Fast Hydrothermal Liquefaction of Nannochloropsis sp. To Produce

Feb 27, 2013 - sealed reactor with temperature data acquisition capabilities. These vessels served as “dummy reactors” and were filled only with a...
0 downloads 4 Views 847KB Size
This is an open access article published under an ACS AuthorChoice License, which permits copying and redistribution of the article or any adaptations for non-commercial purposes.

Article pubs.acs.org/EF

Fast Hydrothermal Liquefaction of Nannochloropsis sp. To Produce Biocrude Julia L. Faeth, Peter J. Valdez, and Phillip E. Savage* Department of Chemical Engineering, University of Michigan, 2300 Hayward Drive, 3074 H. H. Dow Building, Ann Arbor, Michigan 48109, United States S Supporting Information *

ABSTRACT: We investigated the fast hydrothermal liquefaction of green marine alga Nannochloropsis sp. at batch reaction times of 1, 3, and 5 min and set-point temperatures of 300−600 °C. We also performed conventional liquefaction for 60 min at the same temperatures. These experiments cover the broadest range of reaction conditions yet reported for algae liquefaction. The biocrude yield obtained for 1 min reaction times, which was only long enough to heat the reactor from room temperature to about half of the set-point temperature (in °C), increased with an increasing set-point temperature to 66 ± 11 wt % (dry and ashfree basis) at a set-point temperature of 600 °C. The biocrude obtained at this condition contains 84% of the carbon and 91 ± 14% of the heating value present in the dry algae feedstock. This biocrude yield and corresponding energy recovery are the highest reported for liquefaction of this alga. For a reaction time of 1 min, as the set-point temperature increases, light biocrude (e.g., hexane solubles) makes up less of the total biocrude. The biocrudes produced by fast liquefaction have carbon contents and higher heating values similar to biocrudes from the traditional isothermal liquefaction process, which involves treatment for tens of minutes. These results indicate that biocrudes of similar quality may be produced in higher yields and in a fraction of the time previously thought necessary. Such a decrease in the reaction time would greatly reduce the reactor volume required for continuous biocrude production, subsequently reducing the capital costs of such a process. We also show that the reaction ordinate is a useful parameter for interpreting results from algae liquefaction performed at different temperatures and reaction times.



temperature around 300 °C and a pressure high enough to keep the water in the liquid phase.8 The process capitalizes on the interesting properties of water at these near-critical conditions to produce biocrude. For example, there are fewer and weaker hydrogen bonds in hot compressed water than in water at ambient conditions,9 which gives a lower dielectric constant and increased solubility for organic molecules. Hot compressed water also has a higher ion product (Kw); therefore, the acid-catalyzed hydrolytic decomposition of biomolecules is accelerated. HTL is typically performed with slow heating and/or reaction times of tens of minutes or longer,2,10−16 but some recent results suggest that shorter reaction times may be sufficient.17−19 For example, Knežević et al. found that the yield of acetone- and water-soluble organics (referred to as oil by the authors) from HTL of woody biomass reached a maximum at 3−5 min reaction time and then decreased at longer times.19 This reaction time includes the 140 s required to heat the reaction mixture to the set-point temperature of 350 °C. We also draw inspiration from the fast pyrolysis processes used for making biocrudes from dried, millimeter-sized lignocellulosic biomass particles. Fast pyrolysis processes rely on very fast heating rates (from 102 °C/min20 to 102 °C/s21) and short residence times (∼1 s22). These operating conditions minimize undesirable secondary reactions, resulting in higher yields of

INTRODUCTION In 2010, the global energy consumption totaled 8677 megatonnes of oil equivalent (Mtoe), with the transportation sector accounting for 27.3% of the total.1 The preferred fuels within the transportation sector are high-energy-density liquids, most of which are derived from petroleum. The supply of petroleum and other fossil fuels is finite; therefore, alternative sources of high-energy-density liquid fuels, especially those sources that reduce carbon dioxide emissions, are needed. Biocrudes derived from biomass generally have energy densities higher than the initial biomass feedstocks and could reduce carbon dioxide emissions, because atmospheric carbon dioxide is required to grow the biomass. Some biocrudes have higher heating values (HHVs) similar to petroleum-based crudes.2 Additionally, the biocrudes may be upgraded to further improve their energy density and physical properties.3 Fast pyrolysis of dried terrestrial lignocellulosic biomass can produce biocrudes, but aquatic biomass is also an attractive feedstock. Microalgae, for example, are a desirable biomass source because of their potentially high oil content,4 their photosynthetic efficiency,5,6 exceeding that of woody biomass, and their ability to be cultivated on non-arable land and in brackish water.7 Converting wet biomass to biocrude would avoid energyintensive drying steps and potentially increase the overall energy efficiency of biocrude production, especially for biomass with a high moisture content. Hydrothermal liquefaction (HTL) is a process capable of producing high-energy-density biocrudes from wet biomass. It involves subjecting a slurry of biomass and water to a © 2013 American Chemical Society

