Parameter assessment for scale-up of co- and counter-current photochemical reactors using non-collimated LEDs

Highlights

• Design criteria for co-axially illuminated photo chemical reactors were determined.

• A ray tracing model was used to analyse light absorption and reactor performance.

• LED viewing angle significantly influences absorption homogeneity.

• Medium absorbance and viewing angle significantly influences reactor efficiency.

Abstract

Common cross-currently illuminated photochemical reactors lack energy efficiency to make them competitive in industry. The use of co- and counter-current illumination was previously proven to increase reactor performance, but these approaches made use of a collimated LED module whereby the used LED module itself having a lower than typical optical efficiency. In this paper, we study the use of non-collimated LEDs for the use in co- and counter-currently illuminated reactors. A ray tracing model was implemented in COMSOL and was validated using experimental data. Via these experimental data, the regime of no kinetic limitations was observed as conversion is not hampered by increasing light flux. Via the model results, it was determined that the most suitable light source for optimal light absorption by the reagent was the most collimated LED possible, in this case with a total viewing angle of 10°. The optimal reactor set-up uses the most reflective material, preferably aluminium or silver, to recuperate diverging light rays of the used wavelength. Furthermore, the wall thickness of the glass reactor must not be excessively thick, with an optimum at 1.5 mm wall thickness for this case. Regarding reagents and absorbance, it is best to use a higher concentration and reduce reactor length as this increases reactor performance under the condition that quantum yield is stable. Via the use of non-collimated light, it was determined that the entrance efficiency can be increased compared to a fully collimated light source, at the cost of reflection losses that increase with path length.

Introduction

The scale-up of photochemical reactors has been studied extensively in the past years, more specifically in the field of cross-current illumination. With this illumination technique, the incident photons travel perpendicular to the direction of flow, and usually result in short absorption path lengths. Various applications of this commonly used illumination technique as lab setups are shown by Cambié et al. (2016); Knowles et al. (2012), Hook et al. (2005), Plutschack et al. (2017), Gilmore and Seeberger (2014). These reactors all contain a reactor channel illuminated sideways via e.g. a tubular reactor coiled around a mercury lamp. Even in the industrial field, this cross-current illumination technique is commonly applied as an alternative to batch reactors as shown by Oelgemöler and Shvydkiv (2011) and Noël (2017) with similar reactor setups to lab scale. This is contrary to co- and counter-current illumination, which emphasises the use of longer path length by illuminating the reactor in the direction of flow. Scale-up studies using this novel illumination technique are absent and usually non-transferrable from cross-current illumination as different challenges are posed.

Recently, strategies were developed capable of predicting operating conditions for a scaled up cross-current flow reactor. These scale-up strategies are diverse and include experimental variation of operating conditions until decent conversion is obtained (Sambiagio and Noël, 2019; Su et al., 2016; Zhao et al., 2018) as well as the use of several (non-dimensional) parameters for scale up. Some of these parameters do not include absorption of the reagents or photocatalysts, the efficiency of the light source or the illuminated surface area (Elliott et al., 2018) and would therefore be only applicable in similar reactor setups. Other parameters which are more robust take the absorbed light flux into account, which has shown to be much more robust when using a batch reactor and various continuous reactors of different scales (Corcoran et al., 2020). This latter parameter, based on the absorbed light flux, uses a similar approach as the work by Meir et al. (2020a) in which the ‘quantum photon balance’ as a reactor parameter was introduced. In this work, an approach similar to Corcoran et al. (2020) was used to determine the relative absorbed light flux as a reactor parameter for the design of an efficient photochemical reactor.

Meir et al. have shown that the use of co- and counter-current illumination shows significant improvements compared to the state-of-the-art cross-current photoreactor illumination (Meir et al., 2020a,b). Conversion, space-time yield (STY) and photochemical space-time yield (PSTY) increased for equal incident light flux entering the same reactor geometry. The results for co- and counter-current illumination were obtained using a high intensity collimated LED as light source. However, scaling-up such a reactor in an industrial context proves to be difficult as these collimation optics are expensive and the LED efficiency decreases with the presence of these collimation optics. This efficiency decrease is shown in the lamp used by Meir et al., which only had a measured efficiency of 8% in the presence of these collimation optics, whereas a typical LED in this wavelength range (365 nm) has an efficiency of 30–35%. Hence, by omitting the collimation optics, a higher efficiency is achievable. As the derived model explicitly takes collimated light into account, a difference in performance is expected when switching to standard UV-LEDs. However, the use of standard UV-LEDs would significantly improve scalability of the co- and counter-current illumination scheme for homogeneously non-catalysed photochemical reactions.

In this work, we investigated the possibility of using non-collimated light sources to drive homogeneous photochemical reactions to enable easier adoption in industry by decreasing initial cost of the reactor and potentially also operation cost. It is expected that a discrepancy shows up between the derived model by Meir et al. (2020b) assuming ideal conditions regarding light influx and flow regime. In this work, non-collimated light sources are used to irradiate the reactor, which are not considered by the ideal model and are therefore assessed via the use of ray tracing. In this paper, we studied the influence of reactor geometry, reactor material, LED type and absorbance which are important parameters w.r.t. scaling-up co- and countercurrently illuminated photochemical reactors. After optimisation of this set of parameters, the initial cost and operational cost of the co- and counter-currently illuminated photochemical reactor can decrease as no collimation optics are required and the associated energy losses are avoided. This study is performed via the use of experimental and simulation data, with the latter being performed via the ray tracing toolbox in COMSOL.

Note that this study is aimed towards the use of homogeneous photochemical reactions rather than homogeneous photosensitized or photocatalyzed reactions. Single phase photochemistry without a homogeneous catalyst is preferred when the reaction is selective and has a decent quantum yield. Homogeneous photocatalysis is often more selective and has decent quantum yield with commonly used photocatalysts and sensitizers. The disadvantage of homogeneous photocatalysis is the increased need for adequate mixing in the reactor combined with the mismatch between absorption and conversion. At higher conversions, not every absorbed photon (corrected for the quantum yield) yields a converted reagent molecule. This mismatch is absent in case of homogeneous photochemistry as in most cases only the reagent is capable of absorbing light and subsequently converting to its product. Furthermore, when using a catalyst or sensitizer, an extra separation step is often required, thereby increasing operation cost.