![[BioMatNet Database - FAIR Program]](../images/Fair.gif) |
FAIR-CT96-2003 EXTEN: Volume extraction and encapsulation of substances used as flavour chemicals, pharmaceutical raw substances, biochemicals and enzymatic systems |
Source: Final Report January 2000
Introduction
Many of the compounds found in plants have useful applications in the pharmaceutical, food processing and various other industries. They can be employed for aesthetic or physical purposes, to improve a product's performance. As useful biomolecules are usually only present in trace amounts in organic residues, they require careful extraction and downstream processing to render them useful as an active ingredient.
The classical method of extraction involves leaching or percolation from botanical residues, utilising a solvent system compatible with the lipophilic/hydrophilic characteristics of the extract sought. Evaporation of the solvent yields the raw extract, but even if the process takes place under conditions of high vacuum, a residual amount of solvent will always be present within the extract. Concentrations of residual solvent as low as 50 ppm may cause problems for downstream production processes or eventual and use. Many organic solvents have particularly detrimental properties (e.g acetone, dichloromethane, dichloroethane, benzene, hexane etc.) regarding human health and the integrity of various industrial processes.
Objectives
The main aim of the project was to develop a two stage process for the extraction of substances from botanical source matter; the first stage being a hydroalcoholic extraction, and the second stage the extraction of the hydroalcoholic solution with supercritical carbon dioxide. This two stage process should be capable of selecting a single substance at higher yields and purity than obtainable using conventional solvent extractions. In addition, the extracts will be free of any hazardous organic solvent residue, making this technique of considerable interest to the pharmaceutical and foodstuffs industries. The extraction technique is being developed using a list of test extracts for which there is a proven demand.
The project also aimed to develop methods of micro-encapsulation to work in tandem with the extraction process. Microencapsulation is of great interest to the pharmaceutical, food, cosmetics and other industries, giving added value in the following ways:
Microencapsulation can be carded out by varying methods, including coacervation, spray drying and spray coating. During the past ten years a number of techniques have been proposed which utilise the properties of supercritical fluids to produce microscopic particles. The partners are investigating both conventional and supercritical fluid techniques which will allow encapsulation as an in-line process directly after extraction. This will be a considerable advance in terms of process integration, cutting process costs for a number of industries and increasing the stability of unstable products once they are extracted.
The project aimed to use advanced mathematical modelling of the extraction processes involved in order to optimise the extraction parameters from a data set obtained through laboratory scale experiments. Extraction and microencapsulation processes were developed separately in the first half of the project, for integration in the second half.
More specifically the project aimed to develop a three-stage continuous, in-line process for the extraction and encapsulation of botanical substances, with the following specific objectives:
Indirect objectives were as follows:
Activities
These are presented on a task by task basis:
Task 1 Modelling of hydroalcoholic-supercritical fluid solvent system
Task 2 Development of extraction process
Task 3 Development of techniques for separation of extracts
Task 4 Optimisation of encapsulation processes
Task 5 Testing of complete process
Task 6 Project management
The work done under each completed or current technical task is reviewed in more detail in the rest of this section.
Task 1 Modelling of hydroalcoholic supercritical fluid solvent system
The aim was to collate of data in support of new candidate extracts, produce a literature survey to establish physicochemical properties of test substances and generate a list of test extracts as well as define common reference systems for test samples. The actual work performed resulted in distribution of information on the molecular structure of the active ingredients, as well as a literature search covering previous extraction and analytical work on these botanicals. The 'rules of thumb' relating molecular structure to the solubility of a substance in supercritical CO2 were used in addition to the literature search to indicate which test extracts were likely to be extracted well. The list of test extracts for the project was discussed at project management meeting . Additional test extracts of economic interest were proposed to replace those eliminated. The need to use extracts with applications in the food industry was stressed, as exploitation is more straightforward with a less stringent regulatory environment than for the pharmaceutical industry. A list of 9 extracts was drawn up for further work. This included Coenzyme Q10 (from tobacco?) Artemisinin (from Artemsia annual Apigenin-7-g (from camomile?) Capsaicin (from capsicums) Hypericin (from St John's Wort), 1-hyoscyamine (active optical isomer of atropine, from belladonna) Quassin (from Quassia) Resveratrol (from grape skins) Parthenolide (from Feverfew). A sample reference system was agreed.
