Light-driven (cross-)dimerization of terpenes as a route to renewable C 15 – C 30 crudes for fuel and lubricant oil applications†

: Non-fossil hydrocarbons are desirable for transport fuels and lubricant oils to reach a fossil carbon neutral economy. Herein, we show the production of such end-products from crude raw materials via the photosensitized dimerization of terpenes. Terpenes are hydrocarbons originating from renewable sources, such as forestry, industrial bio-waste and photosynthetically active microorganisms. Under irradiation at 365 nm, we observed high conversions of terpenes into dimers ( e.g. 96.1 wt. %, 12 h for α-phellandrene), and remarkable results were obtained using simulated and natural sunlight (90.8 wt .% and 46.6 wt .%, respectively, for α-phellandrene). We show that the lower reactivities of some isomeric monoterpenes could be overcome by a cross-photodimerization with α-phellandrene. We also utilized the cross-photodimerization approach to obtain C 15 and C 30 products, combining mixtures of isoprene, monoterpenes and sesquiterpenes. Hydrogenation of the terpene dimers gave materials with physical properties suitable as high


Introduction
According to the International Renewable Energy Agency (IRENA), liquid biofuels will represent 20% of the total final energy consumption of the transport sector in 2050. 1 The need to develop alternative fuels applies in particular to the shipping and aviation sectors, for which IRENA considers advanced biofuels to be a key solutioni.e.biofuels from nonfood sources. 2 At the same time, other fossil derived products such as solvents and lubricant oils must also be replaced to meet sustainability targets. 3 an example of biomass from a nonedible source, terpenes are major components of essential oils from plants and they can also be extracted from conifer resins (Fig. 1). 4 The main component in the leaf oil of Schinus mole is α-phellandrene 1 (45.7 %), and the eucalyptus oil from some species (e.g.6][7] Other monoterpenes, such as pinenes, are obtained industrially as byproducts of paper pulping processes. 8β-Pinene can be converted to myrcene 2 through pyrolysis, 8,9 while ocimene 3 can be obtained photochemically from α-pinene. 102][13][14][15][16][17][18][19][20] These routes utilize sugars derived from lignocellulosic biomass, or CO2 and water as feedstock, making terpenes potential candidates for sustainable syntheses of biofuels and lubricants.The importance of terpenes, beside the fact that their use does not induce any food security threats, stems from their chemical similarity to fuel molecules derived from crude oil, making terpenes suitable as drop-in surrogate for fossil fuels. 215][26][27][28][29] Such reported methods of dimerization of monoterpenes are acid or metal catalyzed thermal reactions.The two monoterpenes α-and β-pinene have also been used in self-photosensitized cycloaddition with biomass-derived cyclic enones, yielding spiro-compounds suitable as jet fuels. 30All these reactions rely on the reactivity of isolated as well as conjugated π-bonds of monoterpenes.Conjugated dienes have long been known to undergo triplet sensitized reactions promoted by organic photosensitizers. 31,32As such, the precursor molecule of terpenes, isoprene 4, undergoes triplet photosensitized dimerization to give [2+2], [4+2] and [4+4] photocycloaddition products under near-UV light. 33Since cycloalkanes are reported as high energy density combustion fuels, 34 our group has recently reported on a fruitful route to solar jet fuels, where the photobiologically produced isoprene is subject to photosensitized dimerization. 35Herein, we apply the photochemical dimerization approach to mono-and sesquiterpenes bearing conjugated diene, e.g.α-phellandrene 1, myrcene 2 and ocimene 3. The triplet photosensitized dimerization of 1 -3 is reported earlier, [36][37][38] yet, these reactions of 2 and 3 were carried out in dilute solutions in order to favor other intramolecular photoreactions in the triplet statei.e. the internal cyclization of myrcene and the E/Z isomerization of ocimene.On the other hand, the earlier reported triplet sensitized dimerization of α-phellandrene by using naphthalene as photosensitizer was very inefficient (conversion of 12% after 24 h, Scheme 1A). 38Naphthalene could be an inefficient photosensitizer due to a large difference in the triplet energies of α-phellandrene and naphthalene.Matching of the triplet energies of the molecule to be sensitized and that of the photosensitizer is an important factor in photosensitized reactions. 39For conjugated dienes, the triplet energy is around 50 kcal mol -1 and an efficient triplet photosensitizer should have a relatively higher triplet energy to enable effective energy transfer to dienes. 40onveniently, solar irradiation provides wavelengths down to the near-UV range, between 300-400 nm, which is equivalent to the energy of 71 -95 kcal mol -1 .Hence, it is possible to envision a photosensitizer that absorbs in the near-UV spectral region of sunlight and undergoes intersystem crossing, enabling the photosensitized dimerization of terpenes with conjugated diene moieties.
Aryl ketones are good candidates, and extending the π-conjugation in aryl ketones or changing the substituents have been shown to promote a redshifted absorption. 41,42This is the case for some benzophenone derivatives, such as the 1,1-dinaphthylmethanone 8, which is the key photosensitizer in our studies (Scheme 1).
Thereby, we report a route to renewable hydrocarbon crudes from the photosensitized dimerization of monoterpenes, using near-UV light and, ultimately, by using simulated and natural sunlight (Scheme 1B).We find that -phellandrene is especially reactive, and we utilize this in the cross-dimerization between α-phellandrene 1 and the less reactive monoterpenes or isoprene, giving, respectively C15 and C20 cross-dimers.We also describe the dimerization of sesquiterpenes present in ginger oil leading to C30 dimers.Finally, we present the physical properties of some of the hydrogenated crudes, finding that they are suitable as high energy density fuels and lubricant oils.

