Morphological characterisation of ash particles from co-combustion of sewage sludge and wheat straw with X-ray microtomography

Combustion of phosphorus-rich residual streams can produce nutrient-rich ashes and these can be used either in further processing or as materials for direct nutrient recycling. The latter requires knowledge on morphological parameters of such ash particles that may impact plant growth, nutrient availability, and soil physical properties. The present work aims to determine the porosity, pore size, and specific surface area of ash particles, and discuss these properties in light of literature concerning interaction with soil water and plant roots. Bottom ash particles from combustion of sewage sludge and wheat straw and their co-combustion were analysed with X-ray micro-tomography. Image analysis provided information on morphology, specific surface area, porosity, and pore structure on a micrometre scale resolution. Co-combusting sewage sludge with wheat straw resulted in differences in ash particles ’ porosity and pore structure compared to combustion of pure fuels. Pure wheat straw ash displayed 62 vol% porosity while there was no apparent difference between 10 wt% or 30 wt% mixtures of sewage sludge, with a porosity of 29 – 31 vol%. Open pore volume comprise the largest part of the porosity (72 – 99 vol%) enabling interaction between surrounding pore water and nutrients. Overall, the ash particles display large open volume fractions and thin particle walls which may lead to rapid weathering and extensive interaction with soil water. The particles generally contained pore openings over 200 µ m towards the surroundings, which provide opportunities for interaction with microbes and roots from a variety of plant species in addition to nutrient transport by soil water.


Introduction
Phosphorus is an important nutrient, essential for growing and producing food for the ever-increasing world's population.The European Commission has identified phosphorus and phosphate rock as critical raw materials due to its economic importance and high supply risk (European Comission 2017).Today, a large part of the nutrients delivered to forestry and agricultural sectors are lost by leaching and removal of produced biomass.Application of phosphorus fertiliser is frequently used in the agricultural sector to replace the phosphorus removed in products.However, phosphate rock, the raw material for mineral fertiliser production, is a limited resource, and the production is energy-intensive and primarily uses fossil fuel sources.A sustainable society needs to recover phosphorus from residual streams, both from agriculture and society.
The most prominent residual stream in Europe concerning phosphorus output is domestic wastewater, mainly collected at wastewater treatment plants (Schoumans et al., 2015).Using sewage sludge directly as a fertiliser raise concerns about organic contaminants, pathogens, and potentially toxic trace elements such as arsenic, mercury, and cadmium.Sewage sludge contains energy in addition to many macro-and micronutrients, and through co-combustion, nutrient-rich ash can be produced (Rulkens, 2007;Skoglund et al., 2014;Werther and Ogada, 1999).Therefore, thermal conversion of sewage sludge is one way to integrate heat and power production with sustainable and efficient recycling of macro-and micronutrients.Organic contaminants and pathogens are destroyed at combustion temperatures (Kruger and Adam, 2015).Furthermore, it is possible to achieve a significant separation between nutrients and easily volatilized heavy metals (Moller et al., 2007;Nordin et al., 2020;Van de Velden et al., 2008).Several studies show that phosphorus is primarily found in coarse bed/bottom ash fractions which makes it possible to utilise the phosphorus-bound fraction for further processing and/or recycling (Falk et al., 2020a, b;Pettersson et al., 2008;Skoglund et al., 2014).
Monocombustion of sewage sludge may produce ash containing phosphates with low plant availability, requiring additional treatment before being used as fertiliser (Nanzer et al., 2014).However, cocombustion with fuels rich in potassium can produce more plantavailable alkali-containing phosphates (Kumpiene et al., 2016;Zhao et al., 2018).Wheat straw is an agricultural residue rich in silicon and potassium, which can be problematic in monocombustion with the formation of molten ash, i.e. slag.Previous studies have shown that combustion of wheat straw results in extensive melting and slag formation, and the formed slag particles contained many porous bubbles (Parés Viader et al., 2015).Co-combustion of straw and digested sewage sludge results in improved combustion properties (Li et al., 2013;Skoglund et al., 2013;Wang et al., 2014), as well as potassium and phosphorus-rich ash with potential for nutrient recycling (Parés Viader et al., 2015;Skoglund et al., 2013).Phosphorus in the fuel may reduce slagging problems in these systems, partly due to formation of high temperature melting crystalline K-Ca/Mg-phosphates in the ash (Hannl et al., 2020;Lindström et al., 2007).Residual streams from agriculture and society can thereby contribute both to the replacement of fossil energy as well as present an opportunity for resource recovery.