Received: November 26, 2012 Revised: February 8, 2013 Published: February 27, 2013 1391

dx.doi.org/10.1021/ef301925d | Energy Fuels 2013, 27, 1391−1398

Energy & Fuels

Article

Figure 1. Temperature profiles with corresponding biocrude yields (wt % daf) for different set-point temperatures of 300, 500, 400, and 600 °C. microalgae.18 Dummy reactors were inserted into the sandbath alongside loaded reactors, and both were removed at the completion of the desired reaction time (1, 3, or 5 min) and immediately quenched in cold water for 5 min. Cooled reactors were allowed to equilibrate for roughly 60 min before product analysis. The procedure for recovering the biocrude was identical to that described by Valdez et al.18 This procedure involved pouring the reactor contents into a glass centrifuge tube, rinsing the inside of the reactor with 9 mL of dichloromethane, and adding the solution to the centrifuge tube. The contents were centrifuged to facilitate phase separation, and the organic and aqueous phases were transferred to separate vials using a glass pipet. The residual solids were retained in the original test tube, and the organic, aqueous, and solid phases were dried. The mass of each dried product was recorded. The biocrude is defined as the material in the organic phase that remains after solvent removal. The biocrude yield was calculated by dividing the mass of biocrude by the mass of dry and ash-free algal biomass originally loaded into the reactor. Yields of biocrude are reported later in this paper, whereas yields of solids are reported in the Supporting Information. Hexane was added to the biocrude to recover the “light”, hexanesoluble portion of the biocrude. The hexane solubles were transferred to another vial and dried, and the mass was recorded. All reactions were carried out in at least triplicate, and the uncertainty reported for the biocrude yield represents the standard deviation of the population. Unreacted algae, light biocrude, and heavy biocrude samples were sent to Atlantic Microlabs, Inc. for analysis of C, H, N, and S. The O content was calculated by difference.

liquid and vapor products, compared to conventional pyrolysis.22 To the best of our knowledge, HTL of microalgae at short reaction times (a few minutes) and fast heating rates has not yet been explored. This work focuses on what we call “fast HTL”. We report results from experiments at total reaction times as short as 1 min and average heating rates as high as 230 ± 5 °C/min. These conditions have not been explored previously for HTL of any biomass resource.



MATERIALS AND METHODS

Nannochloropsis sp. algae was purchased from Reed Mariculture, Inc. as a preservative-free slurry with roughly 32.5 wt % solids content and used as received. The ash content of dried Nannochloropsis sp. was 6.25 ± 0.23 wt %, which is lower than the ash content of the preservativecontaining Reed Mariculture Nannochloropsis sp. used in some of our lab’s earlier work.12 All chemicals were purchased and are identical to those described by Valdez et al.18 Reactors consisted of a 316 stainlesssteel 3/8 in. Swagelok port connector and two caps. The internal volume of an assembled reactor was approximately 1.67 mL, smaller than the 4.1 mL internal volume reactors used by Valdez et al.18 Additional vessels were constructed using one port connector, one cap, and a bored-through reducing union (from 3/8 to 1/8 in.). Omega Engineering, Inc. 1/8 in. diameter 18 in. long stainless-steel clad thermocouples were inserted in the reducing union to construct a sealed reactor with temperature data acquisition capabilities. These vessels served as “dummy reactors” and were filled only with air. An Omega Engineering, Inc. HH309A datalogger recorded the temperature of dummy reactors during the reactions. We calculated that the algae paste contributed ≤15% of the total thermal mass of a loaded reactor; therefore, the dummy reactors provided a reasonable estimate of the thermal history of the reactors. Reactors were loaded with 0.1−0.62 g of algae paste and 0.19−1.13 g of additional water (see Table S1 in the Supporting Information for details) according to the procedure outlined by Valdez et al.18 and placed in a preheated isothermal Techne IFB-51 fluidized sandbath with a Eurotherm 3216 PID controller. In all cases, the reactors contained 15 wt % algae. Steam tables were used to determine the water density at each set-point temperature that would result in a pressure of roughly 400 bar at the set-point temperature. Differences in water density have little effect on product yields for HTL of



RESULTS AND DISCUSSION Figure 1 depicts the average temperature profile of the reactors at set-point temperatures of 300, 400, 500, and 600 °C. The text boxes overlaid on the profiles at 1, 3, and 5 min indicate the total biocrude yields [wt % on dry and ash-free basis (daf)] obtained at those reaction times. Figure 1 shows that the biocrude yield increased with the batch holding time from 13% at 1 min to 44% at 5 min when the sandbath set-point temperature was 300 °C. In contrast, the biocrude yields reached maxima and then decreased at the other set-point temperatures shown. The highest biocrude yields were realized at the shortest holding time (1 min) at the 1392

dx.doi.org/10.1021/ef301925d | Energy Fuels 2013, 27, 1391−1398

a

1393

N/A N/A 300 300 300 350 350 350 400 400 400 450 450 450 500 500 500 550 550 550 600 600 600

a

0 0 1 3 5 1 3 5 1 3 5 1 3 5 1 3 5 1 3 5 1 3 5

holding time (min)

