1. Introduction
Syngas (synthesis gas) fermentation is the microbiological conversion of CO-, H
2-, and CO
2-rich gases to short chain fatty acids or alcohols. If renewable resources are the feedstock for the gasification to syngas, then the microbial production of commercially relevant chemicals demonstrates higher sustainability than fossil feedstock. The biochemical conversion of syngas represents an alternative to the thermochemical conversion to, e.g., ethanol, given the advantages of milder temperatures and pressures, flexibility to H
2/CO ratios, and higher product selectivity [
1].
Some members of the genus
Clostridium are able to convert CO or CO
2 and H
2 to acetate and ethanol, along with other strain-specific products such as 2,3-butanediol, butanol, and hexanol [
2]. These acetogens use the reductive acetyl-CoA pathway for autotrophic carbon fixation. Energy in the form of ATP is conserved by a membrane-bound ATP-synthase, which uses either H
+ or Na
+ gradients across the membrane [
3].
C. autoethanogenum,
C. ljungdahlii, and
C. ragsdalei belong to the same clade [
4], having a sequence similarity of more than 99% [
5] and producing acetate, ethanol, and 2,3-butanediol from CO or CO
2 and H
2 [
6]. The growth optima are at 37 °C with a pH optimum of pH 5.5–6.0 [
7,
8]. All three strains favor the use of CO over H
2 and CO
2, as theoretically proposed [
9] and shown experimentally for
C. ljungdahlii [
10]. The batch process performances of
C. autoethanogenum and
C. ljungdahlii were studied in anaerobic flasks with 50 kPa, 45 kPa, and 5 kPa of CO, H
2, and CO
2, respectively.
C. autoethanogenum and
C. ljungdahlii achieved similar optical densities and acetate concentrations, whereas the highest ethanol and 2,3-butanediol concentrations were measured with
C. autoethanogenum [
7].
C. ragsdalei achieved lower optical densities, but it was able to produce more ethanol than acetate.
When generated from biogenic materials, syngas contains (along with the main gas components CO, CO
2, H
2, and N
2) a variety of impurities, depending on the gasification process and feedstock. These might include H
2S, NH
3, COS, HCN, and NO
X [
11]. The formation of these impurities can to some extent be controlled by the choice of the gasification conditions. Increasing the equivalence ratio (mass ratio between air and fuel in the gasification process) has been shown to increase the conversion of fuel bound nitrogen to ammonia and N
2 but had no influence on the formation of HCN during the gasification of switchgrass [
12].
It has already been demonstrated that syngas impurities accumulated in the aqueous fermentation broth may, depending on their concentrations, have an impact on the performance of the fermentation processes [
13]. The addition of up to 29.4 mM sulfide, which corresponds to 1.0 g L
−1 H
2S, promoted autotrophic growth and alcohol formation in batch processes with
Clostridium carboxidivorans in a continuously CO/CO
2 gassed stirred-tank reactor [
14]. However, even the addition of 1.9 mM sulfide (0.065 g L
−1 H
2S) slightly decreased cell dry weight (CDW) concentration and promoted ethanol production in anaerobic flasks with
C. ragsdalei using artificial syngas as a substrate [
15]. Sulfide inhibits anaerobic bacteria above a certain threshold and is associated with the undissociated H
2S solved in water, which is membrane-permeable [
16]. Once inside the cell, it can cause DNA damage and protein denaturation [
17]. As a result, there are large differences between the effects of H
2S on different anaerobic strains.
The addition of 93.5 mM NH
4+ (5.00 g L
−1 NH
4Cl) promoted the autotrophic growth and alcohol formation of
C. carboxidivorans in a continuously CO/CO
2 gassed stirred-tank reactor [
14], but the addition of 93.4 mM NH
4+ (1.68 g L
−1) showed no effect on the growth or product formation of
C. ragsdalei in anaerobic flasks with a CO/CO
2 atmosphere [
18]. Ammonium has been shown to decrease the activity of hydrogenases and alcohol-dehydrogenases in acetogens [
19], which catalyze the reactions for the supply of reduction equivalents from H
2 or CO and for alcohol formation, respectively. Therefore, NH
4+ concentrations above a certain threshold could be an obstacle in syngas fermentation processes with the concomitant supply of CO and H
2/CO
2.
