From Inert to Catalytically Active Milling Media: Galvanostatic Coating for Direct Mechanocatalysis

Abstract

The inert milling balls, commonly utilized in mechanochemical reactions, were coated with a layer of Pd and utilized as catalyst in the direct mechanocatalytic Suzuki reaction. With high yields (>80 %), the milling balls can be recycled multiple times in the absence of any solvents, ligands, catalyst‐molecules and ‐powders, while utilizing as little as 0.8 mg of Pd per coated milling ball. The coating sequence, the support material, and the layer thickness were examined towards archiving high catalyst retention, low abrasion and high conversion. The approach was transferred to the coating of milling vessels revealing the interplay between catalytically available surface area and the mechanical energy impact in direct mechanocatalysis.

The galvanostatic coating of milling balls and vessels advances commonly inert materials into catalytically active milling media. The coating sequence, the support material, and the layer thickness are crucial for achieving high activities in the Suzuki cross‐coupling reaction for multiple cycles. The interplay between the mechanical impact of the ball and the catalytically available surface is elucidated.

Introduction

Mechanochemistry describes solid‐state reactions initiated by mechanical energy. ^[1] Commonly, the chemical compounds are brought into reaction by impact or shear forces induced by the collision of milling balls within vibrating or rotating milling vessels. Mechanochemistry proved suitable for the synthesis of small organic molecules,[ ¹ , ² ] polymerization reactions, ^[3] supramolecular synthesis, ^[4] functional materials, ^[5] nanoparticles, ^[6] organometallic compounds, ^[7] pharmaceuticals, ^[8] and biomolecules. ^[9] Besides transferring energy and mixing reagents, the interference of the milling material itself with the reaction mixture is generally considered undesirable. For that reason, chemically inert milling materials such as tungsten carbide (WC), steel, zirconia (ZrO₂), or polymers are applied, and the milling parameters are controlled precisely to suppress their abrasion. A novel concept in which the interaction of the milling ball with the reagents is indeed desired is direct mechanocatalysis. Here, the milling tools are the catalyst, which offers great advantages with respect to catalyst separation and reusability.[ ¹ , ¹⁰ ] Milling balls made of Pd, Ni, or Cu have been used for reactions such as cross‐coupling,[ ¹¹ , ¹² ] cyclotrimerization, ^[13] and hydration. ^[14] A disadvantage of using milling balls fully made of the catalytic metal is their high material cost, particularly for precious metals such as Pd. An even larger drawback is the inability to adjust the catalytic activity of the milling ball independently from their bulk material properties such as density and thus to control important milling parameters such as kinetic energy. Up to now, there is no rational strategy reported on how to manipulate the commonly inactive surfaces of milling balls and milling vessels to make them catalytically active while maintaining their other mechanical properties. On the other hand, the mechanical impact and shearing forces occurring between milling balls and vessel render coating techniques challenging.

In this work, we present a galvanization concept for milling equipment that can withstand abrasion during mechanochemical reactions. With this the catalytically active surface can be tailored towards any metal‐catalyzed reaction while offering the possibility to control the energetic input and significantly increasing the cost‐efficiency (Supporting Information chapter 2.4) and sustainability of direct mechanocatalysis. We demonstrate this for the Suzuki‐, ^[12] and the Glaser coupling, ^[12] identify optimal coating sequences, support materials and layer thicknesses, transfer the concept to the coating of milling vessels and elucidate the interplay between catalytic surface area vs. mechanical impact.

