Ni-Ru/CeO2 Catalytic Hydrothermal Upgrading of Water-Insoluble Biocrude from Algae Hydrothermal Liquefaction

Hydrothermal liquefaction (HTL) of algae is a promising crude bio-oil (biocrude) production technology, which can convert wet algae into water-insoluble biocrude and other coproducts. In this work, algae HTL at 350°C and 20 min was conducted to obtain water-insoluble biocrude (B1), which was then hydrothermally upgraded at 450°C, 60 min, or with added H2 and/or homemade catalyst (i.e., Ni-Ru/CeO2 or Ni/CeO2) for the first time. The characteristics (e.g., yield, elemental component, energy recovery, and molecular and functional group compositions) of upgraded water-insoluble biocrude (B2) as well as light biocrude thereof were analyzed comprehensively. The results show that Ni-Ru/CeO2+H2 led to the highest yield and HHV (higher heating value), the best elemental compositions quality of B2, and the largest fraction and the best light of light biocrude in B2. Ni-Ru/CeO2+H2 had good catalytic desulfurization effect and could transform high-molecular-weight compounds into low-molecular-weight compounds in B1 upgrading. At the condition above, 46.2% of chemical energy in the initial algae could be recovered by B2, while average 54.9% of chemical energy in B2 was distributed in its light biocrude (hexane-soluble) portion. On the whole, Ni-Ru/CeO2+H2 can be considered as the optimal additive in all tested cases.


Introduction
Hydrothermal liquefaction (HTL) of algae for crude biooil (biocrude) production has increasingly drawn much attention due to its benign raw material properties, high biooil yield, avoidance of drying step, and so on. It can chemically convert high moisture algae into water-insoluble biocrude and coproducts (i.e., gases, aqueous phase, and solids) in hot compressed water. However, the obtained biocrude does not meet biodiesel standards in some aspects such as heteroatoms content and heating value, so it requires further biorefinery via proper approaches. An effective method is hydrothermal upgrading, in which biocrude reacts in water, either thermally or with added H 2 and/or catalyst to remove heteroatoms and improve heating value, and so on [1].
Typically, after algae HTL of 1-4 h at around 350 ∘ C, total biocrude (both water-insoluble and water-soluble biocrudes) is collected by organic solvent (e.g., dichloromethane) extraction from all postreaction products and followed hydrothermal upgrading with 2-6 h at a supercritical temperature [2][3][4][5][6][7]. Notably, the water-insoluble biocrude can spontaneously separate from aqueous phase by gravity upon cooling after algae HTL and commonly accounts for more than 85 wt% of the total biocrude and has remarkably distinct characteristics in comparison to the water-soluble biocrude extracted from the aqueous phase by organic solvent [8,9]. Due to pollution and cost restrictions of extraction solvent, it is more feasible to adopt solvent-free method to collect and upgrade the water-insoluble biocrude (as the majority of the total biocrude) in actual industrial production. On the other hand, it is convenient and economic to directly use the aqueous phase instead of water in hydrothermal upgrading, because it is easy to separate part of the aqueous phase after algae HTL and to use the rest for following hydrothermal upgrading. However, relevant investigation about water-insoluble biocrude upgrading with the aqueous phase and catalyst is very scarce now, except for our latest research about catalytic hydrothermal upgrading of waterinsoluble biocrude in supercritical water at 400 ∘ C, 60 min with commercial catalysts (i.e., Mo 2 C, Ru/C, and Pt/C ) [1].
As an extension, this work conducted 20 min of algae HTL at 350 ∘ C to obtain water-insoluble biocrude and then upgraded it at the higher temperature of 450 ∘ C, 60 min with added H 2 and/or Ni-Ru bimetallic catalyst homemade for the first time. The yields, elemental components, energy recoveries, and molecular and functional group compositions of the upgraded water-insoluble biocrudes at various conditions were analyzed systematically. The characteristics concerning the light biocrude in the upgraded biocrude and gaseous products were examined as well. This information is valuable for optimizing processes design and operation parameters of the hydrothermal upgrading of the water-insoluble biocrude from algae HTL.