Dimerization and hydrogenation procedures
General (cross)dimerization procedure.Typically, a mixture of the terpene(s) to be dimerized and the photosensitizer was purged with argon (ca. 10 minutes) under stirring.For the cross-dimerization between α-phellandrene and isoprene, freeze-pump-thaw method was used to degas the mixture.Then the mixture was transferred to the suitable reaction setupaccording to the light source to be used (see below)and irradiated for the desired time.After the photoirradiation was completed, the unreacted starting material was removed by distillation under reduced pressure to afford a mixture containing the photosensitizer and the dimers.When required, the photosensitizer was removed by passing the mixture through silica gel column and using pentane as eluent.Further evaporation of pentane under reduced pressure yielded the pure mixture of dimers.

Dimerization under 365 nm irradiation, in quartz tube. A Rayonet Photochemical
Chamber Reactor RPR-200, from the Southern New England Ultraviolet Company, equipped with a set of 16 x 24W UV lamps at 365 nm was used.Photoreactions were initially carried out in 18 mL quartz test tubes , Ø 13 mm, containing a magnetic stirring bar to allow the sample to be stirred during the photoirradiation.The amount of samples was 2 -4 mL.The quartz tubes were capped with rubber septa.During the photoirradiation, the typical temperature inside the photoreactor was ~40 °C.
Dimerization under 365 nm irradiation, in custom-made setup (CMS).The reactions were carried out in different custom-made setups using a fluorinated ethylene propylene (FEP) tubing coiled around a template.Metal or plastic unions were used when two or more FEP tubing were required: CMS I (Fig. S1A, ESI) consisted of a FEP tubing (O.D. × I.D. : 3.18 mm × 2.1 mm) coiled around a water condenser, with total volume of the loop size of approx.20 mL; CMS II (Fig. S1B, ESI) was constructed to accommodate a larger volume of sample by using a FEP tubing with larger dimensions (6.35 mm x 7.94 mm) and a larger template (2 L jacketed beaker, Ø:130 mm; height: 280 mm), with loop size of approx.400 mL; CMS III (Fig. S1C, ESI) was constructed by using FEP tubing of same dimensions as in CMS I, but the template from CMS II, providing a total loop size of approx.120 mL.For the photoreactions, the required CMS was placed at the center position in the photoreactor (approx.8.5 cm from the lamps for CMS 1, and 2.0 cm for CMS's II and III).During the photoirradiation, the typical temperature inside the photoreactor was ~40 °C.Water was only used for cooling the CMS for reactions containing isoprene.

Dimerization under natural and simulated sunlight. A SS-F5-3A solar simulator, by
Enlitech, with a 300 W Xe lamp, was used.The reactions under natural and simulated sunlight were carried out in a flat version of the CMS's, hereby called CMS IV (Fig. S1D, ESI).In this case, the FEP tubing was shaped into a spiral and fixed over a flat wooden surface.The wooden surface was previously covered with reflective aluminum foil.The flat spiral containing the sample was placed under the Xe lamp at a distance to result in a light intensity of one sun equivalent (AM 1.5 G).We assured this equivalency by measuring the photocurrent produced in a reference photovoltaic cell (47.6 mA).Natural sunlight experiments were run in Uppsala, Sweden (see further details in the results and discussion section).