Concentrations above 40 ppm nitric oxide in the syngas inhibited the autotrophic growth of
C. carboxidivorans with H
2/CO
2, but this inhibition was reversible, and a complete inhibition of the hydrogenase activity was determined at 150 ppm NO [
20].
Other nitrogen species also affect syngas fermentations with
Clostridia. The addition of 0.1 g L
−1 NaNO
3 increased the lag phase of
C. carboxidivorans to 30 h, but higher final CDW concentrations and a strong increase in butyrate concentrations was observed with artificial syngas in a stirred-tank bioreactor [
14]. A total of 15 mM NaNO
3 (1.275 g L
−1) promoted the biomass growth of
C. ljungdahlii in anaerobic flasks with H
2 and CO
2 as the gas phase, but growth was inhibited with CO and CO
2 [
21]. Moreover, the addition of nitrate reduced ethanol formation and promoted formate production. Substituting ammonium with nitrate on a molar basis led to higher optical densities in a chemostat at mean hydraulic residence times between 2 and 3.5 d using
C. ljungdahlii with H
2 and CO
2 as the gas phase, but substituting ammonium with nitrate was associated with “stochastic metabolic crashes” [
22].
C. ljungdahlii is able to reduce nitrate to ammonium, resulting in more than a doubling of the net ATP gain from H
2 when compared to the production of acetate from H
2 [
21].
Information on the effect of syngas impurities on syngas fermentation processes is sparse, and the comparison between different strains is obscured by varying process designs and conditions [
13]. Many of the published results are based on simple batch studies applying non-controlled anaerobic flasks with low gas amounts, low gas-liquid mass transfer rates, and low power input, but these conditions do not reflect scalable continuously gassed syngas fermentation processes and lack the quantitative analysis of gas consumption rates and, therefore, carbon balances. Our report represents an extensive comparison of fully controlled syngas fermentation processes with
C. autoethanogenum,
C. ljungdahlii, and
C. ragsdalei, respectively, in stirred-tank bioreactors at defined power input, gassing rates, and identical medium compositions for studying the strain-specific effects of defined additions of ammonium, sulfide, and nitrate on batch process performance. The process performance of these
Clostridia were compared to published results using
C. carboxidivorans [
14]. Since the syngas components investigated may be present as impurities in syngas, the quantitative criteria for syngas purification processes were identified. This study also provides insights into the choice of preferred strains respecting their individual capabilities of converting CO-rich syngas.
3. Results and Discussion
3.1. Autotrophic Reference Batch Processes in Continuously Gassed Stirred-Tank Bioreactors
All of the strains were first cultivated without the addition of any syngas impurities as autotrophic reference batch processes (
Figure 1). It has to be noted that for the microorganism
C. ljungdahlii, a reduced CO partial pressure of 200 mbar was chosen as opposed to 600 mbar for
C. autoethanogenum and
C. ragsdalei. This reduction was chosen because preliminary studies showed increased growth and ethanol formation for this strain at reduced CO partial pressure (data not shown), and this study aimed to compare the three strains at their best performance. The highest CDW concentrations were observed with
C. ragsdalei (0.56 g L
−1), followed by
C. autoethanogenum (0.52 g L
−1) and
C. ljungdahlii (0.23 g L
−1).