The availability of elements in such ash particles as a plant nutrient is largely affected by physicochemical and biological properties in the soil, root characteristics (e.g.growth rate, specific root length, density and length of root hairs), and biochemical processes occurring at the soilroot interface (Richardson et al., 2009).The root interaction and the root system's growth to intercept new nutrient sources are crucial for phosphorus uptake due to the immobility of phosphorus, compared to other nutrients, e.g.nitrogen, which is more mobile and can reach the roots by mass flow (Barber, 1995;Lynch, 1995).Fine roots, as well as root hairs, is essential for the cycling of nutrients (Keyes et al., 2013).The root systems have a modular nature; morphological plasticity allows roots to encounter nutrient-rich patches and promote formation and elongation of lateral roots in these local nutrient enrichments, while physiological plasticity can allow for changes in the nutrient absorbing capacity of roots (Drew and Saker, 1978;Hodge, 2004;Hutchings and de Kroon, 1994;Jackson et al., 1990).The release of exudates as organic anions and phosphatases, as well as the association of roots with soil microorganisms, like mycorrhizal fungi, are also important (Richardson et al., 2009).
Research in the field of soil properties has shown that wateroccupied soil pores are important for water retention and transport in soils; the pores, together with the particle surface, present sites of physicochemical activity and largely determines the soil's gross physical properties (Childs and George, 1948;Childs and Collis-George, 1950;Falconer et al., 2012;Zaffar and Lu, 2015).The pore size distribution of the soil can give valuable information to describe and model the functioning of soils (Zaffar and Lu, 2015).The ability to transmit and store water depends on the size of the pore.Larger pores with a minimum equivalent diameter over 30 µm are generally air-filled and transmit water, while smaller pores, generally found within soil aggregates, are important for storing water (Cameron and Buchan, 2017).
While the literature on ash-soil interaction is scarce, experiences from thermally treated biomass such as biochars may contain relevant parallels.Porosity, volume and size distribution on pores in biomass chars, has been shown to have significance for, among other things, leaching and the water-retaining ability (Edeh et al., 2020;Rasa et al., 2018;Razzaghi et al., 2020;Zhang and You, 2013).Micrometre-scale pores are important for the uptake of water by plants, and biochars with internal pores in relevant scale have been shown to increase soil porosity and affect the soil water retention properties (Rasa et al., 2018).The total pore volume is important for water holding capacity and the water absorption rate (Zhang and You, 2013).The pore structure is also essential for the absorption of water.The larger pores have a greater absorption rate; they both hold water and act as a passage to small pores (Zhang and You, 2013).
Gas adsorption methods are standard for measuring porosity and pore size and relatively easy to use and have been employed for characterising the porosity of biochars and could be a candidate for ash particles.However, gas adsorption measurements are limited to analysis of small pores (nanometre scale) and lack information about external pores and the porosity and surface area in the micro-metre size range relevant to plant roots, storage of water and plant water uptake (Brewer et al., 2014;Gray et al., 2014;Rasa et al., 2018;Sohi et al., 2010).Microtomography allows for accurate imaging of encapsulated cavities that gas adsorption would not account for, and it can detect pores in a suitable size range, which is crucial when determining the possible interaction surfaces between plant roots and nutrients present in the ash.
The pore structure of several biochars have been studied with X-ray microtomography (Hyväluoma et al., 2018b), showing that a significant part of the biochar volume consisted of pores in size range important for storage of water accessible for plants.
The present work aims to determine specific surface area, porosity, and pore size for different ash particles generated from sewage sludge, wheat straw, and their co-combustion using X-ray microtomography and image analysis.This detailed knowledge of ash particles' porosity and internal microstructure is discussed in relation to nutrient recovery based on their possible interaction with plant roots and pore water.

