reactions changing either operating frequency or vial temperature. Higher-energy Diels Alder reactions were investigated by McKissic et al. while changing variables such as the number of balls and the milling material (ball and vial). Following this, an outstanding study by the Emmerling group demonstrated that activation barriers could be quantitatively determined by controlling the vial temperature. Andersen et al. expanded on all of this via a comprehensive study spanning a wide range of activation barriers while changing material type, vial temperature, and operating frequency (see Fig. 3 for temperature effect). The results suggested thinking of the “A” term as proportional to the mill’s operating frequency, and the energy term as dominated by the Boltzmann energy distribution associated with the bulk temperature of the vial. The key to contrasting this theory with the Hot Spot theory is to recognize the disconnect between the impact energy of the ball and the energy of the reaction. This disconnect is in agreement with a second publication by the Emmerling group that suggested the hot spot theory is insignificant for “soft matter milling syntheses.” The ball’s role is now essentially limited to mixing reaction components and assisting with molecular collisions. This is divorced from the energy available for the reaction. Furthermore, if one considers the earlier discussion on twin-screw extrusion, it is clear that the process of scale up is rather transparent as far as energetics are considered. Refocusing on vibratory mills, consider now the combination of very high effective concentrations (due to lack of solvent) paired with high mixing speeds on top of the fact that the base operating temperatures of even the coolest mills were not below 30 °C, except for brief (<1 hour) reactions, as discussed earlier. With these facts in mind, perhaps the long-lived perpetuity of the Hot Spot theory is not so surprising.
In the end, the conclusion that these papers collectively point to is the ability to speed up reactions (via frequency) while maximizing selectivity (via temperature). However, the evidence of this occurring with an enantioselective reaction remains to be seen in literature despite historical success with enantioselective reactions in mechanochemistry.
The impact of these energy studies cannot be overstated for the practical chemist or engineer. Andersen and Mack described the role of heating and frequency as coarse and fine adjustments for reaction rate, respectively. The practitioner should be enticed by the opportunities to optimize selectivity and/or complete reactions using shorter milling times. Furthermore, this approach can assist in transitioning both low temperature and high temperature reactions to mechanochemical conditions, with the assurance that there will be a reliable path to scale-up the mechanochemistry via TSE. Given the importance of “ice bath” and “refluxing,” the lack of convenient and precise temperature control on commercially available ball milling machines may be a bottle neck to progress in mechanochemistry.
It is by exploiting our understanding of chemical driving forces that chemists are able to achieve profound results. Aromatization is one example routinely cited in sophomore organic chemistry. Another ubiquitous example of a chemical driving force is the formation of the very stable PO bond. One of the earliest mechanochemical organic synthesis papers focused on Wittig reactions involving formation of this bond. However, the peculiar aspect of that article was not the PO bond formation. It was the substitution of K2CO3 in place of “stronger” bases that are conventionally used for the deprotonation step. This provided an early indication that mechanochemists should carefully scrutinize assumptions and preconceptions about reactivity that have historically arisen from conditions involving solvents.
At this point in the development of mechanochemical knowledge, it appears that hard–soft acid–base (HSAB) theory, originally developed by Pearson in 1967, is a convenient guide for understanding reactivity under solvent-free conditions. In this approach, atoms or molecules are classified as either hard (low polarizability) or soft (high polarizability), and a system’s lowest-energy state involves pairing hard with hard and soft with soft. For example, CsF is mismatched since Cs+ is soft and F− is hard. The application of this is evident in the mechanochemical enolate reaction between 2-methylcyclohexanone and bromobenzyl bromide. When LiOH is used as the base (hard–hard pair), there is 0% conversion to any product. On the other hand, if NaOH is used instead, there is 78% conversion to enolate products. Interestingly, LiOH with CsF additive results in 59% conversion to enolate products. The driving force behind this reactivity is the break-up of the mismatched CsF pair: LiOH is unreactive in the absence of CsF, but the opportunity for Li+ and F− to create an even lower energy pair than Li+ and −OH is chemically irresistible. This driving force results in the mismatched CsOH, which can then act as a base. In both cases the base is −OH, but the basicity is enhanced or inhibited by the anion–cation pairing. As it turns out, the enhancement is such that the unexpected mechanochemical deprotonation of alkynes by mistmatched hydroxide anions has been observed. However, if one considers the whole picture, it is not necessarily so surprising. In solution, the solvent is typically responsible for disrupting the ion pair, resulting in a bare (albeit solvated) ion, and thus this step is often overlooked. However, this energy-intensive step cannot be overlooked by the mechanochemist trying to understand basicity in solvent-free conditions.
