FRACTIONAL DISTILLATION

What is fractional distillation?

According to Chen et al. (2025), the application of vacuum fractional distillation (VFD) to separating the constituents of essential oils “has rarely been reported”.

How was kaffir lime oil fractionated in one study?

Warsito et al. (2017) studied kaffir lime oil, from Citrus hystrix DC., obtained from Ngunut, Tulungagung, East Java. According to the authors, the distilled oil, from mixed twigs and leaves, was composed of 20 compounds, with five main components: β-citronellal (46.40 %), L-linalool (13.11 %), β-citronellol (11.03 %), citronellyl acetate (6.76 %) and sabinene (5.91 %).

Fractional distillation was performed on 2 L of this oil with a PiloDist 104-VTU under reduced pressure, with a column 2 m long and 120 stages, monitored by computer and heated by a heating jacket. It was repeated with reflux ratios of 10/10, 20/10 and 60/10, the volume of each collected fraction was measured, and the chemical composition was analysed by GC-MS. The pressures differ between two passages of the paper: the abstract gives 5 and 10 mBar, the methods section 0, 5 and 10 mbar.

Which conditions do the authors give as optimal?

The paper gives two different reflux ratios for the optimum, in two different passages. In the abstract and in the “What this study adds” section, the optimum conditions for fractional distillation were obtained at 5 mBar with a reflux ratio of 10/10. In the results, the optimum pressure for each treatment was determined from the distribution pattern of two components, ß-citronellal and L-linalool, and the optimum conditions applied were a reflux ratio of 20/10 at 5 mbar, the values also given in the caption of Figure 6.

How were the components distributed across the fractions?

According to Warsito et al., kaffir lime oil is composed of oxygenated monoterpenes and hydrocarbon monoterpenes with only a small difference in boiling point. When the oil was heated, the temperature increased gradually from fraction 1 to fraction 9, in line with the total ionic chromatogram profile. The first fraction (F1) was composed of components with a lower molecular mass (boiling point), and subsequent fractions such as F-5 or F-8 of components with a higher boiling point. β-Citronellal and L-linalool were distributed from fraction 1 to fraction 9, hydrocarbon monoterpenes only in fractions 1 to 4, while oxygenated monoterpenes dominated fractions 5 to 9. According to the authors, early fractions “might produce” sabinene at 19.83 % (“increased by 3 times”), and β-pinene, β-myrcene (“β-micren” in the source) and trans-β-ocimene, originally very low in kaffir lime oil, “could reach” 6.99 %, 7.69 % and 7.0 % (“increased by 5-fold”). In their conclusion, the highest levels were 84.86 % for β-citronellal (fraction 7), 20.13 % for L-linalool (fraction 5) and 19.83 % for sabinene (fraction 1); in fraction 9, terpinen-4-ol (“4-terpeneol” in the source), β-citronellol and citronellyl acetate were at 7.16 %, 12.27 % and 5.22 % respectively.

How do the authors relate the order of the fractions to boiling points?

Of fractionation at the optimum conditions, the authors write that the highest level of β-citronellal, in fraction 7, was 84.86 % (“increased by 38.46%”), against 46.40 % in kaffir lime oil, and that sabinene reached its highest level, 19.83 %, in fraction 1 (“increased by 13,92%”), against 5.91 % in kaffir lime oil. According to the authors, this indicates that sabinene has the lowest boiling point among the major components of kaffir lime oil. They give the retention-time sequence of the five major components as sabinene, L-linalool, β-citronellal, citronellol and β-citronellyl acetate, a sequence which, according to them, illustrates the boiling point of the compounds. Apart from sabinene, the four other major components contain oxygen, so that according to the authors it was “possible” for them to form hydrogen bonds that “could” result in a higher boiling point. The authors give 1.52 % of terpinen-4-ol in kaffir lime oil. Terpinen-4-ol, β-citronellol and citronellyl acetate appeared in the final fraction, fraction 9, at 7.16 % (“increased by 5.64%”), 12.27 % (“increased by 1.24%”) and 5.22 % respectively. According to the authors, this indicates that all these compounds have the highest boiling point, which they attribute to the two components containing the highest amount of oxygen and able to form hydrogen bonds.

What did vacuum fractionation separate in an oregano oil?

Rostro-Alanis et al. (2019) fractionated an oregano essential oil from Poliomintha longiflora A.Gray, obtained by steam entrainment. The paper also describes plant material collected 15 days after the rain in May 2018, during and 10 days after flowering, composed of leaf, flower and stem in a proportion of 90:9:1; it does not state that the fractionated oil came from it.