Hydrogenation of selected dimers.
A mixture of terpene dimers (30 -40 mL), PtO2 (50 mg, approx.0.15 wt%), glacial acetic acid (5 mL) and pentane (5 mL) was hydrogenated in a Parr hydrogenation glass reaction vessel.The reaction mixture was stirred under 6 bar H2 pressure for 2 days.During the hydrogenation, the pressure of the reaction vessel was checked several times and refilled with H2 gas to 6 bar until the pressure remained unchanged, indicating the completion of the reaction.Then the reaction mixture was passed through celite using pentane to elute the mixture.The remaining acetic acid was extracted with water, and the organic phase was passed through Na2SO4 and washed with pentane.Pentane was removed under reduced pressure to provide the hydrogenated dimers.Several hydrogenation batches were required to result in 60 -100 mL of hydrogenated samples.These crudes were further tested regarding their fuel and lubricant properties.

Characterization
General characterization.Samples were characterized by NMR spectroscopy and gas chromatographymass spectrometry (GC-MS).The 1 H and 13 C NMR spectra were recorded on a JEOL (400YH magnet) Resonance 400 MHz spectrometer.Chemical shifts δ are reported in ppm, and coupling constants J in Hz. 1 H NMR and 13 C NMR chemical shifts are referenced to the residual solvent signal (CDCl3, 1 H 7.26 ppm and 13 C 77.16 ppm).The GC-MS system used was composed by an Agilent 7890A GC, equipped with a HP-5 capillary column (30 m x 250 μm x 0.25 μm), and an Agilent 5975 mass selective detector (MSD).Helium was used as the carrier gas.Isolated yields of dimers were determined gravimetrically.Quantitative NMR was used in some cases for determining conversions.

Measurement of fuel and lubricant oil properties. Important physical properties of fuels
and lubricant oils were measured for hydrogenated dimers produced in this study, following international and national standard methods.The net heats of combustion (NHOC) were measured according to the ASTM D4809 method.Further property measurements followed Swedish standard methods that are based on international methods: kinematic and dynamic viscosities, and densities were measured at different temperatures (form -20 °C to 100 °C) according to the SS-EN 16896:2016 method; pour point measurements were performed according to the SS-ISO 3016.

Computational methods
All calculations were carried out using the Gaussian 16 program package, 44 at the (U)M06-2X/6-311+G(d,p) level. 45,46Stationary points were characterized as minima through frequency calculations.The heats of combustion were calculated at the M06-2X/6-31+G(d,p) level using the method developed by Pahima et al. 23

Results and Discussion
The initial studies and optimization of the photosensitized dimerization reactions at 365 nm light, and simulated and natural sunlight are presented.Further, the relative reactivities of monoterpenes are discussed, followed by the outcomes of the cross-dimerization between α-phellandrene and other terpenes, and the dimerization of sesquiterpenes in ginger oil.To conclude, the fuel and lubricant oil properties of selected hydrogenated crudes produced in this study are reported, together with a preliminary assessment of the sustainability and commercial viability of the process.