N/A N/A 0.806 4.93 5.91 0.675 6.19 7.45 0.680 7.67 8.60 1.98 8.80 10.1 3.16 10.5 11.4 3.71 11.5 12.8 5.14 13.1 14.4

log(R0) 9.5 13 38 44 23 50 50 32 50 43 52 48 42 55 45 28 51 20 10 66 14 10

N/A ± 2.7b ±7 ± 0.5 ±1 ±4 ±3 ±4 ± 11 ±2 ±6 ±6 ±5 ±4 ± 12 ±7 ±9 ±2 ±7 ±2 ± 11 ±4 ±3

biocrude yield (wt % daf) N/A 71.9 ± 9.4 87 ± 5 53 ± 1 45 ± 6 83 ± 16 46 ± 6 51 ± 5 64 ± 11 54 ± 6 61 ± 1 47 ± 6 59 ± 4 57 ± 3 37 ± 6 44 ± 14 43 ± 20 36 ± 6 15 ± 16 8 ± 14 40 ± 8 24 ± 16 33 ± 27

percentage of light biocrude (of total BC) (%) H (wt %) 7.27 ± 0.020 10.3 ± 0.2 9.61 9.49 9.72 10.05 9.64 9.43 9.39 9.35 9.22 9.03 9.28 9.20 8.99 8.48 8.01 8.93 N/A N/A 9.00 N/A N/A

C (wt %) 52.35 ± 0.005 74 ± 1 71.06 70.43 71.62 70.98 70.56 72.62 68.94 73.00 73.90 68.68 74.25 74.40 68.89 73.36 73.67 69.70 N/A N/A 70.58 N/A N/A

8.88 ± 0.035 1.9 ± 0.3 1.87 4.74 5.45 2.00 6.01 5.84 4.13 6.10 5.62 5.53 5.70 5.80 6.41 6.75 7.05 6.84 N/A N/A 6.92 N/A N/A

N (wt %) 0.66 ± 0.030 0.116 ± 0.089 0.25 0.59 0.58 0.30 0.64 0.63 0.45 0.69 0.74 0.66 0.85 0.89 0.65 1.18 0.94 0.74 N/A N/A 0.74 N/A N/A

S (wt %) 24.60 ± 0.020 13.1 ± 1.1 17.22 14.75 12.63 16.67 13.15 11.48 17.09 10.85 10.52 16.10 9.93 9.72 15.05 10.23 10.33 13.79 N/A N/A 12.75 N/A N/A

O (wt %)

N/A = not available because of insufficient sample available for elemental analysis. bAdjusted from a dry weight basis reported in ref 18 to a dry and ash-free basis.

A B 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21

run

set-point temperature (°C)

Table 1. Summary of Results and Reaction Conditions for Fast HTL

23.75 ± 0.028 N/A 34.69 34.79 35.88 35.39 35.33 36.03 33.71 36.16 36.34 33.30 36.65 36.63 33.50 35.19 34.58 33.92 N/A N/A 34.51 N/A N/A

HHV (MJ/kg)

N/A N/A 18 ± 9 51 ± 0.7 61 ± 2 32 ± 6 70 ± 5 72 ± 6 42 ± 14 72 ± 3 61 ± 9 67 ± 8 70 ± 7 60 ± 6 73 ± 15 62 ± 10 38 ± 12 68 ± 2 N/A N/A 91 ± 14 N/A N/A

energy recovery (%)

Energy & Fuels Article

dx.doi.org/10.1021/ef301925d | Energy Fuels 2013, 27, 1391−1398

Energy & Fuels

Article

Table 2. Summary of Results and Reaction Conditions for Conventional HTL run

set-point temperature (°C)

holding time (min)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25

300 350 400 450 500 550 600 250 250 250 250 300 300 300 300 300 350 350 350 350 350 400 400 400 400

60 60 60 60 60 60 10 20 30 60 90 10 20 40 60 90 10 20 40 60 90 10 20 30 40

a

log(R0)

biocrude yield (wt % daf)

percentage of light biocrude (of total BC) (%)

C (wt %)

H (wt %)

N (wt %)

S (wt %)

O (wt %)

HHV (MJ/kg)

energy recovery (%)

7.53 8.90 10.6 12.0 13.5 15.1 15.5 5.7 5.9 6.2 6.4 6.9 7.2 7.5 7.7 7.8 8.4 8.7 9.0 9.1 9.3 9.8 10.1 10.3 10.4