The fractions were obtained with a fractional distillation system whose column is equivalent to 20 theoretical plates, at a pressure of 5 Torr, over a distillation temperature range of 82 to 140 °C, with about 20 min between the collection of each fraction. The first fraction started to distil at 82 °C and the last at 140 °C, after which undistilled oil was obtained; five fractions resulted: F1, F2, F3, F4 and the undistilled oil (Unoil).

How do the fractions differ?

According to the authors, apart from colour, there are no great differences between the fractions, the undistilled oil and the whole oil (Ooil). At 20 °C, specific gravity was 0.842 g/mL for F1, 0.884 for F4, 0.927 for the undistilled oil and 0.778 for the whole oil, and refractive index 1.47, 1.51, 1.51 and 1.48 respectively. F1 and F4 are colourless, the undistilled oil brown and the whole oil light yellow. As percentage of mean peak area from triplicate determination, monoterpene hydrocarbons account for 95.73 % (F1), 97.75 % (F2), 83.94 % (F3), 2.71 % (F4), 1.91 % (undistilled oil) and 77.65 % (whole oil). In the same order, oxygenated monoterpenes account for 1.51, 1.77, 7.32, 65.11, 68.08 and 15.84 %, and sesquiterpene hydrocarbons are not detected in F1 and F2, then account for 3.31, 25.12, 22.14 and 5.03 %. In the same order, o-cymene accounts for 47.96, 53.97, 38.14, 1.31, 0.973 and 39.13 %, and γ-terpinene for 15.59, 24.43, 40.57, 1.40, 0.94 and 22.34 %. Thymol is not detected in F1, F2 and F3, then accounts for 5.08, 3.77 and 1.71 %; carvacrol is not detected in F1 and F2, then accounts for 4.58, 60.03, 64.31 and 12.60 %. According to the authors, the compounds identified in the oregano oil lie in a volatility range of 100–325 °C and are classified as semivolatiles, and their presence or absence in the fractions is determined by the time and temperature range of the process. The most volatile terpenes of this group are present in F1 and F2, while the less volatile are present in F3 and F4. o-Cymene and γ-terpinene, present in different amounts in F1 to F3, are almost insignificant in F4 and in the undistilled oil; less volatile terpenes such as carvacrol, trans-caryophyllene and α-humulene were detected in the undistilled oil. The whole oil contains all the terpenes found in the fractions and in the undistilled oil, in different proportions. The authors note the absence of oxidised compounds such as aldehydes and ketones, which they read as indicating a self-protection of the oil against oxidation.

How was a rosemary oil fractionated under vacuum?

Silvestre et al. (2019) fractionated a commercial rosemary oil, from Salvia rosmarinus Spenn. (Rosmarinus officinalis L. in the study), obtained by steam distillation of rosemary leaves grown in South Brazil; the oil was kept stored for 3 months in a refrigerated chamber (5–8 °C). The fractional distillation column consisted of 2 stages and a 250 ml flask in the base, which was the reboiler (bottom of the column); the packing was glass Raschig rings with a diameter of 8 mm, and the column was 100 cm in total height and 3 cm in diameter. The system was coupled to a vacuum pump with a maximum vacuum of 10 kPa; the 150 W electrical heating mantle in the bottom of the column was operated in PWM (pulse width modulation) mode, and PT100 sensors monitored the temperature between stages, at the top and at the bottom of the column. Distillation tests were carried out with samples collected from the top, the flask and the first stage, with 120 mL of raw oil in each batch, the column operating under vacuum at an absolute pressure of 10 kPa (the limit of the vacuum pump). The experiments were carried out in triplicate, and stage tests were performed at time intervals of 30 min. Both the raw oil and the fractions were analysed by GC/MS for qualitative analysis and by GC-FID for quantitative analysis, with a column temperature of 60 °C for 8 min, then raised to 180 °C at 3 °C/min and finally to 230 °C at 20 °C/min. Quantification used 1-octanol as internal standard, 755 µg of 1-octanol being injected in each analysis.

What did the raw rosemary oil contain?

According to GC/MS analysis, the raw oil presented a chemotype rich in α-pinene, 1,8-cineole, verbenone and camphene, which the authors call the pineno-cineol chemotype. By quantitative analysis, α-pinene (37.26 wt%) and 1,8-cineole (26.24 wt%) were the major compounds, followed by verbenone (5.53 wt%) and camphene (5.05 wt%).

