*Article* **Acidogenesis of Pentose Liquor to Produce Biohydrogen and Organic Acids Integrated with 1G–2G Ethanol Production in Sugarcane Biorefineries**

**Guilherme Peixoto 1,2, Gustavo Mockaitis 3,4, Wojtyla Kmiecik Moreira 4, Daniel Moureira Fontes Lima 5, Marisa Aparecida de Lima 2, Filipe Vasconcelos Ferreira 4, Lucas Tadeu Fuess 4,\*, Igor Polikarpov <sup>2</sup> and Marcelo Zaiat <sup>4</sup>**


**Abstract:** Second-generation (2G) ethanol production has been increasingly evaluated, and the use of sugarcane bagasse as feedstock has enabled the integration of this process with first-generation (1G) ethanol production from sugarcane. The pretreatment of bagasse generates pentose liquor as a by-product, which can be anaerobically processed to recover energy and value-added chemicals. The potential to produce biohydrogen and organic acids from pentose liquor was assessed using a mesophilic (25 ◦C) upflow anaerobic packed-bed bioreactor in this study. An average organic loading rate of 11.1 g COD·L−1·d−<sup>1</sup> was applied in the reactor, resulting in a low biohydrogen production rate of 120 mL·L−<sup>1</sup> <sup>d</sup>−1. Meanwhile, high lactate (38.6 g·d−1), acetate (31.4 g·d−1), propionate (50.1 g·d−1), and butyrate (50.3 g·d−1) production rates were concomitantly obtained. Preliminary analyses indicated that the full-scale application of this anaerobic acidogenic technology for hydrogen production in a medium-sized 2G ethanol distillery would have the potential to completely fuel 56 hydrogen-powered vehicles per day. An increase of 24.3% was estimated over the economic potential by means of chemical production, whereas an 8.1% increase was calculated if organic acids were converted into methane for cogeneration (806.73 MWh). In addition, 62.7 and 74.7% of excess organic matter from the 2G ethanol waste stream could be removed with the extraction of organic acid as chemical commodities or their utilization as a substrate for biomethane generation, respectively.

**Keywords:** sugarcane bagasse; pentose liquor; dark fermentation; biohydrogen; organic acids; biorefinery

#### **1. Introduction**

Brazilian sugarcane mills are migrating from old concepts of sugar and ethanol production to a new concept of biorefinery. This concept aims to produce not only biofuels but also electricity, food, and other products that use renewable sources from sugarcane biomass [1–3]. One of the main actions to consolidate the biorefinery concept is the implementation of second-generation (2G) ethanol production integrated with first-generation (1G) plants. This industrial concept intends to use surplus bagasse that is currently used as

**Citation:** Peixoto, G.; Mockaitis, G.; Moreira, W.K.; Lima, D.M.F.; de Lima, M.A.; Ferreira, F.V.; Fuess, L.T.; Polikarpov, I.; Zaiat, M. Acidogenesis of Pentose Liquor to Produce Biohydrogen and Organic Acids Integrated with 1G–2G Ethanol Production in Sugarcane Biorefineries. *Waste* **2023**, *1*, 672–688. https://doi.org/10.3390/ waste1030040

Academic Editors: Vassilis Athanasiadis and Dimitris P. Makris

Received: 3 July 2023 Revised: 21 July 2023 Accepted: 31 July 2023 Published: 5 August 2023

**Copyright:** © 2023 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

fuel in cogeneration systems, to produce lignocellulosic ethanol (2G) instead [4]. The utilization of this remaining biomass plays a fundamental role in the biorefinery yield because 1 ton of sugarcane generates 280 kg of bagasse [5], which is very significant considering the production of 610 million tons of sugarcane [6] in the 2022/2023 harvest.