Experimental
Procedures. The adopted algae (Nannochloropsis sp.) slurry had 30.5 ± 2.5 wt% of biomass content, which was determined by drying three separate samples at 65 ∘ C in an oven for 48 h. The dry basis algae (about 23.30 MJ/kg) contained approximately 52.85, 7.43, 8.78, and 0.62 wt% of C, H, N, and S contents, respectively. High purity solvents (i.e., dichloromethane (DCM), n-hexane, and deuterated chloroform (CDCl 3 )) were obtained from commercial sources. High purity (≥99.999%) hydrogen and helium were purchased from Baoguang Gas Co., Ltd. Several 4.1 ml mini-batch reactors were assembled for the HTL experiments by using 1/2-in. 316 stainless steel Swagelok port connectors and caps. Subsequent biocrude upgrading experiments were conducted in the same reactors, but upgrading reactions with added H 2 or with gas analysis were implemented by additionally equipping each reactor with a length of stainless steel tubing and a high-pressure gas valve.
First of all, one set of experiments were carried out to obtain water-insoluble biocrude from algae HTL at 350 ∘ C, 20 min, and 14.1 wt% of algae loading in the reactor (corresponding to 1.2354 g algae slurry and 1.3990 g deionized water). After being sealed, the reactor was placed in a Techne fluidized sand bath (Model SBL-2) preheated up to 350 ∘ C for the desired residence time (20 min). Herein, the corresponding reaction pressure was about 16.5 MPa and the reactor heating-up time was 2-3 min. The postreaction reactor was removed from the sand bath and quenched in an ambienttemperature water bath for 15 min and then equilibrated at room temperature for at least 1 h before products collection and analysis. The reactor was kept in a vertical position as it was cooled so that little biocrude would adhere to the inner surface of the reactor top cap. We then opened the reactor and removed 1.875 g aqueous phase via a pipette. The waterinsoluble biocrude (about 0.1505 g) from algae HTL (with a little solids residue) and the aqueous phase residue (around 0.3750 g) in the reactor were never in contact with DCM and were used in situ for the following upgrading experiments. 0.0225 g catalyst (if necessary, about 15 wt% of B 1 mass) was loaded into the reactor, and then the reactor was vacuumed and filled either with helium (10 KPa) to eliminate the effect of reactor inside air in the test without H 2 addition or with H 2 to 2.0 MPa to check the effect of H 2 addition. The upgrading experiments were carried out in the sand bath at 450 ∘ C, 60 min under different conditions such as catalyst and/or H 2 addition, and the corresponding reaction pressure was approximately 22.5 MPa (based on steam tables). We then removed the reactor from the sand bath and cooled it and followed the previously described method 2 procedure [8] for products collection. In this procedure, the contact between the aqueous phase and DCM is avoided by removing the aqueous product via a pipette, prior to the recovery of the water-insoluble biocrude from the reactor. Eventually, waterinsoluble biocrude (B 2 ) was obtained from B 1 upgrading products, and light biocrude in B 2 (B 2 L ) could be further gained by n-hexane extraction.
The mainly used catalysts were 10 wt%Ni0.1Ru/CeO 2 and 10 wt%Ni/CeO 2 , which were prepared in house and detailed procedures had been described by our previous report [10]. Herein, 10 wt% represents the mass percentage of active metal (Ni+Ru or Ni) in the catalyst, and 0.1 is the mass ratio of Ru/Ni. For simplification, Ni-Ru/CeO 2 , Ni/CeO 2 , and "none" were adopted to separately represent 10 wt%Ni0.1Ru/CeO 2 , 10 wt%Ni/CeO 2 , and the case without catalyst and H 2 in this work. Herein, the cases of Ni/CeO 2 and "none" were tested for comparison with that of Ni-Ru/CeO 2 .

Analysis Methods.
Elemental compositions (i.e., C, H, N, and S) of dry basis algae and all biocrude samples were determined by a cube CHNS elemental analyzer (Elementar Vario EL) with uncertainties of <3% of the reported value. Compound components of biocrude were identified via a gas chromatography-mass spectrometer (GC-MS, Agilent Technologies 6890N) equipped with an autosampler, an autoinjector, a mass spectrometric detector, and an Agilent HP-5 capillary column (50 m × 200 m × 0.33 m). Nuclear magnetic resonance (NMR) and Fourier transform infrared (FT-IR) spectroscopic analyses of biocrude were conducted to characterize functional group compositions. Detailed information on these instruments and procedures was introduced in our previous research [8]. Gaseous products were identified and quantified by an Agilent Technologies model 6890N gas chromatograph equipped with a thermal conductivity detector, following the method reported previously [11].
Biocrude yield and energy recovery were defined as follows: Biocrude yield The higher heating value (HHV) of dry basis algae or biocrude was estimated by the following Dulong formula: where C, H, O, and S are the wt% composition of each element in the material. The O content was calculated by the differences from the C, H, N, and S values and reserving 1 wt% for other elements not assessed.
Three independent experiments were carried out at the same conditions to determine the uncertainties of experimental data. The results reported herein are mean values and their uncertainties are the sample standard deviations.