C. ragsdalei produced considerably more acetate, with a final concentration of 4.61 g L
−1, whereas
C. autoethanogenum produced the highest ethanol and 2,3-butanediol concentrations (2.51 g L
−1 ethanol and 0.53 g L
−1 2,3-butanediol, respectively). No 2,3-butanediol formation was observed with
C. ljungdahlii. CO was the only gaseous substrate to be consumed by all strains. No H
2 uptake was observed in any of the batch processes. Part of the CO was converted to CO
2 for providing the necessary reducing equivalents. Further details on total CO consumption and CO
2 production are given in
Tables S1–S3 in the Supplementary Materials. The maximal CO uptake rates varied considerably at ≈16 mmol L
−1 h
−1 (
C. ragsdalei), ≈12 mmol L
−1 h
−1 (
C. autoethanogenum), and ≈6 mmol L
−1 h
−1 (
C. ljungdahlii). Compared to the high volumetric flow rate of the syngas, the CO uptake rates are relatively low, resulting in maximum CO conversions of 11.9% (
C. ragsdalei), 9.0% (
C. autoethanogenum), and 13.4% (
C. ljungdahlii), which does not support any kind of limitation by the carbon input.
The final alcohol to acetate ratio achieved in the autotrophic reference batch processes exhibited a distinct maximum using
C. autoethanogenum (7.60 g g
−1), as compared to 0.86 g g
−1 with
C. ljungdahlii or 0.41 g g
−1 with
C. ragsdalei. The high CO uptake rates observed with
C. ragsdalei mainly resulted in high acetate production. High organic acid production rates and concentrations have been associated with a failure to trigger alcohol production [
25,
26], as well as with higher ATP maintenance costs [
27].
The observed maximum specific growth rate (0.050 h
−1) and maximum CO uptake rate (6 mmol L
−1 h
−1) of
C. ljungdahlii were in accordance with the published data [
28]. The reported maximum specific growth rates of
C. ragsdalei varied considerably, e.g., 0.175 h
−1 [
8] or 0.065 h
−1 [
29]. The growth rate observed in this study was in line with the reported data (0.116 h
−1). The measured maximum specific growth rate of
C. autoethanogenum (0.065 h
−1) exceeded that of the data from the literature, e.g., 0.042 h
−1 achieved with a gas mixture of 2% CO, 23% CO
2, and 65% H
2 [
30].
3.2. Defined Addition of Impurities: Ammonium
Ammonium was supplemented as NH
4Cl before inoculation with
C. autoethanogenum (+ 1.0 g L
−1, + 3.0 g L
−1, and + 5.0 g L
−1 NH
4Cl),
C. ljungdahlii (+ 3.0 g L
−1, and + 6.0 g L
−1 NH
4Cl), and
C. ragsdalei (+ 3.0 g L
−1, + 6.0 g L
−1, and + 9.0 g L
−1 NH
4Cl) (see
Figure 1). The initial addition of 3.0 g L
−1 NH
4Cl reduced the final CDW concentration of
C. autoethanogenum by 85%, of
C. ljungdahlii by 26%, and of
C. ragsdalei by 45%. Growth inhibition was observed for
C. autoethanogenum after the addition of 5.0 g L
−1 NH
4Cl and for
C. ljungdahlii and
C. ragsdalei after the addition of 6.0 g L
−1 NH
4Cl. The inhibitory effect of NH
4Cl addition was also reflected in the observed decrease in the CO uptake rates in all batch processes after supplementation with ammonium. Alcohol production was reduced after any initial addition of NH
4Cl. The supplementation with 3.0 g L
−1 NH
4Cl prevented any formation of 2,3-butanediol with
C. autoethanogenum and
C. ragsdalei.
3.3. Defined Addition of Impurities: Nitrate
Nitrate was added as NaNO
3 before inoculation with
C. autoethanogenum (0.1 g L
−1, 0.2 g L
−1, 0.5 g L
−1, and 1.0 g L
−1 NaNO
3),
C. ljungdahlii (0.1 g L
−1, and 0.5 g L
−1 NaNO
3), and
C. ragsdalei (0.1 g L
−1, 0.2 g L
−1, and 0.5 g L
−1 NaNO
3) (see
Figure 2). Adding nitrate slowed the growth of
C. autoethanogenum and
C. ragsdalei in all of the nitrate concentrations studied. The complete growth inhibition of
C. ragsdalei was observed after adding 0.5 g L
−1 NaNO
3.