Suitable botanical sources were identified for each test extract under this task. For Coenzyme Q10 and Apigenin-7-g this was not immediately apparent and so work in this area extended throughout the first year of the project. Botanical sources were found for all test extracts in use, but new sources can be evaluated if they are considered superior. Further research on the anti-depressant activity of St John's Wort was published in the second half of the project, indicating that hyperforin was the true active ingredient, rather than hypericin. However, hypericin is still of commercial interest as an anti-viral agent. To provide an empirical basis for the modelling task laboratory extraction studies were carried out resulting in several extracts being eliminated owing to their lack of CO2 solubility. Initial tests were done using pure samples of the active compounds obtained from commercial sources. Supplies of some were difficult to locate and this task was completed only in month 15. Only four of the nine extracts were found sufficiently soluble for purposes of industrial extraction. These were atropine, coenzyme Q10, artemisinin and capsaicin.
It was realised that the test extracts with low solubility in CO2 could be used to demonstrate reverse extraction; removing other substances in the hydroalcoholic extract, and thus enriching the desired active substance. Hypericin, parthenolide and apigenin were therefore retained for this purpose. Some safety problems had been encountered with atropine and so work with this extract was stopped. Since month 15, capsaicin and artemisinin have been used to develop the supercritical fluid extraction process, with hypericin from St John's Wort to demonstrate reverse extraction. Work was planned on the other extracts if time allowed, but this was not the case.
VLE experiments commenced in month 8, with atropine and coenzyme Q10, from month 13 VLE data was obtained for capsaicin and artemisinin in order to develop the second stage extraction process. The enrichment of both target compounds in the gas phase over the liquid phase was significant. Hypericin was also studied in preparation for the reverse extraction development.
Pure samples of each active compound were used to establish the analysis of each extract with SFC-FTIR and SFC-MS. Once the platform analyses of most of the test extracts were complete, SFC-MS was recommended as the main technique to be used in the project by virtue of its speed. However, atropine appeared to break down in the SFC column, requiring development of a novel mobile phase to allow atropine to pass through. Until this was available the required analysis was performed using a conventional wet technique. In addition, hypericin and apigenin required the use of the mass spectrometer's negative ion mode. This entailed some modifications, and this created a delay with respect to the analysis of these extracts. Apigenin was successfully analysed, but progress with hypericin was slow, with difficulty in eluting through the SFC column, even though several types were tried. When the principal target compound of St John's Wort changed from hypericin to hyperforin, a standard sample could not be found. To remedy this problem an HPLC method was developed and hypericin was successfully analysed quantitatively, and by the end of the project quantification of hyperforin content. Alternative botanicals containing apigenin for the demonstration of reverse extraction were also investigated. these were Scharfgarbe, Chrysanthemum and Resula. Hydroalcoholic extracts of these were analysed for apigenin content.
To provide an accurate estimate of the parameters required for maximum performance of each extraction stage the development of mathematical model for mass transfer in hydroalcoholic extraction, a mathematical model for phase equilibrium in second stage supercritical fluid extraction as well as modelling, design and construction of laboratory scale extraction plant were undertaken. Initial two stage extraction experiments were used to evaluate extraction feasibility for each test extract, with the use of mathematical models to establish optimum extraction conditions for each test extract and characterisation of extract samples.
Modelling of the evaporation process was linked to the hydroalcoholic extractor and SFE column. This was completed in month 6 and was used to re-design some of the components in the existing hydroalcoholic extractor. An optimisation algorithm for flow control was then produced, while work on mass transfer modelling of the hydroalcoholic extraction was postponed till year 3 of the project, as it was not critical to the development of the two stage extraction process.
A laboratory extractor was designed with the aid of a simple mathematical model to give the optimum number of extraction stages. This was completed and used in experiments to optimise the extraction conditions for belladonna, feverfew and capsicums.
As far as the mathematical model was concerned, various equations of state were used, but the models could not reproduce the experimental results with enough accuracy. After only limited success, it was decided to cease all modelling work. Polar fluids are notoriously complex and it is clear that the state of the art must advance considerably before mixtures of this type can be successfully modelled. However, there is a large amount of published VLE data on the water-ethanol-carbon dioxide ternary solvent system. This was correlated by hand to produce an algorithm, operated with a spreadsheet. This can calculate the phase composition of the gas and liquid phases of the mixture, from inputs of initial hydroalcoholic extract solvent composition, temperature and pressure. This tool was of use in the development of the two stage extraction process.
