- Email: [email protected]
Hydrothermal upgrading of algae paste in a continuous flow reactor
Bioresource Technology xxx (2015) xxx–xxx
Contents lists available at ScienceDirect
Bioresource Technology journal homepage: www.elsevier.com/locate/biortech
Hydrothermal upgrading of algae paste in a continuous flow reactor Bhavish Patel, Klaus Hellgardt ⇑ Imperial College London, Department of Chemical Engineering, Exhibition Road, South Kensington, London SW7 2AZ, UK
h i g h l i g h t s
g r a p h i c a l a b s t r a c t
Continuous HTL of algal biomass in a
quartz lined plug flow reactor. Production of algal biocrude at short
residence time. Analysis of biocrude from continuous
fast liquefaction.
a r t i c l e
i n f o
Article history: Received 28 January 2015 Received in revised form 1 April 2015 Accepted 2 April 2015 Available online xxxx Keywords: HTL Plug flow reactor Microalgae Biocrude Hydropyrolysis
a b s t r a c t This investigation demonstrates the utility of a novel laboratory scale continuous plug flow reactor for fast Hydrothermal Liquefaction (HTL) of microalgae in a quartz lined chamber. Reactions were carried out between 300 and 380 °C and residence times of 0.5–4 min. Cyclohexane was used as a co-solvent to enhance extraction and prevent char formation. Highest biocrude yield of 38 wt.% was achieved at 380 °C and 30 s as well as Water Soluble Fraction containing up to 60 wt.% matter recovered. Analysis of the biocrude showed that the extent of deoxygenation and denitrogenation after HTL varied and is dependent on the reaction conditions, Fourier Transform Infrared Spectroscopy analysis showed that biocrude contains similar functional moieties with only a small difference observed at different reaction conditions. Conversely, the Simulated Distillation and Size Exclusion Chromatography data showed that harsher conditions produced marginally better biocrude with improved boiling point profile and lower molecular weight compounds, respectively which was confirmed using Gas Chromatography–Mass Spectrometry. Ó 2015 Elsevier Ltd. All rights reserved.
1. Introduction The rise in global population and improvement in living standards is increasing the world’s energy demand. The U.S. Energy Information Administration (EIA) estimates a 56% increase in global energy consumption by 2040 based on 2010 projections of 524 quadrillion Btu (EIA, 2013). Currently the primary source of energy is via combustion of finite fossil fuel resources that has caused considerable increase in atmospheric CO2 with ⇑ Corresponding author. Tel.: +44 207 594 5577. E-mail address: [email protected] (K. Hellgardt).
unprecedented levels reaching 400 ppm (Ewald, 2013). Utilisation of fuel in the transportation sector is particularly problematic as tailpipe emission accounts for the second highest CO2 release. An alternative to fossil fuels for transportation is to convert biomass to liquid fuel. Third generation feedstock such as algae has garnered significant interest recently due to its ability to fix carbon rapidly, high productivity and photosynthetic efficiency as well as its cultivation location not requiring agricultural land, thus not competing with food crops (Patel et al., 2012). A promising processing technology used to convert algal biomass to fuel precursors is Hydrothermal Liquefaction (HTL) which entails treatment of algal slurry at elevated temperature and pressure with water.
http://dx.doi.org/10.1016/j.biortech.2015.04.012 0960-8524/Ó 2015 Elsevier Ltd. All rights reserved.
Please cite this article in press as: Patel, B., Hellgardt, K. Hydrothermal upgrading of algae paste in a continuous flow reactor. Bioresour. Technol. (2015), http://dx.doi.org/10.1016/j.biortech.2015.04.012
2
B. Patel, K. Hellgardt / Bioresource Technology xxx (2015) xxx–xxx
Concentrating and drying of algal biomass is associated with severe energy penalty (Patel et al., 2012). Therefore, eliminating the drying process and directly processing concentrated wet algae using HTL is of great interest for conversion of algal biomass to biocrude by merely exploiting the modified properties of water at high temperature and pressure. As water approaches its critical point, the ionic product (Kw) increases, enabling it to act as an acid or base catalyst during reactions. Furthermore, the increase in dielectric constant and decrease in Hildebrand solubility parameter indicate that water can increasingly act as an organic solvent, enhancing solubility of organic molecules (Savage, 1999). The lipid, protein and carbohydrate macromolecules present in algae are susceptible to degradation/cracking and rapid hydrolysis under hydrothermal conditions. Hence, hydrothermal water is well suited for application as a biomass liquefaction medium. Compared to algal pyrolysis, HTL produces biocrude with better Higher Heating Value (HHV) and also offers the potential to recycle processed water and recover nutrients from the growth medium (López Barreiro et al., 2013). HTL is based on the premise that it is possible to deoxygenate and denitrify biomass, as an initial processing step, without the use of chemicals to produce biocrude. The biocrude produced from HTL has a HHV similar to crude oil (Vardon et al., 2011). Further post processing of biocrude is necessary to either convert it directly to fuel or blending with fossil crude at refineries as demonstrated by Elliott et al. (2013). HTL has been studied by various researchers (Reddy et al., 2013; Valdez et al., 2012; Vardon et al., 2011, 2012; Garcia Alba et al., 2011; Chen et al., 2014) to produce biocrude with variable properties such as energy content, boiling point distribution, (CHNO) elemental content and product yield using several different species of algae under different reaction conditions (heating rate, volume, water:algae ratio and autogenic pressure). But, the majority of research thus far has focused primarily on the use of batch reactor systems. The outcome of these investigations concluded that liquefaction at low Residence Time (RT) could be suitable to produce algal biocrude (Patel and Hellgardt, 2013; Faeth et al., 2013) since catalytic post processing is always necessary and energy expenditure can be limited by shorter contact time during HTL. However, batch reactors are not entirely suitable for biomass liquefaction at short RT due to the variability in heating rate, heating time, uncertainty in autogenic pressure/mixing and post reaction cooling. It is also clear that at scale, only flow reactors are likely to be implemented. Hence, evidence suggests a gradual shift in the research community moving towards implementation of continuous flow reactors (Elliott et al., 2015). Jazrawi et al. (2013) and Elliott et al. (2013) have both successfully demonstrated the application of a pilot scale flow reactor system for HTL of Nannochloropsis and Chorella sp., but the implementation of such system at a much smaller laboratory scale has not yet been attempted for detailed studies of algal HTL. The aforementioned pilot studies only employed neat water during processing and RTs of minutes. From trials in the laboratory it was found that the HTL flow reactor system was susceptible to solid char deposition (especially post reactor) and potential catalytic wall effects (Maiella and Brill, 1998; Chakinala et al., 2009; Potic et al., 2004) arising from the SS316 material under hydrothermal conditions which could contribute towards the HTL reaction. In addition to the charring, the biocrude yield was also consistently low (approximately 20 wt.% at RT between 1 and 5 min and temperature 300–350 °C) during trials, thus further improvement was necessary. To address these issues, firstly the reactor wall was lined with an inert quartz liner, and in order to promote reactive extraction and minimise char deposition, a co-solvent namely cyclohexane was fed into the reactor during HTL. From previous work in the laboratory, cyclohexane was found to be rather stable under
hydrothermal conditions. Short to medium chain hydrocarbons (or kerosene) or alcohols could also be used but due to their potential to skew the product mixture during analysis, these were not selected. The outcome from the co-solvent and quartz liner during further trials of the reactor system showed improvement in the biocrude yields confirming adverse effect in the system in their absence. Thus, based on the above and scoping the literature it can be confirmed that a quartz lined extractive flow reactor at laboratory scale has not been demonstrated previously for algal HTL. This work demonstrates the use of such a system and presents the findings of hydrothermal algal biomass conversion (or hydropyrolysis) to biocrude under various reaction conditions. 2. Method/experimental 2.1. Flow reactor The schematic of the in-house constructed reactor is presented in Fig. 1. The quartz liner was enclosed in a 0.95 cm outer diameter SS316 L tube of length 20 cm to give a total reactor volume of 2 cm3. For a typical reaction, the reactor system was first pressurised to 180 bar by a JASCO 780 piston pump and a ISCO 1820 syringe pump to feed DIW (De-Ionised Water) (later algae paste) and cyclohexane, respectively. Consecutively, the reactor heater was set to the required temperature and the pump flow rates adjusted to achieve the desired RT, and maintain a 10 vol.% cyclohexane concentration through the 2 ml reactor. Once the reaction temperature was attained, the system was allowed to stabilise for approximate 5 RT (or 5 reactor volume) and then the reciprocating High-Performance Liquid Chromatography (HPLC) pump feed switched to pre-prepared algae solution of 1.5 wt.% Nannochloropsis sp. biomass concentration. The algal solution feed tank was stirred to prevent the algae from settling. Based on initial trial runs it was confirmed that the reactor takes approximately 5 RT (or 5 reactor volume (10 ml)) to reach steady-state after which sample collection was initiated continuously for the duration of the reaction. Upon pumping 150 ml of algae solution, the reactor heater was switched off and the reactor flushed thoroughly with a total volume of 50 ml DIW and cyclohexane at the reaction flow rate. The ‘post reaction’ sample of about 50 ml was collected separately whilst the reaction temperature reduced to ambient. The reactor system was further dismantled and the Back Pressure Regulator (BPR) flushed with cyclohexane:water mixture. Only a small quantity of produced residual organic matter was collected upstream to the BPR using this method. The primary 150 ml solution was then filtered, filter paper rinsed with Dichloromethane (DCM) and the filtrate transferred to a separating funnel to obtain the DCM soluble organic phase from which the solvent was evaporated under a steady stream of nitrogen to obtain the biocrude. Given the low concentration of cyclohexane compared to DCM, both solvents evaporated readily at room temperature within 5 h. The filter paper was allowed to dry and a sample of the retained solids collected and weighed for elemental analysis. The leftover material, classed earlier as ‘post reaction’ sample was now treated as above. Typically the ‘post reaction’ organic phase extract contained 10–15 wt.% of the total extracted biocrude. 2.2. Size Exclusion Chromatography (SEC) SEC was used to determine the approximate molecular weight distribution of produced biocrude. Roughly 10 mg of biocrude was dissolved in 2 ml 1-methyl-2-pyrrolidinone (NMP) and filtered through a 0.45 lm PTFE filter. 10 ll of the sample was injected in a
Please cite this article in press as: Patel, B., Hellgardt, K. Hydrothermal upgrading of algae paste in a continuous flow reactor. Bioresour. Technol. (2015), http://dx.doi.org/10.1016/j.biortech.2015.04.012
3
B. Patel, K. Hellgardt / Bioresource Technology xxx (2015) xxx–xxx
Syringe Pump
P1
T2 Check Valve
Back-Pressure Regulator Vent
Cooling Coil
Cyclohexane
Clamp Heater
Phase Separator
Liquid Sample Reactor
HPLC Pump
T1
Algae Slurry
Fig. 1. Schematic of laboratory scale fast HTL reactor.
pump with NMP and chloroform mixture (6:1 vol./vol.%) as the mobile phase flowing at 0.5 ml/min. A ‘‘Mixed D’’ column (Polymer Laboratories, UK) packed with polystyrene/polydivinylbenzene at 50 °C and a Perkin-Elmer LC 290 variable wavelength UV absorbance detector operating at 300 nm was used to take measurements. The data was collected, processed and normalised with respect to polystyrene standard.
2.3. Gas Chromatography–Mass Spectroscopy (GC–MS) A HP GCD 1800A with a J&W scientific HP-5 column (30 m 0.32 mm ID and 0.25 lm film thickness) fitted with an auto sampler operated under constant Helium pressure and equipped with an electron ionisation detector with a source temperature of 300 °C was used. The auto sampler injected 2 ll sample of 3 lg biocrude dissolved in 1 g DCM at 300 °C inlet temperature. After an initial hold time of 2 min, the oven temperature was raised from 60 °C to 300 °C at 10 °C per minute with a final hold time of 4 min. Ion fragmentation from collected data was identified using preinstalled National Institute of Standards and Technology (NIST) Reference database 1A linked to Chemstation 10.
