Hydrothermal Carbonization as a Strategy for Sewage Sludge Management: Influence of Process Withdrawal Point on Hydrochar Properties

Conventional activated sludge systems, still widely used to treat wastewater, produce large amounts of solid waste that is commonly landfilled or incinerated. This study addresses the potential use of Hydrothermal Carbonization (HTC) to valorize sewage sludge residues examining the properties of hydrochars depending on HTC process conditions and sewage sludge withdrawal point. With increasing HTC severity (process residence time and temperature), solid yield, total Chemical Oxygen Demand (COD) and solid pH decrease while ash content increases. Hydrochars produced from primary (thickened) and secondary (digested and dewatered) sludge show peculiar distinct properties. Hydrochars produced from thickened sludge show good fuel properties in terms of Higher Heating Value (HHV) and reduced ash content. However, relatively high volatile matter and O:C and H:C ratios result in thermal reactivity significantly higher than typical coals. Both series of carbonized secondary sludges show neutral pH, low COD, enhanced phosphorous content and low heavy metals concentration: as a whole, they show properties compatible with their use as soil amendments.


Introduction
In 1991, the European Union (EU) Directive 91/271/EEC set new benchmarks for the collection, treatment and monitoring of wastewater in urban areas [1]. In 2000, the EU produced over 10 million dry tons of sewage sludge (latest available official EU data [2]). Since then, production has steadily increased, increasing burdens on municipal wastewater treatment systems [3]. In 2017, the global sewage sludge production reached approximately 45 million of dry tons per year [4]. In Italy alone, over 1100 kilotons of dry sludge matter were produced in 2010, with almost 40% going to landfills and less than 30% used in agricultural applications [2]. In 2016, the European Commission reported that the 1991 directive was successful in terms of current improvements seen in EU water quality despite an increasing population growth. However, the European Commission also underlined the need to develop "innovative solutions to increase resource efficiency, such as solutions for energy recovery, nutrient recovery, and processing to marketable products and water re-use" [5]. Beyond the EU, global As such, direct land application of secondary sludge may not be the optimal nutrient recovery pathway and may well represent a waste of a renewable energy source.
Hydrothermal Carbonization (HTC) is a process to concentrate the carbon in a given biomass, occurring in water at elevated temperatures in the range of 160-280 • C [14], above saturated vapor pressure, where water's dielectric constant decreases so drastically that it catalyzes the carbonization of biomass while acting as an organic solvent [15,16]. HTC proceeds via a series of mechanisms, including hydrolysis, dehydration, decarboxylation, decarbonylation, and demethanation. HTC is performed in 80-95 vol% water [17][18][19][20], making it an ideal processing pathway for wet biomasses [21,22] such as sewage sludge. As well recognized in the literature, HTC not only leads to reduced volume and energy densification of the solid residue, considerably improving its dewaterability [23,24], but also significantly increases its solid fuel properties while stabilizing and disinfecting the sludge [25,26]. Interestingly, HTC of sludge enhances NO + NH 3 reactions during solid combustion, significantly reducing NOx emissions across combustion modes [27].
While multiple studies focused on nutrients recovery such as phosphorous [28][29][30] or probed the impact of processing conditions on the solid and energy yields of hydrochars produced from the HTC of sludge from one withdrawal point in the wastewater treatment (WWT) process [23,[31][32][33][34][35][36][37][38][39], the impact of upstream WWT processes on hydrochars is often not part of the experimental design. Thus, a primary goal of the present work was to determine the most viable point in the WWT process to employ HTC for sludge treatment from a solid fuel production and potential nutrient recovery standpoint. There are two sets of variables of interest in the present work: (1) feedstock withdrawal point from the WWT process; (2) hydrothermal reaction conditions on the composition and oxidation properties of resulting hydrochars. While some research has been done to quantify the environmental and economic benefits of using HTC to treat sewage sludge, understanding the impact of both feedstock and processing conditions on resulting hydrochars is critical to enable better optimization of this waste-to-energy conversion pathway [40].

Materials and Methods
Three kinds of samples were collected at different process points from the municipal Wastewater Treatment Plant (WWTP) of Trento, Italy, which uses a Conventional Activated Sludge (CAS) and Membrane BioReactor (MBR) system (more information about CAS/MBR systems in recent prior works in the field: [41,42]). The primary sludge extracted from the primary settler is sent to a static thickener and mixed with secondary sludge coming from the MBR system (Pipe 1 of Figure 1). The thickened sludge exiting the static thickener (Pipe 3) is the first sample taken for the present work. The secondary sludge coming from the CAS system is sent to a dynamic thickener (Pipe 2). After thickening, both streams are sent to the anaerobic digester. Anaerobic digestion lasts 21 days at 33 • C, after which the digested sludge (Pipe 4) is sent to a dewatering system. The effluent from Pipe 4 is the second sample taken for this work. To the digested sludge from Pipe 4 polyelectrolyte is added; this stream is sent to the dewatering system (centrifuge) where it reaches about 25 wt% solids content to enable solid handling. This dewatered sludge after the centrifuge (Pipe 5) is the third sample used in this work. Each of the three samples taken was characterized and stored in a fridge at 4 • C until use.
Energies 2020, 13, x FOR PEER REVIEW 3 of 22 Three kinds of samples were collected at different process points from the municipal Wastewater Treatment Plant (WWTP) of Trento, Italy, which uses a Conventional Activated Sludge (CAS) and Membrane BioReactor (MBR) system (more information about CAS/MBR systems in recent prior works in the field: [41,42]). The primary sludge extracted from the primary settler is sent to a static thickener and mixed with secondary sludge coming from the MBR system (Pipe 1 of Figure 1). The thickened sludge exiting the static thickener (Pipe 3) is the first sample taken for the present work. The secondary sludge coming from the CAS system is sent to a dynamic thickener (Pipe 2). After thickening, both streams are sent to the anaerobic digester. Anaerobic digestion lasts 21 days at 33 °C, after which the digested sludge (Pipe 4) is sent to a dewatering system. The effluent from Pipe 4 is the second sample taken for this work. To the digested sludge from Pipe 4 polyelectrolyte is added; this stream is sent to the dewatering system (centrifuge) where it reaches about 25 wt% solids content to enable solid handling. This dewatered sludge after the centrifuge (Pipe 5) is the third sample used in this work. Each of the three samples taken was characterized and stored in a fridge at 4 °C until use.