Results and Discussion
All experiments about algae HTL and biocrude upgrading produced water-insoluble biocrude, gases, aqueous phase, and solids, and total mass balance exceeded 91.0 wt% in products collection.

Biocrude Yield.
Algae HTL at 350 ∘ C, 20 min, and 14.1 wt% of algae loading led to 37.5 ± 0.21 wt% of B 1 yield. Figure 1 shows B 2 properties after B 1 upgrading at 450 ∘ C, 60 min, with or without H 2 and/or catalyst. It can be found that Ni-Ru/CeO 2 +H 2 led to the highest B 2 yield (27.0 wt%, corresponding to about 72.0 wt% of B 1 yield), likely due to catalytic hydrogenation effect during B 1 upgrading. The much lower B 2 yield (21.72 wt%) appeared at the Ni-Ru/CeO 2 condition, and this might be attributed to a large amount of gases formation promoted by the catalyst in supercritical water [10]. Based on the B 2 yield (23.94 wt%) in the case of H 2 addition, there seemingly was synergetic effect between H 2 and Ni-Ru/CeO 2 on B 2 yield improvement because their coexistence probably inhibits gases formations. Light biocrude fraction in biocrude, which was defined as the measured mass of light biocrude (hexane-soluble) divided by the biocrude mass, is a proxy for biocrude quality because these hexane-soluble compounds are less polar and of moderate boiling point and thus more desirable for fuel use, compared with heavy biocrude (hexane-insoluble) compounds [9]. As indicated in Figure 1, Ni-Ru/CeO 2 +H 2 could obviously increase the light biocrude fraction in B 2 and led to the highest value (57.14 wt%), meaning the remarkable improvement of B 2 quality. Herein, similar synergetic effect between H 2 and Ni-Ru/CeO 2 on the promotion of the light biocrude fraction in B 2 was found as well, probably due to the conversion of hexane-insoluble biocrude to hexane-soluble biocrude in hydrothermal upgrading at the Ni-Ru/CeO 2 +H 2 condition. Thereby, we could adopt the Ni-Ru/CeO 2 catalyst together with H 2 to regulate the properties of the upgraded waterinsoluble biocrude. It is worthy noticing that the upgrading without H 2 and catalyst (i.e., none) rendered considerably high B 2 yield and light biocrude fraction in B 2 , so it may be considered as a pre-upgrading process of B 1 in actual large scale production. Table 1 indicates elemental compositions and HHVs of B 2 and B 2 L after B 1 upgrading under different conditions. It can be observed that hydrothermal upgrading could increase the C content and simultaneously reduce S and O contents. Deoxygenation can be realized by forming CO 2 , H 2 O, and CO through complex reactions such as decarboxylation and dehydration [12,13]. Note that the Ni-Ru/CeO 2 catalyst had the best catalytic desulfurization effect and Ni-Ru/CeO 2 +H 2 was also much better for sulfur removal. This has good agreement with the findings that the S content in bio-oil by Ni/SiO 2 -Al 2 O 3 catalytic upgrading is below detection limit [14] and Ru/C has good desulfurization performance [15]. Ni-Ru/CeO 2 +H 2 rendered the highest C content, the largest HHV, the lowest O content, and the second lowest N+S content, so totally exhibiting the best catalytic upgrading effect in all tested cases. Consistently, H 2 addition together with catalyst is desirable for hydrogenation and hydrodeoxygenation behaviors in biocrude upgrading [3,4,6]. Apparently, based on the higher C, H contents and HHV but lower heteroatoms (i.e., N, O, and S) contents, each B 2 L had better quality than its corresponding B 2 . Thus, the tested catalysts (i.e., Ni-Ru/CeO 2 and Ni/CeO 2 ) had substantially positive effect on the improvement of B 2 L quality in B 1 upgrading in the presence of H 2 , due to the increase of C, H contents, and HHVs and the decrease of N, S, and O contents. The combined Ni-Ru/CeO 2 +H 2 led to the best B 2 L quality with the highest C+H content, the largest HHV, and the lowest N+S+O content. Overall, the best catalytic effect of Ni-Ru/CeO 2 +H 2 is likely attributed to positive reactions such as hydrogenation, deoxygenation, denitrogenation, and desulfurization in B 1 upgrading.

