Solvent recovery is a form of waste reduction. In–process solvent recovery still is widely used as an alternative to solvent replacement to reduce waste generation. It is attractive, like end–of–pipe pollution control, since it requires little change in existing processes. There is widespread commercial availability of solvent recovery equipment which is another attraction. Availability of equipment suitable for small operations, especially batch operations, make in–process recovery of solvents economically preferable to raw materials substitution.
Commercially available solvent recovery equipment include:
Carbon adsorption of solvent, removal of the solvent by steam, and separation of the solvent for reuse in the operation. Carbon must be regenerated, two or more units are required to keep the operations continuous. Chloric acid formation from chlorinated solvents, carbon bed plugging by particulates, and buildup of certain volatile organics on the carbon and corrosion can be a problem.
Distillation and condensation can be used to separate and recover solvent from other liquids. Removal efficiency can be very high using this process and can be used for solvent mixtures as well as single solvents.
Dissolving the solvent in another material such as scrubbing. Solvents must be then recovered from the resulting solution, through distillation but efficiency of removal is often not high using this method.
Adsorption processes are useful and versatile tools when it comes to waste solvent recovery unit as they can be applied with high efficiency at relatively low cost in cases in which the desired component presents either a fairly small or a fairly high proportion of the stream. The applicable adsorbents vary according to different purposes.108,109 Adsorbents with low polarity (activated carbon, etc.) tend to adsorb nonpolar compounds, whereas ones with high polarity (e.g., silica, alumina) have higher affinity to adsorb polar substances. However, some adsorbents operate via specific binding sites (e.g., molecular sieves, molecularly imprinted polymers) rather than simple hydrophilic-hydrophobic interactions. It is worth mentioning that adsorption cannot easily be installed in a continuous configuration and is usually either a one-bed batch process or a twin-bed process with one bed in the adsorption, whereas the other one in the regeneration phase.
In the regeneration stage of adsorption, high volumes of gas containing organic solvent are produced. Other processes in the chemical industry, such as paint drying or the drying of solid pharmaceutical intermediates or products, also generate a significant amount of solvent vapor.112 This raises another issue, as the recovery of this solvent is highly desired to minimize solvent loss and the environmental burden, as urged by the increasingly strict regulatory environment. For example, the recycling of chlorofluorocarbons has gained a lot of attention since the Montreal protocol.113–115 Incineration of solvent vapors is a widely used solution since it makes use of the solvent's latent heat. However, incineration likely needs supporting fuel to reach the required efficiency and needs continuous solvent vapor feed, not to mention that nonflammable halogenated solvent cannot be eliminated in this manner. Adsorptive systems have proved to be good alternatives. This field of adsorption is dominated by activated carbon adsorbents,116 but molecular sieve zeolites are also employed.117 Polymeric adsorbents are seldom employed in such processes, mainly because of their high price compared with activated carbon and zeolites.118 The choice of adsorbent regeneration technique has a significant effect on the quality of the recovered solvent. Examining the efficiency and applicability of various regeneration processes has been the aim of several studies.112,119 A typical system utilizing activated carbon adsorption to recover solvents from air emissions is shown in Fig. 3.15.11. Steam regeneration is employed to strip solvents from the activated carbon followed by condensation of the steam/solvent mixture through cooling. Eventually the solvent layer is separated by simple decantation.
The integrated production and recovery of ABE using glucose as a substrate and gas stripping as a means of solvent recovery distillation equipment has been reported by Groot et al. , Mollah and Stuckey , Park et al. , and Ezeji et al. [42–44]. Groot et al. produced butanol in a free cell (not immobilized) continuous reactor and removed the product in a separate stripper . As a result of simultaneous product recovery, glucose utilization was improved by threefold, but the selectivity of butanol removal was low at 4 as compared to 19, which is the selectivity at equilibrium, suggesting that the stripper was not efficient. Also solvent productivity in the integrated system was 0.18 g/L h, as compared to 0.17 g/L h in the nonintegrated batch system . Mollah and Stuckey used immobilized cells of C. acetobutylicum to improve productivity and recover butanol by gas stripping . The cells were immobilized in calcium alginate gel and used in a fluidized bed bioreactor. This integrated system achieved a productivity of 0.58 g/L h, which is considered low for an immobilized cell continuous reactor.
Pfizer has redesigned the synthesis of several of its pharmaceutical products to reduce generation of hazardous waste. Changes were made in the synthetic route to sildenafil citrate (see Fig. 9.7), the active ingredient in Viagra® (Dunn et al., 2004), which resulted in a more efficient process that required no extraction and recovery system for solvent steps (see Fig. 9.8). The E-factor (Sheldon, 1992) for the process is 6 kg waste/kg product, which is substantially lower than an E-factor of 25–100, which is typical of pharmaceutical processes. Furthermore, all chlorinated solvents had been eliminated from the commercial process. During the medicinal chemistry stage in 1990, the solvent usage was 1816 L/kg, and the optimized process used 139 L/kg solvent, which was reduced to 31 L/kg during commercial production in 1997 and to 10 L/kg with solvent recoveries. Pfizer plans to replace t-butanol/t-butoxide cyclization with an ethanol/ethoxide cyclization. Combined with other proposed improvements, this is expected to increase the overall yield from 76–80% and further reduce solvent usage and organic waste.
Secondly, the capital investment requirement required to construct facilities to practice each of the three process technologies can be compared. The acid chloride process is a low temperature, atmospheric pressure process and process fluid viscosities are low. Thus, standard design reaction equipment with low cost supporting utilities are used in the reaction area. However, polymer recovery would generally be accomplished by precipitation, washing and drying followed by extruder pelletization — operations which are capital intensive. Also, extensive used solvent recycler for sale is required in the acid chloride process, again leading to increased capital cost. Both the melt or solution diacetate and diphenate processes on the other hand are high temperature, high vacuum processes where process fluid viscosities reach very high values. For these processes, polymer reactors will require some special design features particularly with respect to agitation and heat transfer. Supporting utilities will be rather capital intensive. To balance these costs, however, product recovery is expected to be relatively simple, requiring only one or two melt processing operations most likely using a thin film polymer processor followed by an extruder. Solvent recovery requirements would be modest for the diacetate process but somewhat more costly for the diphenate process where large quantities of phenol (especially from monomer production) will require purification prior to recycle. Some difference in capital investment required for monomer production in the diacetate and diphenate processes is also expected. Diphenyl ester production is less attractive due to the more extreme reaction conditions required and the large phenol recycle streams. However, even with the noted differences, it is estimated that any of the three described processes could be built for approximately the same dollar amount per annual pound of polymer capacity at the 15 Mlb year−1 scale (1 kg = 2.2 lb).