Introduction

The industry of renewable energy is continuously growing as a solution to the climate crisis and fossil fuel depletion. The biodiesel industry produces a surplus of crude glycerol, the main reaction by-product, which represents about 10 wt% of the production. This crude glycerol contains several impurities and null economic value. Catalytic oxidation is one of the several glycerol (GLY) valorization routes known, yielding several high added value products. One of the products is dihydroxyacetone (DHA), with major applications in the cosmetics industry, namely in sunless skin tanning lotions¹.

DHA was obtained by GLY oxidation using commercial catalysts of Pt doped with Bi nanoparticles supported in activated carbon (AC), Pt/AC, and Pt-Bi/AC, with a DHA yield of about 35%. DHA must be separated from the unreacted GLY and reaction byproducts, mostly organic acids. Distillation processes may not be possible as these compounds have high boiling points and are heat sensitive.

A new DHA purification process by continuous chromatographic is herein presented, using two Simulated Moving Bed (SMB) units in cascade packed with poly styrene-divinylbenzene (PS-DVB) ion-exchange resins functionalized with sulfonic groups as stationary phase.

Methods

GLY oxidation in liquid media was performed on a batch reactor operating under base-free conditions, using water as solvent and oxygen as the oxidant agent. Three commercial catalysts were used: Pt5w%/AC from Sigma-Aldrich, Pt5w%-Bi1.5w%/AC from Johnson Matthey, and Pt5w%-Bi5w%/AC from Evonik. The reaction samples were quantified by an HPLC analytical method, using a Hichrom Alltech Organic Acid OA-1000 analytical column on a KNAUER HPLC system.

To design the separation process, the adsorption equilibrium data of the species present in the mixture must be obtained, namely, GLY and the main reaction products oxalic acid (OXA), tartronic acid (TTA), glyceric acid (GCA), glycolic acid (GCO), and DHA. Single-component breakthrough experiments were performed on a stainless steel fixed-bed column (100 x 20 mm), packed with a Dowex® 50WX-2 resin (2% of crosslinking) in H⁺ form and maintained at 293 K using a circulation thermostatic water bath. DHA and GLY adsorption equilibrium data on the Dowex® 50WX-2 resin in Ca²⁺ form was also obtained by single-component breakthrough experiments under similar conditions. The column’s outlet concentration was measured using UV and RI detectors.

The mass of each compound adsorbed in the resin, qi, during the adsorption step was obtained by measuring the area above the adsorption breakthrough curves. Similarly, during the desorption step, qi, was obtained by measuring the area below the desorption breakthrough curves. The adsorption isotherms were determined by fitting the data to well know adsorption equilibrium isotherms models. Multi-component breakthrough experiments were performed on the same semi-preparative fixed-bed column to check for competitive adsorption under the concentrations of interest.

A mathematical model was considered to describe the fixed-bed breakthrough experiments². The model was validated using the single and multi-component breakthrough experiments. Then, the model was extended to the SMB multicolumn process, described elsewhere³. Both models were implemented in the gPROMS model builder software V5.1.1 (PSE, UK).

The SMB cascade was designed based on the separation regions obtained by setting a minimum DHA recovery and purity for the stream of interest, considering the methodology described elsewhere ³. The two SMB unit dimensions and configuration were considered similar, each one with six (100 x 20 mm) columns with the configuration 1-2-2-1, based on the standard operation mode of the FLEX-SMB pilot unit⁴,

Results and Discussion

The catalysts doped with Bi showed higher DHA yields. The highest DHA yield of 36% was achieved with the Pt5%-1.5%/AC catalyst (Johnson Matthey) after two hours of reaction. A mixture with the following composition was considered to study the separation process: 22.0 g L^-1 of GLY, 29.8 g L^-1 of DHA, 12.9 g L^-1 of TTA, 8.2 g L^-1 of GCA, 6.9 g L^-1 of GCO, and 1.2 g L^-1 of OXA.

Adsorption equilibrium data was determined over the resin in H⁺ form. In the range of concentrations studied, the adsorption equilibrium for all species was well described by the linear equilibrium adsorption isotherm, q_i* = K_iC_i, except OXA which was better described by a Freundlich equilibrium adsorption isotherm, q_i* = K_iC_i^1/n_i. The distribution coefficient (K_i) increased in the following order: K_OXA (0.122, n_OXA = 0.7) < K_TTA < (0.554) < K_GCA (0.765) < K_GCO (0.792) < K_GLY (0.809) < K_DHA (0.840).

