Degradation of Components After Exposure in a Biomass Pyrolysis System

Biomass is of increasing interest as a renewable energy source.^1-5  Nonfood biomass feedstocks under investigation include woods, grasses, and agricultural residues. Successful commercialization of biomass pyrolysis and related liquefaction technology to make liquid fuels will require identification of low-cost, corrosion-resistant materials for use in production, transportation, and storage of bio-oils. Although corrosivity of biofuels is an active area of research,^6-10  comparatively few studies of process equipment corrosion have been pursued.^11-13  Process equipment will be exposed to temperatures as high as ∼550°C, with biomass feedstock decomposition products and aggressive O, S, C, H, Cl, etc. containing species potentially encountered. Austenitic stainless steels are the primary alloys of interest for these applications. These environments are challenging to accurately reproduce in a laboratory corrosion study setting. Therefore, in situ alloy test sample exposures in operating biomass pyrolysis and related systems have been pursued, as well as study of operated components run in such systems.^10-12  The present work reports an advanced microstructural characterization study of the corrosion products formed on a 304L (UNS S30403⁽¹⁾) stainless steel component operated in a biomass fast pyrolysis system.

Table 1 shows a summary of the operational history for a 304L stainless steel fluidized bed reactor component run in the Iowa State Pyrolysis Process Development Unit (PPDU).^14-15  The ∼38 cm long nominal schedule 40 pipe section (∼16.8 cm outer diameter and ∼0.7 cm wall thickness) that was deconstructed for analysis contained both the active fluidized bed region of the reactor and the biomass inlet port. The pipe section typically saw ∼0.07 atm conditions, with short durations to ∼0.17 atm. The section was heated using two, half-shell ceramic fiber heaters providing heat input to the process. The reactor was used for multiple feedstock, biomass fast pyrolysis process studies totaling ∼480 h of operation, with the vast majority of the operation conducted at 500°C using red oak feedstock. The reactor segment and locations sectioned for microstructural analysis are shown in Figure 1.

For purposes of comparative benchmarking for the corrosion products formed in the 304L from which the reactor was constructed, a 304L stainless steel test sample (different 304L batch than the 304L the reactor was constructed from) was prepared to a 600 grit surface finish and oxidized isothermally for 500 h at 550°C in air (laboratory furnace open to ambient air). Analyzed compositions for the 304L reactor and 304L oxidation test sample are reported in Table 2.

Characterization of the corrosion products followed procedures described by Brady, et al.¹²  The cross sections were prepared by standard metallographic techniques using nonaqueous polishing media to retain any aqueous soluble corrosion products (particularly chlorides) which may have formed. Characterization was initially accomplished using x-ray diffraction (XRD), scanning electron microscopy (SEM) equipped with energy dispersive x-ray spectroscopy (EDS) and electron probe microanalysis (EPMA) with EDS and wavelength dispersive spectroscopy (WDS) techniques. The corrosion scale as well as locations of internal attack were further analyzed by scanning transmission electron microscopy (STEM) using focused ion beam (FIB) milling lift-out sample preparation procedures.

SEM and STEM analysis of the oxidized cross section of the benchmark 304L test sample exposed for 500 h at 550°C in air are shown in Figure 2. The oxide scale was submicron, consistent with good oxidation resistance by 304L at this relatively benign oxidation condition of 550°C in air. The oxide scale was typical for that of chromia-forming stainless steels,¹⁶  consisting of a duplex oxide structure, with an outer region of discontinuous Fe-rich oxide particles overlying an inner, more continuous Cr-rich region. The STEM EDS maps also suggested a mild degree of Cr depletion and Ni enrichment at the alloy grain boundaries adjacent to the oxide scale.