Initial investigation and optimization of reactions:
On the basis of our previous study, 35 1,1-dinapthylmethanone 8 was used as photosensitizer.Additionally screening with other aromatic ketone derivatives was also performed for dimerization of myrcene in search of a better photosensitizer, however 8 was proved to be the best choice (Table S1, ESI).13]20 Their adiabatic first triplet state energies (E(T1)) were calculated at the (U)M06-2X/6-311+G(d,p) level and found to be lower than 55.5 kcal mol -1 , i.e. the E(T1) of 8 (Table 1 and Fig. S2, ESI).The lower E(T1)'s may enable an efficient energy transfer from the photosensitizer to the monoterpenes.The selected monoterpenes were initially photo-irradiated with 8 for dimerization in quartz test tubes and at 365 nm.The formation of dimers is confirmed by the parent ion m/z 272 in the GC-MS analyses, which indicate a C20H32 molecular formula (Fig. S5, S12 and S18, ESI).The chromatograms of α-phellandrene dimers reveal the presence of at least three major isomers, whereas there is a larger number of dimer isomers of myrcene and ocimene (Fig. 2).While ocimene and myrcene showed rather low conversions after 48 h of irradiation, α-phellandrene exhibited 82 % conversion after 24 h (Table 1, entries 2 -4).This might be a consequence of the endocyclic diene segment, which leads to a longer triplet lifetime compared to acyclic dienes, as the relaxation by twisting is hindered in the former. 47Despite α-terpinene 5 having similar structure to αphellandrene and suitable triplet energy, the reaction of 5 gave a much lower conversion to dimers (Table 1, entry 4).This finding can be correlated either to a preference for the reduction of 8 (Htransfer) leading to the aromatization of 5 and depletion of the photosensitizer, or to steric congestion at the termini of the diene moieties.The first explanation is supported by the observed increase of aromatic content in the reaction (signal at 7.1 ppm in the 1 H NMR spectrum, Fig S3, ESI) and formation of the alcohol derived from 8 (Fig. S4, ESI).A similar result was observed for the reaction of 5 with duroquinone. 37Since this side reaction causes photosensitizer depletion, the reactions with α-terpinene 5 were not further investigated.A similar reduction of 8 was not observed on the reaction with α-phellandrene (Fig. S5, ESI).We also performed control experiments without photosensitizer and the irradiated samples did not show formation of dimers, supporting the triplet photosensitized dimerization process (Fig. S7, ESI).As the most reactive monomer, α-phellandrene was used to assess the effects of varied photosensitizer loading for the reaction in quartz tubes (Table S2, ESI).Starting from 0.50 mol% the conversion to dimers dropped as the photosensitizer loading was decreased.However, the correlation was not linear and similar conversion was observed for three different loadings (0.125 and 0.250 mol%).Interestingly, even as low concentrations as 0.01 mol% enabled the dimerization (12.4 wt.% conversion, 12 h).From those results, 0.125 mol% was indicated to be the optimal loading.
The photoreactions were next carried out in an alternative setup constructed by coiling a FEP tubing onto a glass condenser (Fig. S1A, ESI).This setup, CMS I, provides a higher ratio of surface-area-to-volume exposed to light.We observed the expected increase of the conversion of monomers to dimers for the three monoterpenes 1 -3 (Table 1, entries 1-3, CMS I column).The most remarkable change was noted for the dimerization of α-phellandrene 1.A complete conversion to dimers (12 h, 0.2 mol% and 96 wt.%) was observed at much lower loading of 8 (0.2 mol%) compared to the reaction in quartz tube (12 h, 0.5 mol%, 77 wt.%).The conversion of 63 wt.% was observed after 6 h of photoirradiation with 0.2 mol% loading of 8. Unlike the results in quartz tube, decreasing the amount of 8 resulted in a lower conversion (76 wt.% in 12 h, 0.1 mol% of 8), making the 0.2 mol% the optimal loading in the CMS I.These results show higher efficiency than the earlier reported naphthalene photosensitized dimerization, in which a conversion of only 12 % was reached after 24 h. 38The conversion of myrcene 2 to dimers was increased to 59 wt.%, while there was negligible change in the yield of ocimene 3 dimers.Following these results, we attempted to run dimerization of α-phellandrene in 260 g scale setup by using FEP tubing with larger internal diameter (CMS II, Fig. S1B, ESI, internal diameter = 6.35 mm).A conversion of 99 wt% was observed after 24 h of photoirradiation (Table 1, entry 1.4, CMS I column).The longer reaction time could be attributed to the increased internal diameter of the FEP tubing used, which reduces the light penetration into the sample.
Based on the previously reported photodimerization of conjugated dienes in the presence of triplet photosensitizers, 33 the dimers reported on Table 1 are expected to be cycloalkenes resulting from [2+2], [4+2] and [4+4] photocycloadditions (Fig. 3).Baldwin et al. suggested structure for the dimers of α-phellandrene in the photosensitization promoted by naphthalene, in which the main dimers should be a [2+2] cycloadduct between the methyl substituted double bonds (Fig. 3, structure 1a), derived from the most stable biradical intermediate (Fig. 3, 1BR). 38anwhile, the dimers from reactions of acyclic monoterpenes are more complex in nature and the structures 2a -3l (Fig. 3) are assumed based on the possible outcomes from the aforementioned photocycloadditions.The myrcene dimers possibly consist of at least seven major isomers, while for ocimene this distinction is less straightforward.The 1 H NMR data of reported camphorene (product of the [4+2] cycloaddition, Fig. 3 structures 2g -2j) with myrcene dimers confirmed the presence of camphorene isomers in the mixture (Fig. S13, ESI). 24Further structure assignments are challenging as the 1 H NMR spectra for the mixture of dimers have several overlapping signals.However, the reasoning above strongly supports the notion that the product mixtures consist mainly of cycloalkene adducts.