2.4. Simulated Distillation (SIMDIST) SIMDIST of biocrude samples was carried out on an Agilent 5890 gas chromatograph machine equipped with a flame ionisation detector and a Phenomenex Inferno HT5 column. 2 wt.% sample of biocrude was dissolved in DCM and a 2 ll of sample injected at 350 °C using an auto sampler. The temperature was held for 2 min at 60 °C and ramped at 10 °C per min to 380 °C with an additional holding time of 3 min. Chemstation software was used to obtain integration values of signals and these were processed manually to construct the boiling point curve. Atmospheric equivalent boiling point was obtained by using an alignment chart nomograph and boiling point confirmed using hydrocarbon standards.
2.5. Attenuated Total Reflection–Fourier Transform Infrared Spectroscopy (ATR–FTIR) Approximately 50 lg of DCM free biocrude sample was deposited on the ATR crystal of a Perkin-Elmer Spectrum 100 FTIR. Spectra from 4000–600 cm 1 band and 50 replicate scans were collected for triplicate samples. Freeze dried algal biomass analysis was performed by compacting the powder on the ATR crystal. Background scan was performed under ambient atmosphere prior to all analysis.
2.6. Elemental Carbon, Hydrogen, Nitrogen and Oxygen (CHNO) Samples of DCM extracted biocrude, oven dried filtered char and dried aqueous phase soluble solids were sent for Carbon, Hydrogen and Nitrogen elemental analysis to the University of Sheffield. The O content was found by difference. From previous HTL investigation and subsequent Karl-Fischer titration it can be confirmed that there was negligible water content in the biocrude (Patel and Hellgardt, 2013) and was not analysed. 3. HTL – results and discussion HTL of Nannochloropsis sp. microalgae was carried out in a quartz lined continuous plug flow reactor at 300, 325, 350 and 380 °C at RTs of 0.5, 1, 2 and 4 min. The lipid and ash content of the biomass, as determined through previous investigation (Patel and Hellgardt, 2013) was 20 and 11 wt.%, respectively. 3.1. Gravimetric yields The gravimetric yield of the DCM soluble organic phase, defined as biocrude, DCM insoluble phase classed as Water Soluble Fraction (WSF) and gas (+loss) found by difference is shown in Fig. 2a–c, respectively. Supplementary Information 1 – Fig. 1 shows the mass balance for the yields based on quantified biocrude, WSF and char balanced by gas yield (+experimental loss) found by difference. The highest biocrude yield achieved in this study is 38 wt.% at 380 °C and 30 s RT. The biocrude yield appears to be constant at around 30 wt.% for most reaction conditions and the high yield at short contact time and high temperature is in agreement with other investigations (Faeth et al., 2013; Jazrawi et al., 2013; Patel and Hellgardt, 2013). Faeth et al. (2013) reported that at a temperature of 500 °C and batch holding time of 1 min, biocrude yield of 55 wt.% was obtained compared to 52 wt.% (Valdez et al., 2012) at longer RT. The highest reported yield in this work is similar to previous batch reactor produced biocrude at 380 °C and 5 min RT (Patel and Hellgardt, 2013). Jazrawi et al. (2013) reported a yield of 41.7 wt.% at 350 °C and 3 min RT in a pilot scale continuous reactor. The variability in yields is partly down to the type of algae used, difference in sample reproduction and the reactor temperature profile. It is apparent that high temperature and shorter RT do indeed increase the biocrude yield, and a recent investigation by Bach et al. (2014) on batch HTL of macroalgae demonstrated that at their investigated heating rate of 585 °C/min, biocrude yield of 70 wt.% was achieved. However, the chemistry behind such processes is not fully understood. One potential explanation put forward by Faeth et al. (2013) is that by limiting the heating time, the
Please cite this article in press as: Patel, B., Hellgardt, K. Hydrothermal upgrading of algae paste in a continuous flow reactor. Bioresour. Technol. (2015), http://dx.doi.org/10.1016/j.biortech.2015.04.012
4
B. Patel, K. Hellgardt / Bioresource Technology xxx (2015) xxx–xxx
a
b
80
c
70
Gas and Loss *(wt.-%)
60 50 40 30
30 sec
20
1 min
10
2 min 4 min
0 300
325
350
Temperature (°C) Fig. 2. Gravimetric yield from solvent aided HTL (a) biocrude, (b) Water Soluble Fraction, (c) gas (+loss). ⁄Found by difference.
degradation of algal molecules takes place without further unwanted side reactions. Under hydrothermal conditions, cell rupture will be relatively easy via hydrolysis of the polysaccharides, thus rapidly exposing the cell contents to hydrothermal conditions. Subsequently, the released intracellular material is attacked by
hydrothermal water and only limited degradation occurs. This is similar to pyrolysis, whereby the inert gas is replaced in this case by water, thus resembling hydropyrolysis. Based on previous batch reactor investigations (Faeth et al., 2013; Patel and Hellgardt, 2013; Garcia Alba et al., 2011; Jena
Please cite this article in press as: Patel, B., Hellgardt, K. Hydrothermal upgrading of algae paste in a continuous flow reactor. Bioresour. Technol. (2015), http://dx.doi.org/10.1016/j.biortech.2015.04.012
B. Patel, K. Hellgardt / Bioresource Technology xxx (2015) xxx–xxx
et al., 2011), the observed high yield at low RT and high temperature suggests that at even higher temperature it might be possible to increase the yield substantially provided that short contact time is maintained in addition to the rapid heating rate (Bach et al., 2014). This would be easier to achieve in a flow reactor assuming heat transfer limitation is overcome and char formation limited. The most obvious trend observed in the gravimetric yield is the WSF. At shorter RT and lower temperature, the majority of the mass from the reaction tends to partition into the aqueous phase. At 300 °C and 30 s RT, almost 60 wt.% matter accumulates in the aqueous phase. As the reaction condition becomes more severe, the aqueous soluble matter is observed to decrease and complemented by an increase in produced gas (Fig. 2c). From this study, it is apparent that severe conditions (higher temperature and longer RT) are unlikely to increase the biocrude yield because from Fig. 2a it can be seen that the biocrude fraction of the product pool does not change significantly, but there is a remarkable change in the WSF wt.%. This implies that the conversion of material in the aqueous phase is towards formation of gases. Garcia Alba et al. (2011) made similar observations, achieving 47.4 and 62.8 wt.% gas yield at 450 °C and batch holding time of 5 and 60 min, respectively. The increased gas content caused a minute decrease in char content but significant mass was lost from the WSF. The char formation in their study was fairly constant with only a small decrease at severe conditions, which is similar to this work (approximately 5–10 wt.%). The gas phase was difficult to analyse due to nature of the sample size, but from previous investigations (Patel and Hellgardt, 2013; Garcia Alba et al., 2011), it is clear that produced gas is predominantly composed of CO2 and it is not expected to be different for this work. Additionally, the HTL aqueous phase contains significant amount of nitrogen (N) as NH+4 (Biller et al., 2012) that could potentially be released into the gas phase, partially explaining the increase in gas yield and reduction in WSF. It should be noted that water was evaporated to obtain solid WSF for elemental CHN(O) analysis. It is possible that some of the N based compounds could have been lost during this evaporation. To ensure and confirm that cyclohexane did not degrade during processing, the volume of the solvent fed and recovered post reaction was measured. Generally, a variable error in recovery of approximately 5 vol.% was obtained and no trend was identified with respect to changing reaction conditions. 3.2. ATR–FTIR spectroscopy ATR–FTIR (Supplementary Information 1 – Fig. 2) was used to gain a better understanding of the composition of the produced biocrude by obtaining insight on the functional groups present in the samples. The CAH bond stretching and vibration is represented by the 3000–2800 cm 1 and 1490–1350 cm 1 band in the spectrum and the signal in this region is characteristic of samples with high H and C content as present in the biocrudes (Vardon et al., 2012). On the other hand the OAH stretching vibration between 3500 and 3000 cm 1 from presence of alcohol or moisture is completely absent. However, the CAO stretch at 1320–1210 cm 1 and 1260–1000 cm 1 confirms presence of phenol and branched alcohol species. The most significant property that would contribute towards amenable biocrude for drop in refinery grade feedstock is the lack of N. However, heteroatom functionality and NAH bending (1680–1600 cm 1 and 1575–1525 cm 1) is abundant in all samples which potentially arise from amides, amines and other nitrogenous compounds. The polysaccharide derivatives vibration bands occur from OACAH, CACAH and CAOAH bonds which are found in the 1400–1199 cm 1 region (Leopold et al., 2011). This signal is quite abundant in the samples and the components with aforementioned functionality are responsible for the Maillard
5
reaction that gives rise to the undesirable nitrogen heterocycles (Patel and Hellgardt, 2013), some of which were detected via GC–MS analysis. The C@O stretching at 1730–1700 1 and vibration in the 2000–1500 cm 1 range corresponds to aldehydes, ketones and carboxylic acids (Baloyi et al., 2012). Compared to fossil crude, the functionality of algal biocrude differs remarkably, predominantly due to lack of nitrogen heteroatoms. The prominent CAH functionality in the 3000–2800 cm 1 range arising from significantly higher C and H content is stronger in fossil crude compared to algal biocrude. Additionally, the absorbance in 900–700 cm 1 in crude oil shows greater aromatic content (Carbognani et al., 2003; Wilt et al., 1998), but not much is known about the aromatic content in algal crude (Patel and Hellgardt, 2013), which is potentially a secondary product. 3.3. CHN(O) distribution, van Krevelen diagram and HHV Table 1 shows the elemental composition of algal biomass, biocrude, WSF and char produced at 30 s, 2 min and 4 min RT. In the biocrudes, the O and N levels have both decreased with respect to algal feedstock as expected. O composition in the biocrude range from 14.25–8.94 wt.%, which corresponds to a decrease of at least 55% from unprocessed biomass. This decrease is complemented by denitrogenation of at least 45% to produce biocrude with N content of 2.35–4.11 wt.%. Although overall there is a decrease in N content, the extent to which this takes place is opposite to that of O depending on the reaction condition. For instance at 30 s RT for all temperatures, the O removal is noticed to improve but the amount of denitrogenation decreases. This trait is generally observed at all reaction conditions. The elemental N content tends to reach a stable concentration level after which there is no change in its wt.% thus, elucidating that nitrogen containing compounds are stable under hydrothermal conditions. Faeth et al. (2013) and Garcia Alba et al. (2011) had a similar trend in their HTL biocrude and it is understood that the change is procured by increasing temperature. Hence, this work demonstrates that denitrogenation can be achieved rapidly (in seconds) eliminating the need for more severe reaction conditions. The van Krevelen plot, originally introduced to represent a graphical relationship between C, H and O ratio in coal processing is used to show the O:C, H:C and N:C atomic ratio of the produced biocrude. As seen from plot Fig. 3a and b, it is obvious that there is O and N removal after fast (HTL) of microalgae. The biocrudes produced in this study are concentrated approximately around H:C ratio of 1.6 and an O:C ratio of 0.15. There is a slight increase in deoxygenation at longer RT, but when compared to previous batch reactor HTL study (Patel and Hellgardt, 2013), it is deduced that longer operating time does not significantly improve aforementioned ratios. Based on current HTL studies, algal biocrude produced in this investigation contains the lowest O:C ratio using a continuous flow reactor at short contact time. Biocrude produced in this investigation contains marginally lower O than that from the pilot scale HTL study by Jazrawi et al. (2013). The primary aim of HTL is to degrade macromolecules and produce a product with negligible oxygen and nitrogen content. Compared to fossil crude and coal, the O:C ratio is particularly high in unprocessed biomass. After HTL the O decreases to reach a similar level to that of coal. The extent of O removal is related to processing conditions as seen in Table 1 and previous work (Patel and Hellgardt, 2013). The elemental H composition increases from an initial untreated algal biomass value of 6.6 wt.% to approximately 10 wt.%, for almost all reaction conditions and the increase in H content causes a subsequent increase in C content resulting in formation of stable CAH bonds (McKendry, 2002). Consequently, the accompanying loss of
Please cite this article in press as: Patel, B., Hellgardt, K. Hydrothermal upgrading of algae paste in a continuous flow reactor. Bioresour. Technol. (2015), http://dx.doi.org/10.1016/j.biortech.2015.04.012
6
B. Patel, K. Hellgardt / Bioresource Technology xxx (2015) xxx–xxx
Table 1 HHV, yield and elemental CHNO composition of biocrude, char, WSF and algal biomass (Nannochloropsis sp.).