The external oxide scale retained on the inner surface of the 304L fluidized bed reactor component was ∼10 µm thick (Figure 3), much thicker than the submicron air-formed benchmark 304L oxide scale after comparable total exposure times of 480 h to 500 h (Table 1 and Figure 2). (It should be noted that the comparison 304L test sample oxidized in air was flat, whereas the inner reactor surface was a curved pipe). As with the air-formed oxide scale on the 304L benchmark test sample, the scale formed in the reactor environment was nominally duplex, with an Fe-rich outer region (also containing some Ca), and a Cr-rich inner region evident in SEM EDS mapping. XRD analysis at multiple locations on the inner reactor surface indicated that the oxide was primarily Fe₃O₄ magnetite phase, which can be substituted by Fe_xCr_3-xO₄ and related complex oxides. Carbon in the form of graphite-2H may also have been present, consistent with a black deposit visually observed on much of the inner reactor surface. An amorphous hump was also observed in the XRD data. (The scale formed on 304L in 550°C, 500 h air oxidation was considered too thin for XRD analysis). In further contrast to the air oxidized 304L (Figure 2), regions of internal alloy grain boundary attack extending 5 µm to 15 µm deep were also frequently observed on the inner surface of the 304L reactor (Figure 3). SEM/EDS analysis indicated that the internal attack resulted from internal oxidation and was associated primarily with Cr-rich oxide (the internal oxide may also have contained Fe).

Elemental mapping and backscatter imaging by EPMA of a typical area of oxide scale and internal attack in the 304L fluidized bed reactor inner surface are shown in Figure 4 (different location than that shown in Figure 3). Enhanced backscatter contrast (Figure 4[b]) confirmed that the internal attack was along the alloy grain boundaries. WDS elemental mapping indicated that the external scale and internal attacked regions were O-rich, but also contained some S. The external oxide scale was found to be rich in Ca, K, and Si, which is attributed to the biomass pyrolysis feedstock decomposition products. The regions in the vicinity of the internal alloy grain boundary attack were enriched in Ni at the expense of Cr and Fe (lateral map resolution of approximately several micrometers). No enrichment or depletion of Mo or Mn to the internal attacked region was evident in the EPMA maps, nor was Cl detected.

To further investigate the nature of the oxidation attack in the reactor environment, FIB sample preparation techniques were used to lift-out local regions of the oxide scale and internal grain boundary attack for STEM analysis (Figure 5). Figure 6 shows the STEM cross section of the oxide scale from the C deposit/oxide scale interface at the inner reactor surface to the oxide scale/alloy interface. Although the oxide scale appeared nominally duplex at the SEM image level with an Fe-rich outer layer and Cr-rich inner layer (Figure 2), the STEM analysis indicated a complex, multilayered structure (Figure 6[a]). Fe was present throughout the scale (Figure 6[d]), with the highest levels detected at the outer scale underneath the C-rich deposit, consistent with the SEM analysis. However, several alternating Cr-rich and Cr-lean regions were detected in the middle and inner regions of the scale, with Ni-rich and Fe-containing regions interspersed (Figures 6[e] and [f]). Local regions of S enrichment associated with the Ni-rich regions were observed in the middle of the scale, although they did not appear to directly correlate with a Ni-S phase. The Ni-rich regions also did not appear to be correlated with O, suggestive of entrapped Ni-rich metal within an oxide-based scale. Consistent with the EPMA analysis, the outer scale regions were rich in Si, K, and Ca, with P and Mg also detected (P and Mg maps not shown in Figure 6). Overall, a less protective, multilayered, and multicomponent oxide-based scale structure was formed, without the desired continuous, inner Cr-rich oxide needed for optimal resistance to oxidation.

STEM elemental mapping at the deepest extent of an internal attack region along the alloy grain boundaries (Figure 7) showed local enrichment of O, but not S. Although S was observed in the scale and internal attack regions by EPMA, S was not observed to extend to the attack front at the alloy grain boundary-internal attack interface region. Consistent with the EPMA, Si, Ca, and K observed in the external scale were not detected at the internal attack front by STEM mapping. Immediately adjacent to the internally attacked alloy grain boundary regions, Cr and Mn depletion and Ni enrichment were observed. The local Mn depletion was not detected in the EPMA mapping, likely due to limited resolution compared with the STEM.