C. ljungdahlii exhibited the highest tolerance after nitrate additions, with little difference in the final cell dry weight concentrations in all cases, but a clearly shorter exponential growth phase.
No net production of acetate was observed with
C. ljungdahlii in the final 30 h of the batch processes with 0.1 g L
−1 NaNO
3 and 0.5 g L
−1 NaNO
3, respectively. A strong reduction of acetate production occurred in the batch processes with
C. ragsdalei, which was independent of the initial NaNO
3 concentration. Nitrate is known to increase the ATP/ADP ratio in
C. ljungdahlii [
21]. More ATP can thus be provided through nitrate reduction with less acetate production for biomass formation.
Ethanol production was reduced in all of the batch processes with the addition of NaNO3. C. ljungdahlii showed the lowest reduction of final ethanol concentrations. 2,3-Butanediol formation was strongly inhibited by nitrate, and no 2,3-butanediol was detected with C. ljungdahlii and C. ragsdalei. A reduced formation of 0.10 g L−1 2,3-butanediol was observed with C. autoethanogenum at 0.1 g L−1 NaNO3.
3.4. Defined Addition of Impurities: Hydrogen Sulfide
Hydrogen sulfide was added as thioacetamide (TAA) before inoculation with
C. autoethanogenum (0.1 g L
−1 H
2S, 0.3 g L
−1, and 0.5 g L
−1), as well as with
C. ljungdahlii and
C. ragsdalei (0.1 g L
−1, and 0.5 g L
−1 H
2S) (see
Figure 3).
All of the TAA concentrations investigated decreased the cell dry weight concentrations; the final product concentrations; and, correspondingly, the CO uptake rates of C. autoethanogenum and C. ljungdahlii. Adding 0.3 g L−1 H2S completely inhibited the growth and CO uptake of C. autoethanogenum. No growth of C. ljungdahlii and C. ragsdalei was observed with 0.5 g L−1 H2S. A concentration of 0.1 g L−1 H2S induced a lag phase of 20 h with C. autoethanogenum, but it increased acetate production in the first 30 h. The acetate concentration later decreased until the end of the process. 2,3-Butanediol formation of C. autoethanogenum was strongly inhibited in the batch process with 0.1 g L−1 H2S. The addition of 0.1 g L−1 H2S resulted in less biomass formation with C. ljungdahlii, a delayed acetate production, and no alcohol formation.
The addition of 0.1 g L−1 H2S increased the final cell dry weight concentration of C. ragsdalei by 34%. Higher CO uptake rates were also observed, with maximal CO uptake rates reaching approximately 21 mmol L−1 h−1. Product formation shifted from acetate to alcohol production, with not only higher final concentrations of ethanol and 2,3-butanediol when compared with the reference batch process for each strain, but also higher biomass-related yields of both alcohols. This effect might occur due to the additional presence of a sulfur source or to the reducing effect of H2S, which may lead to a more reduced redox potential in the cultivation medium and, thus, an increase in growth and reduction of acetate to ethanol.
3.5. Comparison of Clostridial Strains
The results with the acetogens
C. autoethanogenum,
C. ljungdahlii, and
C. ragsdalei of this study were compared to previously published reference data with
C. carboxidivorans [
14], which have been measured in batch operated stirred-tank bioreactors at comparable reaction conditions. An overall comparison of the autotrophic reference batch process performances of the three strains studied, including the published results for
C. carboxidivorans [
14], shows that
C. autoethanogenum was able to produce the highest amounts of biomass and alcohols while also maintaining low acid concentrations (
Figure 4).