Nannochloropsis sp.
C 43.2
RT (min)
Temp (°C)
Biocrude
0.50
300 325 350 380 300 325 350 380 300 325 350 380
73.30 74.15 74.25 74.16 73.95 75.75 76.25 75.89 73.81 75.01 75.49 77.02
2.00
4.00
H 6.6
N 7.2
O 32.2
HHV (MJ/kg) 18.28
C
10.10 10.40 10.34 10.34 10.15 10.44 9.92 10.04 10.24 10.37 10.05 10.41
2.35 2.76 3.00 3.25 3.91 3.50 3.78 3.74 4.11 3.70 3.50 3.63
14.25 12.69 12.41 12.25 11.99 10.31 10.05 10.33 11.84 10.92 10.20 8.94
36.63 37.62 37.62 37.61 37.32 38.64 38.12 38.11 37.43 38.18 38.02 39.27
45.05 55.00 55.00 56.12 22.64 22.65 27.21 37.34 21.61 23.00 26.39 27.98
H
N
O
5.68 7.24 3.84 6.42 6.26 3.10 3.18 4.56 3.31 2.50 2.75 2.39
8.39 7.00 8.00 8.15 2.07 2.76 3.05 5.38 2.62 1.65 2.55 3.10
40.88 30.76 33.16 29.31 69.03 71.49 66.56 52.72 72.46 72.85 68.31 66.53
Char
C
H
N
O
Water Soluble Fraction (WSF) 36.63 34.84 31.20 37.04 31.38 30.44 20.63 21.32 29.01 23.02 16.75 14.98
4.21 3.28 3.36 3.97 3.68 3.02 1.61 1.62 2.87 1.88 0.52 0.50
8.24 7.99 6.70 7.69 6.84 6.43 3.05 3.08 6.06 4.61 2.63 2.15
50.92 53.89 58.74 51.30 58.10 60.11 74.71 73.98 62.06 70.49 80.10 82.37
Fig. 3. van Krevelen plot of algal biocrude (a) H:C vs. O:C, (b) H:C vs. N:C.
O potentially reduces the number of CAO bonds present in the biocrude, thus negating any improvement in oxidative stability but at the same time it helps improve the HHV. Typically the N content of fossil crude is <1 wt.% corresponding to a N:C ratio of <0.002 (Speight, 2004). Algal biocrude is far from this figure and HTL alone is not capable of reaching such concentration. The N:C ratio of produced biocrude is low compared to the unprocessed feedstock and the set of experiments conducted in this study also provides some useful insight on N partitioning,
particularly as the WSF at short RT. The N content of the processed water at moderate conditions is high and as conditions become more severe, there is a gradual loss in N mass fraction. Conversely the O content follows the opposite trend. The N is potentially released from the water as a gaseous product (ammonia), hence in order to retain maximum N, reaction conditions of 380 °C and 30 s appear to be most appropriate. The energy content (HHV) of the samples can be calculated based on the elemental composition. It is widely accepted that
Please cite this article in press as: Patel, B., Hellgardt, K. Hydrothermal upgrading of algae paste in a continuous flow reactor. Bioresour. Technol. (2015), http://dx.doi.org/10.1016/j.biortech.2015.04.012
B. Patel, K. Hellgardt / Bioresource Technology xxx (2015) xxx–xxx
the Dulong’s equation which incorporates the elemental C, H, O and S composition, is suitable for algal based HHV approximation. Typically, the last S term is ignored due to the low S concentration in algae. In this study the HHV has a range between 36.64 and of 39.27 MJ/kg, in comparison crude oil has a HHV in the range of 40–50 MJ/kg (Speight, 2004). The highest HHV was produced at the most severe condition investigated (380 °C and 4 min) which was consequently the biocrude with the smallest quantity of O. The decrease in O content of the samples is desirable as it increases the HHV of the fuel. Although biocrude from HTL of algae offers better elemental ratios compared to lignin based fuels (Patel and Hellgardt, 2013), the oxygen and nitrogen content is still problematic. Biocrude with high N cannot be blended with fossil crude unless an additional denitrogenation step is added prior to blending, resulting in additional costs. Catalytic deoxygenation or hydrotreating is also necessary to improve fuel quality, but the presence of nitrogen heterocycles, as identified by GC–MS usually causes catalyst poisoning making denitrogenation challenging.
3.4. Molecular weight analysis – SEC SEC analysis of the biocrude samples shows the average molecular weight of the compounds present in the oil. Supplementary Information 1 – Fig. 3 infers that processing condition has a noticeable, but not significant impact on the average molecular weight of the species present in the biocrude. At reaction condition of 380 °C and 4 min, the average Mw of 200 is detected. For all other analysed biocrudes, a Mw range between 500 and 1000 is observed. The four samples from 30 s RT at 380 °C and 300 °C and the 4 min RT sample at the same temperatures show that a lower Mw distribution is achieved, confirming that the effect of temperature is greater at producing lower average Mw compounds opposed to the RT. From analysed samples it is observed that the Mw distribution of the biocrude is bimodal unlike the biocrude produced by Vardon et al. (2011) which had an approximate Mw of 350. When the reaction condition (particularly RT) is less severe, a second batch of molecules around Mw 1000 kDa is seen. The signal intensity for aforementioned Mw region, for biocrude from least severe reaction condition at 300 °C and 30 s is strongest and predictably weakest at the most severe condition. Presumably polymeric
7
substances present in algae, such as polysaccharides and polypeptides that possess high Mw could give rise to these. Garcia Alba et al. (2011) observed similar ‘clusters’ of compounds in their GPC analysis of batch produced biocrude at 250 °C and 5 min RT, but not as heavy as those in this investigation.
3.5. Boiling point distribution – SIMDIST SIMDIST analysis of selected samples of biocrude produced at various reaction conditions are shown in Fig. 4. To compare the boiling point fraction, the results from previous batch reaction investigation (Patel and Hellgardt, 2013) and North Sea fossil crude are included. From the onset it is evident that the boiling point distribution of the fast liquefaction biocrude is not similar to either crude oil or biocrude formed from batch liquefaction. The atmospheric equivalent boiling point of biocrude from fast liquefaction appears to point towards abundance of compounds with boiling point greater than 265 °C which corresponds to C chain length of C15 or greater. Although the SIMDIST investigation is carried out for compounds with atmospheric equivalent boiling point up to 450 °C, it is highly unlikely that all of the oil is detected within this range. It is expected that an unknown proportion of oil might have a boiling point beyond this and can be found with additional thermograviemtric analysis (not presented in this manuscript). The biocrude produced predominantly has compounds with boiling point within the range of gas oil (265–380 °C) and atmospheric residue (380 °C<). The effect of temperature is not as pronounced as that of RT. With increasing reaction severity the fraction of lower boiling point light gas oil region compounds is greater. Unlike fossil crude and batch reaction produced biocrude, there is a clear lack of short carbon chain length compounds. The lower boiling point fraction present in small quantities is largely in the kerosene range (190–265 °C) and an even smaller quantity in the heavy naphtha (150–190 °C) distillation cut. The biocrude produced in this continuous flow liquefaction investigation is different from that produced by Jazrawi et al. (2013), where more kerosene range fraction was present. The difference could be due to the variability in the lipid, protein and carbohydrate content of the algal species used and also potentially from sample preparation. From the SEC and SIMDIST analysis it is evident that fast HTL potentially takes place in two steps. The first part is the rapid degradation and hydrolysis of algal components as
Fig. 4. SIMDIST of algal biocrude (batch and continuous) and crude oil. ⁄Patel and Hellgardt (2013).
Please cite this article in press as: Patel, B., Hellgardt, K. Hydrothermal upgrading of algae paste in a continuous flow reactor. Bioresour. Technol. (2015), http://dx.doi.org/10.1016/j.biortech.2015.04.012
8
B. Patel, K. Hellgardt / Bioresource Technology xxx (2015) xxx–xxx
suggested by Faeth et al. (2013) and the second being the multitude of reactions that produce significant pool of compounds resulting in formation of lower Mw species. The presence of a negligible quantity of biocrude within the kerosene fraction at longer RT might not necessarily result in more economical operating conditions. The longer RT, lower gravimetric biocrude yield and only a small increase in overall deoxygenation does not seem advantageous enough to pursue considering the energy penalty.
Diels–Alder addition and ring rearrangement (Demirbas, 2009). Similarly, heteroatoms are prevalent at more severe conditions which help compound this multiple reaction pathway suggestion. Furthermore, there appear to be certain compounds such as the fatty acids, which are quite stable. Under the conditions used in this investigation, it was expected that the unstable molecules will degrade quicker, hence producing stable ones, necessating a harsher environment for their removal. Identification of these molecules is important because they prevail for longer and prevent further deoxygenation/denitrogenation.
3.6. Chemical characterisation – GC–MS Table 1 (Supplementary Information 2) shows the list of detected compounds for biocrude produced at three different reaction conditions. Spectra with total chromatogram peak area of greater than 1% were identified using the NIST database. It should be noted that only a proportion of the biocrude produced lies within the temperature range in which the GC–MS operates and thus only a fraction of the components in the biocrude can be deciphered. Additionally, there might have been some loss of lighter compounds during solvent extraction and sample preparation. From the list of compounds in Table 1 (Supplementary Information 2) it becomes apparent that under severe conditions there is more degradation of macromolecules present in algal biomass. It would be expected that proteins, lipids and carbohydrates will rapidly break down to their subsequent products under HTL conditions. The CAN bond present in proteins will easily hydrolyse to form its primary product amino acids along with some hydrocarbons, amines, aldehydes and acids (Brunner, 2009). At longer RTs, these degradation/hydrolysis products can react with carbohydrate hydrolysis product, monosaccharide, to produce a range of products. Thus, there is a larger pool of compounds present at 380 °C and 4 min RT. The lipids (Triacylglycerols – TAG) are expected to hydrolyse to glycerol and their subsequent fatty acids which can further isomerise or hydrolyse to alkanes. At 380 °C and 30 s RT, the presence of triolein (25.6 min) suggests that the reaction condition is not sufficient to hydrolyse all TAG species. Similarly, a direct comparison of the three reaction conditions shows that there is a lack of nitrogen containing compounds at 300 °C and 30 s. Although, the product pool does not show many nitrogen containing compounds, the CHN analysis for the three biocrudes is quite similar. One reason for this might be that the products formed under less severe conditions are not detected within the range of the GC–MS and/ or a multitude of different N containing compounds formed produce a signal similar to the background. There is a significant number of methyl derived alkylated compounds in all compound classes from cleavage of CACH3 bonds, which appear to be easiest to break. The unsaturated species are understood to be derived from dehydration, cyclization,
4. HTL reactor system consideration Table 2 shows a comparison of some key indicators of HTL reaction similar to this investigation (low RT and continuous HTL). Looking at the feed concentration, the 1.5 wt.% biomass concentration used in this work was primarily due to the pumping limitation of the HPCL pump. Higher concentrations caused a blockage of the pump feed valve resulting in a loss of flow. Secondly, the lower biomass concentration combined with solvent extraction aided the reaction by limiting char formation thus preventing the BPR from blocking. Jazrawi et al. (2013) had a similar issue downstream related to the pressure control valve orifice which limited the biomass concentration used in their study (Elliott et al., 2015). Considering that algal biomass:water ratio does not make a substantial impact on HTL reaction (Jena et al., 2011; Garcia Alba et al., 2011) as well as the ability for commercial HTL reactors to be designed with 31.5% (Knorr et al., 2013) biomass handling capacity (for lignocellulosic feedstock), it is of utmost importance to select an appropriate BPR to ensure it does not limit the amount of algae used in the reaction. Elliott et al. (2013) have already demonstrated a slurry concentration of 35 wt.% at pilot scale, so a 40 wt.% commercial algal HTL reactor system is realistically feasible. Pumping higher biomass concentration will result in greater quantity of biocrude yield per unit mass which is essential to maximise the energy return, but this was not necessarily the aim of the reactor used in a laboratory for this investigation. Hence, no provision was made for energy economy at this stage. The primary aim of using cyclohexane was to limit char agglomeration in the system by ensuring there is constant solvation of the formed products. Pilot scale investigation by Jazrawi et al. (2013) and Elliott et al. (2013) did not use a co-solvent but it is expected that the use of solvents would be beneficial as it can aid produce better quality biocrude and increase yield (Xiu and Shahbazi, 2012; Singh et al., 2015). Apart from cyclohexane, solvents such as acetone, n-hexane, 1-4-dioxane or mixture of middle distillates (kerosene/diesel) as well as alcohols (ethanol/methanol or butanol) (Xiu and Shahbazi, 2012; Liu and Zhang, 2008 and Singh et al., 2015) could be used.