Although a pseudo-binary separation is possible, as DHA is the most retained compound, the selectivity for DHA separation from GLY is very low, 1.04, considered a hard separation. If GCO is the light key, the selectivity is slightly increased to 1.06, but a second unit is still necessary to separate GLY from DHA. GLY and DHA single component breakthrough experiments were performed on a similar method on a Dowex® 50WX-2 resin in Ca²⁺ form. The experimental data was well adjusted by the linear adsorption isotherm model, for a K_GLY of 0.962 and K_DHA of 1.131. The separation selectivity is 1.18, thus it is an easier separation.

A two-unit SMB cascade was considered. DHA and GLY were separated from the remaining species on the first SMB packed with the resin in H⁺ form, being collected in the extract. This stream was fed to the second SMB packed with the resin in Ca²⁺ form, with GLY collected on the raffinate stream and DHA in the extract stream. A minimum DHA recovery of 90% was defined on both units. For the first SMB, a minimum extract purity in GLY free-basis of 99% was defined, to guarantee that the organic acids are not collected on the extract stream. A minimum extract purity of 97% was defined for the second SMB, based on the requirements of the cosmetics industry.

The operating conditions of the two units were defined by determining the so-called separation regions, γ_II,γ_III, and regeneration regions, γ_I,γ_IV, which were computed using the gPROMS model, where γ_j is the ratio between the liquid and the solid flow rate in each section j of the SMB. A safety factor of 20% was considered for the regeneration sections of both units, to ensure real process feasibility. Another parameter that must be considered is the switching time t*, which defines the time to switch the inlet and outlet ports. The switching time was optimized to maximize DHA productivity and minimize desorbent consumption.

For the first SMB, a t* of 2.45 min was considered. A small separation region was obtained due to the low selectivity. The operation point, γ_I = 1.807, γ_II = 1.438, γ_III = 1.499, γ_IV = 0.523, was obtained by applying a low safety factor of 1.01 to the region vertex of the first SMB. The performance parameters were fulfilled, with an eluent consumption of 567 L_Des kg_DHA^-1. The extract stream has a DHA concentration of 6.3 g L^-1 and a GLY concentration of 1.6 g L^-1, which is then fed to the second SMB unit.

By defining a t* of 2.12 min, the second SMB operation point, γ_I = 3.167, γ_II = 2.256, γ_III = 2.607, γ_IV = 1.871, was obtained by applying a safety factor of 1.03 to the region vertex of the second SMB. The performance parameters were fulfilled, with an eluent consumption of 651 L_Des kg_DHA^-1. The overall process productivity was 64 kg_DHA (L_Ads·day)^-1, with an overall eluent consumption of 1218 L_Des kg_DHA^-1.

The separation regions obtained with the model are closed to the obtained using the equilibrium theory³ on both units, thus mass transfer limitations may be neglected.

Conclusions

A green DHA process from bio glycerol catalytic oxidation to DHA purification by continuous chromatographic processes was found to be feasible. DHA was obtained by GLY catalytic oxidation with commercial Pt-Bi/AC catalysts, and its purification was successfully studied for the first time, considering a two-units SMB cascade based on the pilot-scale FLEX-SMB unit, with a DHA productivity of 64 kg_DHA (L_Ads·day)^-1 and an eluent consumption of 1218 L_Des kg_DHA^-1.

References

1. Galy, N.; Nguyen, R.; Yalgin, H.; Thiebault, N.; Luart, D.; Len, C., Glycerol in subcritical and supercritical solvents. Journal of Chemical Technology and Biotechnology 2017, 92 (1), 14-26.
2. Coelho, L. C. D.; Filho, N. M. L.; Faria, R. P. V.; Ribeiro, A. M.; Rodrigues, A. E., Selection of a stationary phase for the chromatographic separation of organic acids obtained from bioglycerol oxidation. Adsorption 2017, 23 (5), 627-638.
3. Rodrigues, A., Simulated moving bed technology: principles, design and process applications. Butterworth-Heinemann: 2015.
4. Sá Gomes, P.; Rodrigues, A. E., Simulated Moving Bed Chromatography: From Concept to Proof‐of‐Concept. Chemical Engineering & Technology 2012, 35 (1), 17-34.