Catalytic Rearrangement of α-pinene oxide in a Spinning Disc Reactor (SDR)

Introduction

The future growth of the fine, pharmaceutical and speciality chemical industries are expected to depend heavily on the development of new more environmentally friendly technologies.  Manufacturing in those areas has commonly been associated with the accumulation of large quantities of hazardous waste resulting from the use of mineral acids and Lewis acids used as catalysts. A further serious problem is encountered in the selectivity to desired product, which demands control of isomer formation and minimisation of high molecular weight species.

In the past decade there have been serious efforts in the search for heterogeneous catalysts that would not only enhance reaction rates and product selectivity, but eliminate the problem of separating product from catalyst.  An important industrial example, from the field of fine chemicals, is the rearrangement reaction of α-pinene oxide (3-Oxatricyclo [4.1.1.0(2,4)] octane, 2,7,7-trimethyl-) to campholenic aldehyde (Figure 1).  Campholenic aldehyde is a key intermediate in the synthesis of santalol, the main constituent of natural sandalwood oil.  Currently, a homogeneous catalyst is used for the rearrangement, but selectivity is only moderate and represents optimisation to aldehyde in a reaction that can lead to the formation of more than 100 different products depending on reaction conditions [1].

Fig. 1 - Rearrangment reaction of α-pinene oxide to campholenic aldehyde

Some of the attempts to develop heterogeneous catalysts for this reaction have involved the use of mixed oxide solid acids [2], and US-Y zeolites [3,4] and depending on reaction conditions selectivities towards campholenic aldehyde of 55-80% have been reported.  In the recent reports, researchers have used Ti-Beta to rearrange a-pinene oxide in both the liquid and vapour phase, with the latter conditions giving campholenic aldehyde with an initial selectivity of ~94%, dropping to 80% after 6 hours [5].

New solid acid catalysts based on silica supported zinc triflate have been reported as being active and reasonably selective in the rearrangement of α-pinene oxide to campholenic aldehyde [6].  These catalysts can be recycled without loss of selectivity towards the aldehyde.  The results from an investigation of the use of this type of catalyst fixed to a spinning disc reactor are summarised here.  Should this prove to be successful, then it would provide a flexible process for the a-pinene oxide rearrangement reaction with no catalyst losses and no inorganic waste stream.

Spinning Disc Reactor (SDR)

The SDR produces thin highly sheared films by using the centrifugal force which is created on rotating surfaces7.  Studies have indicated that the fluid dynamics within these films result in significant enhancement in the heat and mass transfer rates [8,9].  Presence of numerous surface waves in the film may product a high degree of mixing and imperfections on the disc have been shown to promote turbulence in the film, and so can further improve the transport characteristics of the thin film flow8.  The SDR technology has been successfully implemented in styrene polymerisation [10] and BaSO4 precipitation [11], to name but a few.  One of the most important characteristics of the SDR process is short and easily controllable residence time of the material on the disc, which can be calculated as:

Where Q is the feed flow rate (m³/s), ω is the angular velocity (s^-1), ν is kinematic viscosity (m²/s) and r is radius (m). Also note that i denotes inner and o - outer diameter of the disc.

Fig. 2 - Spinning Disc Reactor

Experimental Results

The catalyst, prepared by collaborators in the University of York [6], was fixed on the disc surface and the reaction was tried for the range of conditions – disc speeds from 100 to 1500 rpm, flow rates in the range of 4 to 6 cm³/s and disc temperatures ranging from 250°C to 850°C.  A 200 mm diameter smooth disc, as shown in Figure 2, was used to perform the isomerisation reaction. The studies were done by feeding the reactant/solvent mixture which contained 1g of α-pinene oxide (Aldrich 99%) in 100 ml of 1,2-dichloroethane (Aldrich 98%) and 0.5 g of decane (Aldrich 99%) (internal standard).  The feed was introduced at the centre of the disc and streamed across the catalysed surface in the form of thin highly sheared films. Product was continuously collected via pipes located at the bottom of the reactor. Thermocouples were located at selected positions to measure the temperatures of the disc, feed and product. Samples were collected immediately and analysed by GC.

