3.3.1. Amorphous Structures
As shown in
Figure 4, two more obvious structures are present in the mixture of the HTC chars, one white and round, and another one that looks like a grass epidermis, showing stomata and phytoliths. The elemental analysis proved that the round structures are siliceous and about 10–20 µm in size. These siliceous structures were found in many of the chars and appear to be loosely attached to the surface.
In order to identify the provenience of these siliceous structures, EDX measurements were performed on them and on the background material, and the results were compared with those obtained on siliceous structures in wheat and barley. The stems did not show the siliceous structures, but the leaves did, showing that these structures are not caused by torrefaction. Elemental composition in wheat and barley is just C, O, and Si. However, these non-torrefied structures contain more oxygen, which would be expected as combustion reduces oxygen content. The other elements on the char samples may be material deposited on the surface during torrefaction (
Figure 5,
Figures S1 and S2).
Figure 5 shows the results obtained from the TOR 600 sample, while
Figures S1 and S2 present the SEM-EDX results obtained on the oven-dried wheat straw and barley leaves, respectively.
As shown in
Figure S1, the wheat sample presents siliceous structures and stomata. Barley (see
Figure S2) has similar structures, although the elemental composition is more varied, containing K, Cl, and Na. Stomata are also visible.
The size of the siliceous structures was measured for torrefied samples as well and compared with the size of the siliceous structures in PL, wheat, and barley. The average size decreases as the temperature increases, i.e., for TOR 400 it is 20.4 µm, while for TOR 500 the value is 15 µm. The size of the siliceous structures in PL is, on average, 15.6 µm; for wheat it is 17.6 µm, while for barley it is 13.6 µm, respectively. The size comparison and appearance suggest that the siliceous structures are probably wheat. However, the sample is too small to be certain—it is possible that both are present.
What is more, SEM images of PL (
Figure 6) reveal the presence of other siliceous structures that do not look like those found in either oven-dried wheat straw or barley leaves.
For PL, the structures seem to be in hollows. Also visible are what appear as wiggly lines between cells. These are probably phytoliths and can be seen in greater detail on the barley leaf (
Figure S3). They are about 124 µm in length and are composed mainly of silicon. Phytoliths are used in archaeology to identify species as they are resistant to decay, so they could be used to identify the plant material a char is derived from [
36]. Phytoliths are primarily silica and can be a problem if feedstock is used as a fuel [
37].
These phytolith structures remained and were visible in ash samples—not surprising as they are separated from plant material by ashing [
36].
There were other amorphous structures found in torrefied material that looked like feathers (see
Figure S4) and had a similar chemical composition. To test their provenience, feathers and torrefied feathers were analysed.
As can be seen, the actual feather structure is a main rib with side branches emerging, similar to the structure shown in TF135. These branches appear to be strap-like and narrow as they proceed further from the main rib. Torrefaction causes feathers to lose most of their fine structure. The size is also similar. Elemental analysis also suggests that this is a feather. The C and O percentages are similar for the actual feathers and the structures in TF135. Feathers are made of the protein keratin, which has high levels of S, as seen in the elemental analysis. The TF135 feather has a low S value, but this could be a result of natural variation. Another problem is the difference between the feather heated to 450 °C and one heated to 105 °C (oven-dried). Heating seems to cause a loss of Na and Ca; however, the difference may be because the feather was not of the same type; it was taken from a local bird rather than from the PL. Therefore, the long fibres in the HTC chars could have been keratin protein fibres from feathers, which have “an intricate network of connective and hollow fibrous structure [
38]”. However, it was found that HTC would hydrolyse keratin from chicken feathers into small protein [
39].
As for the hydrothermal carbonisation samples, some more and different amorphous structures were shown (see
Figure 7). Many fine white structures, 700 nm–2 µm long, are visible. These could be cellulose fibrils, as the literature suggests that HTC treatment breaks down hemicellulose, which causes individual cellulose fibres to separate. Debris from the cell wall is then deposited on the surface of the hydrochars [
40]. However, microfibrils are 28 nm in diameter, so these fibres are too large—they may be aggregations of microfibrils.
Structures like roots were also observed. The diameter of these structures is about 20 µm for the larger ones, going down to 2 µm for finer branches. These are about the correct size of and similar structure to roots.
3.3.2. Crystalline Structures
Many structures appear crystalline with very regular shapes and distinct angles. These structures, which are found on the surface, seem primarily inorganic and may, therefore, be very important when PL is used as a soil amendment.