Feedstock Characterization
The three feedstocks used were characterized in accordance with accepted environmental practices. All measurements were conducted in triplicate and the average and standard deviation of each data is reported. The total solids content of each sludge was measured by drying in an oven at 105 °C for at least 8 h until constant weight was reached.
The pH of the raw and carbonized samples was measured using a Profi-Line pH 3310 (WTW, Milan, Italy) portable pH-meter by placing 1 g of solids in 20 g of deionized water, shaking for at least 90 min, allowing the mixture to settle for 15 min, and then reading the pH. The total Chemical Oxygen Demand (COD) was measured using a closed reflux titration method [43] using potassium dichromate digestion solution, sulfuric acid reagent, ferroin indicator solution, and a standard ferrous ammonium sulphate titrant according to standard procedures. The same method was used for soluble COD following filtration of the sample through a 0.45 μm filter [43]. Organic nitrogen in the trinegative state was measured via the semi-micro Kjeldahl method. Ammonium nitrogen was measured [43] by first buffering the sample at pH 9.5 with a borate buffer (to decrease hydrolysis), then distilling in a solution of boric acid, and determining the concentration via acid titration with H2SO4. To measure the total phosphorous, samples were first digested in H2SO4, forming molybdophosphoric acid, which was then reduced by stannous chloride to molybdenum blue. The concentration was measured photometrically at 690 nm and compared against a calibration curve [43].
Ultimate analysis to determine elemental composition of C, H, N, S, and O (by difference) was conducted on a LECO 628 analyzer (LECO, Moenchengladbach, Germany) equipped with Sulphur

Feedstock Characterization
The three feedstocks used were characterized in accordance with accepted environmental practices. All measurements were conducted in triplicate and the average and standard deviation of each data is reported. The total solids content of each sludge was measured by drying in an oven at 105 • C for at least 8 h until constant weight was reached.
The pH of the raw and carbonized samples was measured using a Profi-Line pH 3310 (WTW, Milan, Italy) portable pH-meter by placing 1 g of solids in 20 g of deionized water, shaking for at least 90 min, allowing the mixture to settle for 15 min, and then reading the pH. The total Chemical Oxygen Demand (COD) was measured using a closed reflux titration method [43] using potassium dichromate digestion solution, sulfuric acid reagent, ferroin indicator solution, and a standard ferrous ammonium sulphate titrant according to standard procedures. The same method was used for soluble COD following filtration of the sample through a 0.45 µm filter [43]. Organic nitrogen in the trinegative state was measured via the semi-micro Kjeldahl method. Ammonium nitrogen was measured [43] by first buffering the sample at pH 9.5 with a borate buffer (to decrease hydrolysis), then distilling in a solution of boric acid, and determining the concentration via acid titration with H 2 SO 4 . To measure the total phosphorous, samples were first digested in H 2 SO 4 , forming molybdophosphoric acid, which was then reduced by stannous chloride to molybdenum blue. The concentration was measured photometrically at 690 nm and compared against a calibration curve [43].
Ultimate analysis to determine elemental composition of C, H, N, S, and O (by difference) was conducted on a LECO 628 analyzer (LECO, Moenchengladbach, Germany) equipped with Sulphur module for CHN (ASTM D-5373 standard method) and S (ASTM D-1552 standard method) determination. Proximate analyses were done on a LECO Thermogravimetric Analyzer TGA 701 (LECO Corporation, St. Joseph, MI, USA). Samples were heated at 20 • C min −1 to 105 • C in air and held until constant weight (< ±0.05%) to provide a dry baseline. They were subsequently heated at 16 • C min −1 from 105 • C to 900 • C in nitrogen with a hold time of 7 min, where the mass loss was attributed to Volatile Matter (VM). Finally, samples were held at 800 • C in air to oxidize the Fixed Carbon (FC) until the mass change stay within ±0.5% by weight. Mass remaining after this was considered to be ash (inorganic matter) content. The Higher Heating Value, HHV, was measured using an IKA 200C isoperibolic calorimeter (IKA-Werke GmbH, Staufen, Germany) according to the CEN/TS 14918 standard.
Inductively coupled plasma-optical emission spectroscopy (Arcos Ametek, Spectro, Germany) was used to determine the inorganic concentration of the sludge samples. Briefly, samples were oven-dried at 105 • C until constant weight and then acid-digested in concentrated nitric acid (650 mL −1 ; Carlo Erba, Milano, Italy) using a single reaction chamber microwave digestion system (UltraWAVE, Milestone Inc., Sheldon CT USA) and Teflon-lined vials to prevent interference. Elements were quantified using certified multi-element standards (CPI International).