The implementation of 2G ethanol production can improve process sustainability and also ethanol production [7,8]. In this process, the bagasse is pretreated to obtain a solid fraction that is rich in cellulose, while hemicellulose is mainly hydrolyzed to pentoses (pentose liquor). The cellulose fraction obtained from this pretreatment can be enzymatically hydrolyzed to feed the 2G ethanol-producing process, and pentose liquor remains a by-product of the process. Although ethanol production is the natural destination of the hexoses (C6) fraction, the processing of pentose liquor is still not well defined, and alternatives must be studied for the application of this stream. For instance, opportunities for fermenting pentoses to obtain ethanol have been massively studied [9,10]; however, selecting efficient pentose fermenters is still challenging.

The xylose-rich nature of pentose liquor characterizes it as a highly suitable substrate for biohydrogen and organic acid production in acidogenic bioreactors because this process depends primarily on carbohydrate-rich substrates [11,12]. Hydrogen is a renewable clean energy carrier when biomass and its by-products are used as raw materials in fermentation. The biological production of hydrogen from residues and wastewater is less energy intensive and is less expensive than methane steam reforming and electrolysis [11–14]. Moreover, various organic acids and other compounds with different applications (foods, pharmaceuticals, and chemicals) can be generated during the biological production of hydrogen. Most studies have reported that biohydrogen production generates intermediates such as acetic acid, propionic acid, butyric acid, succinic acid, formic acid, butanediol, and acetone [15–20]. These compounds are of special interest because of their high market value [21,22], especially butyric and lactic acid, which are used as precursors of industrial thermoplastics [23] and biodegradable polymers [24,25]. Although only a few studies have addressed hydrogen and organic acid production using sugarcane bagasse (SCB) hydrolysate [26–30], several studies have used pure xylose as a substrate for the production of biohydrogen [31–39], indicating a gap for exploitation within the context of dark fermentation studies. Furthermore, most researchers have used batch reactors in contrast with the few studies conducted in continuous bioreactors. Neither xylose nor SCB hydrolysate have been used in upflow anaerobic packed-bed reactors, which have generally yielded high biohydrogen productivity [40–44].

This study assessed the potential application of sugarcane bagasse-derived pentose liquor as a source of biohydrogen and organic acids in a continuous packed-bed bioreactor, characterizing an alternative to add value to a voluminous by-product which will be inevitably available in integrated 1G-2G sugarcane biorefineries in coming years. A dual approach based on an experimental assessment followed by a scenarization-based exercise through a simple process application analysis in a medium-sized distillery (milling capacity of 500 ton·h<sup>−</sup>1) was carried out, providing the bases to understand the potential of pentose liquor as a substrate for biotechnological applications.

#### **2. Materials and Methods**

#### *2.1. Bioreactor and Support Material*

The upflow anaerobic packed-bed reactor (UAPBR) was built in tubular acrylic with a length-to-diameter (L/D) ratio of 9.4, considering a total height of 750 mm and a working volume of 2.370 L. The bioreactor's bed region was randomly filled with recycled low-density polyethylene (LDPE) pellets to support cell attachment, similar to previous experiments with real wastewater [43–45]. Cylinder-shaped particles with a mean diameter of 12.7 mm and ca. 30 mm in height were used with specific gravity and a surface area of 0.96 g·cm−<sup>3</sup> and 7.94 m2·g−1, respectively. A total of 295 g LDPE·L−<sup>1</sup> was used to build the bed zone, resulting in a bed porosity of 55%. The packed bed occupied 66% of the total volume. A schematic diagram of the reactor is shown in Figure 1.

**Figure 1.** Sketch of the upflow anaerobic packed-bed reactor and details of the LDPE pellets. Legend: 1—inlet port, 2—packed bed, 3.1–3.5—sampling ports, 4—outlet port (liquid phase), 5—outlet port (biogas).