The crystalline structures found on the TOR 350, TOR 450, and TOR 550 samples, respectively, along with those found on the TF 350 and TF 275 samples, were studied in detail. Results are as follows.
Figure 8 presents the SEM images of the crystalline structures observed on the complex 3D network of the TOR 350 char sample along with their composition. These crystals are about 2 µm on one face and appear cuboid. The elemental analysis shows high amounts of K and Cl, so these crystals could be potassium chloride.
In order to support this statement, pure KCl crystals were characterised by SEM/EDX (
Figure S5). The K:Cl ratio is 1.1 for the pure KCl crystal, very close (±1%) to the published value [
41], while for the TOR 350 char sample, this ratio is 1.4. Therefore, even all the Cl is in the form of KCl, that still leaves some K and, with high amounts of carbon and oxygen present, also suggests a possible presence of potassium carbonate, K
2CO
3, which is deliquescent. KCl crystals are also hydroscopic.
Crystalline structures were found in other samples as well, in some cases in very large quantities, as for the TOR 450 char sample. There are two types of crystalline structures observed, some bright white ones and some grey ones, with very similar composition, but with far lower Cl and higher Ca percentages, respectively, for the grey ones. The K:Cl ratio for the bright white ones is 1.5, which supports the assumption that they are most probably KCl crystals. These crystals are about 2 µm on one face and appear cuboid (see
Figure 9).
As for the TOR 550 sample, the crystalline structure measured has a high percentage of P, K, and Mg, confirmed as pyrocoproite (K
2MgO
7P
2) [
42] via XRD. Also, although it has a very high concentration of K but only a little Cl, it means that K is present in forms other than KCl [
3].
Wide-angle XRD measurements were performed in order to identify the nature of all crystals present on the torrefied samples.
Figure 10 shows the diffractograms for the three torrefaction samples, namely TOR 350, TOR 450, and TOR 550, respectively.
The samples are broadly similar, with several peaks at the same positions for all chars. These peaks are considered more reliable and will be analysed below.
Low-angle XRD was measured as well to assess the broad peak at 2θ between 17° and 27°. The broad peak showed a complete amorphous structure. There were no diffraction peaks observed except for a broad band centred at 2θ decreasing from 22.5° for TOR 350 to 20.5° for TOR 550, with the peak maxima at 2θ = 21° for TOR 450, which is a well-known feature for amorphous silica material. The results were in good agreement with those on the JCPDS file for SiO2.
Amorphous carbon also shows a peak in this area [
13,
43].
Table 3 presents the peaks’ identification for the three torrefaction samples. The conclusion from the XRD measurements is that sylvite is present in all samples; calcite and quartz are also present. There are some diffraction peaks in TOR 550, which are difficult to identify. The complex mixture of minerals that might be present causes problems with identification. In rice husk char, for example, as well as sylvite and calcite, a number of complex potassium minerals were present: archerite (KH
2PO
4), chlorocalcite (KCaCl
3), kalicinite (KHCO
3), pyrocoproite (K
2MgO
7P
2), struvite (KMgPO
4·6H
2O) ([
13]). All of these are possibly present given the elemental composition found by EDX but may be present in only small quantities and be undetectable.
Another XRD study of chicken manure identified calcite, hydroxyapatite, struvite, dolomite, quartz, and magnesium phosphate [
44]. This study found calcite and quartz—again, the others may be present in low quantities in the studied PL, but the fact that sylvite is not present is surprising. This implies that the sylvite possibly originates from the plant material, which is not present in manure. This is supported by another study that found sylvite in a 300 °C biochar from straw (
Brassica campestris). In this case, the higher-temperature biochars had no sylvite but did show calcite and dolomite [
45].
The size of the sylvite crystallites was determined by using the Scherrer equation. For 2θ around 28°, the size of the sylvite crystal was 28.5 at 350 °C, 21.4 at 450 °C, and 21.4 at 550 °C, respectively. As for 2θ around 40, the size of the sylvite crystal was 24.6 at 350 °C, 22.1 at 450 °C, and 22.1 at 550 °C, respectively. There is no obvious trend for crystal size with temperature, with all values being similar for a given peak.
One problem with XRD is that it is best suited to the analysis of homogenous material rather than these heterogeneous chars. It has a detection limit of 2% for components in mixtures, so substances present in low concentrations may not be detected [
46].