Task 2 Development of extraction processes
Work was carried out to establish whether pre-treatment of the botanical material was necessary, and if so, what was required. Studies have covered mixing, centrifugation, and vacuum filtration. However, most of the effort has been towards developing effective grinding of plant material to the optimal particle size for extraction. This was established at around 6 mm for almost all plant materials. The importance of minimising the proportion of fines which can affect extraction performance has been emphasised. A variety of milling and cuffing machines were evaluated, using ginger as the primary test material. Generally, these produced a bimodal distribution of particle size, with too many fines and larger pieces of material. Hence the Quadro cone mill was investigated and found to give a high proportion of particles at the optimum size, with little fines. The Kemutec Kibbler, a pre-grinder which was found useful for initial processing of large material which could not be fed directly to the cone mill. These two machines formed the basis for optimal pre-extraction grinding.
A continuous flow extractor was evaluated at the laboratory scale, but it was found that it generated too many fines, which cause problems both for extraction and downstream processing. A static 'coffee-pot" extractor was therefore specified and designed for the demonstration scale plant (500 1). completed in month 21. Commissioning trials were carried out in co-current mode, using ginger root, limeflowers, comfrey and orange peel. A large batch of capsicums was extracted, generating hydroalcoholic extract for use in further SFE experiments and for analysis for capsaicin content.
The work on two stage extraction commenced later than expected due to problems. Some work commenced in month 16 using capsaicin and artemisinin as the primary test extracts. These showed that thick and viscous extracts could be processed in the laboratory scale SFE column. Initial results for enrichment of the active compound were encouraging. The irritant nature of capsaicin meant that extra fans had to be installed in the apparatus to improve conditions in the laboratory.
The reverse extraction experiments using St John's Wort extract required a modified apparatus to allow the use of CO2 entrained with ethanol. The first experiments failed due to foaming In the cell.. Use of a taller cell solved this problem and extraction with entrained CO2 generated samples of both extract and raffinate for analysis. However, as previously explained analysis for compounds present in St John's Wort were delayed till year 3.
A basic specification of the pilot scale SFE column was developed by month 17 resulting in orders for the CO2 compressor in month 23 and the SFE column in month 24. These were the most time critical components, and their projected delivery dates allowed sufficient time for the plant to be completed by the end of the project. The design for the plant was finalised in month 26. In discussion with the equipment manufacturers, a number of modifications to the design were carried out, including changes to the heat exchanger, column packing and support structure, and the raffinate tank. However, various delays in the fabrication of plant components prevented the completion of the SFE plant to after the end of the project (mid-2000).
It was noted that the rate of extraction was suppressed as the concentration of active compound in the liquor increased. A method was devised to reduce the number of extraction stages and therefore the quantity of solvent used, while extracting the same amount of active compound. This was successfully demonstrated at laboratory and pilot scale with a co- current extraction of an unspecified leaf product. A two stage counter-current flow regime was then evaluated at laboratory scale. In order to reduce the requirement for evaporation, this used the wash from the previous batch of plant material for the first extraction, which was then sent for evaporation. The second extraction of the plant material used fresh solvent, which was then retained for the first extraction of the next batch. This approach succeeded in preventing suppression of extraction rate by a high liquor concentration.
Previous theoretical work had shown that significant amounts of active material could be lost in the liquor absorbed by the plant material in the extractor, and that this was dependent on the amount of solvent used. Additional experiments were used to assess the efficiency of a supplementary wash to recover some of this absorbed material. Recirculation of the wash solvent was found to recover more active than in a single pass. The recirculated wash was then combined with the extraction liquor set aside for the next batch. This work showed that a 2 stage counter-current extraction could be operated with the same efficiency as with a larger number of extraction stages. The results were extrapolated and used as a basis for a new specification of the most efficient hydroalcoholic extraction technique.
Further evaluation of the pilot plant was carried out using feverfew, quassia (a hard chip material difficult to extract), bilberry and artemisia.
As a result of the plant described above not being ready the two stage hydroalcoholic extraction process was investigated at pilot scale using the a 4 by 4c column, fitted with two cyclones provided two step depressurisation, and pressurised feed capability. Six runs were carried out using capsicum extract.. Four were diluted with oil and two with water. Positive results were obtained by introducing the feed from the top and the middle of the column. Full analysis of samples to provide the precise degree of capsaicin enrichment by were not completed by the and of the project.
Task 3 Techniques for separation of the extracts from the solvent system
At the beginning of this task, isobaric separation and pressure reduction were evaluated. Each had advantages and disadvantages, but pressure reduction was more reliable and can be applied to all extracts, and it was decided that cyclones would form the basis of the separation stage. However, an equipment manufacturer suggested that a simple gravity separator, would be adequate and more cost effective. However, along with the other SFE plant components, the gravity separator was not delivered by the end of the project. It will be incorporated with the rest of the SFE plant in mid-2000.
