Abstract
As a food processing method, smoking has been used for the purpose of preserving, browning and flavouring food since ancient times. Derived from the thermal decomposition of biomass, normally wood, smoke contains hundreds of chemical compounds covering various classes that convey flavour and colour during the smoking of food. The conditions used to generate wood smoke for the purpose of smoking food greatly affect the formation of the resulting chemicals, and thus the flavour character of the smoked food. However, not much published research focuses on how to control the smoke generation conditions to manipulate levels of specific volatile organic compounds (VOCs) in the gas phase of smoke in order to manipulate the smoke flavour profile. Serious health and regulatory concerns also exist regarding the formation of suspected carcinogens, polycyclic aromatic hydrocarbons (PAHs), during smoke generation. Thus, there is a need to precisely control the smoke generation process to optimise the flavour levels, while minimising the levels of carcinogenic PAHs. In addition, no research has been published on the VOC and odour profiles of kānuka (Kunzea ericoides) wood smoke, a New Zealand indigenous hardwood species that is widely used in the local food industry and for exported food products.
In this thesis, the overall research objective was to characterize the VOC composition, odour-active compounds and PAHs in kānuka smoke as a function of the smoke generation conditions. Three main steps were performed: 1) development of a laboratory-scale smoke generator with which smoke could be generated and collected in a controllable manner, 2) identification of the odour-active compounds present in the gas phase of kānuka smoke, and 3) optimisation of the smoke generation conditions to maximize selected odour-active compounds while minimising the levels of PAHs.
A laboratory-scale smoke generator was developed to enable smoke to be generated and collected in a controlled manner to analyse kānuka wood smoke. The characterisation of the output of this smoke generator was achieved by generating the smoke at two temperatures (280 °C and 480 °C) and supplying either air (21% oxygen) or nitrogen. The VOCs were extracted from the smoke generator using inline-sampling with stir-bar sorptive extraction (SBSE), and analysed using gas chromatography-mass spectrometry (GC-MS). Results showed that reproducible smoke could be generated under controlled temperatures and atmosphere. Quantitative comparison of syringol (2,6-dimethoxyphenol), guaiacol (2-methoxyphenol), phenol, m-cresol (3-methylphenol), acetol (hydroxyacetone), and furfural suggested that the formation of VOCs varied with the bed temperature and applied atmosphere. This demonstrated the possibility of manipulating the smoke composition using this laboratory-scale smoke generator. Eight polycyclic aromatic hydrocarbons (PAHs) with molecular weight no greater than 202 Da were also detected by SBSE. Higher temperature produced higher levels of PAHs, while the impact of atmosphere composition varied in a compound-specific manner.
Using the developed laboratory-scale smoke generator, smoke was generated at four temperatures (180 °C, 280 °C, 380 °C and 480 °C) under either air or nitrogen to further characterise the VOC profile of kānuka wood smoke and to investigate the boundaries of temperature and atmosphere for further experimental design. GC-MS analysis of VOCs extracted by inline SBSE found more than three hundred compounds in kānuka wood smoke, of which phenols were the dominant class, followed by ketones and aldehydes. Principal component analysis (PCA) illustrated that temperature had a greater impact than atmosphere on the VOC formation; however, some large compound specific effects were observed. Comparison of 2,6-dimethylphenol, creosol, furfural, 2-methoxy-4-propylphenol, 2-furanmethanol and 4-hydroxy-3,5-dimethoxybenzaldehyde demonstrated that their intensities varied with the smoke generation temperature under both air and nitrogen, implying that different VOC profiles can be developed.
The odour profile of kānuka wood smoke was investigated by GC-MS/Olfactometry (GC-MS/O) assisted with a trained panel composed of six assessors. Smoke samples were generated under four selected conditions to cover broad enough ranges to avoid omission of potential odour-active compounds. The four conditions included 275 °C under air/nitrogen (50/50, v/v) (S1), 350 °C under air/nitrogen (50/50, v/v) (S2), 350 °C under nitrogen (S3) and 500 °C under air/nitrogen (50/50, v/v) (S4). Sixty-two odour regions were found above the selected threshold in at least one of the treatments. The odour profile of kānuka wood smoke was dominated by phenolic compounds with the highest in number, which was in agreement with findings from the VOC profile. The top five odour-active compounds that contributed most to the odour profile of kānuka wood smoke were: vanillin, guaiacol, creosol, 4-ethyl-2methoxyphenol, 4,4a,5,6,7,8-hexahydro-1-methoxy-2(3H)-naphthalenone and 1,4-dihydro-2,5,8-trimethylnaphthalene. Twenty-nine odour regions exhibited a statistical difference between treatments, demonstrating the possibility of achieving different odour profiles by varying smoke generation conditions. At 350 °C, higher levels of the dominant odour-active compounds such as guaiacol were generated. In contrast, harsher conditions (i.e. at 500 °C) yielded a greater number of compounds and higher levels of odour-active compounds that significantly differentiated the smoke generation treatments, e.g. 2-allyl-4-methylphenol. The GC-MS/O analysis of kānuka wood smoke provided representative odour-active compounds to assist the investigation of how smoke generation conditions influenced the formation of odour-active compounds via a systematic experimental design.
Using a response surface methodology (RSM), a central composite design (CCD) was applied to reveal how four smoke generation parameters (temperature, atmosphere, sweep gas flow rate and moisture content of wood powder) affected the formation of VOCs in kānuka wood smoke. Forty-two compounds were identified depending on the mass spectra and retention indices (RI). Important odour-active compounds were selected based on GC-MS/O results to investigate in more detail. Analysis of variance (ANOVA) of 2-furanmethanol, 3-methyl-1,2-cyclopentanedione, guaiacol, creosol, syringol and vanillin showed that temperature was the most influential factor. The response surface plots were constructed to visually evaluate the effects of selected smoke generation conditions and their interactions on these compounds that represent the intense odour-active compounds and cover the major chemical classes.. The optimized conditions to reach the highest levels of the six compounds were also determined. For instance, the highest response of vanillin that was found with the highest odour response from GC-MS/O analysis was predicted to occur under the following conditions: 500 °C, 100% air, 250 mL of sweep gas and wood powder with 0% moisture content (dried at 105 °C for 2 hours), which was supported by the high level of consistency between the predicted and experimental response.
Along with the investigation of selected odour-active compounds, the relationship between detected levels of PAHs and smoke generation conditions were also researched using RSM. Results demonstrated that temperature had the highest impact on all sixteen priority PAHs assigned by the U.S. Environmental Protection Agency (EPA), though dibenzo(a,h)anthracene was only detected in smoke generated at 500 °C. In contrast, the importance of atmosphere or sweeping gas flow rate varied in a compound-specific way. When the temperature was 350 °C or lower, concentrations of all the U.S. Environmental Protection Agency (EPA) listed PAHs were low or below the detection limit with only a slight increase with temperature regardless of levels of other smoke generation parameters. This observation indicates that the odour profile of kānuka wood smoke could be manipulated while maintaining relatively low level of PAHs, if high temperature (e.g. > 350 °C) could be avoided.
In conclusion, this thesis characterises the VOC profile of kānuka wood smoke and identifies the odour-active compounds that contribute to its odour character for the first time. This thesis provides knowledge on how to tailor levels of odour-active compounds to control flavour, while minimising PAHs by manipulating the smoke generation parameters.