Hydrothermal Carbonization and Product Analysis
Hydrothermal carbonization typically occurs between 180 and 250 • C under autogenous pressure (up to 50 bar) but below the critical point [18]. The present work utilized a 50 mL stainless steel batch HTC reactor rated to withstand 300 • C and 140 bar, with temperature and pressure monitoring and temperature control, as previously described [44,45]. The thickened and digested sludges were used as-received. The reactor was loaded with 35.0 mL ± 0.1 mL biomass, which maintained a biomass (dry biomass)-to-water ratio of about 0.03:1. The dewatered sludge had a solid content of 25 wt% and needed to be diluted to ensure that the biomass was fully submerged. The reactor was filled with 20.00 g ± 0.01 g of dewatered sludge and 15.00 g ± 0.01 g of deionized water to cover the sludge, resulting in a biomass-to-water ratio of 0.17:1. Prior to each run, the reactor was sealed and purged with nitrogen gas; then it was heated up to the desired reaction temperature (190 • C, 220 • C and 250 • C) and held at the set point for the desired reaction time (30 min and 60 min). At least three experimental runs for each of the temperature/time combinations were performed for each of the three sludge samples.
After the reaction time, the reactor was cooled by placing a cold (−25 • C) stainless steel disk under its bottom and by blowing compressed air into its outer walls. The reactor was cooled to ambient temperature in less than 15 min, at which point the produced gas was measured by flowing it into a graduate cylinder filled with water [45]. As reported in the literature, where the CO 2 content is always greater than 90 vol.%, the produced gas was assumed to be comprised entirely of CO 2 [45,46]. The gas yield was estimated according to the ideal gas law under the assumption of standard temperature and pressure as: The liquid and solid HTC products were filtered through a pre-dried and weighed piece of cellulose filter paper. The filter paper was then put in the oven overnight at 105 • C and weighed to calculate the solids produced. The solid yield of the hydrochar, Y hydrochar , was calculated as: The liquid yield was computed as the complement to 1 of the gas and solid yields.
The solid hydrochar was characterized according to the same methods described in Section 2.1. for Higher Heating Value (HHV), proximate, and ultimate analyses. The hydrochar's relative solid reactivity was measured using a Mettler-Toledo Thermogravimetric Analyzer-Differential Scanning Calorimeter (TGA-DSC-1, Mettler-Toledo LLC, Columbus, Ohio, USA) in an oxidative atmosphere. The TGA-DSC was calibrated with NIST-traceable gold, indium, and aluminum and the mass was measured to ±0.1 µg and temperature to ±0.1 • C. Approximately 10 mg of sample was loaded into a 70 µL alumina crucible. Samples were heated at 20 • C min −1 up to 110 • C in air flowing at 50 mL min −1 and held for 30 min to drive off any residual moisture. They were subsequently heated at 20 • C min −1 up to 950 • C and held for 30 min to oxidize all material. The mass fraction of sample converted (X) at any time, t, was calculated as: where m i is the initial mass, m t is the mass at any time, t, and m f is the final mass after the hold at 950 • C. Derivative thermogravimetric (DTG) curves were plotted as dX/dt (s −1 ) versus temperature. Differential scanning calorimeter (DSC) data was normalized as heat flow per sample mass at any given instant (m t ). DTG curves are compared to those from an in-house sample of Illinois No. 6 coal, a high volatile bituminous coal from the Illinois #6 (Herrin) seam from the Argonne Premium Coal Bank [47]. The coal sample is well characterized in the literature and is often used as a standard on which to compare solid fuel oxidation [47][48][49]. Dewaterability-and the improvement due to HTC was determined by measuring the Capillary Suction Time (CST) required for water to be separated from sludge across a filter paper (Whatman 17 CHR, VWR International, Milan, Italy) using a Triton Electronics Ltd. capillary suction timer type 304B according to standard methods [43]. CST provides a quantitative assessment of how readily sludge releases water.
The liquid phase remaining after hydrothermal carbonization was characterized by measuring pH, COD, organic nitrogen, ammonia nitrogen, and phosphorous as described above. Measurements of Readily Biodegradable COD (RBCOD) were performed following the procedure described in literature [50]. To measure NH 4 + nitrogen and soluble COD, samples were screened through a 0.45 µm filter [43].