46 40 34 17 8 0 8 35 35 43 34 51 52 49 41 41 44 41 44 42 44 39 39 35 34

52 75 60 45 6 0 56 48 48 43 38 42 42 45 49 50 42 47 52 58 57 59 67 65 67

N/Aa N/A N/A N/A N/A N/A N/A 70.82 70.03 71.80 71.42 71.19 70.18 74.18 74.53 73.26 71.38 73.07 74.63 74.81 75.58 74.86 74.72 76.07 76.29

N/A N/A N/A N/A N/A N/A N/A 9.51 9.29 9.52 9.19 9.19 9.21 9.66 9.79 9.37 9.11 9.30 9.52 9.62 9.61 9.41 9.63 9.45 9.34

N/A N/A N/A N/A N/A N/A N/A 5.34 5.75 6.13 6.78 6.21 5.94 5.96 5.28 6.32 6.10 6.20 5.61 5.43 5.35 5.91 6.01 5.54 5.37

N/A N/A N/A N/A N/A N/A N/A 0.59 0.64 0.58 0.70 0.64 0.61 0.59 0.52 0.62 0.60 0.62 0.49 0.52 0.43 0.63 0.73 0.71 0.71

N/A N/A N/A N/A N/A N/A N/A 13.14 13.70 11.38 11.31 12.17 13.46 9.02 9.28 9.83 12.21 10.21 9.14 9.02 8.43 8.60 8.31 7.63 7.69

N/A N/A N/A N/A N/A N/A N/A 35.23 34.55 35.88 35.31 35.08 34.53 37.31 37.57 36.45 35.01 36.22 37.24 37.46 37.81 37.26 37.59 37.92 37.83

N/A N/A N/A N/A N/A N/A N/A 52 53 66 52 77 77 78 66 64 65 63 71 67 70 63 63 56 54

N/A = not available.

500 and 600 °C set-point temperatures. The mean biocrude yields obtained under these conditions exceeded 55%. For comparison, biocrude yields from traditional HTL of this alga have reached but not exceeded 52%.18 It is evident that very high biocrude yields are accessible at very short reaction times via fast HTL. Although the chemistry that produces the higher yields during fast HTL is not yet clear, there are multiple potential explanations. The rapid heating may better facilitate lysing of the algal cells, releasing cell contents into the hydrothermal environment for subsequent reaction. Alternatively, hydrolysis and other reactions may simply take place faster than originally thought. Another possible explanation is that undesirable side reactions that become more favorable at more severe conditions are largely avoided during fast HTL. Figure 1 indicates that the temperature within the reactor is roughly 300 °C after being in the 600 °C sandbath for 1 min. This condition led to the highest biocrude yield and is likely desirable for fast HTL. Because the reaction mixture only reached 300 °C, a temperature similar to that used in conventional HTL, we anticipate that fast HTL brings no new concerns regarding corrosion or energy requirements. Table 1 provides a complete listing of the experimental conditions used for fast HTL and a more detailed summary of results. It gives the biocrude yield, the fraction of the total biocrude that was hexane-soluble (light biocrude fraction), the atomic composition of the biocrude, its HHV, and the percentage of biomass heating value recovered in the biocrude (energy recovery). Table 1 also gives the value of the reaction ordinate (R0) for each run. This parameter will be explained fully in a subsequent section. Note that run A provides data for the unreacted, dried algae. Run B provides data from Valdez et

al.18 for the dichloromethane solubles (e.g., “biocrude”) extracted from the unreacted dry algae. All other runs provide data for the biocrude obtained from the reaction conditions indicated. Table 2 provides similar information from experiments with the same alga using conventional rather than fast HTL. Conventional liquefaction reactors reached and maintained the set-point temperature for tens of minutes, whereas in some cases, the fast liquefaction reactors never reached the set-point temperature of the sand bath. Runs 1−7 are new experiments performed in this work. Runs 8−25 provide data calculated from the results of Valdez et al.18 We omit the conventional HTL results from Brown et al.12 because the solids content of the reaction mixtures used varied from 3 to 16 wt % rather than consistently being at 15 wt % as in the present work. Variation in the solids content has been reported to influence biocrude yields for HTL of Nannochloropsis sp.18 The atomic compositions of the total biocrude were calculated from the compositions determined separately for the light and heavy fractions and the relative amounts of each. The HHV was estimated using the Dulong formula (eq 1) and the carbon, hydrogen, oxygen, and sulfur weight percentages in Tables 1 and 2. 0.338C + 1.428(H − O/8) + 0.095S = HHV (MJ/kg) (1)

