**5. Biosynthetic vanillin**

Vanillin (4-hydroxy-3-methoxy-benzaldehyde), the major organoleptic component of natural vanilla, is the most widely used flavour compound in food, cosmetics and pharmaceutical industries [Guzman, 2004]. Natural vanillin, obtained from *Vanilla* pods through a long and expensive process, can supply less than 1% of the market demand. Therefore, most of the vanillin employed is obtained through chemical synthesis from guaiacol or lignin [Ramachandra & Ravishankar, 2000]. The production of vanillin through microbial or enzymatic bioconversion of selected substrates such as ferulic acid or capsaicin is an interesting alternative, as the product can be labelled as "natural" according to the European and US legislation. Studies regarding the production of biotechnological vanillin have shown that several microorganisms have the potential of being used in the bioconversion of ferulic acid into vanillin, such as actinomycetes, *Pseudomonas spp., Bacillus spp*. and *Aspergillus spp.* [Walton et al., 2000]. Vanillin can also be obtained by an enzymatic process from capsaicin in two steps: capsaicin is converted by acilase to vanillylamine, which is then converted to vanillin by amine oxidase. [van den Heuvel et al., 2001]

The recovery of the product from the bioconversion broth is a key point of the whole process. As vanillin has toxic effect on bacterial cells and it can be further transformed to vanillyl alcohol or vanillic acid, the productivity can be potentially enhanced by in-situ removal of vanillin. The addition of adsorbent resins to the culture medium was proposed in the biotransformation of ferulic acid to vanillin by a *Streptomyces* sp. strain [Dongliang et al., 2007] in a fed-batch process. Pervaporation was also a proposed for product removal [Boddeker et al., 1997], however, owing to the low volatility, the vanillin flux was quite low at the bioconversion temperature. The aim of this work is to evaluate the potential of using membrane contactor technology for the recovery of vanillin from bioconversion broths in view of an integrated bioconversion – separation process.

#### **6. Experimental**

The partition coefficients of vanillin as well as of the substrates of the bioconversion (ferulic acid and vanillylamine) in various solvents were determined through batch experiments. The solvent selected was n-butyl acetate. This solvent allows the selective removal of the

non-parallel fibres may lead to higher than expected mass transfer coefficients. Poor fluid distribution may lead to lower than expected values. A number of apparent errors in the

The comparison of the models and correlations is also complicated by the different parameters used to characterize the packing as well as the different definition of the characteristic diameter. In this paper the outer fibre diameter do has been used in the definition of Reynolds and Sherwood numbers; many authors instead used the hydraulic

> <sup>1</sup> <sup>1</sup> *h o d d* φ

Both choices can be accepted, the outer diameter seems more rational to compare the performances of modules made with the same fibres, but with different packing density.

Vanillin (4-hydroxy-3-methoxy-benzaldehyde), the major organoleptic component of natural vanilla, is the most widely used flavour compound in food, cosmetics and pharmaceutical industries [Guzman, 2004]. Natural vanillin, obtained from *Vanilla* pods through a long and expensive process, can supply less than 1% of the market demand. Therefore, most of the vanillin employed is obtained through chemical synthesis from guaiacol or lignin [Ramachandra & Ravishankar, 2000]. The production of vanillin through microbial or enzymatic bioconversion of selected substrates such as ferulic acid or capsaicin is an interesting alternative, as the product can be labelled as "natural" according to the European and US legislation. Studies regarding the production of biotechnological vanillin have shown that several microorganisms have the potential of being used in the bioconversion of ferulic acid into vanillin, such as actinomycetes, *Pseudomonas spp., Bacillus spp*. and *Aspergillus spp.* [Walton et al., 2000]. Vanillin can also be obtained by an enzymatic process from capsaicin in two steps: capsaicin is converted by acilase to vanillylamine,

which is then converted to vanillin by amine oxidase. [van den Heuvel et al., 2001]

view of an integrated bioconversion – separation process.

The recovery of the product from the bioconversion broth is a key point of the whole process. As vanillin has toxic effect on bacterial cells and it can be further transformed to vanillyl alcohol or vanillic acid, the productivity can be potentially enhanced by in-situ removal of vanillin. The addition of adsorbent resins to the culture medium was proposed in the biotransformation of ferulic acid to vanillin by a *Streptomyces* sp. strain [Dongliang et al., 2007] in a fed-batch process. Pervaporation was also a proposed for product removal [Boddeker et al., 1997], however, owing to the low volatility, the vanillin flux was quite low at the bioconversion temperature. The aim of this work is to evaluate the potential of using membrane contactor technology for the recovery of vanillin from bioconversion broths in

The partition coefficients of vanillin as well as of the substrates of the bioconversion (ferulic acid and vanillylamine) in various solvents were determined through batch experiments. The solvent selected was n-butyl acetate. This solvent allows the selective removal of the

⎛ ⎞ <sup>=</sup> ⎜ ⎟ <sup>−</sup> ⎝ ⎠

(49)

literature experimental correlations were also pointed out [Liang & Long, 2005].

diameter dh, which depends on the packing fraction:

**5. Biosynthetic vanillin** 

**6. Experimental** 

produced vanillin from the bioconversion broth removing only a minor extent of the substrates. Indeed the partition coefficient of vanillin is 21 (at pH 7) whereas for ferulic acid it is 0.2 and for vanillylamine 0.01. As shown in Tab 2, the partition coefficient of vanillin depends on pH, making possible to counter-extract vanillin from the vanillin rich solvent phase by using alkaline water.


Table 2. Partition coefficient of Vanillin between water and n- butyl acetate at various pH

Accurel® polypropylene hollow fibres, supplied by Membrana GmbH, Germany, with inner/outer diameter 0.6/1 mm and porosity 60% were used to build the module used in the experiments. The module contained 30 hollow fibres with overall area 158 cm2 , the shell diameter was 10 mm, the packing fraction was thus φ = 0.3.

The experimental set up is shown schematically in Fig. 6. The feed and the solvent were circulated counter-currently through the module and the respective reservoirs by two gear pumps with variable speed. The pressure of aqueous phase was kept a bit larger than that of the organic phase; to this purpose the feed reservoir was kept at a higher level with respect to the solvent reservoir. In addition a suitable overpressure (0.2-0.4 bar) can be created by a valve at the module exit. Both circuits were equipped with instruments for measuring the temperature and the flow rate during the experiments.

Small samples of the organic and aqueous phases were taken at various times and analyzed with HPLC reverse phase system, equipped with a Beckman Ultrasfere 4.6 mm x 250 mm ODS C18 column (particle diameter = 5 μm) , at 35°C. The mobile phase was composed of 70% H20 added with 1% CH3COOH and 30% CH3OH added with 1% CH3COOH; column temperature was 35°C, injection volume was 20 μL. The isocratic elution was performed for 16 minutes.

Four type of extraction experiments were performed: i) extraction from model solutions containing vanillin and ferulic acid or vanillylamine (the substrates of microbial and enzymatic conversion respectively), ii) extraction from the whole bioconversion broth, iii) extraction during bioconversion coupling bioconversion and separation, iv) counterextraction of vanillin from the solvent by NaOH-water solution. The feed flowed either through the lumen or in the shell side. In all the experiments the temperature was 30°C and the pH was 7 in the extraction experiments and 12 in the counter extraction.

The bioconversion broth used in the experiments was obtained in a 3.5 L stirred tank bioreactor by *Pseudomonas fluorescens* BF13-1p at concentration 6 g/L with ferulic acid (2 g/L) as unique carbon source, at 30°C and pH 7.
