Supercritical Extraction of Ylang Ylang (Cananga odorata) Essential Oil at the Near-Critical Region

Abstract

The flowers of the ylang ylang tree contain an essential oil which is utilized in high-quality perfumes. The traditional mode of extraction is by steam distillation but it has been shown that the more modern supercritical fluid extraction (SFE) using carbon dioxide has potential for replacing steam distillation. This technology, however, generally operates under high pressures, up to 500 bar. The work described in this paper examines the possibility of using carbon dioxide at much lower pressures, close to the critical point, i.e., 75 bar and 30 °C. Two series of experiments were therefore carried out under such conditions, the first using carbon dioxide alone and the second utilizing ethanol as a co-solvent, the conditions being chosen by applying the Design of Experiments (DOE) technique over ranges of pressure from 80 to 120 bar and temperatures from 35 to 50 °C. Extraction curves are presented which show the rates of extraction to be significantly increased by the use of the co-solvent, with the measured values being 0.74% to 0.97% with no co-solvent addition, increasing to 0.92% to 1.16% with co-solvent addition. These rates are, however, lower than the rates previously reported at higher pressures, i.e., 0.9 to 1.8%. Better quality oils are, however, produced compared to those at higher pressures, with the major components being benzene benzoate, benzene salicylate, cubebene, and benzyl acetate. It is recommended that an economic study be carried out to evaluate whether it is feasible to utilize this process commercially.

1. Introduction

The ylang ylang tree (Cananga odorata) is native to tropical Southeastern Asian countries, mainly grown to produce its essential oil because of its high floral fragrance. It is well known because of its use in high-quality perfumes [1]. Cananga odorata is of the Annonaceae family, with 125 genera and 2050 species [2]. Currently, the Cananga genus has two known species of plant to date, namely, C. odorata and C. latifolia. C. odorata initially grew in Southeast Asian countries; however, due to its economic importance, it was later introduced into America and China. The ylang ylang flower is a highly fragrant, drooping type of flower where each petal is approximately 7.5 cm long, which is borne in groups of 4–12 hanging in axillary, umbellate clusters. The petals are arranged in two series of three, and the outermost petals are approximately 8–12 mm wide, whereas the inner petals are 5–7 mm wide [3].

There are many extraction methods used in the extraction of oil from raw materials [4]. In the process of separating the essential oil, the mode of extraction affects the quality, yield, and even odor of the oil. In ylang ylang, the main methods of commercial extraction are steam distillation and hydrodistillation [5]. The essential oil is vaporized by the heat from the steam, after which the solute diffuses out of the solid matrix. The oil/steam mixture is then condensed and fed to the collection vessel where it separates from the condensed water by virtue of their immiscibility [4]. The essential oil produced from ylang ylang has a complex combination of fragrance, as it combines floral, spicy, balsamic, fruity, woody, and medicinal, creating a unique after-smell, which is highly desired by the perfume and cosmetics industry [6]. It is used to make products such as cologne, food flavoring, soaps, and lotions, and recent import records show there is still a high demand for the oil [7].

The steam-distilled essential oil was reported to contain over 109 compounds, mainly phenylpropanoids and sesquiterpenes. More specifically, the main components of the essential oil are linalool, geranyl acetate, germacrene-D, beta-caryophyllene, benzyl acetate, geraniol, methyl p-Cresol, methyl benzoate, farnesene, and benzyl benzoate [8], which contribute to ylang ylang’s characteristic odor.