C. ragsdalei achieved 91% (
w/
w) of the maximum cell dry weight concentration measured with
C. autoethanogenum and further produced the highest amounts of organic acids.
C. ljungdahlii showed the lowest production of biomass, organic acids, and alcohols.
The addition of ammonium favored biomass and alcohol formation in the autotrophic batch processes with
C. carboxidivorans [
14] while reducing the productivity of the other three
Clostridial strains (
Figure 5).
C. autoethanogenum demonstrated the lowest tolerance to NH
4Cl, with growth inhibition already occurring with the supplementation of 3.0 g L
−1 NH
4Cl, whereas
C. ljungdahlii and
C. ragsdalei showed inhibition with the supplementation of 6.0 g L
−1 NH
4Cl.
H
2S addition promoted biomass, acid, and alcohol formation in the autotrophic batch processes with
C. ragsdalei at a low initial concentration (0.1 g L
−1 H
2S).
C. carboxidivorans biomass production was increased with both added sulfide concentrations, whereby an increase in the formation of alcohols was solely observed with 0.5 g L
−1 H
2S [
14]. The other strains exhibited reduced productivities after H
2S addition, with
C. ljungdahlii showing a strong inhibition to sulfide. The addition of 0.5 g L
−1 H
2S resulted in strong inhibitions of
C. autoethanogenum,
C. ljungdahlii, and
C. ragsdalei and thus represents a critical impurity in the syngas fermentation with these strains.
Nitrate slowed or inhibited the biomass formation of
C. autoethanogenum and
C. ragsdalei, with the effect increasing at higher initial nitrate concentrations.
C. ljungdahlii was, in turn, less influenced by nitrate, leading to equivalent cell dry weight concentrations with and without nitrate, as well as showing a higher tolerance to this impurity. The alcohol formation in all of the strains studied was suppressed by nitrate, including
C. carboxidivorans [
14], but
C. carboxidivorans was stimulated by the addition of 0.1 g L
−1 NaNO
3 and produced more biomass and organic acids [
14].
Overall, the three
Clostridial strains studied exhibited similar responses to the added impurities. These responses were, in turn, very different from those published with
C. carboxidivorans [
14], which showed more robustness to the effects of the syngas impurities NH
3 and H
2S. Since
C. carboxidivorans is genetically not as closely related to the other three studied strains [
4,
5], it is not surprising that its response to the defined impurities should differ.
C. autoethanogenum, C. ljungdahlii, and
C. ragsdalei were generally inhibited by the addition of H
2S (as TAA), nitrate, and ammonium, with the exception of the addition of 0.1 g L
−1 H
2S to the autotrophic batch process with
C. ragsdalei. As a consequence, the accumulation of 0.1 g L
−1 H
2S, 0.073 g L
−1 NO
3−1 (equivalent to 0.1 g L
−1 NaNO
3), and 2.14 g L
−1 NH
4+ (equivalent to 6.3 g L
−1 NH
4Cl) in fermentation processes with real syngas should be avoided.
Further details on total CO consumption, total CO
2 production, carbon balance recoveries, specific growth rates, and the maximum concentrations of CDW and products are provided in the
Supplementary Materials (Tables S1–S3) for all the batch processes described in this work using the individual microorganisms (