Dimerization using simulated and natural sunlight:
The photosensitizer 8 exhibits a broad absorption band within the range of 300 -400 nm, 35,41 which might enable the dimerization to be carried out under simulated and natural sunlight.Therefore, mixtures of myrcene and αphellandrene with 8 were irradiated under simulated sunlight with an intensity of 1 sun equivalent (AM 1.5 G) and using a flat coiled FEP tubing (CMS IV, Fig. S1D, ESI).The amounts of 8 and the results are summarized on Table 2.The conversion of α-phellandrene to dimers was 90.8 wt.% and achieved after 12 h which is comparable to the experiments using 365 nm light (96.1 wt%, 12 h, CMS I).The conversion of myrcene under solar simulated light was also comparable to the conversion observed under 365 nm irradiation (Table 2, entry 2).Later, a mixture of α-phellandrene and 8 in the CMS IV was irradiated with natural sunlight for 18 h (Uppsala, Sweden 59°51'09.5"N17°39'19.9"E,approx.30 m above sea level on September 19 th and 20 th , 2020), which led to 46.6 wt.% conversion into dimers (Table 2, entry 3).
Some important aspects must be considered before assessing the relevance of this last result.First, the average yearly global horizontal solar irradiance measured for Arlanda-Stockholm (35 km south of Uppsala) is 3,319 MJ/m 2 /year, 48 which in is lower than the average in several other places located in lower latitudes worldwide (Table S3, ESI).Second, our experiment with sunlight was performed in mid-September, which accounts to an average solar irradiance that is half of the solar irradiance during summer months (Table S4, ESI). 48Finally, the sky was not entirely clear during the second day of reaction.Therefore, given these non-optimal conditions, a conversion of 46.6 wt.% is very promising, and one can expect higher conversions to α-phellandrene dimers during optimal year season and in locations with higher solar irradiance.

Less reactive monoterpenes:
Based on what is known from small cyclic and acyclic dienes, 47 myrcene and ocimene have shorter triplet state lifetimes compared to α-phellandrene, due to the former being acyclic conjugated dienes.Additionally, both myrcene and ocimene can undergo competitive side-reactions, a fact that may explain their modest reactivities.For instance, the triplet photosensitized internal cyclization of myrcene to 5,5-dimethyl-lvinylbicyclo[2.1.1]hexane9 (Fig. 4A) has been studied in solution, and the formation of this isomer was also observed in our experiments (18.3 wt%).As soon as the bicyclic structure 9 is formed, it can no longer partake in a photosensitized dimerization, because the reactive diene segment is lost and non-conjugated alkenes show higher triplet energies that are above the triplet energy of the photosensitizer 8. Indeed, the vertical triplet energy of 9 is 103 kcal mol -1 (calculated at the (U)M06-2X/6-311+G(d,p) level), much higher than that of 2 (76 kcal mol -1 ).The adiabatic triplet energy of 9 was not possible to calculate since the compound dissociates during the optimization.Therefore, the hypothesis was that the formation of 9 traps the myrcene that could have reacted in the photosensitized dimerization.In order to verify this hypothesis, we performed an epoxidation of myrcene to protect the isolated double bond and halt the internal cyclization of myrcene (Fig. 4A). 43The myrcene epoxide 10 was then mixed with the photosensitizer 8 and irradiated in the CMS I (0.50 mol%, 365 nm, 48 h).After the irradiation, the formation of an intramolecular cyclization product equivalent to 9 was not observed from the epoxide 10 (Fig. S17, ESI).The isolated yield of dimers of 10 was 80.6 wt%, which is considerably higher than the 59.4 wt% (observed yield in the photodimerization of myrcene 2).The difference between these two outcomes is also close to the yield of 9. Furthermore, the unreacted amounts of myrcene and myrcene epoxide were very similar.These observations support the intramolecular cyclization to 9 as a limiting factor for the lower yield of myrcene dimers.In the case of ocimene, a photosensitized E/Z-isomerization is also possible under the conditions described (Fig. 4B). 31Although in this case the resulting triplet structure is still reactive in the dimerization, the unimolecular intramolecular E/Z isomerization occurs faster than the encounter of one triplet excited ocimene molecule with another ocimene molecule, which would lead to dimers.The E/Z ratio determined by 1 H NMR changed from 2.5:1.0 on the starting material to 3.0:1.0for the ocimene recovered after irradiation.Therefore, the photosensitized E/Z isomerization is possibly a de-excitation method, forming the more stable isomer (E)-ocimene.
Besides the highly competitive E/Z photoisomerization pathway, triplet excited ocimene is also more hindered than myrcene.In the triplet biradical intermediate of myrcene the radical character is always on a terminal carbon.In the case of ocimene, the radical character can reside either at the terminal carbon or at the secondary carbon (Fig. 4C).Indeed, the calculated triplet state structures for ocimene indicate that the most stable biradical configuration occurs when the allyl radical is at the terminal position and the single radical is on in the secondary carbon of the diene (Fig. 4D).This is similar to the case of the biradical intermediate 1BR in the α-phellandrene dimerization (Fig. 3), as in both cases the most stable biradical is the one with allyl radical having a stabilizing methyl group bonded to the allyl radical edge, where the molecular orbital has a nonzero coefficient (see SOMO of the allyl radical in Fig. 4C).Therefore, the non-allylic radical of triplet state ocimene is more hindered to react as it occurs from the secondary carbon.