Materials
Sintered ash particles (bottom slag) from combustion experiments with wheat straw and digested sewage sludge were produced in combustion experiments using an underfed pellet burner in a domestic boiler using 8 mm diameter pellets of four different fuels and co-pelletised fuel mixtures: 100 wt% wheat straw (W100); 100 wt% sewage sludge (S100); 90 wt% wheat straw and 10 wt% sewage sludge (W90S10); and 70 wt% wheat straw and 30 wt% sewage sludge (W70S30).The digested sewage sludge (ash content: 32.8 wt%, dry basis) had been precipitated with both iron sulfate and poly aluminium chloride and was received as dried and granulated material prior to pelletization.Wheat straw had an ash content of 4.1 wt%, dry basis.The fuel mixtures were prepared based on the content of ash forming elements to facilitate co-combustion as well as produce K-containing phosphates, as investigated by Hannl et al. (2020) for these fuels and supported by findings in previous studies (Li et al., 2013;Lindström et al., 2007;Parés Viader et al., 2015;Skoglund et al., 2013;Wang et al., 2014).The under-fed burner (EcoTec, Sweden) is equipped with a rotating ring for removal of accumulated ash, making it suitable for combustion of ash-rich fuels and fuel mixtures; a schematic description is available elsewhere (Falk et al., 2020a).The burner reached a maximum nominal power output of 20 kW and was installed in a reference boiler (Combifire, Sweden).The experiment was performed with excess air resulting in 13-14% O 2 in the flue gas to ensure complete combustion of the fuel particles could be attained.The fuel conversion temperature was monitored with a type N thermocouple in a fixed position in the fuel bed.The process temperature range near the flame front was typically 950 -1100 • C for wheat straw and its two fuel mixtures with peak temperatures around 1200 • C, whereas the sewage sludge combustion displayed a temperature range of 700 -875 • C with peak temperature at 980 • C.

Scanning electron microscopy analysis
The elemental composition of the ashes was determined by analyses with a variable-pressure scanning electron microscope (SEM; Carl Zeiss Evo), equipped with an energy-dispersive X-ray spectroscope (EDS; Oxford Instruments X-Max with an 80 mm2 detector area).The analysis was performed at low vacuum mode at 50 Pa, at an accelerating voltage of 15 kV and probe current of 750 pA.Ash particles were ground to powder before analyses of the ash forming elements, and transferred to carbon tape.Three replicates of each kind of ash were analysed and from each of these, four area analyses were performed of 1.5 mm × 1.1 mm each, given 12 areas per sample type.The C concentration was measured by area analyses of larger particles to avoid including signals from the carbon tape mounting in the analysis result.

X-ray microtomography and image analysis
The X-ray microtomography imaging was performed using a GE v| tome| × 240 device, with a tungsten target and a GE 16 ′′ flat-panel detector with 2024 × 2024 detector crystals, at the X-ray facility at the Department of Soil and Environment at the Swedish University of Agriculture Sciences (Uppsala, Sweden).The tube voltage was 60 kV, and an electron flux of 150 μA was used.The resolution was 6.4 μm.An aluminium sample holder with a 6.5 cm inner diameter was used, working as an optical filter to reduce beam-hardening artefacts.The sample was rotated 360 • , and 2000 radiographs were collected from regular angular intervals with an exposure time of 333 ms, where each radiograph was the average of four exposures.Images were reconstructed using the GE software datos|×.
The obtained reconstructed 3D stacks were analysed by the differences in X-ray attenuation.These differences are caused by differences in density and the atomic number of the elements in the scanned object.Ash appears with bright levels corresponding to high-intensity voxels, pores and background (air) appear with dark levels corresponding to low-intensity voxels (Fig. 1).The data analysis and 3D visualisation for this paper were generated using the software packages Fiji (Fiji Is Just ImageJ) (Schindelin et al., 2012;Schneider et al., 2012) and Avizo 9.3 (Fei VSG Inc., Burlingham, MA).