Three catalysts were available for this study:

• Catalyst 1: 0.05 mmol/g Zn(OTf)₂ supported on SiO₂

• Catalyst 2: 0.01 mmol/g Zn(OTf)₂ supported on K100

• Catalyst 3: 0.05 mmol/g Zn(OTf)₂ supported on HMS₂₄

A batch reactor was used for benchmarking the SDR performance and consisted of a glass vessel with water jacket around for temperature control. The batch temperature was additionally controlled by a thermocouple immersed in the reactant/product mixture.  Samples were taken during the course of reaction, as well as after all the reactant was completely used.  The reason for taking samples after the reaction had achieved completion was to check the stability of the campholenic aldehyde.  Samples were immediately analysed on the GC.  Reaction temperature was chosen to correspond to those used in the SDR experiments for all the catalysts used.  Obtained results are shown in Figures 3 (for Catalyst 1), 4 (for Catalyst 2) and 5 (for Catalyst 3).

Fig. 3 - Batch reaction data (catalyst 1)	Fig. 4 - Batch reaction data (catalyst 2)	Fig. 5 - Batch reaction data (catalyst 3)

SDR experiments

Summary of the conversion/selectivity data obtained for various disc rotational speeds and feed flow rates can be seen in Figures 6 and 7 for Catalyst 1; Figures 8 and 9 for Catalyst 2 and Figures 10 and 11 for Catalyst 3.

When trends for selectivity results are compared (Figures 7, 9 and 11) it is noticeable that both Catalysts 1 and 3 have lower selectivity towards campholenic aldehyde comparing to less active Catalyst 2; in fact when the residence time on the disc was long (e.g. at lower disc speeds [12]), none of the aldehyde was observed when using Catalyst 1.

This could be explained in two ways: either Catalysts 1 and 3 were promoting other reactions, or the aldehyde is being formed, but due to its high reactivity, rearranged itself which resulted in not as much aldehyde being seen at the end of the reaction on the disc.  Knowing that aldehyde that is formed in a batch reaction is easily consumed after its formation, second explanation looks more likely the true.  By increasing the flow rate, residence time on the disc is reduced and therefore higher selectivity is expected if consecutive reactions are taking place, which is experimentally confirmed for Catalysts 1 and 3 (Figures 7 and 11).

Fig. 6 - Summary of the conversion data for all experiments (catalyst 1)	Fig. 7 - Summary of the selectivity data for all experiments (catalyst 1)
Fig. 8 - Summary of the conversion data for all experiments (catalyst 2)	Fig. 9 - Summary of the selectivity data for all experiments (catalyst 2)
Fig. 10 - Summary of the conversion data for all experiments (catalyst 3)	Fig. 11 - Summary of the selectivity data for all experiments (catalyst 3)

Another parameter to be tested for each catalyst was the disc temperature.  In order to assess the influence of the disc temperature on the results, feed flow rate was kept constant at 4 cm³/s, and the disc rotational speed was fixed at 500 rpm.  Disc temperature was varied, ranging from 25 to 850°C.  Results from these runs are given in Figures 12 (in terms of conversion) and 13 (selectivity).  Increased activity at higher disc temperatures was observed for both Catalysts 1 and 2.  At lower temperature reaction has only just started and initially the rate of formation of other products (apart from the campholenic aldehyde) was much higher resulting in low selectivity towards campholenic aldehyde using Catalysts 1 and 2.  In other words, parallel reactions were promoted much more than formation of campholenic aldehyde.

The most selective was Catalyst 2 (around 70%) but it needs to be mentioned that by using this catalyst lowest conversion the α-pinene oxide was observed of all three catalysts used. Catalyst 3 had very reasonable selectivity - around 70%, except at temperatures around 500°C where it was ~20%.

Fig. 12 - Influence of disc temperature on conversion	Fig. 7 - Influence of disc temperature on selectivity

Cascade Simulations

In order to improve the performance of the SDR at 250°C (for the first two catalysts) and 850°C (for Catalyst 3), where conversion was low considering other tried temperatures, a simulation of cascade of three SDRs was performed.  First pass was done in usual way as for all the experiments, but the entire product was collected and fed onto the disc again (pass 2) and product from that pass collected and once more run through (pass 3).  After each pass sample was taken and analysed.  Summary of conversion results for all three catalysts can be seen in Figure 14.  It can be observed that in all cases conversion improved 10–20% after each pass. 

Fig. 14 - Conversion change for all catalysts after three passes	Fig. 15 - Selectivity change for all catalysts after three passes

Considering that residence time, if using a cascade of several discs, would not change dramatically (it would still be in order of seconds), improvements in both conversion and selectivity (Figure 15), are very significant.  Especially in the cases where conversion was very low to begin with (e.g. for Catalyst 1, after first pass conversion was only 27% with no campholenic aldehyde in the product, but after the third pass 52% of α-pinene oxide was converted and almost all of it to campholenic aldehyde (76%). 