The more modern technique for extraction is that of supercritical fluid extraction (SFE), which is seen as more selective than traditional methods [9]. In SFE, the most commonly used solvent is carbon dioxide [10]. It is non-toxic, non-explosive, environmentally benign, and most importantly has low critical conditions (73.8 bar and 31.1 °C). It is also easily attainable, odorless, inert, and cheap. Carbon dioxide extraction, exhibiting almost ambient critical temperature, is very important in that it leads to retaining the thermal stability of the materials during the process. If used on an industrial scale, the extracted carbon dioxide can be easily recycled, reducing the cost as well as minimizing the carbon dioxide released into the atmosphere. Co-solvents can be used to increase extraction. Ethanol is a typical co-solvent for use with CO₂ [11]. It is considered a good pairing since supercritical CO₂ is known to be hydrophobic, and ethanol, being a polar solvent, can be paired with the CO₂ to remove the hydrophilic compounds of the essential oil, such as the sesquiterpenes and oxygenated terpenes, thereby increasing the overall yield. Furthermore, ethanol is an acceptable co-solvent under good manufacturing practices and poses no risk to human health when present in practically unavoidable amounts [11].

In a comprehensive study on SFE from ylang ylang flowers, [12] carried out an experimental program covering a range of temperatures from 35 to 75 °C and pressures from 100 to 500 bar, with this being reported at the 6th International Symposium on Supercritical Fluid Chromatography, Extraction and Processing in 2002 [13]. Extraction was completed in about an hour with yields in the range from 0.9% to 1.8%, with the maximum yield being obtained at 45 °C and 300 bar pressure. [12] also compared hydrodistillation with SFE, with results giving similar yields. When comparing the compositions of the extracts, there were differences in there being no benzyl alcohol and less than half of the benzyl acetate in the oil from hydrodistillation due to thermal degradation at the higher temperatures [14].

The work reported in this paper investigates extraction characteristics at lower pressures at around and just above the critical point, recognizing that the energy costs and turnround times will be lower at the lower pressures. It may, however, be predicted that the yields would be lower than at the higher pressures, so the study around the critical point also investigated the possibility of introducing ethanol as a co-solvent to increase yields.

This study aimed to generate extraction curves near and above the critical region over varying pressure, temperature, and CO₂ flowrates using a bench-scale SFE unit, and to compare the results with those obtained at much higher pressures [12]. In addition, this study aimed to compare the chemical compositions of the extracts with those reported in works at higher pressures. A further and most important aim was to evaluate the effect of adding ethanol as a co-solvent to increase yields.

2. Materials and Methods

2.1. Preparation of Raw Material

The flowers were picked from a single ylang ylang tree. They were harvested each day between 7:20 a.m. and 7:50 a.m. The flowers being harvested were visually yellow in hue and were immediately placed into non-porous, plastic zip-lock bags. For extractions being conducted concurrently, the zip-lock bags containing the flowers were stored at 5 °C, to minimize loss of volatile compounds. The initial preparations after harvesting started by selecting the flowers which had no bruising.

2.2. Measurement of Moisture Content

The moisture content of the flowers was measured daily using the Dean and Stark technique, whereby 15 g of flowers were boiled with toluene, and the amount of water separating from the condensed vapor was measured by volume.

2.3. Extraction Unit

Extraction was carried out using a Bench Scale ‘Speed’ Supercritical Fluid Extraction Unit in batch mode, with the material to be processed being placed in a 100 mL stainless steel vessel. Carbon dioxide was used as the solvent, with CO₂ being continuously circulated through the system. To collect the extract, the effluent from the extraction vessel was expanded to atmospheric pressure in a vessel whereby the extract was separated from the carbon dioxide.

2.4. Experimental Program

Two series of experiments were conducted, one without the co-solvent and one with the co-solvent, with 15 g of ylang ylang being initially charged to the vessel in each case with polypropylene wool put in place in both cases above the charge to hold it in place. The extraction curve was determined by weighing the vessel and sample every 15 min from the start of each experiment. Each extraction was carried out for 210 min by which time all of the extract was removed from the initial charge.

In the case of co-solvent extraction, 1.5 mL of ethanol was added to the ylang ylang flowers prior to extraction.