The image analysis was built on binary segmentation to discriminate solid material (ash) and air (pores and background) as shown in Fig. 1.The definition of pores is here holes or voids filled with air inside the particle, and background is air surrounding the particle.The samples consist of pores that are either connected to the surrounding background (open pores from here on out) or closed pores (discrete pores from here on out).The definition of pore size classification varies in the literature; there is no clear, agreed criterion.In this work, the pore size distribution was divided in size ranges of macropores (>75 µm), mesopores (30-75 µm), and micropores (5-30 µm).This is based on the equivalent diameter and follows the criteria proposed by Cameron and Buchan which is in agreement with the definition used by the Soil Science Society of America (Soil Science Glossary Terms Committee 2008; Cameron and Buchan, 2017).
Three binary models were used for the subsequent step in the image analysis.Here is a brief description; a more detailed description of the steps can be found in the supplementary material (Image analysis, S1).
The first model was used to identify the solid material, i.e. the actual ash particle volume.The second model was a total particle model and consisted of both ash and pores, with the aim of being able to subtract the background from the pore volume.Finally, the third model consist of the pore and background and was used to separate and identify the open respective discrete pores.
Specific surface area, porosity, the share of open respective discrete pores, number of discrete pores, pore size distribution, and the thickness of the particle walls was determined with the following analysis.The specific surface area was determined by generating a surface for the first model (of the ash) and divide the surface area with the volume of the ash.The volume of ash is calculated from the number of voxels identified as ash and the resolution.A polygonal surface module was used to generate a polygonal surface mesh.No smoothing or simplification was used.Specific surface area was determined to enable comparison of surface areas between samples of different sizes.Porosity was calculated as the quotient between the pore volume and the total volume of the total particle model (the second model).The share of open respective discrete pores was calculated by taking the volume of discrete pores and dividing it by the total pore volume.The pore size distribution was estimated by using the equivalent diameter and the pores' volume.A distance map algorithm (Chamfer Distance Map) was used on the first model of the ash particle to determine the thickness of the particle walls.This algorithm calculates the minimal distance from each voxel to its nearest boundary, i.e. pore or particle surface, resulting in an distance intensity for each voxel.The average thickness of the particle walls could then be calculated using this intensity, see supplementary material (Image analysis S1) for further details.
A Pearson correlation test was performed to evaluate the correlation between specific surface area, porosity, open pore volume, number and size of discrete pores, and average thickness of walls.The statistical significance of Pearson Correlation Coefficient (R) was evaluated by the P value, with P < 0.05 indicating a statistical significant correlation.

Ash particle morphology and specific surface area
Tomographic scanning enables in-depth particle analyses as shown in Fig. 2 where examples of ash particle cross-sections are shown in the middle column.The brightest spots are most likely iron-rich melts which have a large attenuation and their presence have been reported in similar samples (Shao et al., 2007;Wang et al., 2012).The images show that the ash particles from co-combustion experiments was smaller and denser, with smaller pores compared to pure wheat straw particles.These ash particles were also more heterogeneous, both in shape and morphology, and the distribution of ash forming elements (shown with differences in optical density) compared to combustion of pure wheat straw.The average relative composition of main ash-forming elements in the ashes on C and O free basis can be seen in Table S1 in Supplementary material.
During the combustion of pure sewage sludge, the experiment was interrupted due to the high ash content (32.8 wt%) causing extensive formation of molten ash which affected airflow to the fuel particles.From this combustion experiment, more or less fully combusted pellets remained; the remaining char was encapsulated by an ash layer in molten form, which solidified after combustion.The carbon content in the samples has been measured with SEM-EDS, indicating a level of 5 -26 At% of carbon in the S100 samples (Table S2, Supplementary material).The homogeneous composition of the ash layer, indicated by an even attenuation and thereby grey-scale, suggests that it has indeed been at least partially molten.This phenomenon can be seen in Fig. 2d-f and is clarified in Fig. 3.The molten ash and char matrix were separated by image analysis (Fig. 3b and d).The melt proportion was defined as the amount of voxels classified as melt divided by the total amount of voxels for the entire particle.S100 ash particles that are visualised with cross- sections in 3a-b (the same particle as Fig. 2), has a melt proportion of 64 vol% and the particle in 3c-d has a melt proportion of 25 vol%.As the main focus of the current work is slag and ash particles, and the fact that S100 samples were not completely combusted, the only results presented from the S100 ash (excluding char) is the pore size distribution.