Table 2 HTL performance comparison of short RT and continuous reactor system.
* #
Reference and reactor type (vol.)
Biomass loading (wt.%)
Reaction conditions (Temp/P/RT)
HHV (MJ/kg)
Biocrude yield (wt.%)
Faeth et al. (2013) Batch (1.67 ml) Elliott et al. (2013) Continuous (0.4–1 L) Jazrawi et al. (2013) Continuous (2 L) Eboibi et al. (2014) Batch (1 L) This study# Continuous (2 ml)
32
300–600 °C/autogenic/1–3 min
33.3–36.7
13–66
17–35
350 °C/200 bar/15 min
39.4–40.1
38–63.6*
1–10
250–350 °C/150–200 bar/3–5 min
27.9–33.8
<10–41.7
16
300–350 °C/autogenic/5 min
32.1–36.7
34–58
1.5
300–380 °C/180 bar/0.5–4 min
36.6–39.3
23.13–38.0
HHV found using Dulong’s equation. Cyclohexane used as a co-solvent.
Please cite this article in press as: Patel, B., Hellgardt, K. Hydrothermal upgrading of algae paste in a continuous flow reactor. Bioresour. Technol. (2015), http://dx.doi.org/10.1016/j.biortech.2015.04.012
B. Patel, K. Hellgardt / Bioresource Technology xxx (2015) xxx–xxx
There is still a wide gap in knowledge for understanding algal HTL in flow reactors, confirmed by lack of relevant literature as shown in Table 2. Thus utilisation of an economically frugal laboratory scale continuous reactor system to perform short RT reactions and investigate the various parameters for algal HTL is important and necessary. 5. Conclusion The HTL process demonstrated in this investigation shows that it is possible to convert algal biomass to biocrude at a relatively short RT of 30 s. The aqueous stream is of utmost importance because the CHNO analysis showed higher concentration of N in the dried WSF. Since the flow characteristics of the system do not seem to have influenced the yield (Supplementary Information 3), theoretically, higher biocrude yields may be achieved at even higher temperatures and shorter RT (Bach et al., 2014) and should be investigated for its feasibility. Whether ultra-short hydropyrolysis can tackle this further remains to be seen. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.biortech.2015.04. 012. References Bach, Q.V., Sillero, M.V., Tran, K.Q., Skjermo, J., 2014. Fast hydrothermal liquefaction of a Norwegian macro-alga: screening tests. Algal Res. 6, 271–276. Baloyi, H., 2012. Algae Liquefaction (Doctoral dissertation). North-West University. Biller, P., Ross, A.B., Skill, S.C., Lea-Langton, A., Balasundaram, B., Hall, C., Riley, R., Llewellyn, C.A., 2012. Nutrient recycling of aqueous phase for microalgae cultivation from the hydrothermal liquefaction process. Algal Res. 1, 70–76. Brunner, G., 2009. Near critical and supercritical water. Part I. Hydrolytic and hydrothermal processes. J. Supercrit. Fluids 47, 373–381. Carbognani, L., Espidel, J., Colaiocco, S., 2003. Characterization of heavy oils and heavy ends. In: Anal. Adv. Hydrocarbon Res. Springer, US, pp. 337–368. Chakinala, A.G., Brilman, D.W., van Swaaij, W.P., Kersten, S.R., 2009. Catalytic and non-catalytic supercritical water gasification of microalgae and glycerol. Ind. Eng. Chem. Res. 49, 1113–1122. Chen, W.H., Lin, B.J., Huang, M.Y., Chang, J.S., 2014. Thermochemical conversion of microalgal biomass into biofuels: a review. Bioresour. Technol. 184, 314–327. Demirbas, A., 2009. Pyrolysis of biomass for fuels and chemicals. Energy Sources Part A 31, 1028–1037. Eboibi, B.E.O., Lewis, D.M., Ashman, P.J., Chinnasamy, S., 2014. Hydrothermal liquefaction of microalgae for biocrude production: improving the biocrude properties with vacuum distillation. Bioresour. Technol. 174, 212–221. Elliott, D.C., Biller, P., Ross, A.B., Schmidt, A.J., Jones, S.B., 2015. Hydrothermal liquefaction of biomass: developments from batch to continuous, process. Bioresour. Technol. 178, 147–156. Elliott, D.C., Hart, T.R., Schmidt, A.J., Neuenschwander, G.G., Rotness, L.J., Olarte, M.V., Zacher, A.H., Albrecht, K.O., Hallen, R.T., Holladay, J.E., 2013. Process development for hydrothermal liquefaction of algae feedstocks in a continuousflow reactor. Algal Res. 2, 445–454. Energy Information Administration (US) (Ed.), 2013. International Energy Outlook 2013 with Projections to 2040. Government Printing Office.
9
Ewald, J., 2013. Carbon Dioxide at NOAA’s Mauna Loa Observatory reaches new milestone: Tops 400 ppm. NOAA Research, 10 May 2013. Faeth, J.L., Valdez, P.J., Savage, P.E., 2013. Fast hydrothermal liquefaction of Nannochloropsis sp. to produce biocrude. Energy Fuels 27, 1391–1398. Garcia Alba, L., Torri, C., Samorì, C., van der Spek, J., Fabbri, D., Kersten, S.R., Brilman, D.W., 2011. Hydrothermal treatment (HTT) of microalgae: evaluation of the process as conversion method in an algae biorefinery concept. Energy Fuels 26, 642–657. Jazrawi, C., Biller, P., Ross, A.B., Montoya, A., Maschmeyer, T., Haynes, B.S., 2013. Pilot plant testing of continuous hydrothermal liquefaction of microalgae. Algal Res. 2, 268–277. Jena, U., Das, K.C., Kastner, J.R., 2011. Effect of operating conditions of thermochemical liquefaction on biocrude production from Spirulina platensis. Bioresour. Technol. 102, 6221–6229. Knorr, D., Lukas, J., Schoen, P., 2013. Production of Advanced Biofuels via Liquefaction – Hydrothermal Liquefaction Reactor Design: April 5, 2013. NREL/SR-5100-60462, National Renewable Energy Laboratory, Golden, Colorado, USA. Leopold, L.F., Leopold, N., Diehl, H.A., Socaciu, C., 2011. Quantification of carbohydrates in fruit juices using FTIR spectroscopy and multivariate analysis. J. Spectrosc. 26, 93–104. Liu, Z., Zhang, F.S., 2008. Effects of various solvents on the liquefaction of biomass to produce fuels and chemical feedstocks. Energy Convers. Manage. 49, 3498– 3504. López Barreiro, D., Prins, W., Ronsse, F., Brilman, W., 2013. Hydrothermal liquefaction (HTL) of microalgae for biofuel production: state of the art review and future prospects. Biomass Bioenergy 53, 113–127. Maiella, P.G., Brill, T.B., 1998. Spectroscopy of hydrothermal reactions. 10. Evidence of wall effects in decarboxylation kinetics of 1.00 m HCO2X (X = H, Na) at 280– 330 °C and 275 bar. J. Phys. Chem. A 102, 5886–5891. McKendry, P., 2002. Energy production from biomass (Part 1): Overview of biomass. Bioresour. Technol. 83, 37–46. Patel, B., Hellgardt, K., 2013. Hydrothermal upgrading of algae paste: application of 31 P-NMR. Environ. Prog. Sustainable Energy 32, 1002–1012. Patel, B., Tamburic, B., Zemichael, F.W., Dechatiwongse, P., Hellgardt, K., 2012. Algal biofuels: a credible prospective? ISRN Renew. Energy 2012. Potic, B., Kersten, S.R.A., Prins, W., Van Swaaij, W.P.M., 2004. A high-throughput screening technique for conversion in hot compressed water. Ind. Eng. Chem. Res. 43, 4580–4584. Reddy, H.K., Muppaneni, T., Rastegary, J., Shirazi, S.A., Ghassemi, A., Deng, S., 2013. Hydrothermal extraction and characterization of bio-crude oils from wet chlorella sorokiniana and dunaliella tertiolecta. Environ. Prog. Sustainable Energy 32, 910–915. Savage, P.E., 1999. Organic chemical reactions in supercritical water. Chem. Rev. 99, 603–622. Singh, R., Bhaskar, T., Balagurumurthy, B., 2015. Effect of solvent on the hydrothermal liquefaction of macro algae Ulva fasciata. Process Saf. Environ. Prot. 93, 154–160. Speight, J.G., 2004. Petroleum Asphaltenes – Part 1: Asphaltenes, resins and the structure of petroleum. Oil Gas Sci. Technol. 59, 467–477. Valdez, P.J., Nelson, M.C., Wang, H.Y., Lin, X.N., Savage, P.E., 2012. Hydrothermal liquefaction of Nannochloropsis sp.: systematic study of process variables and analysis of the product fractions. Biomass Bioenergy 46, 317–331. Vardon, D.R., Sharma, B.K., Blazina, G.V., Rajagopalan, K., Strathmann, T.J., 2012. Thermochemical conversion of raw and defatted algal biomass via hydrothermal liquefaction and slow pyrolysis. Bioresour. Technol. 109, 178– 187. Vardon, D.R., Sharma, B.K., Scott, J., Yu, G., Wang, Z., Schideman, L., Zhang, Y., Strathmann, T.J., 2011. Chemical properties of biocrude oil from the hydrothermal liquefaction of Spirulina algae, swine manure, and digested anaerobic sludge. Bioresour. Technol. 102, 8295–8303. Wilt, B.K., Welch, W.T., Rankin, J.G., 1998. Determination of asphaltenes in petroleum crude oils by Fourier transform infrared spectroscopy. Energy Fuels 12, 1008–1012. Xiu, S., Shahbazi, A., 2012. Bio-oil production and upgrading research: a review. Renew. Sustainable Energy Rev. 16, 4406–4414.