Results and Discussion

First, we focused on the coating of milling balls with Pd. As a benchmark reaction we chose the Suzuki–Miyaura reaction between phenylboronic acid (1 mmol) and iodobenzene (1 mmol) utilizing 1 g K₂CO₃ as base. The reaction was conducted for 60 minutes at 35 Hz with one coated 10 mm milling ball inside a PFA vessel. An extensive substrate screening for this reaction utilizing solid Pd balls has been conducted in our prior work. ^[12] Unfortunately, the direct deposition of Pd on steel support materials appeared unsuitable as it led to full abrasion of the Pd layer during milling (Table 1, Entry 1). In accordance with the literature ^[15] different coating sequences (Table 1, Supporting Information Figure S1) on steel 1.3505 milling balls were investigated and evaluated after 10 cycles of the above‐mentioned Suzuki model reaction. The layer thickness of all used balls was set to 400 nm for better comparability. Those coating sequences based on palladium on a gold layer exhibited advanced durability ^[16] portrayed by an average yield of >55 % after ten reactions. Any other sequence with Pd directly on Ni, Cu or steel led to high abrasion and consequently low yields of less than 20 % after 10 reactions. Similarly, a Cu and Ni layer under the Au layer showed beneficial effects. Therefore, we chose our standard coating sequence to obtain long term stability: Firstly, copper (adhesion mediator), secondly, nickel (diffusion barrier), thirdly gold (support layer) and finally palladium (catalyst) (Table 1, Entry 5). Please note that reference experiments proved that none of the coated interlayers itself contributes to the Suzuki coupling (Supporting Information Table S1), nor is any possible abraded material catalytically active. ICP‐OES measurements showed a loss of 1 μg Pd every hour, which is comparable to a pure Pd‐ball. ^[12] To confirm the latter, we milled Pd balls without substrate (to create abrasion) and then changed the milling balls to inert zirconia balls and added the substrate. Milling for further 60 min does not yield any product confirming that the Pd layer and not any potential abrasion is the spot of catalysis.

Table 1.

Yields and abrasion observed with different coating sequences and supporting materials, averaged over 10 reactions.

Entry	Cu	Ni	Au	Pd	Abrasion [%]	Ø Yield [%]
1	–	–	–	+	99	13
2	–	–	+	+	71	57
3	+	–	–	+	86	19
4	+	+	–	+	99	15
5	+	+	+	+	51	82
Entry	Material			No. Reactions <20 % Yield		Ø Yield [%]
6	Steel 1.3505			10		82
7	Steel 1.4125			2		22
8	Steel 1.4401			5		34
9	K‐Monel			10		77
10	Bronze			8		49

The first reaction cycles show quantitative yield, thus coated balls are comparable to the catalytic activity of a solid Pd ball (Figure 1), while massively reducing the necessary Pd to 0.8 mg (compared to a 4 g full Pd ball and 22.7 mg (10 mol %) for Pd(OAc)₂ or 10.8 mg (10 mol %) for Pd black). This corresponds to an average turn‐over frequency (TOF) of above 640 000 h⁻¹ which is larger than any classical mechanochemically catalyzed reaction (Supporting Information Table S2, Figures S2 and 3). Moreover, PXRD synchrotron in situ measurements (Supporting Information Figure S4) reveal the reaction kinetics using Pd‐coated milling balls to be similar to the usage of solid Pd milling balls. ^[12] The reaction was investigated in different vessel types such as PFA, PTFE, steel or ZrO₂ (Figure S5) showing almost quantitative yield in all of them.

Figure 1.

A) Context and comparison of Pd catalytic Suzuki reactions in mechanochemistry. For details see Supporting Information Section 2. B) Ideal coating sequence for electroplate milling tools. C: Applying the concept of coating milling balls to milling vessels. One inert milling ball in 19 mL coated steel vessel with 1 mmol phenylboronic acid, iodobenzene (1 eq.) and 1 g K₂CO₃ with LAG (EtOH): η=0.15; 35 Hz, 1 h mixer mill (MM500). D: Applying the concept of coated milling tools for direct mechanocatalysis to other catalytic reactions. Glaser: One copper coated ball in PFA vessel (19 mL), 0.28 mmol phenylacetylene, 1.7 g K₂CO₃; 30 Hz, 90 min, mixer mill (MM400).

After establishing the coating sequence, we investigated the influence of different support materials on the durability of the coating. For this comparison we chose five different steels, four alloys, three ceramics and three polymers (Supporting Information Table S3). Independently of which support metal material was used, the first catalytic cycle showed conversion of >95 % in the above‐mentioned reaction protocol (Supporting Information Table S4).