Generated surfaces for the solid material of ash particles are visualised in Fig. 4. The specific surface area, calculated by dividing the surface area with the volume of the solid material, is presented in Table 1.The specific surface area is larger for the W100 samples (31-34 mm 2 / mm 3 ) compared to the mixtures of wheat straw and sewage sludge (17-27 mm 2 /mm 3 ).

Porosity and distribution of pores
Each sample's porosity and pore size distribution were analysed using a combination of binary models of the ash particles, as shown for W90S10 replicate 2 in Fig. 5a-d.3D view of pores in the model are visualised in Fig. 5d, which clarifies the large number of pores that are relatively evenly distributed throughout the particle.The even distribution and the presence of a large number of pores in the particles are similar for W100 and W70S30, although the porosity varies.Table 1 shows a compilation of the results from the image analysis, where the highest porosity was found for W100, with an average 62 vol%.The porosity was on average 29 vol% for W90S10 and 31 vol% for W70S30.In Fig. 5f, pores are shown together with a 3D reconstruction of the ash particle, visualizing the pore openings in contact with the surrounding volume, seen as yellow surfaces (more results in section 3.3).Video 1 (online publication only) provides an overview of the morphology, the porosity and the large pore volume in the 3D reconstructed particle, W90S10 replicate 2.
Fig. 6a-b show cross-section of W90S10 replicate 2, where the pores were separated and labelled with colours and solid material is the black part.The open pore volume and the background were coloured bright blue, which visualises the large pore volume connected the surrounding environment.73 vol% of the total pore volume was open in this sample.Fig. 6c shows a 3D view of the discrete pores of the sample and d the Fig. 2. Ash particles from the combustion of a-c) W100 pellets, d-f) S100 pellets, g-i) W90S10 pellets, j-l) W70S30 pellets.Left column shown camera RGB pictures; the middle column show X-ray tomography cross-sections of ash particles; and the right column shows 3D reconstructions of the individual particles.Grayscale is set after X-ray attenuation (optical density), where brighter parts have higher optical density.discrete pores with a transparent ash particle model.Despite its low volume share, it still consists of many pores.For all samples, between 72 and 99 vol% of the total pore volume was open (Table 1).The porosity and the share of open pore volume were positively correlated (R = 0.81, Table S3 in Supplementary material).The share of the pore volume that consists of open pores was highest for pure wheat straw ash, which also had the highest porosity.The number of discrete pores was negative correlated with the porosity (R = -0.92,Table S3 in Supplementary material).The number increases for W90S10 and W70S30 (average of 558 counts/mm 3 resp.580 counts/mm 3 ) compared to wheat straw samples (average 232 counts/mm 3 ).Fig. 7 shows the average pore size distribution divided in size ranges based on the equivalent diameter and according to the criteria described previously (Soil Science Glossary Terms Committee 2008; Cameron and Buchan, 2017): macropores (>75 µm), mesopores (30-75 µm), and micropores (5-30 µm).For S100, only the discrete pores present in the volume identified as ash (Fig. 7b) could be analysed.For the discrete pores in S100, a larger proportion of small pores (micro-and mesopores) were detected compared to the other samples.There were no significant differences in the pore size distribution of the discrete pores between W100, W90S10, and W70S30.

Particle walls and pore openings
The thickness of particle walls was calculated to gain information concerning the durability of ash particles if they are applied to the soil.The average thickness of particle walls was calculated to 42 µm for W100, 58 µm for W90S10, and 48 µm for W70S30.The average thickness of the particle walls for the respective sample replicate is presented in Table 1.A histogram over the numbers of voxels located at a certain distance from the nearest pore or surface (in voxels) are shown for W90S10, replicate 2 in supplementary material (Figure S1).