#### *2.2. Substrates*

The UAPBR was fed with two different substrates in distinct operating periods, namely, a xylose-based lab-made wastewater (experimental condition I) and pentose liquor derived from 2G ethanol production (experimental condition II), as indicated in Figure 2. The former was prepared using 99.9% pure D-(+)-Xylose (Sigma-Aldrich®, San Luis, MO, USA), whilst pentose liquor was obtained from the processing of residual SCB collected from a Brazilian sugarcane mill. Bagasse was subjected to physicochemical pretreatment (121 ◦C and 1.1 atm) utilizing a mass ratio of solid (g dry weight) to liquor (g) at 1:10 with 2% sulfuric acid (H2SO4) (*v*/*v*) for 60 min [46]. This pretreatment resulted in a hydrolysate with an organic matter concentration in terms of COD (chemical oxygen demand) of 14 g·L−1. Prior to feeding the bioreactor, the hydrolysate was neutralized with sodium hydroxide (NaOH) and diluted with tap water to obtain a COD of approximately 2 g·L<sup>−</sup>1. The compositional characteristics of the wastewater utilized in both experimental conditions are shown in Table 1.

**Figure 2.** Processing of sugarcane bagasse to produce second generation ethanol. Sugarcane bagasse constituents by Aguilar et al. [46].

**Table 1.** Compositional characterization of the wastewater used in reactor feeding.


#### *2.3. Experimental Procedure*

Bioreactor inoculation was carried out prior to experimental conditions I and II and consisted of the natural fermentation of xylose and pentose liquor, respectively. The natural fermentation was performed by exposing 30 L of each medium for 3 days to the atmosphere at ca. 25 ◦C. In previous studies, such as those published elsewhere [16,40,41], this procedure yielded mainly microorganisms similar to *Clostridium* sp. (91%), *Klebsiella* sp. (97%), and *Enterobacter* sp. (93%), which are directly related to hydrogen and organic acid production [41]. Naturally fermented substrates were recirculated into the UAPBR for 7 days [16] as a strategy to promote the attachment of microorganisms to the packed bed, after which a continuous operation was started. The volumetric flow rate of the bioreactor feed was maintained at ca. 1.19 L·h−<sup>1</sup> using a positive displacement pump (Concept Plus, ProMinent Brasil Ltd.a., São Bernardo do Campo, Brazil), resulting in a hydraulic retention time (HRT) of 2 h. The operating temperature was controlled at 25 ± 1 ◦C in a thermostatic chamber (410-DRE, Nova Ética, Vargem Grande Paulista, Brazil). In condition I, xylose was used as the only substrate for 36 days, whilst pentose liquor was evaluated as the substrate for 39 days in condition II. The pH of both substrates (lab-made and real wastewater) was maintained at ca. 6.0 by dosing with NaOH (500 mg·L<sup>−</sup>1) or hydrochloric acid (HCl) (10 mol·L−1) solutions. Additionally, both media were supplemented with a macro and micronutrient solution containing (in mg·L−1): CH4N2O (7.7), NiSO4·6H2O (0.15), FeSO4·7H2O (2.5), FeCl3·6H2O (0.25), CoCl2·2H2O (0.04), CaCl2·6H2O (2.06), SeO2 (0.036), KH2PO4 (1.3), KHPO4 (5.36), and Na2HPO4·2H2O (2.76).

#### *2.4. Monitoring Procedure and Analytical Methods*

Liquid phase monitoring was based on periodic measurements of the pH, COD, and concentrations of the total carbohydrates, organic acids, and solvents. pH and COD were determined according to the Standard Methods for the Examination of Water and Wastewater [47]. Potentiometric measurements (pH) were carried out using a pH meter with a standard glass electrode (Digimed Instrumentação Analítica, São Paulo, Brazil). Total carbohydrates were measured as proposed by Dubois et al. [48], whilst organic acids and solvents were analyzed with a high-performance liquid chromatography (HPLC) system (Shimadzu Scientific Instruments, Kyoto, Japan) using the same configuration and protocols described elsewhere [49].