Task 4 Optimisation of encapsulation processes
The work commenced early on with an evaluation of encapsulant materials for use in the project. These were tested for CO2 solubility and hydrogenated palm oil, hydrogenated castor oil and polyethylene glycol were found to be of potential use for the RESS technique. The spray cooling technique was selected to act as the main conventional microencapsulation technique because it produces microspheres similar to those produced by RESS, rather than microcapsules. Laboratory experiments were carried out using various model active ingredients and the encapsulant materials listed above. The microspheres produced were studied for their release properties, and each material was found to exhibit slow or instant release characteristics. The samples produced by the spray cooling technique were intended as references for comparison with the microspheres produced by the RESS experiment. Work on the RESS experiments were delayed due to problems in obtaining of components, owing to the high specification of the experimental design. Benzoic acid, as a well-studied test material, was used as a test substance, while atropine was also used and microparticles were obtained of both substance. SEM analysis was limited at first, owing to problems of chemical reactions with the coating of the fixing tape, but this problem was soon solved.
The RESS technique was evaluated and concerns were raised regarding the likely low yield of the technique. Alternatives techniques using supercritical fluids were proposed which could also be operated in-line, including PGSS, ASES and impregnation of pre-formed silica or cellulose spheres. It was also realised that the spray cooling technique could be operated in-line with the SFE column, with the CO2 stream depressurised into the hot melt of encapsulant. Further RESS experiments with encapsulant HPO gave poor results In comparison with atropine. These runs were carried out under identical conditions to the atropine runs, and this indicated that microencapsulation with this technique was not possible for most pairings of active ingredient and encapsulant. The technique was only feasible where the solubilities of each material in SCCO2 were very similar.
As a result two new process concepts for in-line encapsulation which were more likely to be feasible were investigated. The first was a two step process whereby the active ingredient, dissolved in SCCO2 , was sprayed by RESS into an encapsulant material liquefied with CO2 under pressure. Spraying of the Impregnated liquid through a nozzle in a PGSS type process would give composite particles. The second concept, involve a concentric dual nozzle which brought together streams of active and encapsulant, both dissolved in SCCO2 and expanded by RESS.
Additional laboratory scale trials of the spray cooling process were carried out with two types of nozzles evaluated; air nozzles and ultrasonic nozzles. A wide range of encapsulant materials and active ingredients were used in the trials, many of which had not been tested before. Particle size analysis and dissolution tests were carried out on samples of microspheres produced by the spray cooling unit. The ultrasonic nozzle gave a more uniform distribution of microspheres with similar size and shape, although ft had a lower throughput than the air nozzle. It also has some limitations with abrasive feeds, but was selected as the better option. Of the encapsulant materials, a mix of HPO and monumuls gave the best results.
On the strength of these results, a pilot scale spray cooling plant at month was designed and testing and optimisation commenced. The RESS apparatus was modified with a new high pressure vessel and other components, and commissioned with the dual nozzle component. However, again there were considerable delays in the fabrication of the dual nozzle and work on this concept did not proceed in the remainder of the project.
For optimisation of encapsulation experimental plants theophylline was selected as a suitable test material, since it is because it is well documented in the European Pharmacopoeia and analysis is straightforward. Optimisation of the spray cooling process involved the use of nine fats currently used in the food and pharmaceutical industries as encapsulants. The hydrophilio-lipophilic balance (HLB) of each fat was found to be critical for the release profile of the resulting microsphere, with the more hydrophilic fats having shorter half lives in dissolution tests. The encapsulant. materials GV60 (a brand of HPO) and monomuls were selected for further studies because, when mixed, they gave a large range of release profiles, and were suitable for human consumption. A matrix of experiments were designed in order to correlate the three factors assumed to influence the microencapsulation function: ratio of GV60/monomuls, % content of theophylline active and air pressure in the spray cooling process. This trial showed that most rapid release was obtained using monomuls with a high theophylline content, and the slowest release using GV60 with a low theophylline content.
Micronised theophylline was also used to compare with the standard powder. Microspheres of about half the size obtained before were made with spray cooling, and these had slightly faster release rates. Tests with theophylline in benzoic acid were not successful, due to precipitation of the active on mixing with the fat. Separate trials were used to evaluate the encapsulant ethyl cellulose, using coacervation. This material gives a permeable encapsulation suitable for drug delivery.
The work with theophylline was completed with a thorough investigation of release kinetics. Release profiles with several lipids, as well as ethyl cellulose, were compared against theoretical models based on polynomial functions. Later microencapsulation by polling, a similar technique to spray cooling, was developed where an ultrasonic nozzle breaks the flow of fat into droplets before solidification, and liquid nitrogen is used as the cooling medium. This was found more suitable for large microspheres, which required more intensive cooling. A final optimisation exercise to improve the economy of both spray cooling and polling techniques involved minimisation of their cooling requirements. No real improvements were gained for spray cooling, but with manipulation of feed and N2 flow rates, it was possible to reduce the liquid N2 consumption of the polling technique tenfold.