Please cite this article in press as: Patel, B., Hellgardt, K. Hydrothermal upgrading of algae paste in a continuous flow reactor. Bioresour. Technol. (2015), http://dx.doi.org/10.1016/j.biortech.2015.04.012
Contents lists available at ScienceDirect
Bioresource Technology journal homepage: www.elsevier.com/locate/biortech
Hydrothermal upgrading of algae paste in a continuous flow reactor Bhavish Patel, Klaus Hellgardt ⇑ Imperial College London, Department of Chemical Engineering, Exhibition Road, South Kensington, London SW7 2AZ, UK
h i g h l i g h t s
g r a p h i c a l a b s t r a c t
Continuous HTL of algal biomass in a
quartz lined plug flow reactor. Production of algal biocrude at short
residence time. Analysis of biocrude from continuous
fast liquefaction.
a r t i c l e
i n f o
Article history: Received 28 January 2015 Received in revised form 1 April 2015 Accepted 2 April 2015 Available online xxxx Keywords: HTL Plug flow reactor Microalgae Biocrude Hydropyrolysis
a b s t r a c t This investigation demonstrates the utility of a novel laboratory scale continuous plug flow reactor for fast Hydrothermal Liquefaction (HTL) of microalgae in a quartz lined chamber. Reactions were carried out between 300 and 380 °C and residence times of 0.5–4 min. Cyclohexane was used as a co-solvent to enhance extraction and prevent char formation. Highest biocrude yield of 38 wt.% was achieved at 380 °C and 30 s as well as Water Soluble Fraction containing up to 60 wt.% matter recovered. Analysis of the biocrude showed that the extent of deoxygenation and denitrogenation after HTL varied and is dependent on the reaction conditions, Fourier Transform Infrared Spectroscopy analysis showed that biocrude contains similar functional moieties with only a small difference observed at different reaction conditions. Conversely, the Simulated Distillation and Size Exclusion Chromatography data showed that harsher conditions produced marginally better biocrude with improved boiling point profile and lower molecular weight compounds, respectively which was confirmed using Gas Chromatography–Mass Spectrometry. Ó 2015 Elsevier Ltd. All rights reserved.
1. Introduction The rise in global population and improvement in living standards is increasing the world’s energy demand. The U.S. Energy Information Administration (EIA) estimates a 56% increase in global energy consumption by 2040 based on 2010 projections of 524 quadrillion Btu (EIA, 2013). Currently the primary source of energy is via combustion of finite fossil fuel resources that has caused considerable increase in atmospheric CO2 with ⇑ Corresponding author. Tel.: +44 207 594 5577. E-mail address: [email protected] (K. Hellgardt).
unprecedented levels reaching 400 ppm (Ewald, 2013). Utilisation of fuel in the transportation sector is particularly problematic as tailpipe emission accounts for the second highest CO2 release. An alternative to fossil fuels for transportation is to convert biomass to liquid fuel. Third generation feedstock such as algae has garnered significant interest recently due to its ability to fix carbon rapidly, high productivity and photosynthetic efficiency as well as its cultivation location not requiring agricultural land, thus not competing with food crops (Patel et al., 2012). A promising processing technology used to convert algal biomass to fuel precursors is Hydrothermal Liquefaction (HTL) which entails treatment of algal slurry at elevated temperature and pressure with water.
http://dx.doi.org/10.1016/j.biortech.2015.04.012 0960-8524/Ó 2015 Elsevier Ltd. All rights reserved.
Please cite this article in press as: Patel, B., Hellgardt, K. Hydrothermal upgrading of algae paste in a continuous flow reactor. Bioresour. Technol. (2015), http://dx.doi.org/10.1016/j.biortech.2015.04.012
2
B. Patel, K. Hellgardt / Bioresource Technology xxx (2015) xxx–xxx
Concentrating and drying of algal biomass is associated with severe energy penalty (Patel et al., 2012). Therefore, eliminating the drying process and directly processing concentrated wet algae using HTL is of great interest for conversion of algal biomass to biocrude by merely exploiting the modified properties of water at high temperature and pressure. As water approaches its critical point, the ionic product (Kw) increases, enabling it to act as an acid or base catalyst during reactions. Furthermore, the increase in dielectric constant and decrease in Hildebrand solubility parameter indicate that water can increasingly act as an organic solvent, enhancing solubility of organic molecules (Savage, 1999). The lipid, protein and carbohydrate macromolecules present in algae are susceptible to degradation/cracking and rapid hydrolysis under hydrothermal conditions. Hence, hydrothermal water is well suited for application as a biomass liquefaction medium. Compared to algal pyrolysis, HTL produces biocrude with better Higher Heating Value (HHV) and also offers the potential to recycle processed water and recover nutrients from the growth medium (López Barreiro et al., 2013). HTL is based on the premise that it is possible to deoxygenate and denitrify biomass, as an initial processing step, without the use of chemicals to produce biocrude. The biocrude produced from HTL has a HHV similar to crude oil (Vardon et al., 2011). Further post processing of biocrude is necessary to either convert it directly to fuel or blending with fossil crude at refineries as demonstrated by Elliott et al. (2013). HTL has been studied by various researchers (Reddy et al., 2013; Valdez et al., 2012; Vardon et al., 2011, 2012; Garcia Alba et al., 2011; Chen et al., 2014) to produce biocrude with variable properties such as energy content, boiling point distribution, (CHNO) elemental content and product yield using several different species of algae under different reaction conditions (heating rate, volume, water:algae ratio and autogenic pressure). But, the majority of research thus far has focused primarily on the use of batch reactor systems. The outcome of these investigations concluded that liquefaction at low Residence Time (RT) could be suitable to produce algal biocrude (Patel and Hellgardt, 2013; Faeth et al., 2013) since catalytic post processing is always necessary and energy expenditure can be limited by shorter contact time during HTL. However, batch reactors are not entirely suitable for biomass liquefaction at short RT due to the variability in heating rate, heating time, uncertainty in autogenic pressure/mixing and post reaction cooling. It is also clear that at scale, only flow reactors are likely to be implemented. Hence, evidence suggests a gradual shift in the research community moving towards implementation of continuous flow reactors (Elliott et al., 2015). Jazrawi et al. (2013) and Elliott et al. (2013) have both successfully demonstrated the application of a pilot scale flow reactor system for HTL of Nannochloropsis and Chorella sp., but the implementation of such system at a much smaller laboratory scale has not yet been attempted for detailed studies of algal HTL. The aforementioned pilot studies only employed neat water during processing and RTs of minutes. From trials in the laboratory it was found that the HTL flow reactor system was susceptible to solid char deposition (especially post reactor) and potential catalytic wall effects (Maiella and Brill, 1998; Chakinala et al., 2009; Potic et al., 2004) arising from the SS316 material under hydrothermal conditions which could contribute towards the HTL reaction. In addition to the charring, the biocrude yield was also consistently low (approximately 20 wt.% at RT between 1 and 5 min and temperature 300–350 °C) during trials, thus further improvement was necessary. To address these issues, firstly the reactor wall was lined with an inert quartz liner, and in order to promote reactive extraction and minimise char deposition, a co-solvent namely cyclohexane was fed into the reactor during HTL. From previous work in the laboratory, cyclohexane was found to be rather stable under
hydrothermal conditions. Short to medium chain hydrocarbons (or kerosene) or alcohols could also be used but due to their potential to skew the product mixture during analysis, these were not selected. The outcome from the co-solvent and quartz liner during further trials of the reactor system showed improvement in the biocrude yields confirming adverse effect in the system in their absence. Thus, based on the above and scoping the literature it can be confirmed that a quartz lined extractive flow reactor at laboratory scale has not been demonstrated previously for algal HTL. This work demonstrates the use of such a system and presents the findings of hydrothermal algal biomass conversion (or hydropyrolysis) to biocrude under various reaction conditions. 2. Method/experimental 2.1. Flow reactor The schematic of the in-house constructed reactor is presented in Fig. 1. The quartz liner was enclosed in a 0.95 cm outer diameter SS316 L tube of length 20 cm to give a total reactor volume of 2 cm3. For a typical reaction, the reactor system was first pressurised to 180 bar by a JASCO 780 piston pump and a ISCO 1820 syringe pump to feed DIW (De-Ionised Water) (later algae paste) and cyclohexane, respectively. Consecutively, the reactor heater was set to the required temperature and the pump flow rates adjusted to achieve the desired RT, and maintain a 10 vol.% cyclohexane concentration through the 2 ml reactor. Once the reaction temperature was attained, the system was allowed to stabilise for approximate 5 RT (or 5 reactor volume) and then the reciprocating High-Performance Liquid Chromatography (HPLC) pump feed switched to pre-prepared algae solution of 1.5 wt.% Nannochloropsis sp. biomass concentration. The algal solution feed tank was stirred to prevent the algae from settling. Based on initial trial runs it was confirmed that the reactor takes approximately 5 RT (or 5 reactor volume (10 ml)) to reach steady-state after which sample collection was initiated continuously for the duration of the reaction. Upon pumping 150 ml of algae solution, the reactor heater was switched off and the reactor flushed thoroughly with a total volume of 50 ml DIW and cyclohexane at the reaction flow rate. The ‘post reaction’ sample of about 50 ml was collected separately whilst the reaction temperature reduced to ambient. The reactor system was further dismantled and the Back Pressure Regulator (BPR) flushed with cyclohexane:water mixture. Only a small quantity of produced residual organic matter was collected upstream to the BPR using this method. The primary 150 ml solution was then filtered, filter paper rinsed with Dichloromethane (DCM) and the filtrate transferred to a separating funnel to obtain the DCM soluble organic phase from which the solvent was evaporated under a steady stream of nitrogen to obtain the biocrude. Given the low concentration of cyclohexane compared to DCM, both solvents evaporated readily at room temperature within 5 h. The filter paper was allowed to dry and a sample of the retained solids collected and weighed for elemental analysis. The leftover material, classed earlier as ‘post reaction’ sample was now treated as above. Typically the ‘post reaction’ organic phase extract contained 10–15 wt.% of the total extracted biocrude. 2.2. Size Exclusion Chromatography (SEC) SEC was used to determine the approximate molecular weight distribution of produced biocrude. Roughly 10 mg of biocrude was dissolved in 2 ml 1-methyl-2-pyrrolidinone (NMP) and filtered through a 0.45 lm PTFE filter. 10 ll of the sample was injected in a
Please cite this article in press as: Patel, B., Hellgardt, K. Hydrothermal upgrading of algae paste in a continuous flow reactor. Bioresour. Technol. (2015), http://dx.doi.org/10.1016/j.biortech.2015.04.012
3
B. Patel, K. Hellgardt / Bioresource Technology xxx (2015) xxx–xxx
Syringe Pump
P1
T2 Check Valve
Back-Pressure Regulator Vent
Cooling Coil
Cyclohexane
Clamp Heater
Phase Separator
Liquid Sample Reactor
HPLC Pump
T1
Algae Slurry
Fig. 1. Schematic of laboratory scale fast HTL reactor.
pump with NMP and chloroform mixture (6:1 vol./vol.%) as the mobile phase flowing at 0.5 ml/min. A ‘‘Mixed D’’ column (Polymer Laboratories, UK) packed with polystyrene/polydivinylbenzene at 50 °C and a Perkin-Elmer LC 290 variable wavelength UV absorbance detector operating at 300 nm was used to take measurements. The data was collected, processed and normalised with respect to polystyrene standard.
2.3. Gas Chromatography–Mass Spectroscopy (GC–MS) A HP GCD 1800A with a J&W scientific HP-5 column (30 m 0.32 mm ID and 0.25 lm film thickness) fitted with an auto sampler operated under constant Helium pressure and equipped with an electron ionisation detector with a source temperature of 300 °C was used. The auto sampler injected 2 ll sample of 3 lg biocrude dissolved in 1 g DCM at 300 °C inlet temperature. After an initial hold time of 2 min, the oven temperature was raised from 60 °C to 300 °C at 10 °C per minute with a final hold time of 4 min. Ion fragmentation from collected data was identified using preinstalled National Institute of Standards and Technology (NIST) Reference database 1A linked to Chemstation 10.