Subsequently, the different support materials were tested towards their long‐term catalytic activity and abrasion resistance (Table 1, entry 6 to 10, Supporting Information Table S4), both key parameters in terms of catalyst performance. Therefore, the above‐mentioned catalysis was run several times until the catalytic activity dropped below a product yield of 20 %. In this study, none of the polymer or ceramic supports showed activity beyond the first cycle, which is due to the abrasion of the conductive metal adhesive. At the same time, steel and other alloys resist the mechanical stress of milling significantly better, with steel 1.3505 and K‐Monel (Table 1, entry 6 and 9) even maintaining a catalytic activity of above 20 % after 10 cycles. It became also apparent that the decline in catalytic activity depends strongly on the material composition. Due to their stable oxide layers, the materials with a high amount of chromium (Supporting Information Tables S3 and 4) showed a decreased longevity of the electroplated layers compared to materials with a low chromium amount. In‐depth investigations of this behavior showed that the mechanical weak point is the interface between the steel and the first layer ‐ copper. For the first reactions, only slight scratches are visible in the palladium layer (Figure 2). However, after multiple experiments the steel surface is visibly growing on the milling ball (Figure 2), indicating a breakage between the copper and the steel as the main mechanism of milling ball degeneration. Among the best performing support materials, steel 1.3505 was set as standard material in all following reactions as it is comparably cheap and readily available.

Figure 2.

SEM pictures of the surface's development of the cycled ball: Before usage, with 100 % Palladium on the surface. After three reactions, with 77.9 % Palladium, 15.7 % Steel and 6.4 % Gold, Nickel, and Copper on the surface. After six reactions, with 61.3 % Palladium, 36.9 % Steel and 1.8 % Gold, Nickel, and Copper on the surface. After ten reactions, with 27.5 % Palladium, 69.1 % Steel and 3.4 % Gold, Nickel, and Copper on the surface. Green: Palladium, Yellow: Gold, light blue: Nickel, Orange Copper, dark blue: Iron.

As the layer thickness also plays a vital role in electroplating and is a parameter that can be easily controlled, we investigated the influence of the Pd layer thickness and modified the coating protocol to prepare six different Pd layers with different thicknesses between 150 and 1200 nm (Supporting Information Table S5). Regardless of the Pd film thickness, the conversion after one catalytic cycle is >95 % (Supporting Information Figure S6), which, however, decreases with each cycle. Particularly thick (>1000 nm) and thin (<250 nm) layers seem most affected by abrasion. For the thicker Pd layers, it is possible that hydrogen is absorbed into the Pd layer during the acidic electroplating process, which causes thick Pd layers to fracture faster than thinner layers. ^[16] While thin palladium layers also chip more easily due to their low thickness. These theories are supported by SEM images (Supporting Information Figures S7 and 8), showing two types of abrasion. On the one hand, small notches are created through which the layers are thinned and subsequently abraded, and smaller craters that reach the steel surface of the sphere. The latter was dominant for very thick and thin balls. Thus, the abrasion is not evenly distributed over the ball, but rather punctual (SEM) (Figure 2). Interestingly, medium layer thicknesses of about 250–500 nm show the best results, as they exhibit longer activity retention with medium loss of catalyst layer. Depending on the reaction, they achieved a yield of up to 80 % up to the 8th reaction and showed a yield of over 20 % for the 10th reaction in each case. These results are further corroborated by SEM images where the percentage of palladium on the surface was determined after ten reactions. While the milling ball with an initial film thickness of 445 nm had an area of about 61 % palladium, the 1200 nm and 161 nm balls had palladium areas of about 41 % and 53 %, respectively (Supporting Information Figure S8). We thus conclude, that the long‐term catalytic activity does depends on the integrity of the layer. This also shows that the Pd‐film thickness affects the overall durability. The root cause of the abrasion could also be narrowed down to the utilized base. The use of softer bases such as sodium acetate instead of K₂CO₃ indeed minimize the damage to the ball surface, but did not lead to any product formation.