Cross-dimerization of monoterpenes:
To improve the yield of ocimene and myrcene dimerization, we explored a cross-dimerization strategy between these less reactive compounds and α-phellandrene.Ocimene and myrcene were mixed with α-phellandrene in different molar ratios and with 0.25 mol% of photosensitizer 8.The results from the cross-dimerization is tabulated in Table 3.The individual conversions (column "Conversion (wt%)") express how much of the starting myrcene or ocimene ended up in the C20 mixtureas dimers to themselves or as crossdimers formed with α-phellandrene.Reactions performed under 365 nm irradiation in the CMS I by using 8 as the photosensitizer (0.25 mol%).
The best results were found when the 2:1 ratio of α-phellandrene:ocimene or αphellandrene:myrcene was used.The yield of the cross-dimers was determined by the quantitative 1 H NMR spectrum of the crude irradiated mixture by comparing with the unreacted 2 and 3.
Ocimene showed almost a fivefold increase in its conversion to C20 hydrocarbons, reaching 76.6 wt.% after 48 h (Table 3, entry 5).Myrcene conversion exceeded 90 wt.% already after 24 h (Table 3, entry 3).The improvement of the conversion of myrcene into C20 products is evident from the analysis of the 1 H NMR spectrum of the crude irradiated mixture, in which protons correspond to unreacted myrcene are barely visible due to its conversion into dimers and/or the intramolecular cyclized product 9 (Fig. S22, ESI).The values from cross dimerization with α-phellandrene in Table 3 are plotted in Fig. 5 and compared to the dimerization of ocimene and myrcene (Table 1).Data from Table 3 compared to their conversions when neat conditions were used (Table 1).Labels 1.0 eq. and 0.5 eq.describe the amount of monoterpene used as starting material related to α-phellandrene (mol).
The higher conversions of myrcene and ocimene resulted from the long-lived triplet state of α-phellandrene that attacks not only other α-phellandrene molecules, but also the diene segments of myrcene and ocimene.Interestingly, the cross-dimerization attempt using 1:1 ratio of α-phellandrene and myrcene did not significantly improve the conversion of myrcene to C20 products.However, the implicit dilution led to enhance the formation of 9 (30 wt.%).

Route to C15 and C30 hydrocarbons:
Following the cross-dimerization of α-phellandrene with other monoterpenes, we tested the reaction with isoprene and monoterpenes in order to obtain C15 hydrocarbons.The cross-dimerization with isoprene led mainly to the formation of C15 and C20 compounds (Fig. 6A).However, a minor amount of C10 hydrocarbons (dimers of isoprene) was also observed.Conjugated dienes larger than monoterpenes are also available in nature.Examples of those fall in the continuation of the terpene homologue series (sesquiterpenes, diterpenes, etc.), and their dimerization could also provide hydrocarbon precursors to possible applications as lubricant oils.
The observed reactivity of α-phellandrene makes endocyclic dienes the natural candidates for further investigations, such as the sesquiterpenes α-zingiberene 6 and β-sesquiphellandrene 7.
These compounds are two of the main components present in ginger oil.Therefore, we irradiated a C15 fraction sample of ginger oil (87 g) mixed with photosensitizer 8 (0.25 mol%) in the CMS III, under 365 nm for 48 h.The viscosity of the crude mixture produced was visually higher and almost complete conversion of α-zingiberene 6 to its dimers was observed by comparing the GC-MS trace of starting material and irradiated sample (Fig. 6B).The isolated yield of C30 dimers was 61 wt.% (considering the amount of conjugated dienes in the starting material, determined by 1 H NMRsee Fig S33 , ESI).Together with previous results reported herein, this is a further piece of evidence that endocyclic diene structures are a key feature for efficient photosensitized dimerization of hydrocarbon biomolecules towards renewable fuels and materials.

Fuel and lubricant related properties of selected C15 to C30 crudes:
We computed the heats of combustion for the hydrogenation products of some of the compounds in Fig. 3 by following the procedure reported by Pahima et al. 23 As expected, these compounds have high heats of combustion between 42.3 and 43.3 MJ kg -1 , which are suitable for high energy density fuel applications (Table S5, ESI).Hence, we synthesized hydrogenated samples of α-phellandrene dimers (HAPD), myrcene dimers (HMD), α-phellandrene-isoprene cross-dimers (HAPID) and ginger oil dimers (HGOD), and some physical properties relevant to fuels and lubricant oils were measured (Table 4).