A variety of crystalline structures were observed as well on the surface of the biochar samples prepared in the tube furnace.
Figure 11 presents the SEM micrograph along with EDX composition for the TF 275 sample, while
Figure 12 shows SEM/EDX results from the TF 350 biochar sample.
The large crystal observed on the TF 275 sample is probably calcium carbonate (calcite), whilst the flat crystal is possibly some type of phosphate. The round crystal could be potassium carbonate (K2CO3) or possibly pyrocoproite (K2MgP2O7), which was found to be present in the XRD results for the TOR samples.
The crystalline structure shown on the TF 350 sample is most probably aluminosilicate.
To the best of our knowledge, this is the first time that the crystalline structures on the surface of the poultry litter biochar were identified, and their composition determined.
Some of the feedstock nutrients, namely P, K, Ca, and Mg, were concentrated on the surface, and their content increases with increasing the pyrolysis temperature. It is important to know the type and composition of the crystalline structures on the surfaces of the biochars as these crystals are the first to be released into the soil, to cycle nutrients back into agricultural fields. As a measure of the direct nutrient value of biochars, it is not the total content but, rather, the availability of the nutrient that is an important consideration.
Hydrochar samples showed no crystals on the surface as they, presumably, dissolved in the process water.
3.3.3. Macroporosity and Elemental Composition (Surface, Subsurface, and Bulk) of Biochar Samples
The SEM/EDX technique was also employed to visualise the porous structure and to determine the pore size and the elemental composition of the biochar samples, and to assess their changes with the biochar production conditions. As seen in
Figure 13 and
Figures S6–S10, distinct morphologies of increasing porous structure are present.
The represented microscopic honeycomb-like structures, typical of fibrous plant materials, are present in the PL biochar from the wood shavings used as bedding material. These microstructures evolve in shape and complexity as the torrefaction temperature increases. The TOR 600 sample, for example, shows a very regular pore structure, but still has a wide range of pore sizes, from 4.2 µm to 12.3 µm (
Figure S10A).
Along with the external macroporosity between the biochar’s particles and the residual macroporosity based on the plant cellular structure shown in the SEM images, a pyrogenic nanoporosity develops within the solid biochar volume and increases with production temperature but constitutes only a small portion of total porosity, even in higher-production-temperature biochars [
47]. What is more, the pyrogenic nanopores are formed because of chemical changes at higher pyrolysis temperatures, higher than 600 °C, which was the maximum torrefaction temperature used to prepare the samples in this study.
Due to this, our study focused mainly on the macroporosity of the biochar samples, which is a key parameter influencing their water uptake.
Using the SEM software, the pore size was directly measured.
Figure 14 presents the variation in the mean pore size with temperature.
Mean pore size decreases with torrefaction temperature as expected but TOR 500 seems anomalous. In fact, there are many problems with this method of measuring pore size. It is often difficult to decide where the maximum diameter is, as this involves subjective judgement. Also, some pores are seen in oblique view, so that the measurement is larger. Most studies that find a decrease in pore size with temperature are considering a single feedstock [
47]. One of the main features of poultry litter is its heterogeneous nature. Pore size is linked to the cell size of the feedstock—here, there are many different plant species present, including grasses and wood shavings as well as inorganic material. The SEM images and EDX composition of such an inorganic structure from TOR 350 are presented in
Figure 15. Its elemental analysis shows very little carbon, suggesting its inorganic nature. It looks mineral and contains both calcium and potassium at high levels, with magnesium and sodium also present.
The heterogeneity of the PL means that not all SEM measurements are from the same material; thus, the observed variation in morphology. However, there are some advantages of the SEM direct measurement of the pore size method. The pores are seen, so the wide variation in size, in even a single site, can be seen and measured.
In a heterogeneous feedstock like PL, it potentially allows the contributions of different materials to pore size to be identified. It is also possible to see how the pores are formed from cell walls. Perhaps a more representative measurement of pore size would use only one type of pore, such as what will be called a “true honeycomb”, as seen for TOR 600 (
Figure S10B). The outer layer of plant material has been lost, exposing the cellular structure. The SEM pore size distribution measurement results are presented in
Figure 16.
TOR 350 has the widest range of pore sizes, and the most frequent size is between 10 and 19.9 µm. TOR 600 has a small range, and the most frequent size is 0–9.9 µm. TOR 500 seems anomalous with a peak frequency at 20–29 µm.