Results
To assess the optimal point to withdraw sludge from the wastewater treatment process for hydrothermal carbonization in terms of resulting hydrochar properties, three samples were pulled from various points along the process: Thickened, Digested, and Dewatered ( Figure 1). The feedstocks characteristics are presented in Tables 1 and 2. These three samples were subjected to hydrothermal carbonization at three temperatures (190 • C, 220 • C, 250 • C) and two residence times (30 min, 60 min) each, producing a total of 18 hydrochar samples for analysis.
Looking at the data of the different raw sludges, it is clear that they differ substantially. Thickened sludge contains about 46 wt% elemental carbon and 15 wt% ash. In digested sludge, the carbon decreases to about 26 wt%, and the ash content increases to 45 wt%, due to stabilization during anaerobic digestion. Dewatered sludge contains 36 wt% carbon and about 28 wt% ash. Thus, even if dewatering is a mechanical process, it greatly modifies the sludge characteristics. The supernatant from sludge dewatering has high concentrations of inorganic compounds such as N-NH 4 + , P compounds, CaCO 3 , Mg, K, Na, and other minerals that contribute to the ash content [51]. The dewatering unit washes away these inorganics strongly decreasing the ash content of the dewatered sludge: this reflects on an increase in elemental C, H, N, and O, and also in FC and VM. VM variation is extremely significant, passing from a value of 50 wt% in the digested sludge to about 66 wt% in the dewatered sludge. These differences are also due to the fact that the digested sludge, immediately upstream of the dewatering operation, is chemically conditioned with organic polyelectrolyte (1-10 g/kg dry solids [52]) that is quickly adsorbed on the sludge particles.

Ammonia Nitrogen in Liquid (g/L)
aras received, n.a.-not applicable, a Based on 1g of dry biomass dissolved in one liter of distilled water, HTC-Hydrothermal Carbonization.

Hydrochar Properties Depend on Sludge Withdrawal Point and Carbonization Conditions
Solid hydrochar yields and properties are strongly influenced by both sewage sludge feedstock and HTC temperature and residence time, as shown in Table 2 (see also Supplementary Materials Figure S1). In general, as the harshness of carbonization increases, the solid hydrochar yield decreases, as is the case with biomasses across the literature [53]. The one exception to this general trend was the digested sludge sample carbonized at 220 • C for 30 min, which had a lower solid yield than its 250 • C and 190 • C counterparts. This anomaly was observed for multiple experimental runs. Such behavior could be attributed to potential re-condensing of tarry materials onto the hydrochar matrix, increasing the observed "solid" yield at 250 • C. Such materials were previously identified on the surface of heterogeneous biomasses that can be maximized at carbonization conditions specific to each biomass [17]. As the solid hydrochar is collected on a 10-100 µm filter, any particles smaller than this size range would be considered part of the liquid yield, which may also explain this anomaly.
As a result of the carbonization process, which releases organic compounds from the solid matrix and forces them into the water phase while generating CO 2 , the total COD present substantially decreased upon carbonization (at any time, temperature) of the dewatered sludge sample, with a similar behavior noted for the thickened and digested samples (the effect on these two samples is less pronounced given the lower initial concentration in the raw samples) as shown in Table 1 and Figure S2 in Supplementary Materials. Conversely, soluble COD increased with increasing severity, most notably for the thickened and digested samples, with a significantly higher soluble COD that is maintained throughout carbonization for the dewatered samples. Such behavior has been described in the literature for other biomasses, including sludge, secondary sludge, wood waste, and dairy waste [54]. Given the nature of the carbonization process, where water acts as both a solvent and transport medium, simultaneously drawing organics into the water phase and generating CO 2 , this COD mass balance suggests that indeed the insoluble fraction is being converted to CO 2 (and lesser amounts of CO and other non-condensable gases) while the soluble portion is retained in the process water.
The decrease in hydrochar pH for increasingly severe carbonization of the thickened sludge is to be expected; Volatile Fatty Acids (VFAs) present in primary sludge and formed via hydrolysis of triacylglycerols [55], especially acetic acid, are known to remain stable in the liquid phase, only breaking down at higher carbonization temperatures [54]. It was found here that the pH for the thickened sludge dropped from 7.2 (raw) to 6.0 for both the 190 • C and 220 • C conditions and to 5.8 for the 250 • C condition of the 30 min HTC samples and to 5.3, 6.1, and 6.0 for the 190 • C, 220 • C, and 250 • C 60 min samples, respectively (Table 1; plotted in Figure S3 in Supplementary Materials). This suggests that carbonization is able to release some VFAs from the primary matrix, but that the overall acidic nature of the hydrochar increases slightly. It should be noted that these measurements were taken in the same manner used to measure biochar pH, by equilibrating the hydrochar with deionized water and reading the pH of the liquid following settling. Anecdotally, when the pH of the process water is measured immediately following carbonization, the pH increases slightly upon severe carbonization, which would indicate hydrolysis of released VFAs occurs in the process water. For the digested samples, carbonization at 250 • C yields hydrochars with slightly higher pH than their 190 • C and 220 • C counterparts, suggesting a reduced acid content of the char. For the dewatered secondary sludge, the pH decreases with increasing carbonization but is overall lowest to begin with, suggesting that the acid content decreased during anaerobic digestion.
The proposed loss of semi-volatile compounds from the sludge matrices upon carbonization is further supported by the proximate analysis ( Figure 2; Table 2; data of Table 2 plotted in Figure S4 in Supplementary Materials). The sludge treatment of WWTP results in a decrease in fixed carbon and volatile matter content upon digestion, with a slight increase in both upon dewatering. Changes in proximate analysis as a function of carbonization conditions versus the raw sludge withdrawn at each point are given in Figure 2. Figure 2 reports the percentage change in value of the various variables (FC, VM, Ash, all data on a dry basis) due to HTC, in respect to the raw biomasses. As compared to the raw sludge sample from each withdrawal point, the volatile matter content decreased with increasingly harsh carbonization. The fixed carbon content also decreases upon HTC, shifting the balance of the hydrochar composition to the inorganic (loosely termed "ash" phase). The change in proximate analysis appears to be more heavily temperature dependent than time dependent for the thickened samples. In this case, the ash content increased by 50 wt%, 100 wt%, and 150 wt% over the raw thickened sample as the carbonization temperature increased from 190 • C to 220 • C to 250 • C, respectively. This increase in inorganic content is offset more so by decreases in fixed carbon than volatile matter. For the digested samples, the increase in ash content-ranging from 25 wt% to 50 wt% as temperature increases-is offset by decreases in both VM and FC, with a slightly higher decrease in FC than VM across HTC conditions. The behavior of the dewatered samples is somewhat different. While for HTC at 30 min there is a slight increase in ash (especially at 220 • C and 250 • C carbonization), the 220 • C 30 min sample is offset more by changes in FC, whereas the 250 • C 60 min sample's higher ash content is due to a decrease in VM. The dewatered 60 min HTC samples show even further divergent behavior; for all temperatures, the FC and ash contents increase, offset by a decrease in VM content. For the 190 • C and 220 • C samples, the change in FC weight percent is actually greater than that of the ash, whereas the 250 • C sample has a larger increase in ash than fixed carbon. This is likely due to the preparation of the samples. Both thickened and digested samples were used as received. The dewatered samples had too low of a moisture content to ensure all solids were submerged in the reactor, such that deionized water was added to the samples prior to carbonization. It is possible that some of the inorganics migrated into the water phase (to establish a concentration equilibrium), and thus, the relative change in ash content was lower for these samples.     This behavior is echoed by the ultimate analysis, whereby HTC temperature has a greater impact on organic element composition than time (Figure 3; Figure 4; Table 2; data of Table 2 plotted in Figure S5 in Supplementary Materials). As expected, the biological transformations of the primary sludge lead to overall lower oxygen and carbon contents of the raw digested and dewatered samples [56]. Figure 3 reports the percentage change in content of atomic species C, H, N, and O (all data on a dry basis) due to HTC, in respect to the raw biomasses. As shown in Figure 3, in general the weight percent of the organic elements decrease upon carbonization of the sludge samples, the balance made up by an increasing (relative) inorganic content. The elemental carbon contents for the thickened 190 • C 30 min and 60 min and all dewatered samples remain constant within statistical significance. Interestingly, the relative percent changes of C, H, and O were fairly minimal for the thickened sludge carbonized at 190 • C and 220 • C, suggesting minimal decarboxylation and dehydration, both of which would lower the O:C and H:C ratios, respectively [57]. However, this is not to imply a stagnant system; for the thickened (and indeed all sludge samples) the nitrogen content of the solids decreases considerably upon even mild carbonization, ranging from hydrochars with 40 wt% to those with 80 wt% lower N content than their raw counterparts (Figure 3). This loss of nitrogen supports the hypothesis that there is significant hydrolysis of fat and protein components present in the sludge. The organic nitrogen in the process liquid tends to decrease by 15-85 wt% upon carbonization of all samples. The ammonia nitrogen present in the liquid varies in a certain range, without any clear trend with time or temperature, as shown in Table 1. Given the substantial elemental decrease in N of the hydrochar, and liquid-phase organic nitrogen upon carbonization, closure of a mass balance would suggest either production of nitrogen gases (N 2 , NO x ) or formation of nitrate or nitrite in the liquid phase, which could precipitate out as salts onto the hydrochar. As reported by Kruse and co-workers, this concentration would be too low to detect with N content measurements [58].   with time or temperature, as shown in Table 1. Given the substantial elemental decrease in N of the 101 hydrochar, and liquid-phase organic nitrogen upon carbonization, closure of a mass balance would 102 suggest either production of nitrogen gases (N2, NOx) or formation of nitrate or nitrite in the liquid phase, which could precipitate out as salts onto the hydrochar. As reported by Kruse and co-workers, this concentration would be too low to detect with N content measurements [58].