The role of HSAB is readily observed with respect to enhancing/inhibiting nucleophilicity, as well. For example, as Table 1 shows, the reactivity of various halogens was strongly controlled by their counter-ion pairing in nucleophilic attacks on benzylic leaving groups. A second example of this is the reduction activity of NaBH4, which was shown to be effective for aldehydes and ketones, and upon including LiCl, esters were reduced as well without any modification to reaction conditions. Theoretically, the logic should apply to defining leaving groups too (i.e., what is the anion–cation pair that will result?). This may have profound impacts with respect to the need for protecting groups, which directly violate a key principle of green chemistry. Contemplating the effects of HSAB theory on the contours of reaction energy diagrams, such as Fig. 4, while interpreting/predicting mechanochemical reactions will be very important.
Table 1 Nucleophilicity enhancement/inhibition resulting from HSAB theory match/mismatch of ions. Reprinted from ref. 74, Copyright (2009), with permission from Elsevier
Entry Metal Nucleophile Conversion (%) Yield (%) 1 Na F — — 2 K F — — 3 Cs F 50 50 4 Na Cl — — 5 K Cl 14 14 6 Na I >99 94 7 K I 64 58 8 Na SCN >99 97 9 K SCN >99 96 10 Na N3 75 75 11 Na OAc — — 12 K OAc — — 13 Na CN — — 14 K CN — —
In general, much more work remains to be done towards completing the understanding of the driving forces observed so far, identifying new ones, and finally exploiting them for the development unique chemical opportunities. For further discussion on other reactions yielding surprising results, the reader is referred to Bolm’s recent review on altered product selectivities.
This reaction or that reaction
“Can you do a Wittig reaction?” Questions of this form are natural and common. It is important, however, to keep in mind the chemistry: Does a chemist want to do a Wittig reaction? Or is the real desire to click together two molecules while leaving behind alkene functionality? If the purpose is synthesis, then the latter is probably the real desire. There are countless examples of named reactions that work wonderfully under mechanochemical conditions. Indeed, five key publications spell this out. Three of these publications are in-depth reviews of mechanochemical organic reaction literature. The first was authored in 2013 for Chem Soc Rev by Wang. Two years later Hernández and Friščić wrote a review that focused on metal-catalyzed organic reactions. Most recently, Tan and Friščić expanded on those prior publications in 2018. The other two publications are books, published in 2014 and 2016, both devoted entirely to organic mechanochemical synthesis. Readers interested in an exhaustive foray into the reactions that have been observed are directed to look into these five publications. Despite the success in copying and mimicking conventional reactions, some important reactions have remained elusive and may need a “mechanoequivalent.”
Enolate reactions involving regioselectivity challenges may be an example of this. Despite notable successes, recent work towards taming the mechanochemistry of enolates has proved difficult. This is, in part, due to the high effective concentrations of starting materials and reagents and a current logistical limitation of being unable to slowly introduce reagents. However, revisiting enolate chemistry with a matured understanding of both mechanochemical energetics and driving forces should inevitably produce a controllable “classical” enolate reaction or a mechanochemical equivalent. Another set of challenging reactions would be those proceeding through an SN1 mechanism. With no solvent to stabilize the separation of charged intermediates, reaching any kind of product is unreasonable. Again, understanding and controlling both energy and driving forces will go a long way towards creating mechanoequivalent approaches.
In summary, there are numerous incentives to explore mechanochemistry. One may either incorporate it simply as another tool in their toolbox, or they may immerse themselves fully into the development of the field. For those seeking a reliable and predictable tool, the knowledge is now at hand in the mechanochemistry field to provide that for a vast array of reactions. Furthermore, benchtop results now have a solvent-free, continuous process of industrial significance for scaling up reactions. For those pursuing a pioneering mechanochemistry experience, there are also numerous places to plant a flag in the looming “Gold Rush” era of reactions unique to mechanochemical organic synthesis. The two motivations will form positive feedback loops with one another, furthering the commercialization and standardization of mechanochemical equipment.