C. autoethanogenum,
C. ljungdahlii, and
C. ragsdalei, respectively).
The identified inhibiting concentrations of impurities in the liquid phase were used to estimate the corresponding concentrations in the gas phase and, therefore, provide a quantitative goal respecting the quality requirements of real syngases. The typical orders of magnitude for trace impurities in real biogenic syngas from entrained flow gasification of biogenic residues are 4500 ppm NH
3, 500 ppm H
2S, and 200 ppm NO
x [
31,
32]. The corresponding concentrations in a syngas were estimated given the assumption of complete absorption of the syngas trace component in the liquid phase within a process time of 60 h, as previously described [
14]. Typical solubilities of the investigated gas impurities in pure water at 25 °C and 1013.25 mbar partial pressure are 0.1876 mol NH
3 mol
−1 H
2O, 1.830∙10
−3 mol H
2S mol
−1 H
2O, 3.477∙10
−5 mol NO mol
−1 H
2O, and 1.488∙10
−4 mol NO
2 mol
−1 H
2O [
33]. These concentrations correspond to 199.7 g L
−1 NH
3, 3.90 g L
−1 H
2S, 0.065 g L
−1 NO, and 1.28 g L
−1 NO
2, respectively. However, it has to be noted that these concentrations do not take any chemical reaction into account [
33]. All investigated concentrations of NH
4+, H
2S, and NO
3− are in the range of these typical solubilities. It should be noted that the impurities threshold identified in this study represents a conservative limit, since in a continuously gassed process, the impurities would gradually accumulate in the cell broth. Thus, the growth phase would occur at lower impurity concentrations than the one in this study, leading presumably to higher CDW and product concentrations and the adaption of the cells.
The initial addition of 3.0 g L
−1 NH
4Cl in the liquid phase was found to be inhibiting for growth and product formation of
C. autoethanogenum,
C. ljungdahlii, and
C. ragsdalei in the autotrophic batch processes. Under the given assumptions, 3.0 g L
−1 NH
4Cl would be reached at a concentration of 4560 ppm NH
3 in the gas phase within a total batch process time of 60 h. Since the standard medium in all of the autotrophic batch processes already contained 3.3 g L
−1 NH
4Cl [
23], a reduction of the initially supplied NH
4Cl could reduce the inhibiting effect of NH
3 provided by a typical biogenic syngas (4500 ppm NH
3). It has already been shown that a reduction of the initial ammonium concentration in the medium by 50% did not influence biomass growth or product formation of
C. ragsdalei [
18]. Additionally, with the continuous gassing, the ammonium concentration would increase with process time rather than remaining constant, enabling a possible adaption of the cells. The growth of
C. ragsdalei resumed with 6.0 g L
−1 and 9.0 g L
−1 NH
4Cl after 100 h (data not shown), and growth of
C. autoethanogenum resumed with 3.0 g L
−1 NH
4Cl after 78 h (data not shown), indicating that an adaption to increasing ammonium concentrations may be possible.
The initial addition of 0.1 g L−1 NaNO3 was found to inhibit growth and alcohol production with C. ragsdalei, and C. autoethanogenum. An amount of 0.1 g L−1 NaNO3 would correspond to a gas phase concentration of 118 ppm NOx after a process time of 60 h if the entire amount of nitrogen from the NOx was absorbed in the liquid phase and converted into NO3−. Therefore, the purification of a typical biogenic syngas at 200 ppm NOx would be necessary to ensure a stable process without reduced alcohol production. However, the growth of C. ljungdahlii was not affected, and its alcohol production was only slightly reduced by the addition of NaNO3 concentrations of up to 0.5 g L−1 NaNO3. Under the same assumptions, a concentration of 0.5 g L−1 NaNO3 would be reached after 60 h with a syngas at 588 ppm NOx. Thus, a biogenic syngas at 200 ppm NOx might not need a further purification for this trace component if C. ljungdahlii were applied for syngas fermentation.
An initial concentration of 0.1 g L−1 H2S was found to inhibit growth as well as alcohol production of C. autoethanogenum and C. ljungdahlii. An amount of 0.1 g L−1 H2S corresponds to 108 ppm H2S in the syngas within a batch process time of 60 h. The inhibiting concentration for C. ragsdalei of 0.5 g L−1 H2S corresponds to a concentration of 540 ppm H2S in the gas phase for 60 h. Thus, a biogenic syngas at a typical concentration of 200 ppm H2S would not be critical in autotrophic batch processes with C. ragsdalei. If C. autoethanogenum or C. ljungdahlii were applied for syngas fermentation, H2S separation from the biogenic syngas would be necessary.