LBO = lubricant base oil
The crudes HMD, HAPD and HAPID have the experimental net heats of combustion (NHOC) higher than 43 MJ kg -1 , which agrees with our computed values for some of the hydrogenated dimers (Table S5, ESI).The volumetric NHOC values were higher than those for D2 diesel and biodiesel.Such values are desired for high energy density fuel applications. 34wever, the obtained kinematic viscosities may restrict their use in such applications.While HAPD and HMD crudes have too high viscosities to be used stand-alone as diesel, they could be mixed in blends to afford the required viscosities. 24,29Yet, these blends could possibly be made entirely from non-fossil sources, using for example the C15 fraction produced in the crossdimerization of α-phellandrene and isoprene.Indeed, adjustments of HAPID viscosity would be required to a lesser extent, as its viscosity is closer to that of D2 diesel (6.45 mm 2 s -1 , at 40 °C).
Moreover, the volumetric NHOC of HAPID was 8.8 % higher than the one for a typical D2 diesel.
Alternatively, the C15 fraction in HAPID can be partially or entirely separated from the C20 fraction, affording a crude with higher proportion of C15 compounds, which would reduce the kinematic viscosity and likely approach the properties of hydrogenated sesquiterpenes, such as bisabolene. 19The pour point values found for the HAPID crudes were even lower than those for D2 diesel and biodiesel.Furthermore, earlier reports proposed to use C18-C20 hydrocarbons blends suitable for jet fuels. 25,26,34,53Therefore, this could also be an alternative end-use to the crudes HAPD and HMD.
The high viscosities of HAPD and HMD makes it possible to use these crudes for lubricant oil formulations.The kinematic viscosity at 40 °C (KV40) of HMD is 12.7 mm 2 s -1 , which is comparable to C30 poly alpha-olefins (POA) mixtures (15.23 mm 2 s -1 ). 50HAPD crudes have values for KV40 comparable to those of base oils belonging to API Group II 6 (41.5 mm 2 s -1 ). 51The dimers derived from ginger oil (HGOD) have a very high KV40, and both HAPD and HGOD KV40 values are comparable to those of base oil synthesized from triesters and estolides derived from vegetable oils, 52 showing a lower pour point in some cases.On the other hand, the viscosity index (VI) values were very low for both crudes HAPD, HMD and HGOD, as it was expected considering their cyclic structures.Yet, the use of viscosity index improvers or further refining for ring opening of some of the structures could increase the VI values. 51While the presence of rings negatively affects the VI, it should be considered that the cyclic structures also afford better solvency, 51 therefore, the choice for keeping the rings could be interesting for applications where good solvency is required.

Outlines in GHG mitigation and economic feasibility:
Our studies were carried out in a laboratory scale with maximum production rate at 11 g/h (Table 1, entry 1.4) of dimers under 365 nm light.Therefore, further work on scaling-up the process is required for a robust life cycle assessment (LCA) and techno-economic analysis (TEA).Yet, we briefly discuss the potential strengths and impacts of the photosensitized dimerization of terpenes.
The non-edible evergreen trees of the Eucalyptus spp.correspond to 20 million hectares of forests worldwide. 54Among them, the E. dives (phellandrene variant) has 60 -80 % of αphellandrene in its extracted oil. 7Assuming the highest oil yield of the E. dives to be similar to the one for E. globulus (900 kg ha -1 yr -1 ), the highest potential of global production of α-phellandrene would be 14 million tonnes per year.The actual yield would be lower since there is a large variation of constituents, oil yields, harvest windows, and other properties among different species. 6Because high α-phellandrene-content eucalyptus oils are used for inexpensive cleaning products, 6,7 their use instead as a feedstock to lubricant oils and fuel production could be an added value to the eucalyptus oil market.On that regard, a recent work showed the economic potential of producing jet fuel from the leaf oil of Eucalyptus spp. 55udies on the environmental impact of using eucalyptus tree to produce biomass for energy purpose have shown an opportunity to reduce GHG emissions up to 80 % when compared to fossil fuels, 56,57 especially due to the sequestration of CO2 by the crops offsets the emissions during the process, associated to their fast growth and short-rotation.The main source in these studies is the wood from the eucalyptus trees.Therefore, obtaining α-phellandrene from the leaf oil could be an additional route to valuable and sustainable goods.On the other hand, the high demand of land-use is a drawback, as well as the risk of threatening of biodiversity, as specific species that produce high amounts of α-phellandrene would be the optimal choice.
The α-phellandrene production from eucalyptus trees can also be a side route in the paper manufacturing, as well as it is for other terpenes.For instance, the pulp from paper manufacturing is a source to the bio-waste sulfate turpentine, which has a worldwide production of 330 000 tonnes/year. 8The pyrolysis of β-pinene, one of the main components in the sulfate turpentine, produces myrcene, 9 making turpentine a bio-waste source for myrcene production.
Alternatively, α-phellandrene and myrcene can also be produced by metabolically engineered cyanobacteria.Formighieri and Melis reported on the cyanobacterial production of αphellandrene to reach titers of 0.15 mg g -1 dcw (dry cell weight), 20 while titers of 0.33 mg of βphellandrene g −1 dcw were achieved, over an incubation period of 48 h.Another strain reported on the same work produced mainly β-myrcene (62 %)but the overall amount of terpenes produced by that strain was low.
9][60] These studies reveal the possibility of reducing GHG emissions and they also indicate commercial viability.Their results depended heavily on the use of clean energy as input, optimal photobioreactors and improvements on the metabolic engineering of cyanobacteria in order to enhance cell productivity. 58,59The high volatility of small molecules enables an easier escape from the cell culture, 59,60 but in the case of monoterpenes and larger molecules, more energy demanding methods may be required to extract molecules from culture such as the use of organic solvents, which might increase the environmental impact. 17,20The toxicity of monoterpenes to the cells represents another issue. 61 the sources examined and the ways to obtain them are considered to be renewable, we now reflect on the later steps to transform such feedstock into fuels and lubricant oils.Our approach uses low amounts of photosensitizer, and heating or cooling is not required during the dimerization step.Indeed, the main resource consumption would be electricity to power UV LEDs.
While this could be provided by clean energy sources, such as photovoltaic cells or hydropower, it is also righteous to consider natural sunlight, as we have demonstrated the significant conversions obtained under simulated and natural sunlight.In this case, energy consumption would be much lower for the dimerization step.However, the land-use could be significant if one considers that the flat FEP setup (CMS IV) must be very wide to accommodate a multi-liter reaction mixture.
The steps after the dimerization are existing downstream processes in the oil industryhydrogenation and distillation.Optimal and greener hydrogenation methods, with cheaper catalysts, and use of clean energy inputs should be considered to afford a final product that accounts for a reduced environmental impact as well as being commercially feasible. 62