112
The carbon and fixed carbon values reported in Table 2 and the trends of the same variables 113 shown in Figures 2 and 3 testify to the peculiarity of these kinds of substrates which behave 114 differently during HTC in respect to the vast majority of the other biomasses. While HTC applied to sludges ( Figure 2). The C content in the hydrochars is equal to or lower than C content in the raw 119 sludges ( Figure 3). This apparently strange behavior depends on the very high ash content of the 120 sludge samples and was also previously reported in the literature [31,57,61]: in all these cases, the C 121 and FC contents were expressed on a dry basis. Reverse C and FC trends were reported when the 122 data was expressed on a dry ash free (daf) basis [23]. Here, on a daf basis, thickened and dewatered 123 sludge increase their C content after HTC, and this applies to the digestate treated at the highest HTC   The carbon and fixed carbon values reported in Table 2 and the trends of the same variables shown in Figures 2 and 3 testify to the peculiarity of these kinds of substrates which behave differently during HTC in respect to the vast majority of the other biomasses. While HTC applied to agro-waste, lignocellulosic feedstock, organic fraction of municipal solid waste, or compost [20,45,53,59,60] results in an increase in the values of C and FC, this is not the case for the majority of the sludge samples here investigated. The FC content decreases after HTC for thickened and digested sludges (Figure 2). The C content in the hydrochars is equal to or lower than C content in the raw sludges ( Figure 3). This apparently strange behavior depends on the very high ash content of the sludge samples and was also previously reported in the literature [31,57,61]: in all these cases, the C and FC contents were expressed on a dry basis. Reverse C and FC trends were reported when the data was expressed on a dry ash free (daf) basis [23]. Here, on a daf basis, thickened and dewatered sludge increase their C content after HTC, and this applies to the digestate treated at the highest HTC temperature of 250 • C, too; FC data, conversely, even on a daf basis, remain still lower in the hydrochars in respect to the parent biomasses. To summarize, the organics in the sludge carbonized as should be expected in an HTC process: the increase in C content is evident when the basis of calculation is represented by the organics themselves (i.e., data on a daf basis), and the HTC was sufficiently severe. On a dry basis, conversely, the relative increase in ash content often prevails over the relative increase in C, as recently also discussed by Ferrentino et al. [62].