The calculated HHV was not validated with calorimetry because of insufficient sample mass. However, HHV calculation models based on elemental analysis have been considered to be sufficient for different types of biomass.23 As indicated in Table 1, the Nannochloropsis sp. biomass had a HHV of 23.75 ± 0.03 MJ/kg on an ash-free basis. The HHV 1394

dx.doi.org/10.1021/ef301925d | Energy Fuels 2013, 27, 1391−1398

Energy & Fuels

Article

reaction conditions both much milder and much more severe than those previously examined. Figure 2 shows how the biocrude yields from fast and conventional HTL vary with R0. The yields at the lowest values

of all biocrudes produced by fast HTL range between 33 and 37 MJ/kg, which is consistent with the 34−38 MJ/kg HHV range in Table 2 for biocrudes from conventional HTL of Nannochloropsis sp. Fast HTL can produce biocrudes with the same heating value obtained from conventional HTL. The energy recovery reported in Table 1 was calculated as the energy content (product of HHV and mass) in the biocrude relative to that in the initial algal biomass. An energy recovery of 100% indicates that all of the chemical energy in the algal biomass has been transferred to the biocrude. The energy recovery for fast HTL varies from 18 to 91% within the reaction conditions investigated. It generally increases with an increasing set-point temperature for the 1 min reaction times (see runs 1, 4, 7, 10, 13, 16, and 19), reaching a maximum of 91 ± 14% at a set-point temperature of 600 °C. Table 2 shows that the highest energy recovery achieved by conventional HTL of this same alga was 81%. The energy recovery was more typically 60−80% for the different conditions examined. A majority of the energy recoveries displayed in Table 1 for fast HTL are also within this range, but all were acquired at shorter reaction times. The data in Tables 1 and 2 provide a wealth of information about the biocrude formed from liquefaction of the same alga over a very wide range of conditions. Thus, they provide the opportunity to examine how the yield and properties of the biocrude vary with the severity of the reaction conditions. The data in Table 1 are from non-isothermal fast HTL experiments, whereas the data in Table 2 are from nominally isothermal, conventional HTL experiments. To facilitate a meaningful comparison of such results, we employ the reaction ordinate, R0,24 which has been used in the biomass processing (wood pulping) literature to combine the effects of both reaction time and temperature into a single parameter that serves as a metric for the severity of the reaction conditions. We note that the literature includes other similar parameters, such as the severity factor,25 prehydrolysis factor,26 and H factor.27 The reaction ordinate increases with both time and temperature, and its value is unity at the reference temperature, often 100 °C,27 and reference time (1 min in this work). The values of R0 for each run in Table 1 and for runs 1−7 in Table 2 were calculated from the experimental temperature profile by numerically integrating the expression in eq 2, where t represents reaction time in minutes.25 The reaction ordinate value has no physical meaning in isolation. It is strictly used to compare relative thermal treatment intensities using a single parameter. R0 =

∫0

t

e(T(t ) ° C − 100 ° C)/14.75 ° Cdt

Figure 2. Variation of the biocrude yield with reaction ordinate for fast and conventional HTL.

of R0 are 10−30%, although the uncertainties in these data are large enough that we cannot be certain that the differences in yield are statistically significant. The lowest yield shown is comparable to the 9.5 wt % yield available via extraction from the unreacted algae, as indicated in Table 1. This similarity suggests that very little hydrothermal treatment has occurred at this very mild condition (heating at 300 °C for 1 min). As the severity of hydrothermal treatment increases, the biocrude yield increases to 50% or more, but mean yields this high were not observed for any runs with R0 > 108. The biocrude yield decreases steadily as the reaction ordinate exceeds 108, most likely the result of increased gasification reactions under such severe reaction conditions. The availability of high biocrude yields at values of R0 between 102 and 105 indicates that there is no need to use the longer reaction times of conventional HTL to achieve high yields. The results from conventional HTL in Figure 2 show reasonable agreement with the data from this present investigation into fast HTL. This similarity in the yields at similar values of the reaction ordinate suggests that R0 might be a useful parameter for judging the severity of algae liquefaction conditions in both the conventional and fast modes. A final observation that we make from Figure 2 is that the biocrude yields obtained from fast HTL at the two conditions that give R0 values of about 105 differ by more than twice the sum of their standard deviations. The higher yield of 66 ± 11 wt % was obtained by heating the reactor at 600 °C for 1 min, whereas the lower yield of 38 ± 0.5 wt % was obtained by heating at 300 °C for 3 min. Figure 1 shows that the final reactor temperatures were similar in both experiments. These two points indicate that the reaction ordinate alone, although useful for making qualitative comparisons of the combined effects of time and temperature, might not be sufficient to fully characterize fast HTL. It appears that the heating rate or power input during fast HTL might also be important. The three data points at R0 values of around 100.75 are also consistent with this view. Higher yields appeared when using higher set-point temperatures (faster heating rates), even though the values for the reaction ordinate were all similar. Of course, as was mentioned above, the larger uncertainties associated with these data prevent one from using them to draw a firm conclusion in this regard. Nevertheless, it does appear that the influence of