Gas phase monitoring was carried out by measuring both the biogas flow rate (BFR) and composition. BFR was obtained by directly coupling a gas meter (MilliGascounter-1 V30, Dr.-Ing. Ritter Apparatebau GMBH & Co. KG, Bochum, Germany) to the headspace of the reactor. The biogas composition in terms of hydrogen (H2), nitrogen (N2), methane (CH4), and carbon dioxide (CO2) was carried out using a gas chromatography set (GC-2010) equipped with a thermal conductivity detector (GC/TCD; Shimadzu Scientific Instruments, Kyoto, Japan) and argon as the carrier gas, as described elsewhere [50]. Gas samples were collected from UAPBR's headspace with 1000 μL-insulin syringes equipped with Teflon body two-way valves (Supelco® Analytical—Sigma-Aldrich, Bellefonte, PA, USA).

#### *2.5. Scenario Assessment*

Four different scenarios, namely, 1 to 4, were considered for a medium-sized 2G ethanol plant (milling capacity of 500 ton·h<sup>−</sup>1), providing a preliminary analysis of pentose liquor as raw material for producing bioenergy and biochemicals. Scenarios 1 and 2 did not include pentose as raw material, whilst scenarios 3 and 4 considered the utilization of pentoses for the energy recovery and/or generation of value-added products. In Scenario 1, 50% of SCB was considered to be used for 2G ethanol production, with the remaining fraction used in the cogeneration of electricity and steam. In Scenario 2, 100% of SCB was directed to 2G ethanol production. In Scenario 3, 2G ethanol was produced with the utilization of all available SCB. Additionally, hydrogen and volatile organic acids (VOA) were derived from pentose liquor. In Scenario 4, 2G ethanol, hydrogen, and organic acids

were produced. In this last scenario, organic acid consumption in biomethane production for cogeneration was considered despite the use of chemicals, as reported in Scenario 3.

#### **3. Results and Discussion**

#### *3.1. Biohydrogen Production*

Biogas generation profiles obtained with xylose-based wastewater and pentose liquor are depicted in Figure 3, showing distinct patterns according to the type of substrate. In experimental condition I (xylose as the substrate), biogas production reached 193.8 mL·h−<sup>1</sup> (4.65 L·d−1) on day 10. This production peak was followed by continuous decay until the end of the operation (Figure 3A). This drop in biogas generation has also been observed in previous studies regarding hydrogen production from carbohydrates [40–42] and complex wastewaters [44,51]. A coherent explanation can be found in the major growth of homoacetogenic organisms, which are capable of using the Wood– Ljungdahl pathway because the applied specific organic load decreased with the increase in the biomass in the packed-bed reactor [52,53]. The homoacetogenic pathway explains the biogas production decrease because both H2 and CO2 produced by hydrogen-producing bacteria (HPB) were converted into acetate, as demonstrated in Reaction (1) [54]. Numerous studies dealing with the dark fermentation of sugarcane-derived substrates, namely, vinasse, molasses, and juice, have reported the occurrence of homoacetogenesis when applying mesophilic temperature conditions [19,55–58], which supports this hypothesis. Homoacetogenic bacteria belonging to genera *Moorella*, *Oxobacter,* and *Sporomusa*, as well as to the family *Lachnospiraceae,* were identified in vinasse-fed reactors [19,59,60]. Some clostridial groups [61], as well as some sulfate-reducing bacteria belonging to the genus *Desulfotomaculum* [62], are also capable of utilizing H2 and CO2.

$$2\text{CO}\_2 + 4\text{H}\_2 \rightarrow \text{Acetate} + 2\text{H}\_2\text{O} \ (\Delta \text{G}^\circ = -104 \text{ kJ} \cdot \text{mol}^{-1}) \tag{1}$$

**Figure 3.** Biogas flowrate and composition and substrate utilization in experimental conditions (**A**) I (xylose-based wastewater) and (**B**) II (pentose liquor).