What did the authors’ simulation predict?

Through a software simulation, the authors estimated the boiling temperature of the main compounds of rosemary oil, which made it “possible” to evaluate “roughly” the distillation pattern of the raw oil. In this estimate, the most volatile components were α-pinene and camphene, having higher vapour pressures, with a boiling temperature of about 75–80 °C at 10 kPa. The least volatile were verbenone and camphor, with almost the same boiling point near 130 °C at 10 kPa; from this simulation, the authors expect that these compounds “will likely remain at the bottom”, not distilling through the column or, if distilling, in a very small amount.

What did fractionation by temperature give?

In the samples collected at the top, monoterpenes such as camphene, β-pinene, sabinene and p-cymene presented appreciable recoveries (higher than 30 %) in the first temperature range (up to 70 °C), while other classes had smaller recoveries: 19.83 % for 1,8-cineole, 1.48 % for verbenone and 6.81 % for linalool. At 125 °C, α-pinene, camphene, β-pinene, myrcene, limonene, 1,8-cineole, p-cymene and 3-carene were nearly completely distilled (recoveries higher than 97.74 %); only γ-terpinene presented a lower recovery (89.15 %), and aldehydes (neral) were not detected at the top. Alcohols and ketones were the components with the highest content retained at the bottom of the column. At 125 °C, verbenone had the highest concentration, 24.42 wt%, more than four times its concentration in the raw oil (5.53 wt%), followed by β-caryophyllene with 11.77 wt% and neral with 11.63 wt%; verbenone tended to concentrate at the bottom, although a small amount was distilled and collected at the top. β-Caryophyllene, a sesquiterpene, remained at the bottom of the column and was present at more than 5 times its initial concentration, from 2.26 wt% in the raw oil to 11.77 wt% at the bottom at 125 °C. The only aldehyde, neral, was not distilled and remained at the bottom of the column, its content increasing from 2.23 wt% in the raw oil to 11.63 wt% at the end of the process; linalool went from 2.36 wt% in the raw oil to 4.41 wt%. Two compounds of the bottom fraction, verbenol and caryophyllene oxide, were not detected in the raw oil; according to the authors, their presence “could” be due to thermal effects (degradation, isomerization) or to concentrations in the raw oil lower than the limit of detection of the GC/MS and GC/FID systems. The paper words this in two ways: its conclusion says it “was unclear” whether the origin of such compounds was thermal degradation or a limitation of the GC/MS and GC-FID systems, while its abstract states that “there was no evidence of thermal degradation in any of the experiments”.

What did fractionation by time in the first stage give?

In the fractionation by time with collection in the first stage, camphene, β-pinene, sabinene, myrcene, limonene and p-cymene were mainly distilled in the first stage, leaving little or almost no amount in the flask, with recoveries greater than 90 % for some of these terpenes. After 90 min of processing, 90.37 % of the hydrocarbon terpenes and 60.71 % of the ethers (1,8-cineole) had been distilled. Alcohols, ketones and aldehydes were retained mostly in the flask, “probably” due to intermolecular interactions according to the authors; at the end of the process, 75.15 % of alcohols, 85.45 % of ketones and 98.37 % of aldehydes remained in the bottom of the column.

What do the authors conclude?

In their conclusion, the authors write that through a vacuum distillation process using a packed column “it was possible” to separate rosemary essential oil into components of different chemical classes: oxygen-containing classes such as alcohols and ketones remained at the bottom, separated from the hydrocarbon terpenes, which were mostly collected at the top stage. According to them, an additional stage would be necessary in the separation by stages to remove more hydrocarbon terpenes from the raw oil without removing the other chemical classes from the bottom, and separating the bottom fraction “would probably require” a column with more stages, because the relative volatility of these compounds is similar.

How was an orange terpene oil fractionated?

Ambrosio et al. (2021) used a commercial citrus essential oil, Brazilian Orange Terpenes (BOT), a by-product of orange juice processing supplied by a factory in São Paulo State, Brazil.