In order to increase the lifetime of the catalyst, recoating the Pd layer was investigated, and it was possible to counteract the yield decrease. If the upper Pd layer is recoated every cycle, the catalytic activity can be maintained constantly above 94 % (Supporting Information Table S6). Recoating every fifth cycle retains the catalytic activity above 65 % for 25 cycles. SEM images (Supporting Information Figure S7) show the reason behind catalytic activity diminution; once a destruction penetrates below the Au layer, even recoating cannot repair this fracture and irreversible layer abrasion takes place. Another approach to counteract the decrease in yield is to increase the reaction time incrementally by 30 min every five reactions. It turned out that the yield could be kept above 60 % in up to 15 reactions. In the final reactions, the yield dropped rapidly, as the ball surface was exposed longer and the coatings abraded completely (Supporting Information Figures S7 and 8). The examination of the ball surface at the end of the experiment showed a Pd surface of less than 5 %.

While solid Pd milling vessels were unthinkable for monetary reasons, the coating approach now allows to coat milling ball, milling vessel or even both. Applying our standard coating protocol to a steel milling vessel and conducting the model catalysis using steel balls provides the same high conversions of >95 % as compared to the usage of coated milling balls. Expanding our concept even further we decided to investigate the influence of the area of coating, while at the same time keeping the introduced energy constant. Since reactions times of 30 minutes gave quantitative yields in all systems we decided to shorten the reaction time to 10 minutes in order to determine the influence of the catalyst surface (Figure 3). The obtained results are of great importance for the understanding of direct mechanocatalysis as a whole, since they show that even a 14‐fold increase in theoretically available catalytic surface area, when coating a vessel compared to only a milling ball, has only a minor impact on the yield (Figure 3, Table 2). The only important factor for this type of reaction is the presence of the catalyst at the point of impact and this is a given for both the ball and the vessel.

Figure 3.

Yields with catalyst surfaces of different sizes. Coated ball in steel vessel (19 ml); Steel ball in coated steel vessel (19 mL) coated steel ball in coated steel vessel (19 mL); 1 mmol phenylboronic acid, iodobenzene (1 eq.), 1 g K₂CO₃ Lag (EtOH): η=0.15; 35 Hz, 10 min mixer mill (MM500).

Table 2.

Yields obtained from the coated steel vessel, with different milling balls and their theoretical kinetic energy calculated with ${\mathbf{E}}_{\mathbf{k}\mathbf{i}\mathbf{n}}=\frac{1}{2}\mathbf{m}·{\mathbf{v}}^2$ . One ball in coated steel vessel (19 mL), 0.1 mmol phenylboronic acid, iodobenzene (1 eq.), 1 g K₂CO₃ Lag (EtOH): η=0.15; 35 Hz, 1 h mixer mill (MM500). Steel ball highlighted as future reference method.

Entry	Ball material	Kinetic energy [mJ]	Density [g cm−3]	Yield [%]
1	WC	15.2	15.6	69
2	Steel [a]	7.9	7.8	99
3	Steel [b]	7.9	7.8	97
4	ZrO2	5.9	5.6	77
5	Si3N4	3.4	3.2	76
6	PTFE [c]	3.0	2.2	70
7	PP	0.8	0.9	14
8	No Ball	–	–	28
9	No Ball (5 h)	–	–	43
10	No Ball (10 h)	–	–	74
11	No Ball (30 h)	–	–	78
12	No Ball (48 h)	–	–	98

[a] Coated steel ball. [b] Non‐coated steel ball. [c] PTFE‐ball with an iron core.

The importance of this impact event also becomes apparent when the ball material is varied within a coated milling vessel. The yield correlates with the milling ball density and thus the gained kinetic energy during milling. This indicates that mechanical energy is indeed crucial for direct mechanocatalysis and the role of mechanochemistry goes beyond pure mixing. Two experiments seem to fall out of this series; firstly, WC balls with the highest density show a slightly decreased yield as compared to the lighter steel (Table 2, Entry 1). This is due to the extreme energy impact provided during milling that has led to the complete destruction of the Pd layer in the coated vessel (Supporting Information Figure S9) and thus not enough palladium being present at the point of impact to catalyze the reaction. Secondly, even shaking without any ball provides a considerable yield of 28 % after 1 h (Figure S10). This observation is of outmost relevance for three reasons: It shows that i) the acceleration of the substrate particles and their self‐diffusion during shaking is already sufficient for a decent conversion at the catalytically active vessel surface ii) since milling with light PP balls show a lower yield than milling without any balls, suggests that a hard enough material is needed to get a crushing effect, but not too hard so as not to destroy the layers. In this approach it seems like the PP‐ball hindered the mixing effect, most likely by compacting the powder and thus hindering diffusion. iii) By increasing the milling time to 30 h even full conversion can be reached without any detectable abrasion. This highlights that the catalyst surface plays a subordinated role opposite to a typical heterogeneous catalysis perspective. The movement of the substrates provides a continuously refreshed surface and hence, allows for smaller active catalyst surfaces.