For the experimental program, as planned and conducted, a fractional factorial design was used due to the resources available. This type of design is useful in predicting the effects that the operating conditions—temperature, pressure, and flowrate—would have on the response—the percentage yield of extract (based on mass of fresh material used). This generated a planned series of experiments with 13 runs for the co-solvent runs and 8 runs without the use of a co-solvent. The pressure ranged from 80 to 120 bar, the flowrate from 1.5 to 3 L/min, and the temperature from 35 to 50 °C. Linear regression and Analysis of Variance (ANOVA) are the tools used in the factorial Design of Experiments (DOE) analysis.

The general model for this has the following form:

$$Y={\beta }_0+{\beta }_1{X}_1+\dots +{\beta }_p{X}_p+\epsilon$$

where Y is the response (which is the dependent variable), X_i is the predictor (which is the independent variable), and ε is the random error or noise.

It should be noted that the factorial DOE does not strictly require replicates to perform ANOVA, although replicates can improve the accuracy and consistency of the results.

2.5. Component Analysis

The gas chromatography–mass spectrometry unit was used to analyze the compositions of the various samples obtained through the SFE unit, using Agilent system model 6890N (Agilent, Santa Clara, CA, USA) as the gas chromatography unit, a mass selective detector model 5973N (Agilent, Santa Clara, CA, USA), and an Agilent Chem Station data system(Agilent, Santa Clara, CA, USA) to analyze the oil compositions. The gas chromatography column which was used was an HP-5ms fused silica capillary(Agilent, Santa Clara, CA, USA) with stationary phase using a 5% phenyl-methylpolysiloxane, with a film thickness of 0.25 μm and a length of 30 m as well as a 0.25 mm internal diameter.

The temperature program used was as follows: the starting oven temperature was held at 60 °C for a time of 1 min, ramped at 10 °C min⁻¹ to 180 °C, further ramped at 20 °C min⁻¹ to 280 °C, and held for 15 min. The temperature of the injector was sustained at 270 °C and the temperature of the detector was 300 °C. An amount of 1 μL of the sample was injected neat, with a split ratio of 1:10. The carrier gas for the GC-MS used was helium exhibiting a flowrate of 1.0 mL min⁻¹. For the mass spectrometry, the ionization electron energy was adjusted to 70 eV. The transfer line was maintained at 250 °C, the ion source was maintained at 230 °C, and the electron multiplier was set to 1568. The total run time was 33 min.

Most compounds were tentatively identified by comparing their mass spectra with those in the NIST 2004 and Wiley 275 libraries(Wiley Science Solutions, Hoboken, NJ, USA), as well as CAS Scifinderⁿ. Without the use of the correction factor, the compounds’ relative percentages (%) were calculated with reference to the normalization method.

3. Experimental Results

3.1. Moisture Content of Flowers

The average moisture content of the flowers charged to the extraction vessel was measured to be 53% wb.

3.2. Experiments without Co-Solvent

Eight experiments were carried out without the co-solvent, with the extraction curves being shown in Figure 1. The conditions for each run and the maximum extractions achieved are shown in

Table 1. Experimental results for supercritical fluid extraction yields.

Non-Co-Solvent Experimental Results Uncoded Variables
Runs	Pressure (bar)	Temperature (°C)	Flowrate (L/min)	Yield (wt %)
1	80	35	3	0.791
2	120	50	1.5	0.829
3	120	35	1.5	0.826
4	120	35	3	0.965
5	80	35	1.5	0.735
6	120	50	3	0.971
7	80	50	1.5	0.738
8	80	50	3	0.802

.

Figure 1 and Table 1 show the maximum extraction to be 0.97%, but this was at a pressure of 120 bar. The yield at the critical point was 0.8%.

The equation generated from ANOVA was

Yield % = 0.8321 + 0.0656A + 0.0501C + 0.0201AC + 0.0014BC − 0.0006ABC

where A is pressure, B is temperature, and C is flowrate.

Presentation of compositions.