(2)

The values of R0 for runs 8−25 in Table 2 were calculated by assuming that the reactor was at the temperature indicated for the entire reaction time. That is, we assumed that the reactors heated instantaneously to the set-point temperature. Given the long reaction times in Table 2, this assumption has a negligible effect on the values of R0. Table 2 shows that previous work on HTL of this particular Nannochloropsis sp. has encompassed reaction conditions resulting in R0 ranging from 105 to 1010. Brown et al.12 used conditions that lead to a R0 range from 104 to 1014, but we omit these results because the solids content of the reaction mixtures varied from 3 to 16 wt %, which is known to affect the biocrude yield.18 The present work investigates HTL reaction conditions covering a range of R0 values from 101 to 1016. Thus, this report provides information about HTL at 1395

dx.doi.org/10.1021/ef301925d | Energy Fuels 2013, 27, 1391−1398

Energy & Fuels

Article

the heating rate during fast liquefaction deserves additional attention. Figure 3 illustrates how the HHV of the biocrude varies with the reaction ordinate. Within the range of R0 from 102 to 108,

Figure 5. Variation of light biocrude yield with reaction ordinate.

undergone secondary reactions to form heavy biocrude (presumably higher molecular weight compounds), although the literature contains conflicting views about this point.18,28 The yields of light and total biocrude are important metrics for assessing HTL, but the composition of the biocrude is also important. Biocrudes with more hydrogen and carbon and less oxygen, nitrogen, and sulfur are more desirable products. Accordingly, we next examine the influence of the processing conditions, as characterized by the reaction ordinate, on the atomic composition of the total biocrude. Figure 6a shows that the H/C ratio of the total biocrude decreases gradually with the reaction ordinate up to a value of about 109, at which point the decline becomes more pronounced. These harsher reaction conditions probably promote reactions that lead to the production of aromatic compounds.12 For comparison, the H/C ratio of petroleum crudes ranges from 1.4 to 1.9.29 Figure 6b shows that the O/C ratio of the biocrude is less than 0.2, even with very mild hydrothermal treatment. The O/ C ratio of the initial algal biomass is 0.35. A similar sharp initial decline in the O/C ratio between the algae and the biocrude has been noted in previous studies with conventional HTL.12 The O/C ratio in Figure 6b decreases steadily to approximately 0.08−0.1 as the reaction ordinate increases. More severe liquefaction conditions drive more of the oxygen atoms out of the biocrude, but the O/C ratio still greatly exceeds the 0.0004−0.01 O/C ratio for petroleum crude oils.29 This reduction in the oxygen content is also the main reason that the HHV increased with R0 in Figure 3. Figure 6b also shows how the N/C ratio varies with the reaction ordinate. This ratio is 0.14 for the initial algal biomass, but it is as low as 0.02 for the biocrude formed at mild conditions (R0 < 101). The N/C ratio then increases sharply with the reaction ordinate, however, and reaches a value of about 0.07 at R0 = 102. It remains near this value even as the liquefaction conditions become more severe. N/C ratios for petroleum crudes are 0.001−0.02.29 Figure 6c shows that the S/C ratio exhibits trends similar to the N/C ratio. The biocrudes formed at the mildest conditions have the lowest ratios, and these fall below that in the original algal biomass (0.0047). For reference, the range of S/C ratios for petroleum crudes is 0.0002−0.03.29 The ratio then quickly rises to a nearly constant value at R0 = 102. Unlike the N/C data, the S/C data indicate that this ratio can increase again at the most severe conditions investigated. The similar trends in panels b and c of Figure 6 for the N/C and S/C ratios led us to plot these two ratios together. Figure 7 shows the correlation between the N/C and S/C ratios for both the light and heavy biocrudes. It appears that these atomic ratios may be correlated, and further replication may clarify the

Figure 3. Variation of the biocrude HHV with reaction ordinate.

where yields of 50% or more are reported in Tables 1 and 2, the HHV increases steadily from about 33 to 37−38 MJ/kg. This increase in HHV with R0, although lesser in relative magnitude than the decrease in biocrude yield with R0, could compensate, at least in part, for the decrease in biocrude yield at higher values of the reaction ordinate. Figure 4 illustrates the

Figure 4. Variation of energy recovery in biocrude with reaction ordinate.