As for the TF biochar samples, SEM measurements showed that the mean pore size is increasing with temperature (see
Figure 17), which was an expected result as well. Since the poultry litter feedstock was a mixture of wood, barley, wheat straws, feathers, chicken excreta, and spilled feed, etc., the microscopy images of TF samples show, as for the TOR samples, the presence of different types of biochar particles of varying size and morphology, which led to external pores, i.e., those pores between the biochar particles, of different sizes. The residual macroporosity is formed by the evolution of volatiles from the solid during the thermal degradation of the poultry litter. Knowing that the corresponding temperatures for maximum decomposition of hemicellulose and cellulose are 300 °C and 355 °C, respectively, while lignin decomposition begins at about 280 °C with a maximum rate occurring between 350 and 450 °C and the completion of the reaction at 450 and 500 °C, the increase in the pore size with temperature for the TF sample is most probably due to the volatilisation of hemicellulose and cellulose within the temperature range used in the tube furnace.
As the temperature of the conversion of poultry litter to biochar increases to produce the TOR samples, the pore size decreases since almost all the volatile compounds were lost at temperatures of around 350 °C.
There is a slight difference between the mean pore size for TF 350 and TOR 350, 25 µm compared to 14 µm. It might be due to the different amount of poultry litter used to obtain the two samples: two grams were used for the TF samples’ production, while for the TOR samples, 1 kg of poultry litter was used, with the heat transfer for carbonisation being faster for the surface samples (TF) than that for the bulk samples (TOR).
Figures S11–S14 show the SEM images and pore size measurements of the TF samples. There is still a wide range of pore sizes in each temperature as in TOR samples. For TF 135, the residual macropores are not very wide; some of them are not completely open, so the surface area and volume, respectively, of the pores is low (
Figure S11). As the conversion temperature is increased at 200 °C, pores with different sizes and appearance (far more open), similar to those observed for the TOR samples, are more likely to be detected along with pores such as those on the TF 135 sample (
Figure S12).
At higher temperatures, i.e., 275 and 350 °C, respectively, pores are now similar to those found in the TOR samples.
Figure 18 shows the frequency distribution of pore size with temperature for the tube furnace biochar samples. There is a smaller range of pore sizes at lower temperatures, with most pores in the smaller size range for TF 135 and TF 200.
SEM images were taken for the HTC samples as well.
Figures S15–S18 show the morphology and pore size measurements of the carbonised samples. As seen in
Figure S15, for the HTC 80 sample, despite quite a low temperature, there is a large amount of honeycomb morphology with many pores in the process of formation.
True honeycomb morphology, but with no cavity in the centre, was observed for the HTC 95 sample (see
Figure S16).
For the HTC 120 sample (
Figure S17), the honeycomb has a different structure to previous samples; these are arguably not true pores, or they are pores in the process of forming. The pores may be formed from a different plant material. It is also possible that the differences in pore morphology in the HTC samples are due to the different decomposition of polymers in HTC, where hydrolysis is the most important mechanism [
33]. The difference could also be due to pressure, although this is a low-pressure sample (2 bar).
If they are pores in the process of formation, a higher carbonisation temperature should be beneficial. Therefore, an extra HTC biochar sample was prepared at 221 °C, HTC 221.
Figure S18 shows the SEM micrograph and pore size measurements for this sample. The microscopic pore morphology seems to be improved but there still are closed cavities. Similar observations were made about PL hydrochars obtained at temperatures of 150 and 175 °C, respectively; SEM micrographs showed incomplete decomposition and a corrugated surface with holes [
48].
As expected, the mean pore size of the HTC samples (slightly) increases with temperature, from 10 µm at 80 °C to 15µm at 221 °C.
Figure 19 presents the frequency distribution for the HTC samples’ pore size against temperature showing a lower range than for the torrefied material with very few pores 20 µm or above. However, HTC 221 has the highest number of large pores. HTC 80 and HTC 95 are quite similar, but HTC 120 appears different. This could simply be sample variation; however, HTC 120 was prepared at a lower pressure of about 2 bar, as opposed to up to 20 bar for other points. This might suggest that pressure is an important factor in pore size.
The elemental composition of the poultry litter biochars is another key parameter in their application as soil amendments. As mentioned above, PL biochars are heterogeneous materials, with complex structure and morphology. Therefore, a heterogeneous composition is expected as well. First, SEM/EDX measurements were used to assess the elemental composition at the surface and subsurface level (about 1 micron depth) of the different biochar’s samples prepared in this study.