Empirical Model

All the data obtained on the spinning disc were analysed using Microsoft Excel Regression Analysis Tool.  Conversion, xA, can be represented as:

Where N is rotational speed (rpm) of the disc, T is the disc temperature (K) and Q is liquid feed flow rate (m³/s).  Or:

Where I, a, b and c are regression parameters in the equation.

The following model equations which describe conversion on the spinning disc reactor can be written:

The model equations were developed under the following conditions:

150 ≤ N ≤ 1500;  250°C ≤ T ≤ 850°C;  3 cm³/s ≤ Q ≤ 6 cm₃/s

Conclusions

It has been shown that the Spinning Disc Reactor (SDR) can be used to perform organic catalytic reactions, such as that of isomerisation of α-pinene oxide.  SDR proved to be capable of enhancing the overall rate of reaction in comparison to reaction in a batch reactor, due to an intense mixing mechanism within the thin film in a SDR.  Additionally, the selectivity towards campholenic aldehyde is not only as high as in batch processes, but also easily controlled (i.e. by disc diameter and disc speed).

In a batch reactor, both catalyst and reactant are mixed and processed.  Times may vary, depending on conditions, but are usually in order of 30-60 minutes for this particular reaction.  After the reaction is finished, the catalyst and solvent have to be removed from the mixture in order to get clean product.  This is not the case for the process carried out on the spinning disc.  Not only that reaction times are very low (in order of seconds, rather than hours), but there is no need for the separation process to be completed, since there is no catalyst in the product mixture and the process is fully continuous which is a huge advantage of its own.

Comparison Table 1 demonstrates the advantages of using Spinning Disc Technology

Table 1 - Comparison between batch reactor and Spinning Disc Reactor

High conversions (full conversions most of the time) of α-pinene oxide achieved in a SDR compromised even better selectivity towards campholenic aldehyde; however, campholenic aldehyde is not the only valuable product of this reaction.  Almost every product is an important component in pharmaceutical industry.

Primary parameters studied for each catalyst used were disc rotational speed and feed flow rate with the aim of fully assessing the influence of residence time on conversion and selectivity [12,13].  A rise in selectivity values from 0 to 75% was observed by changing the rotational speed from 200 to 1500 rpm, while influence on conversion was not as significant since the conversion values were high for most of the conditions, varying from 80 to 100% (this is valid for Catalysts 1 and 3; for Catalyst 2 conversion varied from 40 to 90%).  It was also observed that increase of the flow rate by 1 cm³/s resulted in the increase of selectivity of at least 10% in cases of more active catalysts 1 and 3.  Different disc temperatures were also studied, as it was one of the main factors that influence the rate of reaction.  From the results it can be concluded that higher disc temperatures increased the conversion of α-pinene oxide in cases of the first two catalysts, whilst optimum temperature for accomplishing highest conversion for third catalyst was seen at 450°C.

From the performed batch processes it was clear that reaction does not stop after all the α-pinene oxide was used, a range of consecutive reactions take place and campholenic aldehyde disappears as well, therefore even when conversion is 100% on the SDR, reaction has gone much further and for this reason selectivity towards campholenic aldehyde is reduced.  It is expected that more campholenic aldehyde can be obtained at full conversion if smaller discs are used.

Optimal SDR conditions can be determined for all the catalysts used in this study for accomplishing highest conversion/selectivity.  These conditions are as follows:

• Catalyst 1: disc temperature of 850°C, disc rotational speed of 1500 rpm and feed flow rate of 6 cm³/s for achieving 62% selectivity at conversion of 77%;

• Catalyst 2: disc temperature of 850°C, disc rotational speed of 850 rpm and feed flow rate of 4 cm³/s for achieving 83% selectivity at conversion of 59%;

• Catalyst 3: disc temperature of 450°C, disc rotational speed of 1500 rpm and feed flow rate of 6 cm³/s for achieving 75% selectivity at conversion of 85%.

By using a simulation of cascade of three reactors it was shown that reaction could be carried out further even when the rate of reaction was lower (e.g. at lower temperature).

Regression analysis was used to obtain SDR models which can describe dependency of conversion of -pinene oxide upon the SDR variables (disc rotational speed, feed flow rate and disc temperature).  From the developed models can be seen that the parameter which influence the conversion of α-pinene oxide in a SDR the most is the disc temperature.

Other potential benefits, apart from enhanced reaction rate to be gained from the proposed process may include an improvement in its intrinsic safety as a result of reduced inventory in the system at any given time and minimal risk of thermal runaways at high operating temperatures due to short residence times and enhanced heat removal rates.