A higher magnification view of the internal attack regions in Figure 7 are shown in Figure 8. The STEM dark field and secondary electron images revealed that local nanoscale porosity was present in the attacked alloy grain boundaries (it was not possible to discern if these voids were in addition to the attacked grain boundary, or part of the attacked grain boundary through a 3D network). Again, Cr and Mn depletion and Ni enrichment in the vicinity of the alloy grain boundaries were evident. A semiquantitative estimate of the degree of local depletion yielded a decrease from ∼19 wt% to ∼9 wt% Cr and ∼2 wt% to ∼1 wt% Mn, with concomitant increase in Ni of ∼7 wt% to ∼14 wt% and Fe from ∼72 wt% to ∼76 wt% (note that beam overlap with underlying alloy matrix may have impacted this estimate). The attack on the grain boundaries was associated with O, but not S. The STEM maps were inconclusive with regards to the local internal oxide(s) formed at the alloy grain boundaries.

Compared with 500 h oxidation at 550°C in air, exposure of 304L in a biomass fast pyrolysis fluidized bed system operated for ∼480 h from 350°C to 550°C (Table 1) resulted in a greatly increased extent of external scaling and internal attack along alloy grain boundaries. The faster oxidation scaling rate is attributed to impacts of the S, K, Ca, Mg, and P species from the biomass feedstock decomposition and the inability to establish a locally protective inner continuous Cr-rich oxide layer in the resultant complex C-H-O gas mixtures. In recent work, similar enhanced scaling and internal attack of 316L (UNS S31603) and 304L stainless steels have been observed in a range of biomass liquefaction related process environments.^10-12  In particular, the internal attack was related to both internal oxidation and internal sulfidation, with Ni-enriched alloy areas at the internal attack-alloy interface surrounding discrete Cr-S particles.¹²  The regions of internal attack (both at alloy grain boundaries and intragrain regions) also resulted in local, porous Cr oxide formation. Several instances of Mn-S and Mo-S particles, rather than Cr-S, were also observed. Based on those findings, it was hypothesized that the internal attack was triggered by the S preferentially tying up Cr.¹²  However, the findings of the present work suggest that the internal attack in biomass pyrolysis environments can also occur without preferential local sulfidation of Cr at the advance of the internal attack front.

Sulfur was present in the fluidized bed reactor environment of the PPDU examined in the present work, as evidenced by the detection of S by EPMA analysis in the external scale and the internal attack regions (Figure 4). However, the absence of discrete sulfide phases at the advance of the internal attack front in the STEM imaging suggests that this S did not play a direct role in the internal attack. S would be expected in small amounts in the environment from the source biomass feedstock. A typical batch of the red oak and corn stover biomass feedstock contained S in the range of 0.01 wt% and 0.15 wt%, respectively.¹⁵  Biomass feedstocks can also introduce other aggressive species such as Cl,¹⁷  as well as various ash and related deposits including Ca, K, Si, Mg, and P species observed in the external scale in the present work (these were not detected ahead of the internal attack front).

At relatively low temperatures such as the 350°C to 550°C reactor environment, alloy grain boundary diffusion plays a major role in transport of the protective scale-forming element, Cr in the case of 304L stainless steel, toward the external scale oxidation interface.^18-21  This preferential outward transport of Cr in light of the relatively rapid external scaling encountered, attributed to impacts of the Ca, K, S, Mg, P, and Si biomass decomposition product species, and the complex C-H-O gas mixtures in the fast pyrolysis environment resulted in the alloy grain boundary regions depleted in Cr. This depletion of Cr resulted in local nanoscale voids, effectively enriched the surrounding alloy grain boundary regions in Ni and left the grain boundaries vulnerable to internal attack. These microstructural internal attack features have analogs in a wide variety of environments. Local Ni enrichment due to high-temperature oxidation and corrosion of stainless steels has been observed in bio-oil + vacuum oil coprocessing, simulated biomass firing environments, and superheater tubes from biomass boilers.^12,22-24  Recently, selective oxidation and alloy grain boundary Cr depletion and local nanoporosity have also emerged as key features in high-temperature water environments relevant to pressurized water reactors (Ni-Cr base alloys), as well as model alloy systems for study of intergranular stress corrosion cracking.^25-27 