Figure S12 shows the SEM images and EDX results of the TF 200 sample, as an example of the heterogeneous nature of the PL biochar’s samples.
The elemental composition of the torrefied samples was measured twice by EDX in a six-month time interval. Low-temperature chars showed little difference; however, at higher temperatures, there are big differences. In TOR 600, carbon has dropped from 70% to 47%, whilst oxygen has gone up from 22% to 28%. The most likely explanation is that the stored chars have taken up water, so oxygen increases. This alters all the percentages, but is most obvious for the largest element, carbon.
This explanation is supported by the hydrophobicity experiments, which showed that the low-temperature biochars take up little water compared to the high-temperature ones.
The mean EDX composition of the TF samples is given in
Table 4.
As for the HTC biochar samples, the EDX measurements, shown in
Table 5, reveal that carbon decreases slightly and oxygen increases slightly above 80 °C, but there is little change after this temperature. All other elements are present at lower levels and seem to increase slightly with temperature from 80 °C but after this stay relatively constant. This is because the chars lose water and, hence, weight, so the solid material is present at a lower level; thus, elements appear to increase. All minor elements are below 1%, apart from Ca, and so cannot be considered reliable. The tolerance of the system according to Oxford instruments is 0.1% but below 1% the peaks should be checked manually, which was not done. The higher levels of Ca may mean that Ca present on the surface is not water soluble. K shows an increase, but it is not significant and is very small compared to torrefied chars. This is presumably because the K is present as a soluble salt and dissolves into the surrounding HTC liquor.
The bulk composition of the biochar samples prepared in this study was measured by atomic absorption spectroscopy (AAS) and inductively coupled plasma mass spectrometry (ICP-MS), with the purpose to assess the presence of potentially toxic elements (heavy metals) such as Cu, Pb, and As. Both methods were considered. However, as the concentration of the heavy metals in the biochar samples determined by AAS was very low, ICP-MS was then used, as its sensitivity is better. The results of the ICP-MS measurements are shown are shown in
Table 6.
This study also showed a concentration effect in the char, with double the values in the char compared to the PL [
49].
Concentration levels in the low-temperature chars are not much different to those in PL. However, at higher temperatures, levels of magnesium, manganese, and aluminium increase. This is presumably a simple concentration effect. Standards were taken through the whole method to give an indication of both method and machine error. Errors are due to dilutions, variable loss of water during reflux, and metals taken up from anti-bumping granules, nitric acid, and glassware. This gave a total error of about 1% for Cu, 7% for Zn and Ni, 10% for Cd, and 35% for Fe. Levels of most metals are low and are very close to detection limits, so values may not be accurate. As expected, there is good agreement between the 350 °C results for the TF and those for TOR. The highest values were for Ca, Mg, Na, and K, but these are all very susceptible to environmental contamination, and the blank is high for all four. However, K shows a very different trend. The other three elements increase and then level off; presumably this is an artefact caused by the fact that the char weighs less, so the percentage for a given weight is higher. K drops sharply, which may be important and, therefore, will be checked using a flame photometer; Zn, Al, Mn, and Cu, though present at lower concentrations, show the same trend as K.
As KCl crystals have been found on the surface of the char, it is important to know K levels; therefore, the flame photometer was used on both water-extracted and total-digestion samples.
The results are presented in
Figure 20. As can be seen, over 60% of the total K in the TOR samples is water extractable (WE). This is important in terms of plant nutrients as water-extractable K is available to plants, but it may also be leached before it can be used. Soluble K may also affect germination as it is mainly on the biochar surface. This was tested using a KCl solution in the germination test. Other studies have found lower levels of water-extractable K of about 46% at 300 °C going up to 49% at 600 °C [
10]. A similar trend was observed, with WE % staying fairly constant until 550 °C when it increases, whilst in this study the increase was observed above 550 °C. TF and TOR biochars show an increase in both water-extractable and total K with increased production temperature. The TF 350 value is slightly higher than the TOR 350 value.
The HTC material has little K, presumably because soluble K has been extracted into the HTC liquor. Flame photometry measurements showed a maximum of 1% K (TD) for the HTC 120, while for the PL, the value was 2.3% or 22.9 mg g
−1. This value is lower than the published data of 41.8 mg g
−1 for raw PL [
10]. Another study measured between 5.65% and 7.59% total K in PL; again, the value measured in this study is lower [
50], as the source of the PL is different.