The oil was fractionated in a batch vacuum distillation system, a BR–800 micro fractional distillation apparatus with a Teflon spinning band column (20 cm × 7 mm) of 30 theoretical plates covered with a silvered vacuum jacket, an automatic reflux valve (solenoid) at the top of the column, and a condenser. 20 g of BOT oil were weighed into a 50 mL round-bottom flask, and the initial temperatures in the boiling flask (T1) and at the top of the fractionating column (T2) were 27 and 23 °C. Distillation started once T1 reached 54 °C, given as the boiling temperature of BOT, and heating was held at the point where T1 and T2 stabilised, about 40 min later. The first fraction was collected at T1 = 90 °C, T2 = 51 °C and 17.5 mbar, the following fractions at regular 10 min intervals, and the fraction remaining in the boiling flask was considered the bottom fraction, “the terpene less one”. Four fractions, F1 to F4, were obtained, with yields related to the initial oil of 15.2, 34.0, 20.4 and 29.0 % respectively. F1 was the most volatile fraction and F4, the fraction remaining in the boiling flask, the least volatile.

What did the BOT fractions contain?

In Table 1 of Ambrosio et al. (2021), the relative amount of limonene, by internal standardisation with n-tetradecane, is 11.06 (non-polar DB-5MS column) and 16.41 (polar DB-WAX column) in F1, 10.86 and 15.80 in F2, 11.81 and 15.66 in F3, and 0.13 and 0.08 in F4. According to the authors, the fractional distillation process “could” separate BOT oil into four fractions, F1 to F4 (F4 being the bottom fraction), with a collection interval of 10 min. Their discussion gives for the first three fractions 10.86–16.41 ≈ 86.95–93.20 % and for F4 a very low relative amount (0.13/0.08 ≈ 0.96/0.61 %) of a compound whose figures match the limonene row of Table 1. According to their discussion, minor compounds of BOT oil (carvone, cis-carveol, trans-carveol, cis-p-mentha-2,8-dien-1-ol and trans-p-mentha-2,8-dien-1-ol) were detected at relatively higher amounts in F4 than in F1, F2 and F3.

What did vacuum fractional and molecular distillation produce from Litsea cubeba oil?

In the study by Chen et al. (2025), fruits of Litsea cubeba (Lour.) Pers. were collected in Anshun City, Guizhou Province, China, in mid-June 2022 and dried in the shade at room temperature. 5.0 kg of fruits were hydrodistilled for 12 h in a Clevenger-type apparatus with 15.0 L of water, and the upper oil layer was collected and dried with sodium sulfate. The oil was obtained by hydrodistillation with a yield of 3.5 %. In total, 36 compounds were detected by GC-MS across all the oils studied, and the main components of the starting oil were β-citral (14.69 %), citronellal (14.00 %), limonene (11.40 %), linalool (9.25 %), α-citral (7.56 %) and 1,3,4-trimethyl-3-cyclohexene-1-carboxaldehyde (7.18 %).

How were the fractions obtained?

For VFD, 231.40 g of oil were placed in a distillation flask fitted with a fractionating column, with cooling water at 3 °C, and heating started once the pressure had stabilised at 0.4 kPa. Three distillates were collected: a low-boiling-point fraction (VFD1) at 38 to 44 °C, a medium-boiling-point fraction (VFD2) at 44 to 76 °C and a high-boiling-point fraction (VFD3) at 76 to 97 °C.

The double molecular distillation was carried out on a 2-inch wiped-film molecular evaporator. In the first stage, the distillation temperature was 55 °C, the cooling temperature 3 °C, the vacuum 0.6 kPa, the wiper speed 350 rpm and the feed rate 4 mL/min; fraction MDH1 then fed the second stage, run at 65 °C, 3 °C, 0.5 kPa, 400 rpm and 4 mL/min. VFD gave fractions VFD1, VFD2 and VFD3 with yields of 24.36, 37.45 and 26.88 %; the first molecular distillation gave MDL1 and MDH1 with 20.47 and 19.32 %, and the second, fed with MDH1, gave MDL2 and MDH2 with 9.22 and 48.44 %.

How did the composition change?