realization of this expectation by showing that the energy recovery at large values of R0 (e.g., 108) can be as high as it is at milder conditions (e.g., R0 = 102). As mentioned in the Materials and Methods, the total biocrude was fractionated into light (hexane-soluble) and heavy (hexane-insoluble) components. The light fraction generally has a higher H/C ratio and lower O/C, N/C, and S/C ratios (see the Supporting Information), which makes it a more desirable product than the heavy fraction. Figure 5 shows that the light biocrude yield (wt % daf) increases from about 10% at R0 values less than 101 to roughly 30% at R0 of approximately 109. The yield for R0 values less than 101 is only slightly higher than the 6.7% light biocrude yield available without HTL via the dichloromethane extraction of the raw algae. Large run-torun variability is observed, possibly because of the difficulty of working with the small sample masses. At higher values of the reaction ordinate, however, the yield of light biocrude dramatically decreased, and it is 1012. This decrease in yield is most likely due to conversion of light biocrude into gaseous products, as discussed in relation to Figure 2. It is also possible that some light biocrude may have 1396

dx.doi.org/10.1021/ef301925d | Energy Fuels 2013, 27, 1391−1398

Energy & Fuels

Article

fraction become incorporated into the biocrude, both its sulfur and nitrogen contents increase. These ratios were lowest at the mildest conditions examined, which suggests very little incorporation of protein-derived products in the biocrude at those mild conditions. In addition to exploring how the composition of the total biocrude varied with the reaction ordinate, we also examined this variation for the light and heavy biocrude. To the best of our knowledge, such detailed information about the composition of these two biocrude fractions has not been previously reported. Figure 8 depicts the H/C and S/C atomic ratios for

Figure 8. Variation of light (LBC) and heavy biocrude (HBC) atomic ratios with reaction ordinate: (a) H/C and (b) N/C. Figure 6. Variation of biocrude atomic ratios with reaction ordinate: (a) H/C, (b) O/C and N/C, and (c) S/C.

the light and heavy biocrude. Figure 8a shows that the H/C ratio for the light biocrude consistently exceeds the H/C ratio for the heavy biocrude produced at the corresponding R0 value. A higher H/C atomic ratio is indicative of a higher energy content, making light biocrude preferable to heavy biocrude. Figure 8b reinforces this idea, because the N/C ratio for the heavy biocrude exceeds that of the light biocrude at all of the reaction conditions examined here. The difference between the N/C ratios for the light and heavy biocrude is greatest at the mildest liquefaction conditions, and as the reaction ordinate increases, the difference becomes much smaller. Low N/C atomic ratios are preferred, because a high nitrogen content leads to undesirable NOx emissions upon combustion. Overall, the data in Figures 2−6 and Figure 8 show consistent trends in several important biocrude quality metrics (e.g., yield, composition, and HHV) over the range of R0 values explored in this work. Thus, the reaction ordinate or some similar metric may be a useful way to consolidate the effects of time and temperature into one parameter for the HTL of microalgae. The consolidation of time and temperature is particularly useful when considering non-isothermal conditions, such as those used in this study of fast HTL.

Figure 7. Comparison of atomic ratios (N/C and S/C).

trend. Generally, low N/C and low S/C occur together as do high N/C and high S/C. This correlation suggests a common origin and/or common reaction paths for the N and S atoms in the biocrude. The protein fraction in algae is generally where most of the N atoms reside, and the protein fraction can contain S atoms as well. As more molecules from the protein 1397