Elemental Composition and HHV.
Energy recovery is a key indicator of the effectiveness of a hydrothermal process in capturing the chemical energy of initial algae in the produced biocrude [16]. Due to little  Figure 1). Moreover, the energy recoveries of B 2 L ranged from 23.2 to 32.1% (average 27.9%) of the heating value of the algae loaded initially. Notably, it can be calculated that 49.7-58.6% (average 54.9%) of chemical energy in B 2 was distributed to B 2 L (i.e., its light biocrude portion). Hence, taking into account the yields, elemental compositions, HHVs, and energy recoveries of both B 2 and B 2 L , Ni-Ru/CeO 2 +H 2 can be regarded as the optimal additive in all tested cases.

Gas Analysis.
Note that Ni-Ru/CeO 2 is favorable for organic matters gasification to produce a flammable gas mixture in supercritical water [10], so gaseous products are also concerned in this research. Table 2 provides gaseous products after hydrothermal upgrading of B 1 at different conditions, so major gases compositions included H 2 , CH 4 , CO 2 , C 2 H 2 , C 2 H 6 , CO, and C 2 H 4 at the condition in absence of H 2 . Compared with the case without catalyst and H 2 (i.e., none), the Ni-Ru/CeO 2 catalyst could significantly improve H 2 mole fraction (from 29.6% to 41.1%), and meanwhile the postreaction pressure increased from about 2.2 to 3.5 bar. This suggests that Ni-Ru/CeO 2 led to much more H 2 and C-containing gases formation during B 1 upgrading probably mainly via promoting water-gas shift and decarbonation reactions. This outcome is in good accord with the findings that Ru/C [17] and Ni-Ru/CeO 2 [10] are effective for hydrothermal gasification and the relatively low B 2 yield in Figure 1 as well. Notably, it reminds us that these flammable gases formed in the water-insoluble biocrude upgrading should be disposed properly in large scale production processes. In the case of H 2 addition, major identified gaseous products were CH 4 , CO 2 , C 2 H 6 , C 2 H 4 , CO, and unreacted H 2 , but there was no C 2 H 2 due to hydrogenation effect. Similar gases compositions are also found in catalytic upgrading of duckweed biocrude in subcritical water [18]. The far high H 2 mole fraction (about 70%) is mainly attributed to initial H 2 addition (2.0 MPa). Nonetheless, the remarkable reduction of the postreaction pressure in contrast to that before reaction suggests a large amount of H 2 consumption for heteroatoms removal (likely in the forms of H 2 O, NH 3 , and H 2 S) and hydrogenation in B 1 upgrading. Certainly, at the same time, there is also H 2 formation due to biocrude gasification in supercritical water (see the "none" group).
Apparently, the total ion chromatogram of B 2 looks completely different from that of B 1 , because most compounds appearing in the range of 45-75 min in B 1 disappeared or reduced. Major compounds in B 2 appeared in the range of 28-55 min, where the peaks areas sum of all identified compounds took up 95.7% of the total peak area. Main compounds involved pentadecane, tridecane, tetradecane, heptadecane, and dodecane, in which pentadecane and tridecane seem to be the most abundant two species. Thus, hydrothermal upgrading of B 1 with Ni-Ru/CeO 2 +H 2 is able to transform high-molecular-weight compounds into low-molecular-weight (i.e., low-boiling-point) compounds, meaning the quality improvement of the upgraded biocrude. More importantly, this catalytic upgrading helps to remove heteroatoms (e.g., O and N) in B 1 to form the upgraded biocrude with abundant presence of a series of aliphatic saturated hydrocarbons. Reactions such as decarboxylation, dehydroxylation, and deamination contribute to the heteroatoms removal, hydrocarbons formation, and so corresponding increase in carbon content and HHV after hydrothermal upgrading. 13 C-NMR analysis of biocrude was conducted to identify the types of functional groups. Chemical shift provides information concerning functional group identification, and corresponding peak area can roughly reflect its relative abundance. Figure 3 illustrates 1 H NMR spectra of B 1 and B 2 derived from B 1 upgrading at the Ni-Ru/CeO 2 +H 2 condition. The resonances at 0.8 and 1.2 ppm are characteristics of protons in terminal methyl groups and methylene groups in alkyl chains, respectively [23]. The peak at around 2.1 ppm is consistent with resonance expected from protons on carbon atoms to an acyl group [15]. There was a peak near 7.2 ppm in B 1 and B 2 , which arises from aromatic protons or conjugated dienes [15,24], and herein the much smaller peak area in the B 2 spectrum suggests that the catalytic hydrothermal upgrading by Ni-Ru/CeO 2 +H 2 can effectively reduce the compounds containing the functional groups above. This is consistent with the findings in the GC-MS analysis. Differently, in the B 2 spectrum, a clear peak emerged at around 5.2 ppm representing phenolic -OH [24], possibly due to the existence of hydroxyketone in the upgraded biocrude. The appearance of these peaks above reveals the  existence of alkyl moieties, carbonyl functionalities, and aromatic or unsaturated molecules in the biocrudes, though in different amounts [9]. Overall, in B 1 and B 2 obtained from B 1 upgrading with Ni-Ru/CeO 2 +H 2 , alkane functionality (0.5-1.5 ppm) and aliphatic -to-heteroatom/unsaturated functionality (1.5-3.0 ppm) [25] were very abundant, due to the presence of alkanes and other compounds with aliphatic methylene and methyl groups. about C-related functional groups. The two spectra possessed some peaks in the 0-55 ppm region, where aliphatic methyl and methylene carbon (such as alkane carbons) atoms appear [4]. This suggests that aliphatic linkage variability lies with chain length or degree of branching within the linkers [24]. In detail, the region above can be divided into 0-28 ppm (short aliphatics) and 28-55 (long and branched aliphatics), and these aliphatic carbon atoms contribute significantly to energy content [25]. Except for the solvent peak at around 77 ppm, there was no signal in the 55-95 ppm region, where carbohydrate carbons appear [25]. Moreover, each biocrude also indicated a peak at about 207 ppm, where carbonyl carbons in ketones and aldehydes appear [8]. The drop of the peak area implies the content reduction of the O-containing compounds above after B 1 upgrading with Ni-Ru/CeO 2 +H 2 . Figure 5 elucidates FT-IR spectra of B 1 from algae HTL and B 2 obtained from hydrothermal upgrading of B 1 with Ni-Ru/CeO 2 +H 2 . GC-MS and NMR analysis results show that the two biocrudes contained methylene groups in alkanes, which is proved by further FT-IR analysis herein. Asymmetrical and symmetrical C-H stretching vibrations in aliphatic methylene groups appear in the range of 2800-3000 cm −1 [15]. The existence of carbonyl carbon (such as carboxylic acids and esters) is characterized by the bands at 1650-1760 cm −1 [8]. The high intensity in these two regions is in accord with a significant amount of hydrogen in the biocrudes being aliphatic [15]. The spectrum of B 2 also exhibited strong absorbance at approximately 1450 cm −1 , where the scissoring band in methylene groups emerges [23]. This is probably attributed to some mononuclear aromatic compounds (e.g., naphthalene, 1,2,3,4-tetrahydro-1,1,6-trimethyl-in Table 3) present in the biocrude. In addition, the two biocrudes also had other different functional groups such as methyl, methylene, aromatic, and alkyne groups appearing at ∼1378, ∼1271, ∼1036, and ∼735 cm −1 , respectively, and their different absorbance intensities imply content variations. Different from B 1 , B 2 presented certain intensity of bands within 2000-2500 cm −1 representing triple bonds or cumulative double bonds and large amounts of absorption peaks within 3600-3900 cm −1 (e.g., O-H stretching vibrations in hydroxyketone [26]). Hence, hydrothermal upgrading of B 1 with Ni-Ru/CeO 2 +H 2 led to obvious variations of functional groups compositions and contents in B 2 .