As mean area percentage of three independent injections, limonene accounts for 11.40 % of the starting oil, 27.45 % of VFD1, 13.18 % of VFD2 and 0.19 % of VFD3. In the same order, citronellal accounts for 14.00, 3.26, 14.04 and 16.23 %, β-citral for 14.69, 2.39, 17.01 and 30.03 %, and α-citral for 7.56, 0.53, 6.06 and 17.04 %. In the VFD fractions, the relative content of monoterpene hydrocarbons decreased from 77.92 % in VFD1 to 23.28 % in VFD2 and 0.35 % in VFD3, while that of oxygenated monoterpenes increased: 12.69 % in VFD1, 65.56 % in VFD2 and 83.17 % in VFD3. In the molecular distillation, the monoterpene content of the light phase fractions decreased from 76.34 % in MDL1 to 57.89 % in MDL2, while the percentage of oxygenated monoterpenes increased from 71.72 % in MDH1 to 79.75 % in MDH2. Compared with the starting oil, the authors write that “oxygenated compounds” in VFD3 and MDH2 “increased by 27.07 and 23.65%” and that “monoterpenes decreased by 28.95 and 22.94%”. Their table gives oxygenated monoterpenes at 56.10 % in the starting oil, 83.17 % in VFD3 and 79.75 % in MDH2, and monoterpene hydrocarbons at 29.30 %, 0.35 % and 6.36 %. The contents of sesquiterpenes in VFD3 and MDH2 were about twice as high as in the starting oil. The relative contents of β-citral and α-citral increased from 14.69 and 7.56 % in the starting oil to 30.03 and 17.04 % in VFD3, and to 22.64 and 10.82 % in MDH2. The seven distillates lay at a certain distance from the starting oil in the authors’ Figure 1, which they read as indicating that VFD and MD “could” effectively separate the compounds of the oil. According to the authors, overall, VFD and MD greatly increased the relative contents of oxygenated monoterpenes and decreased those of monoterpene hydrocarbons in VFD3 and MDH2, results which indicated that VFD and MD changed the composition and relative content of the oil.

How did one study separate citrus oils by wiped-film molecular distillation?

Yang et al. (2021) separated crude citrus essential oils on a lab-modified wiped-film molecular distillation apparatus. The crude oil, 300 ml, was fed at 2.5 ml/min under an operating pressure of 6 × 10⁻³ mbar; the film-forming system was set at 250 rpm, and the evaporation and condensation temperatures were kept at 65 and 5 °C. According to the authors, colourless, low-molecular-weight volatiles reached the condenser plate while heated under vacuum and were collected in the distillate stream, while high-molecular-weight components returned to the heating plate, giving an enriched residue stream.

What does an EFSA opinion on “star anise terpenes” report?

A 2023 opinion of the EFSA FEEDAP Panel concerns “star anise terpenes”, produced by fractional distillation of star anise oil obtained from the fruit (or seeds) and leaves of Illicium verum Hook.f., for all animal species. According to the opinion, the star anise oil from which the additive is obtained is produced from the fruits and leaves of this species, with the mention “(sourced from China)”. The opinion describes the additive as a pale yellow to yellow mobile liquid with a characteristic aroma (sweet spicy anise). In five recent batches, all originating from China, the refractive index (20 °C) ranged from 1.4802 to 1.4907 (specification 1.475–1.560), the density (20 °C) from 882 to 898 kg/m3 (specification 860–930 kg/m3), and the optical rotation (20 °C, three batches) from 3.4° to 4.1° (specification 2.0–15.0).

What do the specifications of the additive contain?

According to the opinion, four components contribute to the specifications of star anise terpenes, with trans-anethole selected as the phytochemical marker. Expressed as percentage of gas chromatographic peak area, the specifications and measured values are: estragole < 25 (mean 23.59, range 23.3–24.3), limonene 5–25 (22.79; 22.7–23.0), linalool 1–15 (10.18; 9.9–10.6) and trans-anethole 5–15 (11.47; 11.3–11.9). The number of batches differs within the opinion: its text states that the analysis of three batches by GC-FID showed compliance with these specifications, and the note on the total refers to the three batches analysed, while the table caption gives a batch-to-batch variation based on four batches. The applicant’s GC-MS characterisation of four additional batches detected up to 88 constituents, 83 of which were identified and accounted on average for 98.6 % (97.9–99.0 %) of the GC area, the four specified compounds accounting for about 51.5 % on average (34.1–64.0 %).

According to the opinion, analysis of the seven batches of star anise terpenes confirmed the presence of estragole in all batches (13.5–24.4 %), and safrole was not detected (LOD 0.001 %). The FEEDAP Panel notes that the concentration of estragole is enriched 4- to 12-fold in star anise terpenes compared with the star anise oil from which the additive is obtained.

What does the opinion conclude?