2.4. Simulated Distillation (SIMDIST) SIMDIST of biocrude samples was carried out on an Agilent 5890 gas chromatograph machine equipped with a flame ionisation detector and a Phenomenex Inferno HT5 column. 2 wt.% sample of biocrude was dissolved in DCM and a 2 ll of sample injected at 350 °C using an auto sampler. The temperature was held for 2 min at 60 °C and ramped at 10 °C per min to 380 °C with an additional holding time of 3 min. Chemstation software was used to obtain integration values of signals and these were processed manually to construct the boiling point curve. Atmospheric equivalent boiling point was obtained by using an alignment chart nomograph and boiling point confirmed using hydrocarbon standards.
2.5. Attenuated Total Reflection–Fourier Transform Infrared Spectroscopy (ATR–FTIR) Approximately 50 lg of DCM free biocrude sample was deposited on the ATR crystal of a Perkin-Elmer Spectrum 100 FTIR. Spectra from 4000–600 cm 1 band and 50 replicate scans were collected for triplicate samples. Freeze dried algal biomass analysis was performed by compacting the powder on the ATR crystal. Background scan was performed under ambient atmosphere prior to all analysis.
2.6. Elemental Carbon, Hydrogen, Nitrogen and Oxygen (CHNO) Samples of DCM extracted biocrude, oven dried filtered char and dried aqueous phase soluble solids were sent for Carbon, Hydrogen and Nitrogen elemental analysis to the University of Sheffield. The O content was found by difference. From previous HTL investigation and subsequent Karl-Fischer titration it can be confirmed that there was negligible water content in the biocrude (Patel and Hellgardt, 2013) and was not analysed. 3. HTL – results and discussion HTL of Nannochloropsis sp. microalgae was carried out in a quartz lined continuous plug flow reactor at 300, 325, 350 and 380 °C at RTs of 0.5, 1, 2 and 4 min. The lipid and ash content of the biomass, as determined through previous investigation (Patel and Hellgardt, 2013) was 20 and 11 wt.%, respectively. 3.1. Gravimetric yields The gravimetric yield of the DCM soluble organic phase, defined as biocrude, DCM insoluble phase classed as Water Soluble Fraction (WSF) and gas (+loss) found by difference is shown in Fig. 2a–c, respectively. Supplementary Information 1 – Fig. 1 shows the mass balance for the yields based on quantified biocrude, WSF and char balanced by gas yield (+experimental loss) found by difference. The highest biocrude yield achieved in this study is 38 wt.% at 380 °C and 30 s RT. The biocrude yield appears to be constant at around 30 wt.% for most reaction conditions and the high yield at short contact time and high temperature is in agreement with other investigations (Faeth et al., 2013; Jazrawi et al., 2013; Patel and Hellgardt, 2013). Faeth et al. (2013) reported that at a temperature of 500 °C and batch holding time of 1 min, biocrude yield of 55 wt.% was obtained compared to 52 wt.% (Valdez et al., 2012) at longer RT. The highest reported yield in this work is similar to previous batch reactor produced biocrude at 380 °C and 5 min RT (Patel and Hellgardt, 2013). Jazrawi et al. (2013) reported a yield of 41.7 wt.% at 350 °C and 3 min RT in a pilot scale continuous reactor. The variability in yields is partly down to the type of algae used, difference in sample reproduction and the reactor temperature profile. It is apparent that high temperature and shorter RT do indeed increase the biocrude yield, and a recent investigation by Bach et al. (2014) on batch HTL of macroalgae demonstrated that at their investigated heating rate of 585 °C/min, biocrude yield of 70 wt.% was achieved. However, the chemistry behind such processes is not fully understood. One potential explanation put forward by Faeth et al. (2013) is that by limiting the heating time, the
Please cite this article in press as: Patel, B., Hellgardt, K. Hydrothermal upgrading of algae paste in a continuous flow reactor. Bioresour. Technol. (2015), http://dx.doi.org/10.1016/j.biortech.2015.04.012
4
B. Patel, K. Hellgardt / Bioresource Technology xxx (2015) xxx–xxx
a
b
80
c
70
Gas and Loss *(wt.-%)
60 50 40 30
30 sec
20
1 min
10
2 min 4 min
0 300
325
350
Temperature (°C) Fig. 2. Gravimetric yield from solvent aided HTL (a) biocrude, (b) Water Soluble Fraction, (c) gas (+loss). ⁄Found by difference.
degradation of algal molecules takes place without further unwanted side reactions. Under hydrothermal conditions, cell rupture will be relatively easy via hydrolysis of the polysaccharides, thus rapidly exposing the cell contents to hydrothermal conditions. Subsequently, the released intracellular material is attacked by
hydrothermal water and only limited degradation occurs. This is similar to pyrolysis, whereby the inert gas is replaced in this case by water, thus resembling hydropyrolysis. Based on previous batch reactor investigations (Faeth et al., 2013; Patel and Hellgardt, 2013; Garcia Alba et al., 2011; Jena
Please cite this article in press as: Patel, B., Hellgardt, K. Hydrothermal upgrading of algae paste in a continuous flow reactor. Bioresour. Technol. (2015), http://dx.doi.org/10.1016/j.biortech.2015.04.012
B. Patel, K. Hellgardt / Bioresource Technology xxx (2015) xxx–xxx
et al., 2011), the observed high yield at low RT and high temperature suggests that at even higher temperature it might be possible to increase the yield substantially provided that short contact time is maintained in addition to the rapid heating rate (Bach et al., 2014). This would be easier to achieve in a flow reactor assuming heat transfer limitation is overcome and char formation limited. The most obvious trend observed in the gravimetric yield is the WSF. At shorter RT and lower temperature, the majority of the mass from the reaction tends to partition into the aqueous phase. At 300 °C and 30 s RT, almost 60 wt.% matter accumulates in the aqueous phase. As the reaction condition becomes more severe, the aqueous soluble matter is observed to decrease and complemented by an increase in produced gas (Fig. 2c). From this study, it is apparent that severe conditions (higher temperature and longer RT) are unlikely to increase the biocrude yield because from Fig. 2a it can be seen that the biocrude fraction of the product pool does not change significantly, but there is a remarkable change in the WSF wt.%. This implies that the conversion of material in the aqueous phase is towards formation of gases. Garcia Alba et al. (2011) made similar observations, achieving 47.4 and 62.8 wt.% gas yield at 450 °C and batch holding time of 5 and 60 min, respectively. The increased gas content caused a minute decrease in char content but significant mass was lost from the WSF. The char formation in their study was fairly constant with only a small decrease at severe conditions, which is similar to this work (approximately 5–10 wt.%). The gas phase was difficult to analyse due to nature of the sample size, but from previous investigations (Patel and Hellgardt, 2013; Garcia Alba et al., 2011), it is clear that produced gas is predominantly composed of CO2 and it is not expected to be different for this work. Additionally, the HTL aqueous phase contains significant amount of nitrogen (N) as NH+4 (Biller et al., 2012) that could potentially be released into the gas phase, partially explaining the increase in gas yield and reduction in WSF. It should be noted that water was evaporated to obtain solid WSF for elemental CHN(O) analysis. It is possible that some of the N based compounds could have been lost during this evaporation. To ensure and confirm that cyclohexane did not degrade during processing, the volume of the solvent fed and recovered post reaction was measured. Generally, a variable error in recovery of approximately 5 vol.% was obtained and no trend was identified with respect to changing reaction conditions. 3.2. ATR–FTIR spectroscopy ATR–FTIR (Supplementary Information 1 – Fig. 2) was used to gain a better understanding of the composition of the produced biocrude by obtaining insight on the functional groups present in the samples. The CAH bond stretching and vibration is represented by the 3000–2800 cm 1 and 1490–1350 cm 1 band in the spectrum and the signal in this region is characteristic of samples with high H and C content as present in the biocrudes (Vardon et al., 2012). On the other hand the OAH stretching vibration between 3500 and 3000 cm 1 from presence of alcohol or moisture is completely absent. However, the CAO stretch at 1320–1210 cm 1 and 1260–1000 cm 1 confirms presence of phenol and branched alcohol species. The most significant property that would contribute towards amenable biocrude for drop in refinery grade feedstock is the lack of N. However, heteroatom functionality and NAH bending (1680–1600 cm 1 and 1575–1525 cm 1) is abundant in all samples which potentially arise from amides, amines and other nitrogenous compounds. The polysaccharide derivatives vibration bands occur from OACAH, CACAH and CAOAH bonds which are found in the 1400–1199 cm 1 region (Leopold et al., 2011). This signal is quite abundant in the samples and the components with aforementioned functionality are responsible for the Maillard
5
reaction that gives rise to the undesirable nitrogen heterocycles (Patel and Hellgardt, 2013), some of which were detected via GC–MS analysis. The C@O stretching at 1730–1700 1 and vibration in the 2000–1500 cm 1 range corresponds to aldehydes, ketones and carboxylic acids (Baloyi et al., 2012). Compared to fossil crude, the functionality of algal biocrude differs remarkably, predominantly due to lack of nitrogen heteroatoms. The prominent CAH functionality in the 3000–2800 cm 1 range arising from significantly higher C and H content is stronger in fossil crude compared to algal biocrude. Additionally, the absorbance in 900–700 cm 1 in crude oil shows greater aromatic content (Carbognani et al., 2003; Wilt et al., 1998), but not much is known about the aromatic content in algal crude (Patel and Hellgardt, 2013), which is potentially a secondary product. 3.3. CHN(O) distribution, van Krevelen diagram and HHV Table 1 shows the elemental composition of algal biomass, biocrude, WSF and char produced at 30 s, 2 min and 4 min RT. In the biocrudes, the O and N levels have both decreased with respect to algal feedstock as expected. O composition in the biocrude range from 14.25–8.94 wt.%, which corresponds to a decrease of at least 55% from unprocessed biomass. This decrease is complemented by denitrogenation of at least 45% to produce biocrude with N content of 2.35–4.11 wt.%. Although overall there is a decrease in N content, the extent to which this takes place is opposite to that of O depending on the reaction condition. For instance at 30 s RT for all temperatures, the O removal is noticed to improve but the amount of denitrogenation decreases. This trait is generally observed at all reaction conditions. The elemental N content tends to reach a stable concentration level after which there is no change in its wt.% thus, elucidating that nitrogen containing compounds are stable under hydrothermal conditions. Faeth et al. (2013) and Garcia Alba et al. (2011) had a similar trend in their HTL biocrude and it is understood that the change is procured by increasing temperature. Hence, this work demonstrates that denitrogenation can be achieved rapidly (in seconds) eliminating the need for more severe reaction conditions. The van Krevelen plot, originally introduced to represent a graphical relationship between C, H and O ratio in coal processing is used to show the O:C, H:C and N:C atomic ratio of the produced biocrude. As seen from plot Fig. 3a and b, it is obvious that there is O and N removal after fast (HTL) of microalgae. The biocrudes produced in this study are concentrated approximately around H:C ratio of 1.6 and an O:C ratio of 0.15. There is a slight increase in deoxygenation at longer RT, but when compared to previous batch reactor HTL study (Patel and Hellgardt, 2013), it is deduced that longer operating time does not significantly improve aforementioned ratios. Based on current HTL studies, algal biocrude produced in this investigation contains the lowest O:C ratio using a continuous flow reactor at short contact time. Biocrude produced in this investigation contains marginally lower O than that from the pilot scale HTL study by Jazrawi et al. (2013). The primary aim of HTL is to degrade macromolecules and produce a product with negligible oxygen and nitrogen content. Compared to fossil crude and coal, the O:C ratio is particularly high in unprocessed biomass. After HTL the O decreases to reach a similar level to that of coal. The extent of O removal is related to processing conditions as seen in Table 1 and previous work (Patel and Hellgardt, 2013). The elemental H composition increases from an initial untreated algal biomass value of 6.6 wt.% to approximately 10 wt.%, for almost all reaction conditions and the increase in H content causes a subsequent increase in C content resulting in formation of stable CAH bonds (McKendry, 2002). Consequently, the accompanying loss of
Please cite this article in press as: Patel, B., Hellgardt, K. Hydrothermal upgrading of algae paste in a continuous flow reactor. Bioresour. Technol. (2015), http://dx.doi.org/10.1016/j.biortech.2015.04.012
6
B. Patel, K. Hellgardt / Bioresource Technology xxx (2015) xxx–xxx
Table 1 HHV, yield and elemental CHNO composition of biocrude, char, WSF and algal biomass (Nannochloropsis sp.).