The pore openings towards the surroundings were measured manually for each sample (see example Figure S2 in supplementary material).The "openings" refere to the area of the open pore volume that are the connection between the pore and the surrounding (see Fig. 5f).They were analysed to investigate if plants root might have capabilities to penetrate the insides of the ash particle in the search for nutrients.Overall, the mixtures of wheat straw and sewage sludge were similar with just a few larger openings (app.500 µm or larger), quite many openings over approximately 200 µm, and many small openings under 100 µm.The pure wheat straw samples contained more and larger openings and had quite many larger openings (over 500 µm), the largest openings were approximately 2000 µm.Most of the largest openings were exposed where particles had been split into two or more pieces, thereby exposing new surfaces.The samples also consisted of many small openings under 100 µm.Summary comments on results of the measurements of the pore openings are compiled in Table S4 in supplementary material.

Ash particle morphology and specific surface area
To our knowledge, this is the first study of internal microstructure of ash particles including statistical measurements of specific surface area, porosity, and pore size distribution using X-ray tomography with relevance for future valorization of ashes.There was no apparent morphological difference between 10 wt% or 30 wt% mixtures of sewage sludge, but these displayed different characteristics compared to wheat straw (Fig. 2).This similarity for co-combustion cases is likely caused by similar shares of melt produced at these mixture ratios, and these findings are consistent with results from thermodynamic equilibrium modelling of the exact fuels used in this study (Hannl et al., 2020).Incomplete combustion of sewage sludge (Fig. 3) makes the morphological comparison difficult, although the ash particle analysis displayed a larger number of discrete micropores and mesopores for this material (Fig. 7).
Surface area is an important property for understanding how material behaves in soils that has not been reported in literature for ash particles, to our knowledge.For further discussion, our results are compared with literature values for specific surface area of biochars (Hyväluoma et al., 2018b).They reported specific surface area values of 87-132 mm 2 /mm 3 , produced with X-ray tomography.The biochars had a comparable porosity with the ashes in this study.The relatively low specific surface area of the ashes in this study compared to biochars, may be explained by that the largest part of the pore volume consists of large pores over 75 μm in diameter and pores connected to the surrounding, in contrast to the biochars, which mainly consists of pores with a diameter under 30 μm.Apparent differences in structure between biochar and ash can also be seen comparing Fig. 3 (b and d), where the blue part is remaining char, with the ashes in Fig. 2 (e, h and k), which also may explain a higher specific surface area for the biochars compared to the ashes in this study.
For the samples in this study, an interesting negative correlation between the specific surface area and the thickness of the particle walls was found (R = -0.80).With thicker particle walls, the particle seems to be more compact and has a smoother surface, which can give a lower specific surface area.The cause for the relationship between thickness of particle walls and surface area development during conversion does require deeper studies including chemical analysis in order to elucidate the mechanisms responsible.There were also a correlation (positive, R = 0.81) between the fraction of open pore volume and the specific surface area.A high open pore volume appears to increase the roughness of the particle surface and in some case the surface extends into what probably has been a large pore (see example for wheat straw Fig. 4a).Fig. 3. Cross-section of ash and char residue from combustion experiments with 100% sewage sludge from tomographic scanning of the particles.The images show two replicates a-b) particle consisting mostly of ash that has been melted with some parts of char residue and c-d) pellet shaped particle with poor fuel conversion.In Figures b respective d, the models have been segmented by optical density (X-ray attenuation), the blue part is the remaining carbon matrix, and the brighter parts are ash-forming elements with higher optical density.(For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) That may explain the larger specific surface area of W100 compared to W90S10 and W70S30, and it can be tentatively seen by comparing the surface reconstruction of wheat straw ash and ash from co-combustion in Fig. 4. Hyväluoma et al. (2018b) concluded that the image resolution strongly affects the specific surface area by limiting the visibility of smaller pores and pore wall roughness.That is probably the case in this study with 6.4 μm pixel size compared to that of 1.14 μm in the study by Hyväluoma et al., which might have resulted in an underestimated surface area.Higher resolution and additional methods may be needed to obtain more accurate values.These results indicate that tomographic methods are restricted by resolution concerning detection of sub-micron pores and small fluctuations in pore wall roughness but can appropriately identify encapsulated voids.Surface area as measured by gas adsorption is influenced by nanometre-scale resolution, less important for plant roots, microbes, water transport, but cannot account for volumes with no channels for gas to pass through.Combined, this means that surface area determined by gas adsorption methods are not directly  Fig. 5. a) Cross-section of W90S10 replicate 2, b) a cross-section from the ash particle model, where solid material (ash) is coloured in blue, and the black part is air (pores and background) c) cross-section from the total particle model, d) a cross-section from the subtracted model where the pores are coloured yellow, e) is a 3D view of the pores and f) is a 3D view of the pores together with a rendered model of the ash particle.(For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)comparably with the specific surface area from tomographic images (Rasa et al., 2018;Sohi et al., 2010).