The opinion describes estragole as a compound with experimentally proven genotoxicity and carcinogenicity in rodents, as reviewed by the EMA in 2021, and as one of the main components of the additive. According to the opinion, the data submitted by the applicant showed that the production process of star anise terpenes results in an enrichment of estragole from star anise oil. The opinion states that the additive contains up to 25 % estragole by specification and is obtained by a manufacturing process that results in the enrichment of this genotoxic carcinogen, which “is not in line with” the principles of the general approach to assessing botanical preparations containing genotoxic and/or carcinogenic compounds used as feed additives. The FEEDAP Panel therefore considered it inappropriate to perform an assessment of the safety and efficacy of star anise terpenes for its use as a feed additive. The opinion states that any manufacturing processes of botanical feed additives “should avoid selective extraction and enrichment of genotoxic and/or carcinogenic substances and should aim at the reduction of these substances”.

What does the opinion say about impurities?

The applicant referred to the “periodic testing” of some representative flavourings premixtures for mercury, cadmium, lead, arsenic, fluoride, dioxins and polychlorinated biphenyls, organochlorine and organophosphorus pesticides, aflatoxins (B1, B2, G1, G2) and ochratoxin A, but no data were provided on the presence of these impurities. Since the additive is produced by fractional distillation of star anise oil, the opinion considers the likelihood of any measurable carry-over of all these elements low, except for mercury.

References

• Warsito W., Palungan M.H., Utomo E.P., « Profiling study of the major and minor components of kaffir lime oil (Citrus hystrix DC.) in the fractional distillation process », The Pan African Medical Journal 2017;27:282, doi:10.11604/pamj.2017.27.282.9679 — https://www.ebi.ac.uk/europepmc/webservices/rest/PMC5660901/fullTextXML
• Rostro-Alanis M.J., Báez-González J., Torres-Alvarez C., Parra-Saldívar R., Rodriguez-Rodriguez J. et al., « Chemical Composition and Biological Activities of Oregano Essential Oil and Its Fractions Obtained by Vacuum Distillation », Molecules 2019;24(10):1904, doi:10.3390/molecules24101904 — https://www.ebi.ac.uk/europepmc/webservices/rest/PMC6571842/fullTextXML
• Silvestre W.P., Medeiros F.R., Agostini F., Toss D., Pauletti G.F., « Fractionation of rosemary (Rosmarinus officinalis L.) essential oil using vacuum fractional distillation », Journal of Food Science and Technology 2019;56(12):5422-5434, doi:10.1007/s13197-019-04013-z — https://pmc.ncbi.nlm.nih.gov/articles/PMC6838235/
• Ambrosio C.M.S., Diaz-Arenas G.L., Agudelo L.P.A., Stashenko E., Contreras-Castillo C.J. et al., « Chemical Composition and Antibacterial and Antioxidant Activity of a Citrus Essential Oil and Its Fractions », Molecules 2021;26(10):2888, doi:10.3390/molecules26102888 — https://www.ebi.ac.uk/europepmc/webservices/rest/PMC8152727/fullTextXML
• Chen Q., Lin X., Zhang J., Liu S., Zou Z. et al., « Chemical Compositions and Antibacterial Activities of Litsea cubeba Essential Oil and Its Distillates Prepared by Vacuum Fractional Distillation and Molecular Distillation », Journal of Agricultural and Food Chemistry 2025;73(12):7270, doi:10.1021/acs.jafc.4c11955 — https://www.ebi.ac.uk/europepmc/webservices/rest/PMC11951146/fullTextXML
• EFSA Panel on Additives and Products or Substances used in Animal Feed (FEEDAP), « Safety and efficacy of a feed additive consisting of a fraction of the essential oil from the fruit and leaves of Illicium verum Hook.f. (star anise terpenes) for use in all animal species (FEFANA asbl) », EFSA Journal 2023;21(10):e08341, doi:10.2903/j.efsa.2023.8341 — https://www.ebi.ac.uk/europepmc/webservices/rest/PMC10588559/fullTextXML
• Yang F., Zhang H., Tian G., Ren W., Li J. et al., « Effects of Molecular Distillation on the Chemical Components, Cleaning, and Antibacterial Abilities of Four Different Citrus Oils », Frontiers in Nutrition 2021;8:731724, doi:10.3389/fnut.2021.731724 — https://www.ebi.ac.uk/europepmc/webservices/rest/PMC8440794/fullTextXML
• Govaerts R. (ed.), World Checklist of Vascular Plants (WCVP), Royal Botanic Gardens, Kew — https://sftp.kew.org/pub/data-repositories/WCVP/wcvp.zip