A sample chromatogram (with no co-solvent) is shown in Figure 2 and the component compositions determined by GCMS across all experimental runs are presented in the range shown in

Table 2. Range of component compositions with no co-solvent.

Component	Retention Time (min)	Percentage Composition
Benzyl Alcohol	5.55	0.66–2.90
p-Cresol	6.12	0.65–1.67
Benzyl Acetate	7.45	0.58–14.41
Phenylethyl Acetate	8.73	0.66–1.34
Cinnamyl Acetate	9.41	0.50–13.24
Geranyl Acetate	10.39	0.39–0.95
Beta Caryophyllene	11.00	0.35–1.22
Gamma Cadinene	11.81	6.67–9.82
Isoeugenol	11.92	1.17–5.24
Farnesene	12.00	1.34–4.54
Ocimene	12.03	2.84–7.34
Cubebene	12.22	3.15–9.84
Bourbonene	12.93	0.71–2.07
Farnesol	14.67	2.03–5.40
Benzyl Benzoate	15.34	12.76–24.14
Farnesyl Acetate	15.84	2.97–9.68
Benzyl Salicylate	16.20	12.16–16.15

.

Table 2 shows the major components as benzyl benzoate, benzyl acetate, benzyl salicylate, cubibene, farnesol acetate, benzyl alcohol, and ocimene.

3.3. With Co-Solvent

Thirteen experiments were carried out using ethanol as a co-solvent, with typical extraction curves being shown in Figure 3. The conditions for each run and maximum extractions achieved are shown in

Table 3. Co-solvent experimental results for supercritical fluid extraction yield.

Co-Solvent Experimental Results Uncoded Variables
Runs	Pressure (bar)	Temperature (°C)	Flowrate (L/min)	Yield (wt %)
1	80	35	3	0.942
2	120	35	1.5	1.084
3	100	42.5	2.25	1.042
4	80	50	1.5	0.924
5	80	50	3	0.971
6	80	35	1.5	0.919
7	100	42.5	2.25	1.027
8	120	35	3	1.157
9	120	50	3	1.164
10	100	42.5	2.25	1.037
11	120	50	1.5	1.091
12	100	42.5	2.25	1.051
13	100	42.5	2.25	1.032

Figure 3 and Table 3 show the maximum yield as 1.164% at 120 bar of pressure and 50 °C.

The equation generated from ANOVA was given as follows:

Yield % = 1.03 + 0.0925A + 0.0270C + 0.0095AC + 0.0030BC − 0.0030AB

where A is pressure, B is temperature, and C is flowrate.

Presentation of Compositions.

A sample chromatogram (with co-solvent) is shown in Figure 4 and the component compositions determined by GCMS across all experimental runs are presented in the range shown in

Table 4. Range of component compositions with co-solvent.

Component	Retention Time (min)	Percentage Composition
p-Cresol	6.12	0.13–1.35
Benzyl Alcohol	5.55	0.31–2.58
Benzyl Acetate	7.45	0.55–12.65
Geranyl Acetate	10.39	0.68–1.49
Cinnamyl Acetate	9.41	8.06–15.21
Beta Caryophyllene	11.00	0.30–2.85
Cubebene	12.22	0.70–15.45
Gamma Cadinene	11.81	0.23–22.57
Isoeugenol	11.92	0.47–7.73
Ocimene	12.03	0.25–5.84
Farnesene	12.00	1.75–16.52
Bourbonene	12.93	1.06–2.04
Farnesol	14.67	3.21–5.84
Benzyl Benzoate	15.34	17.43–36.62
Farnesyl Acetate	15.84	1.82–15.13
Benzyl Salicylate	16.20	7.50–18.16

.

Table 4 shows the major components as benzyl benzoate, gamma cadinene, benzyl salicylate, benzyl acetate, cubebene, farnesyl acetate, and farnesene.