From an engineering viewpoint, it is not yet clear if 300 series stainless steels such as 304L (and 316L) are sufficiently corrosion resistant for long term use (∼5 y to 10 y) in biomass pyrolysis and related liquefaction process equipment. Assuming linear reaction kinetics, 10 µm of scaling in 500 h yields 1 mm of scaling in 50,000 h, which may be tolerable. This is a simplistic estimate, and does not consider a possible transition to accelerated break away kinetics or the possibility of slower-growing parabolic kinetics. Of potentially greater concern is the internal attack observed on alloy grain boundaries, which extended 5 µm to 15 µm in the 500 h exposure of the present work and could detrimentally impact mechanical integrity, especially if the component were pressurized. Additional studies of operated equipment or in situ alloy test samples with much longer-term exposures are greatly needed to more definitively answer these questions, and provide a sound kinetic basis for lifetime modeling inputs. Finding such exposure opportunities is difficult, as most pilot scale systems are devoted to short-term process optimization studies to understand and improve the quality and cost effectiveness of the resultant bio-oils, rather than long-term operation with continuous loading of biomass feedstock. The development of laboratory furnace corrosion tests capable of simulating complex biomass pyrolysis environments would also be most welcome, although the extent to which the biomass decomposition products and resultant C-H-O gas mixtures could be duplicated in a laboratory furnace setting remains to be demonstrated. Linked efforts of in situ plant exposures and laboratory simulation studies have succeeded in providing new insights into corrosion mechanisms in biomass-fired power plants.²⁸ 

Advanced microstructural characterization by EPMA and STEM of the corrosion products formed on a 304L fluidized bed reactor segment run in the Iowa State University PPDU was pursued.

• •

  The extent of scaling and internal attack observed for 304L stainless steel after ∼480 h operation in biomass fast pyrolysis at ∼500°C (ranging from 350°C to 550°C) was significantly greater than that observed for a 304L control test sample oxidized in air for 500 h at 550°C.

• •

  Internal attack in the biomass fast pyrolysis environment proceeded along alloy grain boundaries. Although O, S, Ca, K, Mg, and P were detected in the external scale, only O was definitively detected at the internal oxidation front along the alloy grain boundaries.

• •

  The internal attack of the 304L alloy grain boundaries was associated with local nanoscale porosity, and Cr (and Mn) depletion and Ni enrichment. The vulnerability to internal attack is a consequence of the relatively low operation temperatures resulting in dependence on alloy grain boundary diffusion to support selective oxidation of Cr, in light of the high scaling rates encountered in the fast pyrolysis environment due to impacts of Ca, K, S, Mg, and P species and the complex C-H-O gas mixtures formed.

This research was sponsored by the U.S. Department of Energy, Bioenergy Technologies Office. This manuscript has been authored by UT-Battelle, LLC under Contract No. DE-AC05-00OR22725 with the U.S. Department of Energy. The United States Government retains and the publisher, by accepting the article for publication, acknowledges that the United States Government retains a nonexclusive, paidup, irrevocable, worldwide license to publish or reproduce the published form of this manuscript, or allow others to do so, for United States Government purposes. The Department of Energy will provide public access to these results of federally sponsored research in accordance with the DOE Public Access Plan (http://energy.gov/downloads/doe-public-access-plan). The authors thank Jay Jun, Sebastien Dryepondt, and Bruce Pint for helpful comments on the manuscript. Adam Willoughby, Tyson Jordan, and Tracie Lowe are thanked for their assistance with the experimental work, and Donna Baltrip for formatting this paper.