The mean biogas production rates obtained in conditions I and II were 0.026 L·L−1·h−<sup>1</sup> (624 mL·L−1·d−1) and 0.017 L·L−1·h−<sup>1</sup> (408 mL·L−1·d−1), respectively. The volumetric hydrogen production rate (VHPR) followed the same trend, with values of 0.016 (384 mL·L−1·d−1) and 0.005 L·h−1·L−<sup>1</sup> (120 mL·L−1·d−1) observed in experimental conditions I and II, respectively. The main reason for a higher hydrogen production rate in the bioreactor processing xylose was the production peak reported at the beginning of the operation (days 8–11). This behavior could be explained considering that the substrate in condition I consisted only of readily available xylose, which could potentially contribute to a faster and more efficient biomass adaptation [63].

By contrast, experimental condition II contained a mixture of pentoses (xylose, arabinose, and rhamnose; Table 1). These different carbon sources and the presence of furfural and acetic acid could be responsible for lower hydrogen production rate at the beginning of the bioreactor operation. In the study by Aguilar et al. [46], the sugarcane bagasse

subjected to the same diluted acid pretreatment released approximately 1.2 g·L−<sup>1</sup> of furfural and 4.5 g·L−<sup>1</sup> of acetic acid, which are considered toxic to fermentation in these ranges [46,64]. However, the hydrolysate generated in this study was diluted by a factor of 7 (COD decreased from 13,989 to 1989 mg·L<sup>−</sup>1) before its utilization in the bioreactor, which consequently resulted in lower concentrations of both constituents. Thus, the best explanation for the worst performance in biohydrogen production in condition II was the lower rate of biomass adaptation. Nevertheless, after 11 days of operation, the highest peak of substrate utilization (83.2%) was noticed in this experimental condition (pentose-containing hydrolysate). According to the biogas profile in Figure 3B, xylose was not utilized for hydrogen production. Rather, this substrate was diverted for the production of soluble metabolites. Overall, in both conditions, the reported efficiency in xylose conversion (Figure 3A,B) was considered high because UAPBR was operated at a short hydraulic retention time (2 h).

Because the pentose liquor derives from sugarcane, it is pertinent to compare the obtained results with those found in the dark fermentation of other sugarcane-derived substrates, focusing on fixed-film reactors. Overall, the VHPR observed in condition II (120 mL·L−1·d−1) exceeded only the value reported by Ferraz Jr. et al. [51], who achieved a mean VHPR of 84 mL·L−1·d−<sup>1</sup> while processing vinasse in an LDPE-filled UAPBR at mesophilic conditions (25 ◦C). Using the same type of substrate and reactor but at thermophilic conditions (55 ◦C), much better results were reported, namely, 526.8 mL·L−1·d−<sup>1</sup> [65], 761.7 mL·L−1·d−<sup>1</sup> [43] and 1604 mL·L−1·d−<sup>1</sup> [44]. Changing the bed conformation, i.e., replacing the random packing by orderly placing the support media further produced even better results, with VHPR reaching 2074 mL·L−1·d−<sup>1</sup> [66] and 3477 mL·L−1·d−<sup>1</sup> [67] while still considering vinasse as the substrate and temperature conditions. Comparing the results obtained herein with the utilization of vinasse is pertinent because both the pentose liquor and vinasse are highly complex materials. The use of molasses, a much simpler carbohydrate-rich sugarcane-derived by-product, led to much higher VHPR values, reaching 4504 mL·L−1·d−<sup>1</sup> [68] and 8479 mL·L−1·d−<sup>1</sup> [69]. Apart from the temperature and support material arrangement differences, in most of the comparative studies, much higher organic loading rate (OLR) levels were used (>50 g COD·L−1·d−1) compared to the one applied in the UAPBR (11.1 g COD·L−1·d−1), which could explain the high discrepancy among the VHPR values. Future studies with pentose liquor should focus on optimizing the OLR, as carried out for vinasse [65,66] and molasses [68].

