D7, Z9 were gifted from Lepu Pharmaceuticals and used without further purification. All solvents and reagents were obtained from Energy Chemical and used without further purification.

Mechanochemical Wittig reaction in a customized stainless steel vial (3.0 mL volume) was carried out by adding 1 mmol D7, 1mmol Z9 and 1.3 mmol base. The mixture was ball-milled for 16 hrs. For liquid-assisted grinding experiments, 1 mL solvent was also added. The vial was shaked at 20 Hz in a Spex8000M Mixer/Mill. HPLC was performed to evaluate the composition of filtrate. Typically,

0.670 g D7, 0.258 g Z9 and 0.197 g 1,8-Diazabicyclo[5.4.0]undec-7-ene (DBU) were added at temperature around 25˚C, then ball-milled for 16 hrs. After the reaction was completed, 2 mL 95% ethanol was added to the solution in the reflux state. Then the temperature was lowered to 0˚C - 5˚C in order to precipitate the off-white crystalline solid, which was dried in an oven at temperature 55˚C to obtain about 0.4 g product.

In a 250 mL 4-necked round bottom flask, DMSO (150 mL), potassium carbonate (33.54 g), Z9 (50 g) and D7 (20.87 g) were added at temperature range of 25˚C to 35˚C under stirring with speed 250 rpm/min [10]. The reaction mixture was heated to 70˚C - 75˚C for 5 to 7 hours. After completion of the reaction as determined by TLC, the reaction mixture was cooled to a temperature range of 25˚C to 35˚C. Toluene (250 mL) was added for dilution of the reaction mixture and stirred for 30 minutes. The organic layer was added to water (100 mL) under stirring and maintained for 30 minutes. The organic layer was separated and washed with water (2 × 100 mL) in the same manner as described above. The organic layer was distilled in Rotavapor bath at temperature from 50˚C to 60˚C under vacuum. After distillation, isopropanol (100 mL) was added to the residue immediately, and the mixture was maintained for 30 minutes. Then the reaction mixture was brought to a temperature range of 25˚C to 30˚C by circulating room-temperature water slowly. In this period, the emerging precipitate was further cooled to 10˚C for 30 minutes, and then filtered by using filter paper. The filter cake was washed with prechilled (at temperature 10˚C) isopropanol (50 mL). The product was dried in an oven at 55˚C until the moisture content was no more than 1%. The dried product appeared as an off-white crystalline solid weighing about 45 g.

To focus the study, Z9, D7 and potassium carbonate were ball-milled in a stainless-steel vial with solvents at opposite ends of the dielectric spectrum, as well as a control without any solvent (Table 1).

In general, we noticed that more polar solvents (high dielectric constants) favor Z selectivity and a higher overall conversion, whereas the use of less polar solvents (lower dielectric constants) favors E selectivity and a lower overall conversion. It was very exciting that in the absence of solvent, the conversion could reach 99.1%, and the E/Z ratio could achieve 70:30. The reason may be that D7 gradually turned into liquid state during the grinding process. In addition, under the grinding conditions, the longer the potassium carbonate is ground, the finer the particles, new potassium carbonate would be exposed continuously to promote the reaction. Therefore, this reaction can be performed without solvents.

Witting reaction’s selectivity depends on a number of factors. In the presence of lithium and sodium salts, highly selective (E)-olefins were obtained [11]. The strength of bases is crucial to the reaction [12] [13]. To study the influence of bases in the ball mill, some commonly used bases were applied (Table 2).

It was surprising that the common inorganic bases, potassium hydroxide and sodium hydroxide, played a good role in the solvent-free process with high yield (entries 1 - 2). From the perspective of conversion, potassium carbonate was much better than sodium carbonate. The main reason is that the alkalinity of potassium carbonate is stronger than that of sodium carbonate (entries 3 - 4). However, the reaction exhibited worse performance when the stronger bases such as CH₃CH₂OK or CH₃CH₂ONa were used (entries 5 - 6). Furthermore, the organic base DBU initiated the reaction well under the conditions (entries 7, pKa for DBU = 24.13) [14]. Wittig reaction by using DBU as the base is a good choice, so DBU was chosen for the following optimization. The acquired data confirmed the solid-state character of the events leading to the mechanochemical generation of phosphorus ylides [15] [16] [17], which include: (i) breaking the crystal lattice of a phosphonium compound and the formation of an amorphous phase; (ii) deprotonation of an amorphous phosphonium salt by DBU in a homogeneous solid-state reaction.

E-factor is an important index to evaluate the environmental impact of a reaction [18]. It enables us to estimate the amount of waste that could be generated per kilogram of product produced. The higher the E-factor value, the more waste will be generated.

E-factor = mass ( waste ) mass ( product ) (1)

where mass (waste) and mass (product) represent the amounts of waste and product generated in the same weight unit. Unlike the atom economy, the waste accounts for all nonrecyclable postreaction components including unreacted starting materials, byproducts and solvents. For example, the oil industrial processes have a lower E-factor because they are highly optimized for their tasks, and various products generated from the refinery are utilized very efficiently, which minimizes deposals. Higher value products such as pharmaceuticals always have a higher E-factor due to the multistep synthesis procedures. The essence of green chemistry is to reduce the use and generation of hazardous substances through careful design; E-factor serves as a parameter for the chemists to achieve waste reduction that is based on the outcome of reaction.

The E-factors for both procedures were calculated using Equation (1) and shown in Table 3. Although solvents could be reduced to improve the E-factor value for the general procedure, however, the solvent cannot be recycled in many cases. On the contrary, the mechanochemical route achieved the lower E-factor because of the higher production volume of the product at the rarely using a highly toxic solvent or reagent. It is important to realize that the E-factor is often strongly dictated by the amount of solvent used relative to the production. This is consistent with the practice that solvents account for the highest amount of waste produced in the chemical industry.