Porosity and distribution of pores
The authors could not find that similar results of sample porosity had previously been presented for biomass ash particles, so again comparison is made with the more frequently investigated biochars.In the present work the porosity was determined to be 61-63 vol% for wheat straw and 24-38 vol% for the co-combustion cases.Even though the porosity of ash particles from wheat straw and the co-combustion cases are different, the values are similar to those previously presented for a variety of biochars.Different biochar samples show considerable variation in porosity, which generally varies greatly depending on raw material, with low effect of process temperature (for low-temperature pyrolysis processes) (Brewer et al., 2014;Gray et al., 2014;Hyväluoma et al., 2018a;Hyväluoma et al., 2018b;Jones et al., 2015).Brewer et al. (2014) showed that biochar porosities, calculated from differences in density, varied from 55 vol% to 86 vol%, and were dependent on biochar feedstock (wood < grass), and with a slight increase in porosity with pyrolysis temperature.Hyväluoma et al. (2018b) used a tomographic scan with a pixel size of 1.14 μm on biochars samples from scots pine bark, salix, and coffee cake, and found that the porosity were mostly affected by type of raw material and varied between 34 vol% and 68 vol% (salix < scots pine bark < coffee cake).Other studies using tomography resulted in detected porosities of 32-37 vol% for willow char (Hyväluoma et al., 2018a), and 23-29 vol% for cottonseed hull chars (Jones et al., 2015).Pycnometry was used by Gray et al. (2014) resulting in measured porosities of 83-85 vol% for douglas fir and 63-69 vol% for hazelnut shells.
The large open pore volumes (Table 1) indicate that ash particles are quite open for interaction with the surrounding environment, e.g.soil water and roots.Pores in soil with a minimum equivalent diameter over 30 µm generally transmit water while smaller pores can store water (Cameron and Buchan, 2017).The micrometre-scale porosity of biochar and the pore structure is important for water holding capacity and water absorption (Rasa et al., 2018;Zhang and You, 2013).The porosity of the ash is comparable to that of many biochars.However, unlike biochars, which often consist of pores with a diameter smaller than 30 μm (Hyväluoma et al., 2018b;Rasa et al., 2018;Zhang and You, 2013), the ash has a large part of its volume as mesopores and macropores (Fig. 7), indicating that these ashes have lower water-holding properties compared to biochars.The porosity was similar between W100, W90S10 and W70S30.In terms of all pores, there was a slightly larger proportion of mesopores for W90S10 and W70S30 (approx.0.03).But given the large proportion of macropores (approx.0.96), this is probably of little importance.However, even if the largest volume fraction consists of larger or open pores, there are still a significant number of small pores that comprise 69-79% of the total number of pores.There is also a great variety of physical properties between soils and how ash added to soil affects soil physical properties, as e.g.leaching and the water-retaining ability will vary depending on the type of soil.

Particle walls and pore openings
The relatively thin particle walls together with the large open pore volume fraction of the ash particles (72-99 vol%) pointing at that the ash particles probably will weather upon soil application.This process may also create access to nutrients in discrete pores that cannot interact with soil water or roots prior to weathering.
Fine roots have often been defined historically according to a diameter-based cutoff, usually ≤ 2 mm, but more recently with increasingly smaller diameter cutoffs (e.g.1.0 or 0.5 mm).Fine roots can be categorised using descriptions of stream-order, where the most distant, unbranched roots are first order, which has high uptake potential of water and nutrients and increased mycorrhizal colonisation compared to higher-order roots (McCormack et al., 2015).In addition to roots the microbial activity is of great importance for nutrient transport in soil.Root morphology, and root diameter, varies greatly among species and is usually stated as the mean diameter of the root segments, here are some examples.Many woody species have root diameter in first-order roots of 100-200 µm or larger (Pregitzer et al., 2002;Wells and Eissenstat, 2001;Xia et al., 2010).In herbaceous plants the root diameter in first-order roots can be less than 100 μm (Eissenstat, 1997).