4. Discussion

4.1. Extraction Curves without Co-Solvent

The extraction curves without the co-solvent, shown in Figure 1, show similar-shaped trends with changing variables, with maximum yields being achieved after about 3.5 h. Figure 1 shows a slowing of the rate of extraction between 30 and 60 min, with the curve steepening before dropping off after about 3 h. This is a little different from SFE curves at higher pressures, which are more rounded [12]. This is probably because the lower pressures give rise to slower penetration of the carbon dioxide through the bed. In addition, the maximum yield at 0.97% was much lower than those reported at higher pressures, 1.8% [12], and the extraction times to completion were higher. It is appropriate to note that, in both investigations, the ylang ylang flowers were sourced from the same tree, so the comparison is direct.

The results of the factorial DOE indicated that pressure and flowrate were the only factors with a significant effect on percentage yield, consistent with the findings of [12]. This aligns with the results observed with the relatively low yield at the critical point of 0.79% increasing with pressure and flowrate to 0.97% at 120 bar.

4.2. Extraction Curves with Co-Solvent

The extraction curves with the co-solvent shown in Figure 2 are a little different from those without the co-solvent in that the rate of extraction is much higher in the first hour, much more like the curves at higher pressures. In fact, after about 1 h, the extraction is almost double that without the co-solvent.

The yield at the critical point with the co-solvent was 0.94%, with this increasing to 1.16% at 120 bar and 50 °C. This was again confirmed by the results of the DOE indicating pressure and flowrate as the only factors with significant effects on the response.

4.3. Comparison of Extract Compositions

The extracts from the experiments with both no co-solvent and co-solvent possessed similar components, mainly benzyl benzoate, benzyl acetate, benzyl salicylate, and cubebene. However, the concentrations in the extracts using the co-solvent tended to be higher. Scents of the various compounds in perfumes were classified by [15] to be alcohols, aldehydes, esters, ether, phenols, and hydrocarbons, and components from all of these classifications are present in the extracts, except for the ether, p-methylanisole. When compared to the reported chemical composition of steam-distilled ylang ylang oils as quoted by [16], many of the components are common, but there are differences such as the lack of linalool in the oils from the experimental program and the lower values of farnesene. This, however, is offset by the much higher values of the benzyl compounds. Whether the difference in compositions gives rise to a better perfume can only be decided by making up final perfumes and carrying out the relevant tests for commercial acceptance.

4.4. Comparisons with SFE at Higher Pressures

As may be expected, the yields at around and just above the critical point are somewhat lower than those reported at higher pressures, where the yield was as high as 1.8% at 300 bar and 45 °C [12]. The compositions of the extracts were, however, somewhat different with values up to ~25% of eugenol and nerol for the higher pressures. It is worth noting that the oils collected at around the critical point are of higher quality than those obtained at pressures between 200 and 500 bar, which result in the extraction of mainly oleoresins [17]. The comparison may be deemed reasonably exact as the flowers in both investigations were from the same ylang ylang tree.

4.5. Potential Commercialization

Since extraction using carbon dioxide around the critical point was deemed to be technically feasible, producing what could be a good-quality oil, it is recommended that

(1)

Larger amounts of extracts be produced for the purpose of making up potentially commercial perfume samples for evaluation and market testing.

(2)

An economic feasibility evaluation be carried out to determine the capital and operating costs. This should then be compared to the costs of operating the traditional steam distillation extraction to evaluate if extraction at or around the critical point is commercially viable.

5. Conclusions and Recommendations

Extraction of the essential oil of ylang ylang using carbon dioxide at or around the critical point is feasible with yields of 0.8 to 1%, and the yield was observed to increase to 1.2% using ethanol as the co-solvent. Although the yields are lower than those by extraction at higher pressures, they are said to be preferable as outlined by [17]; further analyses are required to verify this quantitatively. Some of the major components tentatively identified were benzyl benzoate, benzyl salicylate, benzyl acetate, and cubibene. It is recommended that an economic feasibility be carried out to compare costs of production with the traditional process of steam distillation.