In order to provide references for the supercritical fluid encapsulation development, spray cooling was used to process a capsicum extract. Samples of the extract and encapsulated material were analysed and further tests were made of a specially purchased water-free capsicum extract, to simulate the processing of a CO2 extract. This proved to be more successfully encapsulated by the process, probably as a result of the absence of water.
Experiments with the RESS-PGSS concept for microencapsulation proceeded throughout year 3. GV60 was used as the encapsulant, with theophylline as the test substance for the first experiments. The following process parameters were investigated for their influence on particle size and morphology;
Particles were obtained on all runs and were imaged using scanning electron microscopy. Yields were good, and the particles ranged from I to 100 microns in diameter. Spherical particles were observed as well as fibrous structures. In order to understand more about the structure of the generated material, experiments were carried out with GV60 and theophylline separately. The pure theophylline samples were the only ones observed by SEM to include fibrous structures, and so it was proved that the fibres previously viewed in the composite samples were entirely theophylline.
The technique was then adapted, involving expansion from a lower pressure (-20 bar) with a larger nozzle, and a greater path length for the particles. The apparatus was re-designed with commercial nozzles which could be interchanged easily for different results. It was decided to use capsaicin as the test active, to integrate more with the extraction work, and over 20 trials were carried out. However, it was not possible to carry out release tests of any of the samples generated, due to the need for larger sample size that required for SFC-MS. However, this is the only way of proving that the particles have a microencapsulation functionality. It is expected that further samples will be generated after the project and analysed fully.
Task 5 Testing of the complete process
The aim was to test the whole process from the extraction of the botanical material to the final encapsulated test products dispersed within their carder substances. However, this task was not attempted, mainly because the two stage extraction plant was not completed in time, and work with the RESS-PGSS technique was still at the laboratory level.
Conclusions
Overall, the EXTEN project has been a success, although many of the deliverables were not achieved as originally scheduled hydroalcoholic process have been redeveloped from first principles. It is an ancient production process, with some companies continuing to use traditional methods. After a thorough investigation throughout the three years, it has been concluded that the basic principles of traditional hydroalcoholic extraction are the best approach. However, significant optimisation has taken place, particularly in the pre-treatment grinding of plant material, the extraction duration and amount of solvent required. The new hydroalcoholic extractor is designed to take advantage of the new, shorter extraction times and provide greater production efficiency. The other main contribution to understanding of hydroalcoholic extraction has been the development of a mass transfer model. Surprisingly, this has not been done for this type of extraction before, and so the results will be freely disseminated.
The principal innovation in the second stage supercritical fluid extraction has been the processing of hydroalcoholic extracts, which can be very viscous, in columns which are designed for extraction of free-flowing liquids, such as wine or olive oil. There have been problems with blockages, but it appears that the ability to process viscous concentrates effectively and safely now exists. Several of the test extracts have been processed with significant enrichments of the active compounds. Full- scale demonstration of the technology will only be possible once the SFE plant is completed.
The project planned to use computer modelling of the phase equilibrium of the tertiary CO2 -water-ethanol mixture, to assist with development of the SFE process. This proved elusive even after 12 months of sustained effort to find an approach which could correctly predict phase behaviour. The state of the art in thermodynamic modelling is simply not advanced enough to deal with this solvent system. Fortunately, the alternative of manual correlation of VLE data proved sufficient for the requirements of the project.
The development of supercritical fluid microencapsulation processes specifically to be integrated in-line with an extraction process was always innovative and high-risk. The original concept was based on RESS, and this appeared to be feasible, until it was realised that encapsulant and active compounds had such different solubilities that they could not be simultaneously processed in the same phase.
However, the know-how gained in spray cooling and related techniques contributed to the experience gained in supercritical fluid processes and generated a new concept which has the potential to give very good microencapsulation with high yields. The use of fats is a particular innovation, as previous studies on supercritical fluid microencapsulation have generally used polymers, with limited success. With the developed process, a truly continuous extraction-encapsulation process is not feasible, but semi-continuous operation will attain the desired result. It is clear that there is much scope for further research, and alternative process designs will allow complete in-line integration of extraction and encapsulation.
© Copyright 2006 Policy Statements
Updated
by CPL Press
3 July, 2007
- biomatnet@biomatnet.org
 |
 |
 |
News |
 |
Events |
|
|