dx.doi.org/10.1021/ef301925d | Energy Fuels 2013, 27, 1391−1398

Energy & Fuels



Article

(6) Li, Y.; Horsman, M.; Wu, N.; Lan, C. Q.; Dubois-Calero, N. Biotechnol. Prog. 2008, 24, 815−820. (7) Hu, Q.; Sommerfeld, M.; Jarvis, E.; Ghirardi, M.; Posewitz, M.; Seibert, M.; Darzins, A. Plant J. 2008, 54, 621−639. (8) Savage, P. E.; Levine, R. B.; Huelsman, C. M. In Thermochemical Conversion of Biomass to Liquid Fuels and Chemicals; Crocker, M., Ed.; Royal Society of Chemistry (RSC) Publishing: Cambridge, U.K., 2010; RSC Energy and Environment Series, pp 190−219. (9) Akiya, N.; Savage, P. E. Chem. Rev. 2002, 102, 2725−2750. (10) Changi, S.; Matzger, A. J.; Savage, P. E. Green Chem. 2012, 14, 2856−2867. (11) Biller, P.; Ross, A. B. Bioresour. Technol. 2011, 102, 215−225. (12) Brown, T. M.; Duan, P.; Savage, P. E. Energy Fuels 2010, 24, 3639−3646. (13) Jena, U.; Das, K. C.; Kastner, J. R. Bioresour. Technol. 2011, 102, 6221−6229. (14) Li, D.; Chen, L.; Xu, D.; Zhang, X.; Ye, N.; Chen, F.; Chen, S. Bioresour. Technol. 2011, 104, 737−742. (15) Valdez, P. J.; Dickinson, J. G.; Savage, P. E. Energy Fuels 2011, 25, 3235−3243. (16) Yang, Y. F.; Feng, C. P.; Inamori, Y.; Maekawa, T. Resour., Conserv. Recycl. 2004, 43, 21−33. (17) Garcia Alba, L.; Torri, C.; Samorì, C.; Van Der Spek, J.; Fabbri, D.; Kersten, S. R. A.; Brilman, D. W. F. W. Energy Fuels 2012, 26, 642−657. (18) Valdez, P. J.; Nelson, M. C.; Wang, H. Y.; Lin, X. N.; Savage, P. E. Biomass Bioenergy 2012, 46, 317−331. (19) Knežević, D.; van Swaaij, W.; Kersten, S. Ind. Eng. Chem. Res. 2010, 49, 104−112. (20) Mayor, J. R.; Williams, A. J. Energy Resour. Technol. 2010, 132, 041801-1−041801-6. (21) Miao, X.; Wu, Q. J. Biotechnol. 2004, 110, 85−93. (22) Bridgwater, A. V.; Meier, D.; Radlein, D. Org. Geochem. 1999, 30, 1479−1493. (23) Vargas-Moreno, J. M.; Callejón-Ferre, A. J.; Pérez-Alonso, J.; Velázquez-Martí, B. Renewable Sustainable Energy Rev. 2012, 16, 3065− 3083. (24) Overend, R. P.; Chornet, E.; Gascoigne, J. A. Philos. Trans. R. Soc., A 1987, 321, 523−526. (25) Rogalinski, T.; Ingram, T.; Brunner, G. J. Supercrit. Fluids 2008, 47, 54−63. (26) Brasch, D. J.; Free, K. W. Tappi J. 1965, 48, 245−24848. (27) Vroom, K. E. Pulp Paper Res. Inst. Can. 1957, 38, 228−231. (28) Changi, S.; Zhu, M.; Savage, P. E. ChemSusChem 2012, 5, 1743−1757. (29) Speight, J. G. The Chemistry and Technology of Petroleum, 4th ed.; CRC Press (Taylor and Francis Group): Boca Raton, FL: 2007; p 243.

CONCLUSION A biocrude yield of 66 ± 11 wt % daf was obtained from HTL of Nannochloropsis sp. at a reaction time of 1 min and set-point temperature of 600 °C. This yield exceeds any reported previously for the HTL of this Nannochloropsis sp.12,18 Up to 91 ± 14% of the energy in the dry algae was retained in the biocrude produced at these conditions. HHVs for the biocrude produced by fast HTL at 1, 3, and 5 min reaction times were 33−37 MJ/kg, consistent with the 34−38 MJ/kg range for conventional HTL of this Nannochloropsis sp. Further, light biocrude was preferable to heavy biocrude obtained using identical reaction conditions, from an atomic composition perspective. This work indicates that energy-dense biocrudes may be obtained in high yield in a matter of minutes by rapidly heating algae slurries to perform HTL. This new approach, which we term fast HTL, is a significant improvement over the tens of minutes traditionally used to perform HTL. Reducing the reaction time decreases the reactor volume and, hence, the capital costs for a HTL biofuel process. This work also demonstrated the utility of the reaction ordinate as a means to combine the effects of time and temperature into one parameter for the HTL of algae. We find that the yield, composition, and heating value of biocrudes produced under different conditions can be correlated using the reaction ordinate. Additionally, results obtained by both fast and conventional HTL exhibit the same trends with the reaction ordinate, which indicates that this parameter might be generally useful for uniting and comparing results from liquefaction experiments performed under different reaction conditions.



ASSOCIATED CONTENT

S Supporting Information *

Reactor loadings (Table S1), elemental composition of light and heavy biocrude (Table S2), and solids yields from fast HTL experiments (Table S3). This material is available free of charge via the Internet at http://pubs.acs.org.



AUTHOR INFORMATION

Corresponding Author

*Telephone: +1-734-764-3386. Fax: +1-734-763-0459. E-mail: [email protected]. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS The authors acknowledge the National Science Foundation Grant EFRI-0937992 for financial support of this research. Peter J. Valdez thanks the Rackham Graduate School for financial support.



REFERENCES

(1) International Energy Agency (IEA). Key World Energy Statistics 2012; IEA: Paris, France, 2012. (2) Vardon, D. R.; Sharma, B. K.; Scott, J.; Yu, G.; Wang, Z.; Schideman, L.; Zhang, Y.; Strathmann, T. J. Bioresour. Technol. 2011, 102, 8295−8303. (3) Duan, P.; Savage, P. E. Bioresour. Technol. 2011, 102, 1899−1906. (4) Chisti, Y. Biotechnol. Adv. 2007, 25, 294−306. (5) Costa, J. A. V.; de Morais, M. G. Bioresour. Technol. 2011, 102, 2− 9. 1398

dx.doi.org/10.1021/ef301925d | Energy Fuels 2013, 27, 1391−1398