Distribution of Nutrients and Inorganics: Potential for Use as Soil Amendment
As shown in Table 1, the total phosphorous concentrated in the solid hydrochar (all samples/processing conditions) from 5 wt% in the raw samples to up to 10 wt% in the carbonized samples, with the thickened and dewatered 250 • C 30 min and 60 min samples having the highest amounts. The thickened sludge (all times/temperatures) showed lower organic nitrogen concentrations in the liquid phase, suggesting it remains in the solid phase or exits as a gas, as mentioned previously. This, coupled with the higher FC and VM contents of the thickened and dewatered hydrochars, suggests they would make better soil amendments than the digested sludge hydrochars [63][64][65].
Thermally treated biomasses tend to show enhanced P fertilizer values as a result of various mechanisms, including structural surface changes and improved association of P to inorganics such as Mg, Ca, and Al [66]. In the present work, the type of sludge sample had a larger impact on the retention and concentration of these nutrients than the carbonization conditions. As shown in Figure 5, the dewatered samples had higher overall Al concentrations-from 4 mg Al /g sludge for the raw sample as compared to~2 mg Al /g sludge for the thickened and digested raw samples. Carbonization at 220 and 250 • C for both 30 min and 60 min doubled all of these concentrations, to 8 mg Al /g sludge for the carbonized dewatered sample and~4 mg Al /g sludge for the others. Conversely, the digested sludge had almost twice the concentration of magnesium than the thickened or dewatered samples (which again almost doubled upon carbonization at 220 • C and 250 • C). The calcium concentration was quite similar for all sludge samples at 23, 33, and 28 mg Ca /g sludge , for the thickened, digested, and dewatered raw samples, respectively. Upon carbonization at 250 • C for 30 min, the Ca content increased by over 100% for the thickened sample but only by 39% and 43% for the digested and dewatered raw samples, respectively. While soil incubation studies are beyond the scope of the present work, prior work by Thomsen and co-workers [66] suggests that the more heavily oxidized hydrochar samples (e.g., those that released more CO 2 and thus had higher gas yield) containing more Mg, Ca, and Al would have a higher P availability. This corresponds to the digested 250 • C 60 min and dewatered 220 • C and 250 • C for 30 min and 60 min samples. Future work will investigate the degree to which hydrothermal carbonization plays a role in P plant availability, as well as the impact of HTC processing conditions itself on the volatilization and re-condensation of P species on the hydrochar surfaces. Thermally treated biomasses tend to show enhanced P fertilizer values as a result of various 140 mechanisms, including structural surface changes and improved association of P to inorganics such 141 as Mg, Ca, and Al [66]. In the present work, the type of sludge sample had a larger impact on the 142 retention and concentration of these nutrients than the carbonization conditions. As shown in Figure   143 5, the dewatered samples had higher overall Al concentrations-from 4 mgAl/gsludge for the raw sample Thomsen and co-workers [66] suggests that the more heavily oxidized hydrochar samples (e.g., those 154 that released more CO2 and thus had higher gas yield) containing more Mg, Ca, and Al would have  The concentrations of a series of additional inorganic elements in the sludges and hydrochars were measured. Figure 5 shows a representative set of these elements (all data available in Supplementary Materials: Tables S1 and S2, Figures S6 and S7). Silver was included in the analysis, given its increasing prevalence in consumer materials as an incorporated nanomaterial and resulting detection in wastewater treatment systems [67]. Prior work indicates that silver will more likely accumulate in the biosolids than in the WWTP effluent [68]. Here, considerably higher levels of silver in the digested sludge are found as compared to the thickened or dewatered sludge, suggesting that the digestion indeed concentrates the silver in the biosolid, but that the dewatering process shifts the silver to the liquid phase. The silver concentration in the digested hydrochar decreases with increasing carbonization. Both of these indicate that the silver is in an easily mobile state in the digested sludge, equilibrating with its aqueous phase to lower solid concentration.
Arsenic and vanadium were included in this analysis given their high prevalence in Italian drinking water sources; the acceptable limit for arsenic in the Trentino-Alto Adige region increases to 50 µg L −1 over the nationwide 10 µg L −1 limit due to natural lithology [69]. As seen in Figure 5, the arsenic concentrations in the solid samples decreased with increasing harshness for the thickened and digested samples (with the exception of the 190 • C and 250 • C, 60 min digested samples). All other hydrochars had arsenic concentrations below 2 µg g −1 . The International Biochar Initiative (IBI) [70] suggests an acceptable range for As in biochars of between 13 and 100 µg g −1 if they are to be applied as a soil amendment.
In Sicily, concentrations of V in drinking water sources are routinely above E.U. and U.S. maximum contaminant levels (MCLs) due to the underlying geology [71][72][73]. As such, it is important to ensure that concentrations of these metals were below MCLs before considering the potential for hydrochars to be land-applied, which could exacerbate the metal contamination issue. The vanadium concentration was slightly increased upon carbonization, but never exceeded 10 µgg −1 ; the MCL for drinking water in Italy is 50 µgL −1 . V is known to co-precipitate with iron (III) [74]; while the oxidation states are not known here, the concentrations of iron in all solids exceeded 2 mgg −1 (therefore higher by several orders of magnitude), such that it is possible that V is present in the hydrochars in an iron (hydr)oxide precipitate.
As shown in Figure 5, the concentrations of chromium, cobalt, nickel, molybdenum, and lead (and a series of additional inorganics, as given in SI) are only modestly affected by carbonization and sludge withdrawal point. The cobalt concentration of the digested sludge was notably higher than the thickened or dewatered, but upon carbonization reached the same levels as the other samples of hydrochars, all at less than 3 µgg −1 . The concentrations of all of these metals are below IBI maximum allowable thresholds [70], making them reasonable candidates for use as soil amendments on the basis of heavy metal content.