Pore openings have to be larger than root tips for the root to be able to enter and penetrate further into the particle.The possibility of firstorder roots getting into the inside of the ashes may therefore vary between species.W90S10 and W70S30 have a few larger openings (app.500 µm) and quite many over 200 µm, indicating that the fine roots of many plant species could reach inner parts of an ash particle in their search for nutrients.This probability increases when roots can seek out nutrient-rich places (Drew and Saker, 1978;Hodge, 2004;Hutchings and de Kroon, 1994;Jackson et al., 1990).In addition to fine roots, root hairs are important for phosphate uptake as Keyes et al. (2013) showed that hairs and roots contribute equally to the phosphate uptake.Root hairs are thin with a diameter of approximately 6-10 μm (Leitner et al., 2010), indicating that root hairs can reach further into small pores of the ash particles investigated herein.Additionally, if root hairs can penetrate a particle, so can microbes since they are in similar size range.
The morphology of ash particles seemingly admits interaction with both soil water through pores as well as at least some direct contact with plant roots.The driving force for such interaction is the chemical composition of these ash particles where phosphorus and potassium are important elements, as reported elsewhere (Hannl et al., 2020;Kumpiene et al., 2016;Li et al., 2013;Lindström et al., 2007;Moller et al., 2007;Nordin et al., 2020;Parés Viader et al., 2015;Rulkens, 2007;Skoglund et al., 2013Skoglund et al., , 2014;;Van de Velden et al., 2008;Wang et al., 2014;Werther and Ogada, 1999;Zhao et al., 2018).Further studies are required to link ash particle morphology to the distribution of elements and compounds in ash particles.For example, determining the vicinity of phosphate compounds to surface areas in pores and particles could be valuable to understand the actual water or root interaction with phosphates in ash particles upon soil application.X-ray microtomography plays a crucial role in such studies, especially with higher resolution.This would increase our understanding of how nutrient recycling from residual streams, such as wheat straw and sewage sludge, can be a key component in a sustainable society, facilitating the integration of renewable energy with returning nutrients to the soil.

Conclusion
This novel X-ray microtomography study of ash from combustion of sewage sludge, wheat straw, and their fuel mixtures, showed that the resulting ash particles generally have high porosity with relatively thin particle walls and a large open volume fraction, indicating that the particles may weather upon soil application.Ash particles from cocombusted sewage sludge and wheat straw displayed an abundance of openings>200 µm.
The specific surface area, porosity and pore structure of ashes were different when co-combusting sewage sludge with wheat straw compared to combustion of pure fuels.The specific surface area was lower for the co-combustion cases compared to pure wheat straw.The porosity was on average 62 vol% for pure wheat straw ash and 29 vol% and 31 vol% for the ash with a mixture of 10 respective 30 wt % of sewage sludge with wheat straw.The open pore volume comprised a large part of the total pore volume (72-99 vol%).
The high open pore volume suggests that soil water may readily permeate the phosphorus rich ash particles from co-combustion of sewage sludge and wheat straw and thereby contribute to nutrient transport from ash to soil, microbes, or plants.The large openings may also facilitate direct interaction between the inner surfaces of ash particles and the fine roots of many plant species.This in combination with soil water access to pores suggest that ash particles from co-combustion of the residual streams sewage sludge and wheat straw may present an interesting option for nutrient recovery by direct field application.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Fig. 1 .
Fig. 1.The figure shows cross-section of W90S10 replicate 2, with descriptions and arrow pointing on features discussed in the text.

Fig. 6 .Fig. 7 .
Fig. 6. a) cross-section of W90S10 replicate 2, where the ash particle is black and the pores and background are coloured.The coloured part was separated into background and open pores (bright blue) and discrete pores (yellow).b) 3D view of the discrete pores and c) 3D view of discrete pores together with a transparent model of the ash particle.(For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Table 1
Compilation of results from image analysis consisting of specific surface area, porosity, open pore volume, number and size of discrete pores, and average thickness of walls.