Conclusions
We have applied a solid organic photochemical knowledge to the conversion of key hydrocarbon biomolecules into renewable crudes for fuels and lubricant oils.By combining reactor design, selection of photosensitizer and use of reactive substrates, we have described the sunlightpromoted dimerization and cross-dimerization of terpenes to obtain C15-C30 hydrocarbon mixtures as potential precursors to drop-in diesel fuel and lubricant oil surrogates.Our preliminary assessment depicted the sustainability and viability of this process, which still requires further modification and scaling up in order to be submitted to a robust LCA and TEA in future.
Making use of isoprene, monoterpenes and sesquiterpenes, a full sunlight-driven process to C10 -C30 hydrocarbons would be possible, encompassing a diversity of sustainable liquid biofuels and lubricant oils produced.

Fig. 1
Fig. 1 Terpenes that were used in this work and their possible sources: evergreen trees and plants (essential oils), industrial waste (turpentine), sugars derived from lignocellulosic biomass (converted into terpenes by microorganisms) and production by metabolically engineered photosynthetic microorganisms (cyanobacteria).

Scheme 1 A
Scheme 1 A) Previous work by Baldwin and Nelson on the photosensitized dimerization of α-phellandrene using naphthalene as the photosensitizer.B) Our present approach on the photosensitized dimerization of α-phellandrene and other terpenes as key step to produce high energy density fuels and base oils.NHOC (Net Heat of Combustion); KV40 (Kinematic Viscosity at 40 °C).

Fig. 2 A
Fig. 2 A) Myrcene and myrcene epoxide dimerization reactions with their respective isolated yields of dimers and recovered byproducts and reactants after distillation.B) Ocimene dimerization with the isolated

Figure 3
Figure 3 Conversion dimers of ocimene and myrcene in the cross-dimerization with α-phellandrene.

Fig. 4 A
Fig. 4 A) GC trace of cross-dimerization between α-phellandrene and isoprene (1:1), where C10 -C20 hydrocarbons can be observed.B) GC trace of the ginger oil sample after irradiation under 365 nm, mixed with photosensitizer 8 (0.5 mol %) showing C30 dimers.Inset: chromatogram for the region of C15, before and after irradiation show that α-zingiberene peak (6) is reduced after irradiation, as well as 7, though to a lower extent.

Table 1
Conversions of monoterpenes to dimers in the photosensitized dimerization.

Table 2
Dimerization under simulated and natural sunlight.
a A qualitative experiment in which the sample was exposed to natural sunlight.Total time of exposure to light outdoor was of 18 h over two consecutive days.

Table 4
Selected fuel and lubricant oil properties of HAPD, HMD, HAPID and HGOD (all produced in this work), as well as some commercially available diesel and lubricant oils for comparison.