Energy Content and Oxidative Reactivity: Potential for Use as Solid Fuel
As Table 2 shows, the solid hydrochars have similar higher heating values to their raw counterparts. In this case, HTC has not substantially improved the "energy density" of the solid fuel. However, HTC does enable a more efficient solid-liquid separation and dewaterability. For the raw dewatered sludge, CST was not complete after 60 min, while it was complete for the 190 • C 60 min hydrochar after only 380 s. With an increase in HTC temperature, CST was 209 and 90 s at 220 • C and 250 • C (60 min), respectively. Images of the CST trials are available in Supplementary Materials Figure S8. This improved dewaterability demonstrates a reduction of volume for transport of the solid waste and reduced moisture content for potential combustion applications. Figure 6 plots the DTG curves of the hydrochar samples produced at each of the three carbonization temperatures at 30 min, alongside an Illinois No. 6 coal sample. As can be seen, the hydrochar sludge samples are considerably more reactive than the coal sample. Their peak DTG temperatures (highest conversion rates) occur at hundreds of degrees less than the coal sample and at higher conversion rates. The highest peak mass loss rates occur for all three hydrochars produced from the thickened sludge, prior to any anaerobic digestion. Post-digestion, while the shape of the DTG curves changes, the peak rates are quite similar for both the thickened and dewatered sludge. For all three samples, the 190 • C carbonized hydrochars display the highest reactivity compared to the other HTC temperatures. Especially at such mild conditions, HTC does not significantly carbonize the sample-oftentimes, the original materials microstructure is preserved, whereas higher temperatures lead to a more complete destruction of the carbon matrix [53,75].

220
Given the relatively high reactivity of any of the sludge hydrochars, it may be difficult to 221 combust them for electricity generation in current boilers designed for solid fuels such as coal [76].

222
While the higher heating values of especially thickened sludge (18-21 MJ kg −1 ) are suitable for such 223 combustion schemes, their lower ignition and peak reactivity temperatures are considerably lower 224 than that of most bituminous coals, and therefore may lead to loss of efficiency in boilers [77]. That, 225 combined with the higher ash content that could result in slagging and fouling, suggests that the 226 sludge hydrochars may perform better in co-combustion scenarios [78].

227
It was previously shown that blending biofuels with similar characteristics at ratios less than 20 228 wt% with coal mitigates fuel segregation and efficiency loss issues while increasing the share of 229 renewables in energy generation portfolios [48,49,76]. Recent work in the literature suggests that 230 hydrochars can be co-combusted with a variety of coals in economically, environmentally, and 231 energetically viable schemes in existing infrastructure [79][80][81] and may even improve the emissions 232 profile at optimized blending ratios [82]. It was recently demonstrated that hydrochars with similar 233 reactivities can be oxidized with Illinois No. 6 coal at ratios of 10 wt% hydrochar, balance coal, 234 without causing significant fuel segregation [83]. In summation, the high reactivity and ash content  Given the relatively high reactivity of any of the sludge hydrochars, it may be difficult to combust them for electricity generation in current boilers designed for solid fuels such as coal [76]. While the higher heating values of especially thickened sludge (18-21 MJ kg −1 ) are suitable for such combustion schemes, their lower ignition and peak reactivity temperatures are considerably lower than that of most bituminous coals, and therefore may lead to loss of efficiency in boilers [77]. That, combined with the higher ash content that could result in slagging and fouling, suggests that the sludge hydrochars may perform better in co-combustion scenarios [78].
It was previously shown that blending biofuels with similar characteristics at ratios less than 20 wt% with coal mitigates fuel segregation and efficiency loss issues while increasing the share of renewables in energy generation portfolios [48,49,76]. Recent work in the literature suggests that hydrochars can be co-combusted with a variety of coals in economically, environmentally, and energetically viable schemes in existing infrastructure [79][80][81] and may even improve the emissions profile at optimized blending ratios [82]. It was recently demonstrated that hydrochars with similar reactivities can be oxidized with Illinois No. 6 coal at ratios of 10 wt% hydrochar, balance coal, without causing significant fuel segregation [83]. In summation, the high reactivity and ash content (though high HHV) of the thickened and dewatered hydrochars temper enthusiasm for their use as a combustible fuel. The low HHV of the digested sludge-below low-rank coals-makes it difficult to envision a combustion scenario where this would be a valuable solid fuel. As such, the potential for using sludge-based hydrochars as a drop-in solid fuel is likely minimal, though this could be accomplished without the need for anaerobic digestion of the sludge if used in a co-fired fuel scenario.