In terms of cyclability (Figure 4) the coated vessel performed better than a coated steel ball, since the vessel has a higher amount of catalytically active surface and thus, small amounts of surface destruction have little effect on the overall catalytic activity of the vessel.

Figure 4.

Catalyst retention over 10 cycles for a coated ball in steel vessel (19 ml) and Steel ball in coated steel vessel (19 ml).1 mmol phenylboronic acid, iodobenzene (1 eq.), 1 g K₂CO₃ Lag (EtOH): η=0.15; 35 Hz, 10 min mixer mill (MM500).

To fully realize the potential of the coating approach we did a screening with several halogenated aromatic substrates (Figure 5). This demonstrates that the coated ball is on par with a solid palladium ball and can tolerate a wide array of functional groups. In addition, the introduction of various ligands typical for C−C‐coupling reactions (Supporting Information Chapter 2.9) showed no beneficial effects. To highlight the versatility of the presented method, another reaction requiring a different metal coating was also investigated. Our choice fell on the copper‐catalyzed Glaser coupling with electroplated steel milling balls. While the softer copper suffering under slightly increased abrasion of the catalytic layer (Figure 1D, Supporting Information Table S7), the general versatility of our approach could be demonstrated by achieving similar yields (42 %) to the direct mechanochemical approach utilizing bronze milling balls (47 %).

Figure 5.

Suzuki coupling using coated steel balls with different substrates. A) Standard reaction with iodobenzene and bromobenzene. B) Different synthesized products with their yield. The yields of a solid palladium ball are given in grey brackets. ^[12] One coated steel ball in PFA vessel (19 mL), 0.1 mmol phenylboronic acid, halogenated aromatic substrate (1 eq.), 1 g K₂CO₃ Lag (EtOH): η=0.15; 35 Hz, 1 h mixer mill (MM500).

Conclusion

In summary, we demonstrated that conventional steel milling equipment can be coated with a thin Pd layer to achieve high catalytic activity for several cycles in the direct mechanocatalytic Suzuki cross‐coupling reaction, comparable or even superior to the use of Pd(OAc)₂, Pd black, or even solid Pd balls. With this, the amount of required Pd was reduced to 0.8 mg per ball leading to an outstanding TOF of more than 600 000 h⁻¹ and thus significantly reduced costs as compared to any of the above‐mentioned alternatives. We identified the coating sequence steel‐copper‐nickel‐gold‐palladium, a low chromium‐containing support steel, as well as a Pd layer thickness between 250–600 nm to be beneficial to achieve minimum abrasion and high cyclability. We transferred the coating sequence to the milling vessels achieving even higher catalyst retention. With this we were able to uncouple the catalytic activity from the mass of the milling balls and thus the introduced kinetic energy. In doing so, we identified only the catalyst at the spot of the mechanochemical impact as important, rather than the total available Pd surface. We finally transferred this novel coating approach to the Cu‐catalyzed Glaser coupling showcasing its versatility.

Conflict of interest

The authors declare no conflict of interest.

Supporting information

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re‐organized for online delivery, but are not copy‐edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

Supporting Information

M. Wohlgemuth, M. Mayer, M. Rappen, F. Schmidt, R. Saure, S. Grätz, L. Borchardt, Angew. Chem. Int. Ed. 2022, 61, e202212694; Angew. Chem. 2022, 134, e202212694.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.