Nannochloropsis sp.
C 43.2
RT (min)
Temp (°C)
Biocrude
0.50
300 325 350 380 300 325 350 380 300 325 350 380
73.30 74.15 74.25 74.16 73.95 75.75 76.25 75.89 73.81 75.01 75.49 77.02
2.00
4.00
H 6.6
N 7.2
O 32.2
HHV (MJ/kg) 18.28
C
10.10 10.40 10.34 10.34 10.15 10.44 9.92 10.04 10.24 10.37 10.05 10.41
2.35 2.76 3.00 3.25 3.91 3.50 3.78 3.74 4.11 3.70 3.50 3.63
14.25 12.69 12.41 12.25 11.99 10.31 10.05 10.33 11.84 10.92 10.20 8.94
36.63 37.62 37.62 37.61 37.32 38.64 38.12 38.11 37.43 38.18 38.02 39.27
45.05 55.00 55.00 56.12 22.64 22.65 27.21 37.34 21.61 23.00 26.39 27.98
H
N
O
5.68 7.24 3.84 6.42 6.26 3.10 3.18 4.56 3.31 2.50 2.75 2.39
8.39 7.00 8.00 8.15 2.07 2.76 3.05 5.38 2.62 1.65 2.55 3.10
40.88 30.76 33.16 29.31 69.03 71.49 66.56 52.72 72.46 72.85 68.31 66.53
Char
C
H
N
O
Water Soluble Fraction (WSF) 36.63 34.84 31.20 37.04 31.38 30.44 20.63 21.32 29.01 23.02 16.75 14.98
4.21 3.28 3.36 3.97 3.68 3.02 1.61 1.62 2.87 1.88 0.52 0.50
8.24 7.99 6.70 7.69 6.84 6.43 3.05 3.08 6.06 4.61 2.63 2.15
50.92 53.89 58.74 51.30 58.10 60.11 74.71 73.98 62.06 70.49 80.10 82.37
Fig. 3. van Krevelen plot of algal biocrude (a) H:C vs. O:C, (b) H:C vs. N:C.
O potentially reduces the number of CAO bonds present in the biocrude, thus negating any improvement in oxidative stability but at the same time it helps improve the HHV. Typically the N content of fossil crude is <1 wt.% corresponding to a N:C ratio of <0.002 (Speight, 2004). Algal biocrude is far from this figure and HTL alone is not capable of reaching such concentration. The N:C ratio of produced biocrude is low compared to the unprocessed feedstock and the set of experiments conducted in this study also provides some useful insight on N partitioning,
particularly as the WSF at short RT. The N content of the processed water at moderate conditions is high and as conditions become more severe, there is a gradual loss in N mass fraction. Conversely the O content follows the opposite trend. The N is potentially released from the water as a gaseous product (ammonia), hence in order to retain maximum N, reaction conditions of 380 °C and 30 s appear to be most appropriate. The energy content (HHV) of the samples can be calculated based on the elemental composition. It is widely accepted that
Please cite this article in press as: Patel, B., Hellgardt, K. Hydrothermal upgrading of algae paste in a continuous flow reactor. Bioresour. Technol. (2015), http://dx.doi.org/10.1016/j.biortech.2015.04.012
B. Patel, K. Hellgardt / Bioresource Technology xxx (2015) xxx–xxx
the Dulong’s equation which incorporates the elemental C, H, O and S composition, is suitable for algal based HHV approximation. Typically, the last S term is ignored due to the low S concentration in algae. In this study the HHV has a range between 36.64 and of 39.27 MJ/kg, in comparison crude oil has a HHV in the range of 40–50 MJ/kg (Speight, 2004). The highest HHV was produced at the most severe condition investigated (380 °C and 4 min) which was consequently the biocrude with the smallest quantity of O. The decrease in O content of the samples is desirable as it increases the HHV of the fuel. Although biocrude from HTL of algae offers better elemental ratios compared to lignin based fuels (Patel and Hellgardt, 2013), the oxygen and nitrogen content is still problematic. Biocrude with high N cannot be blended with fossil crude unless an additional denitrogenation step is added prior to blending, resulting in additional costs. Catalytic deoxygenation or hydrotreating is also necessary to improve fuel quality, but the presence of nitrogen heterocycles, as identified by GC–MS usually causes catalyst poisoning making denitrogenation challenging.
3.4. Molecular weight analysis – SEC SEC analysis of the biocrude samples shows the average molecular weight of the compounds present in the oil. Supplementary Information 1 – Fig. 3 infers that processing condition has a noticeable, but not significant impact on the average molecular weight of the species present in the biocrude. At reaction condition of 380 °C and 4 min, the average Mw of 200 is detected. For all other analysed biocrudes, a Mw range between 500 and 1000 is observed. The four samples from 30 s RT at 380 °C and 300 °C and the 4 min RT sample at the same temperatures show that a lower Mw distribution is achieved, confirming that the effect of temperature is greater at producing lower average Mw compounds opposed to the RT. From analysed samples it is observed that the Mw distribution of the biocrude is bimodal unlike the biocrude produced by Vardon et al. (2011) which had an approximate Mw of 350. When the reaction condition (particularly RT) is less severe, a second batch of molecules around Mw 1000 kDa is seen. The signal intensity for aforementioned Mw region, for biocrude from least severe reaction condition at 300 °C and 30 s is strongest and predictably weakest at the most severe condition. Presumably polymeric
7
substances present in algae, such as polysaccharides and polypeptides that possess high Mw could give rise to these. Garcia Alba et al. (2011) observed similar ‘clusters’ of compounds in their GPC analysis of batch produced biocrude at 250 °C and 5 min RT, but not as heavy as those in this investigation.
3.5. Boiling point distribution – SIMDIST SIMDIST analysis of selected samples of biocrude produced at various reaction conditions are shown in Fig. 4. To compare the boiling point fraction, the results from previous batch reaction investigation (Patel and Hellgardt, 2013) and North Sea fossil crude are included. From the onset it is evident that the boiling point distribution of the fast liquefaction biocrude is not similar to either crude oil or biocrude formed from batch liquefaction. The atmospheric equivalent boiling point of biocrude from fast liquefaction appears to point towards abundance of compounds with boiling point greater than 265 °C which corresponds to C chain length of C15 or greater. Although the SIMDIST investigation is carried out for compounds with atmospheric equivalent boiling point up to 450 °C, it is highly unlikely that all of the oil is detected within this range. It is expected that an unknown proportion of oil might have a boiling point beyond this and can be found with additional thermograviemtric analysis (not presented in this manuscript). The biocrude produced predominantly has compounds with boiling point within the range of gas oil (265–380 °C) and atmospheric residue (380 °C<). The effect of temperature is not as pronounced as that of RT. With increasing reaction severity the fraction of lower boiling point light gas oil region compounds is greater. Unlike fossil crude and batch reaction produced biocrude, there is a clear lack of short carbon chain length compounds. The lower boiling point fraction present in small quantities is largely in the kerosene range (190–265 °C) and an even smaller quantity in the heavy naphtha (150–190 °C) distillation cut. The biocrude produced in this continuous flow liquefaction investigation is different from that produced by Jazrawi et al. (2013), where more kerosene range fraction was present. The difference could be due to the variability in the lipid, protein and carbohydrate content of the algal species used and also potentially from sample preparation. From the SEC and SIMDIST analysis it is evident that fast HTL potentially takes place in two steps. The first part is the rapid degradation and hydrolysis of algal components as
Fig. 4. SIMDIST of algal biocrude (batch and continuous) and crude oil. ⁄Patel and Hellgardt (2013).
Please cite this article in press as: Patel, B., Hellgardt, K. Hydrothermal upgrading of algae paste in a continuous flow reactor. Bioresour. Technol. (2015), http://dx.doi.org/10.1016/j.biortech.2015.04.012
8
B. Patel, K. Hellgardt / Bioresource Technology xxx (2015) xxx–xxx
suggested by Faeth et al. (2013) and the second being the multitude of reactions that produce significant pool of compounds resulting in formation of lower Mw species. The presence of a negligible quantity of biocrude within the kerosene fraction at longer RT might not necessarily result in more economical operating conditions. The longer RT, lower gravimetric biocrude yield and only a small increase in overall deoxygenation does not seem advantageous enough to pursue considering the energy penalty.
Diels–Alder addition and ring rearrangement (Demirbas, 2009). Similarly, heteroatoms are prevalent at more severe conditions which help compound this multiple reaction pathway suggestion. Furthermore, there appear to be certain compounds such as the fatty acids, which are quite stable. Under the conditions used in this investigation, it was expected that the unstable molecules will degrade quicker, hence producing stable ones, necessating a harsher environment for their removal. Identification of these molecules is important because they prevail for longer and prevent further deoxygenation/denitrogenation.
3.6. Chemical characterisation – GC–MS Table 1 (Supplementary Information 2) shows the list of detected compounds for biocrude produced at three different reaction conditions. Spectra with total chromatogram peak area of greater than 1% were identified using the NIST database. It should be noted that only a proportion of the biocrude produced lies within the temperature range in which the GC–MS operates and thus only a fraction of the components in the biocrude can be deciphered. Additionally, there might have been some loss of lighter compounds during solvent extraction and sample preparation. From the list of compounds in Table 1 (Supplementary Information 2) it becomes apparent that under severe conditions there is more degradation of macromolecules present in algal biomass. It would be expected that proteins, lipids and carbohydrates will rapidly break down to their subsequent products under HTL conditions. The CAN bond present in proteins will easily hydrolyse to form its primary product amino acids along with some hydrocarbons, amines, aldehydes and acids (Brunner, 2009). At longer RTs, these degradation/hydrolysis products can react with carbohydrate hydrolysis product, monosaccharide, to produce a range of products. Thus, there is a larger pool of compounds present at 380 °C and 4 min RT. The lipids (Triacylglycerols – TAG) are expected to hydrolyse to glycerol and their subsequent fatty acids which can further isomerise or hydrolyse to alkanes. At 380 °C and 30 s RT, the presence of triolein (25.6 min) suggests that the reaction condition is not sufficient to hydrolyse all TAG species. Similarly, a direct comparison of the three reaction conditions shows that there is a lack of nitrogen containing compounds at 300 °C and 30 s. Although, the product pool does not show many nitrogen containing compounds, the CHN analysis for the three biocrudes is quite similar. One reason for this might be that the products formed under less severe conditions are not detected within the range of the GC–MS and/ or a multitude of different N containing compounds formed produce a signal similar to the background. There is a significant number of methyl derived alkylated compounds in all compound classes from cleavage of CACH3 bonds, which appear to be easiest to break. The unsaturated species are understood to be derived from dehydration, cyclization,
4. HTL reactor system consideration Table 2 shows a comparison of some key indicators of HTL reaction similar to this investigation (low RT and continuous HTL). Looking at the feed concentration, the 1.5 wt.% biomass concentration used in this work was primarily due to the pumping limitation of the HPCL pump. Higher concentrations caused a blockage of the pump feed valve resulting in a loss of flow. Secondly, the lower biomass concentration combined with solvent extraction aided the reaction by limiting char formation thus preventing the BPR from blocking. Jazrawi et al. (2013) had a similar issue downstream related to the pressure control valve orifice which limited the biomass concentration used in their study (Elliott et al., 2015). Considering that algal biomass:water ratio does not make a substantial impact on HTL reaction (Jena et al., 2011; Garcia Alba et al., 2011) as well as the ability for commercial HTL reactors to be designed with 31.5% (Knorr et al., 2013) biomass handling capacity (for lignocellulosic feedstock), it is of utmost importance to select an appropriate BPR to ensure it does not limit the amount of algae used in the reaction. Elliott et al. (2013) have already demonstrated a slurry concentration of 35 wt.% at pilot scale, so a 40 wt.% commercial algal HTL reactor system is realistically feasible. Pumping higher biomass concentration will result in greater quantity of biocrude yield per unit mass which is essential to maximise the energy return, but this was not necessarily the aim of the reactor used in a laboratory for this investigation. Hence, no provision was made for energy economy at this stage. The primary aim of using cyclohexane was to limit char agglomeration in the system by ensuring there is constant solvation of the formed products. Pilot scale investigation by Jazrawi et al. (2013) and Elliott et al. (2013) did not use a co-solvent but it is expected that the use of solvents would be beneficial as it can aid produce better quality biocrude and increase yield (Xiu and Shahbazi, 2012; Singh et al., 2015). Apart from cyclohexane, solvents such as acetone, n-hexane, 1-4-dioxane or mixture of middle distillates (kerosene/diesel) as well as alcohols (ethanol/methanol or butanol) (Xiu and Shahbazi, 2012; Liu and Zhang, 2008 and Singh et al., 2015) could be used.