Further Considerations
This work examined the impact of WWTP withdrawal point and hydrothermal carbonization conditions on resulting hydrochar properties. Table S3 in Supplementary Materials provides a summary of the overall results discussed. While there is considerable potential to use HTC to convert sewage sludge to renewable fuels and/or use the hydrochars for nutrient recovery redistribution, several questions about the feasibility of implementing such a process remain unanswered.
One question is what to do with the process water remaining after treatment, which contains unreacted feedstock and/or chemical intermediates that are potentially hazardous [84]. While this is beyond the scope of the present work, others have previously demonstrated several potential management options. Process liquors remaining after hydrothermal treatments have been shown to be suitable feedstocks for aerobic, anoxic, and anaerobic processes [85][86][87]. The experimental data of the present investigation testify that RBCOD (readily biodegradable COD) for HTC performed at 190 • C reached values above 85% of total COD, while increasing the temperature led to values below 15 % at 250 • C. While wet air oxidation liquors have shown some inhibitory behaviors for strictly anaerobic treatments [86], this has not been the case for some studies of HTC liquors [87,88], although others suggest methanogenesis inhibition at high inoculum concentrations (>25 g COD L −1 ) [89]. As demonstrated by Qiao and co-workers [90], HTC process water may actually increase the efficiency of the solid hydrolysis step, a rate-limiting step in the digestion process [91,92], and enhance methane production in the digester [93]. The use of nitrates as oxidants can catalyze COD and dissolved organic carbon removal [94], and, as it was recently demonstrated, hydrochars themselves can be used to enhance anaerobic digestion [95].
Second, the fate of some heavy metals during HTC, especially Cu and Cr, may complicate the use of these materials as a soil amendment, as successive WWT and carbonization concentrates them within the hydrochar. While it was not found here that heavy metal concentrations were above recommended IBI limits in the hydrochars, the concentrations of heavy metals in wastewater solids and effluents vary widely across the globe and can pose health risks in certain areas [96]. To address this potential concern, Shi and co-workers [97] demonstrated that Cd in sludge hydrochars can be immobilized by the synergistic nature of apatite P present in the chars and addition of hydroxyapatite. However, as both acid and alkaline leaching have been shown to dissolve phosphate in sludges and sludge ashes [98], the long-term stability of the Cd immobilized by the method of hydroxyapatite addition is not clear. Prior work by Yoshizaki and Tomida [99] demonstrated that such heavy metals could be removed by phosphoric acid and hydrogen peroxide for downstream reuse and recovery. Their method had enhanced recovery and was more environmentally and economically viable than treatment with hydrochloric acid or sulfuric acid, opening a potential pathway for extraction of the metals from hydrochars. Such treatment would likely act as a porogen to increase the surface area of the hydrochars [100], increasing their ability to retain water and slow-release nutrients when used as a solid amendment [101], and may open up possibilities for conversion of the materials to activated carbons for use in water treatment, battery electrodes, and other high-value materials [102][103][104][105].

Conclusions
The present study probed the impact of hydrothermal processing conditions and sewage sludge withdrawal point on resulting hydrochars. The hydrothermal carbonization (HTC) of sludge proceeds similarly to many other wet biomasses; as harshness (time and temperature) of the process increases, the solid yield decreases, the ash (inorganic) content increases, total COD decreases but soluble COD increases, and solid pH decreases. However, there are distinct differences between the hydrochars produced from primary (thickened) versus secondary (digested and dewatered) sludge. Thickened sludge carbonized at moderate conditions (220 • C, 30 min) produced the most viable solid fuel with the highest HHV, moderate ash content, and high volatile matter content. However, with O:C and H:C ratios higher than typical bituminous coals of similar heating content, the thermal reactivity of the hydrochar was significantly higher than coals typically combusted. This suggests that sludge hydrochars could be co-fired with coal but are not ideal solid fuels. On the other hand, hydrochars produced from secondary sludge are more viable as potential soil amendments. The carbonized digested sludges show relatively neutral pH, low COD, and enhanced phosphorous, along with enhanced Ca, Mg, and Al concentrations to help mobilize P. Their heavy metal composition is well below International Biochar Initiative standards, though the elemental oxygen content and lower volatile matter content warrant future inquiry into this pathway.
Supplementary Materials: The following are available online at http://www.mdpi.com/1996-1073/13/11/2890/s1, Figure S1: Product distribution among solid, gas and liquid (by difference) phases following hydrothermal carbonization of three sludge samples, Figure S2: Impact of hydrothermal carbonization on total COD (top) and soluble COD (bottom) of three sludge samples, Figure S3: pH of raw and hydrothermally carbonized sludge samples, Figure S4 Figure S8: Visual evidence of hydrochar dewaterability, Table S1: ICP-OES analysis of inorganics present in three sludge feedstocks, Table S2: ICP-OES analysis of inorganics present in hydrothermally carbonized sludge samples, Table S3: Summary of observations of sludge hydrochars.