Conclusions
In all tested hydrothermal upgrading (at 450 ∘ C, 60 min and with added H 2 and/or catalyst) of B 1 from algae HTL at 350 ∘ C, 20 min, Ni-Ru/CeO 2 had the best catalytic desulfurization effect and could significantly improve H 2 mole fraction and rendered much more H 2 and C-containing gases formation after B 1 upgrading. Ni-Ru/CeO 2 +H 2 led to the highest B 2 yield (27.0 wt%), the best elemental compositions, and the largest HHV (39.94 MJ/kg), as well as the highest B 2 L fraction in B 2 (57.1 wt%) and the best B 2 L quality. At the Ni-Ru/CeO 2 +H 2 condition, 46.2% of chemical energy in initial algae could be recovered in B 2 , and average 54.9% of chemical energy in B 2 was in its light biocrude portion. Ni-Ru/CeO 2 +H 2 was able to transform high-molecular-weight compounds into low-molecular-weight compounds and led B 2 to containing a series of abundant aliphatic saturated hydrocarbons such as pentadecane, tridecane, tetradecane, heptadecane, and dodecane. Overall, Ni-Ru/CeO 2 +H 2 can be regarded as the optimal additive in all tested cases, and there seemingly was synergetic effect between Ni-Ru/CeO 2 and H 2 on B 2 yield and quality improvement after B 1 upgrading.

Conflicts of Interest
The authors declare that they have no conflicts of interest.