Table 2 HTL performance comparison of short RT and continuous reactor system.
* #
Reference and reactor type (vol.)
Biomass loading (wt.%)
Reaction conditions (Temp/P/RT)
HHV (MJ/kg)
Biocrude yield (wt.%)
Faeth et al. (2013) Batch (1.67 ml) Elliott et al. (2013) Continuous (0.4–1 L) Jazrawi et al. (2013) Continuous (2 L) Eboibi et al. (2014) Batch (1 L) This study# Continuous (2 ml)
32
300–600 °C/autogenic/1–3 min
33.3–36.7
13–66
17–35
350 °C/200 bar/15 min
39.4–40.1
38–63.6*
1–10
250–350 °C/150–200 bar/3–5 min
27.9–33.8
<10–41.7
16
300–350 °C/autogenic/5 min
32.1–36.7
34–58
1.5
300–380 °C/180 bar/0.5–4 min
36.6–39.3
23.13–38.0
HHV found using Dulong’s equation. Cyclohexane used as a co-solvent.
Please cite this article in press as: Patel, B., Hellgardt, K. Hydrothermal upgrading of algae paste in a continuous flow reactor. Bioresour. Technol. (2015), http://dx.doi.org/10.1016/j.biortech.2015.04.012
B. Patel, K. Hellgardt / Bioresource Technology xxx (2015) xxx–xxx
There is still a wide gap in knowledge for understanding algal HTL in flow reactors, confirmed by lack of relevant literature as shown in Table 2. Thus utilisation of an economically frugal laboratory scale continuous reactor system to perform short RT reactions and investigate the various parameters for algal HTL is important and necessary. 5. Conclusion The HTL process demonstrated in this investigation shows that it is possible to convert algal biomass to biocrude at a relatively short RT of 30 s. The aqueous stream is of utmost importance because the CHNO analysis showed higher concentration of N in the dried WSF. Since the flow characteristics of the system do not seem to have influenced the yield (Supplementary Information 3), theoretically, higher biocrude yields may be achieved at even higher temperatures and shorter RT (Bach et al., 2014) and should be investigated for its feasibility. Whether ultra-short hydropyrolysis can tackle this further remains to be seen. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.biortech.2015.04. 012. References Bach, Q.V., Sillero, M.V., Tran, K.Q., Skjermo, J., 2014. Fast hydrothermal liquefaction of a Norwegian macro-alga: screening tests. Algal Res. 6, 271–276. Baloyi, H., 2012. Algae Liquefaction (Doctoral dissertation). North-West University. Biller, P., Ross, A.B., Skill, S.C., Lea-Langton, A., Balasundaram, B., Hall, C., Riley, R., Llewellyn, C.A., 2012. Nutrient recycling of aqueous phase for microalgae cultivation from the hydrothermal liquefaction process. Algal Res. 1, 70–76. Brunner, G., 2009. Near critical and supercritical water. Part I. Hydrolytic and hydrothermal processes. J. Supercrit. Fluids 47, 373–381. Carbognani, L., Espidel, J., Colaiocco, S., 2003. Characterization of heavy oils and heavy ends. In: Anal. Adv. Hydrocarbon Res. Springer, US, pp. 337–368. Chakinala, A.G., Brilman, D.W., van Swaaij, W.P., Kersten, S.R., 2009. Catalytic and non-catalytic supercritical water gasification of microalgae and glycerol. Ind. Eng. Chem. Res. 49, 1113–1122. Chen, W.H., Lin, B.J., Huang, M.Y., Chang, J.S., 2014. Thermochemical conversion of microalgal biomass into biofuels: a review. Bioresour. Technol. 184, 314–327. Demirbas, A., 2009. Pyrolysis of biomass for fuels and chemicals. Energy Sources Part A 31, 1028–1037. Eboibi, B.E.O., Lewis, D.M., Ashman, P.J., Chinnasamy, S., 2014. Hydrothermal liquefaction of microalgae for biocrude production: improving the biocrude properties with vacuum distillation. Bioresour. Technol. 174, 212–221. Elliott, D.C., Biller, P., Ross, A.B., Schmidt, A.J., Jones, S.B., 2015. Hydrothermal liquefaction of biomass: developments from batch to continuous, process. Bioresour. Technol. 178, 147–156. Elliott, D.C., Hart, T.R., Schmidt, A.J., Neuenschwander, G.G., Rotness, L.J., Olarte, M.V., Zacher, A.H., Albrecht, K.O., Hallen, R.T., Holladay, J.E., 2013. Process development for hydrothermal liquefaction of algae feedstocks in a continuousflow reactor. Algal Res. 2, 445–454. Energy Information Administration (US) (Ed.), 2013. International Energy Outlook 2013 with Projections to 2040. Government Printing Office.
9
Ewald, J., 2013. Carbon Dioxide at NOAA’s Mauna Loa Observatory reaches new milestone: Tops 400 ppm. NOAA Research, 10 May 2013. Faeth, J.L., Valdez, P.J., Savage, P.E., 2013. Fast hydrothermal liquefaction of Nannochloropsis sp. to produce biocrude. Energy Fuels 27, 1391–1398. Garcia Alba, L., Torri, C., Samorì, C., van der Spek, J., Fabbri, D., Kersten, S.R., Brilman, D.W., 2011. Hydrothermal treatment (HTT) of microalgae: evaluation of the process as conversion method in an algae biorefinery concept. Energy Fuels 26, 642–657. Jazrawi, C., Biller, P., Ross, A.B., Montoya, A., Maschmeyer, T., Haynes, B.S., 2013. Pilot plant testing of continuous hydrothermal liquefaction of microalgae. Algal Res. 2, 268–277. Jena, U., Das, K.C., Kastner, J.R., 2011. Effect of operating conditions of thermochemical liquefaction on biocrude production from Spirulina platensis. Bioresour. Technol. 102, 6221–6229. Knorr, D., Lukas, J., Schoen, P., 2013. Production of Advanced Biofuels via Liquefaction – Hydrothermal Liquefaction Reactor Design: April 5, 2013. NREL/SR-5100-60462, National Renewable Energy Laboratory, Golden, Colorado, USA. Leopold, L.F., Leopold, N., Diehl, H.A., Socaciu, C., 2011. Quantification of carbohydrates in fruit juices using FTIR spectroscopy and multivariate analysis. J. Spectrosc. 26, 93–104. Liu, Z., Zhang, F.S., 2008. Effects of various solvents on the liquefaction of biomass to produce fuels and chemical feedstocks. Energy Convers. Manage. 49, 3498– 3504. López Barreiro, D., Prins, W., Ronsse, F., Brilman, W., 2013. Hydrothermal liquefaction (HTL) of microalgae for biofuel production: state of the art review and future prospects. Biomass Bioenergy 53, 113–127. Maiella, P.G., Brill, T.B., 1998. Spectroscopy of hydrothermal reactions. 10. Evidence of wall effects in decarboxylation kinetics of 1.00 m HCO2X (X = H, Na) at 280– 330 °C and 275 bar. J. Phys. Chem. A 102, 5886–5891. McKendry, P., 2002. Energy production from biomass (Part 1): Overview of biomass. Bioresour. Technol. 83, 37–46. Patel, B., Hellgardt, K., 2013. Hydrothermal upgrading of algae paste: application of 31 P-NMR. Environ. Prog. Sustainable Energy 32, 1002–1012. Patel, B., Tamburic, B., Zemichael, F.W., Dechatiwongse, P., Hellgardt, K., 2012. Algal biofuels: a credible prospective? ISRN Renew. Energy 2012. Potic, B., Kersten, S.R.A., Prins, W., Van Swaaij, W.P.M., 2004. A high-throughput screening technique for conversion in hot compressed water. Ind. Eng. Chem. Res. 43, 4580–4584. Reddy, H.K., Muppaneni, T., Rastegary, J., Shirazi, S.A., Ghassemi, A., Deng, S., 2013. Hydrothermal extraction and characterization of bio-crude oils from wet chlorella sorokiniana and dunaliella tertiolecta. Environ. Prog. Sustainable Energy 32, 910–915. Savage, P.E., 1999. Organic chemical reactions in supercritical water. Chem. Rev. 99, 603–622. Singh, R., Bhaskar, T., Balagurumurthy, B., 2015. Effect of solvent on the hydrothermal liquefaction of macro algae Ulva fasciata. Process Saf. Environ. Prot. 93, 154–160. Speight, J.G., 2004. Petroleum Asphaltenes – Part 1: Asphaltenes, resins and the structure of petroleum. Oil Gas Sci. Technol. 59, 467–477. Valdez, P.J., Nelson, M.C., Wang, H.Y., Lin, X.N., Savage, P.E., 2012. Hydrothermal liquefaction of Nannochloropsis sp.: systematic study of process variables and analysis of the product fractions. Biomass Bioenergy 46, 317–331. Vardon, D.R., Sharma, B.K., Blazina, G.V., Rajagopalan, K., Strathmann, T.J., 2012. Thermochemical conversion of raw and defatted algal biomass via hydrothermal liquefaction and slow pyrolysis. Bioresour. Technol. 109, 178– 187. Vardon, D.R., Sharma, B.K., Scott, J., Yu, G., Wang, Z., Schideman, L., Zhang, Y., Strathmann, T.J., 2011. Chemical properties of biocrude oil from the hydrothermal liquefaction of Spirulina algae, swine manure, and digested anaerobic sludge. Bioresour. Technol. 102, 8295–8303. Wilt, B.K., Welch, W.T., Rankin, J.G., 1998. Determination of asphaltenes in petroleum crude oils by Fourier transform infrared spectroscopy. Energy Fuels 12, 1008–1012. Xiu, S., Shahbazi, A., 2012. Bio-oil production and upgrading research: a review. Renew. Sustainable Energy Rev. 16, 4406–4414.
Please cite this article in press as: Patel, B., Hellgardt, K. Hydrothermal upgrading of algae paste in a continuous flow reactor. Bioresour. Technol. (2015), http://dx.doi.org/10.1016/j.biortech.2015.04.012