#### *3.2. Organic Acid Production*

The distribution of soluble metabolites in experimental conditions I and II showed a higher amount of organic acids and solvents in the latter (Figure 4D,E). It is likely that hydrolysate compounds, especially the macro and micronutrients released during the hydrolysis of SCB, stimulated the metabolism of immobilized bacteria toward soluble metabolite production. Lactate was produced in large quantities in both experimental conditions. One plausible explanation for its high concentration is the short HRT (2 h) utilized in the bioreactor operation. The reduced HRT could have led to the accumulation of lactic acid because the reaction time was not sufficient to convert lactate and acetate to butyrate and hydrogen. This hypothesis is supported by the hydrogen profiles (Figure 3A,B) and the metabolic pathway suggested elsewhere [70], as demonstrated in Reactions (2)–(4). The results observed during the dark fermentation of other sugarcane-derived substrates, such as vinasse [20,66,67,71] and molasses [68,69,72], also corroborate this hypothesis because in all cases, lactate was identified as the primary precursor of biohydrogen and butyrate.

Lactate + 0.4Acetate + 0.7H+→0.7Butyrate + 0.6H2 + CO2 + 0.4H2O (ΔG◦ <sup>=</sup> <sup>−</sup>183.9 kJ·mol<sup>−</sup>1) (2)

Lactate + Acetate + H+→Butyrate + 0.8H2 + 1.4CO2 + 0.6H2O (ΔG◦ <sup>=</sup> <sup>−</sup>59.4 kJ·mol<sup>−</sup>1) (3)

$$2\text{Lactate} + \text{H}^+ \rightarrow \text{Butyrate} + 2\text{H}\_2 + 2\text{CO}\_2 \text{ (}\Delta \text{G}^\circ = -64.1 \text{ kJ}\cdot \text{mol}^{-1}\text{)}\tag{4}$$

**Figure 4.** Condition I (xylose-based wastewater): (**A**,**B**) soluble phase products and (**C**) hydrogento-propionate (H2:HPr) and hydrogen-to-ethanol (H2:EtOH) ratios. Condition II (pentose liquor): (**D**,**E**) soluble phase products and (**F**) hydrogen-to-propionate (H2:HPr) and hydrogen-to-ethanol (H2:EtOH) ratios. Nomenclature: HCi—citric acid, HMa—malic acid, HSu—succinic acid, HLa lactic acid, HFo—formic acid, HAc—acetic acid (resulting from fermentation in condition I), HAc Total—total acetic acid (resulting from both bagasse hydrolysate and substrate fermentation in condition II); HAc Bio—acetic acid (resulting from fermentation in condition II), HPr—propionic acid, HBu—butyric acid, HVa—valeric acid, HCa—caproic acid, EtOH—ethanol. Note: HAc Total calculated according to Aguilar et al. [46].

The higher concentration of malate detected in experimental condition II could possibly be linked to the release of malic acid, which is part of the natural composition of plants, including sugarcane and citrus fruits. Malic acid has been previously detected in sugarcane-derived substrates, such as vinasse [73,74]. According to Figure 4D, malate presented a marked decay profile after day 10 because it could be readily metabolized by the acclimatized bacterial consortium. Malate was most likely converted to pyruvic acid, which was integrated into the ethanol fermentation reactions involving decarboxylation to acetaldehyde with a subsequent reduction in alcohol [75]. The increasing profile of ethanol production, as shown in Figure 4D, supports this hypothesis. Malate may also have been converted to lactate following malolactic fermentation [76], which has been previously suggested to occur during vinasse fermentation [77].

Caproate was another relevant soluble metabolite produced in the fermentation of pentose liquor. According to the metabolite profiles shown in Figures 3B and 4D, caproic acid production is likely related to a pathway described elsewhere [78], in which simultaneous hydrogen and caproic acid production was feasible but resulted in low amounts of biohydrogen. These authors reported that caproic acid could also be formed through the

secondary fermentation of ethanol and acetate or ethanol and butyrate. Because acetate was one of the major products found in experimental condition II (Figure 4E), it is likely that acidogenic bacteria used acetate as an alternative terminal acceptor for the reducing equivalents during caproic acid synthesis [79]. Moreover, during the fermentation of pentoses, elevated production of ethanol (Figure 4D) potentially resulted from the abovementioned metabolic pathway. This observation is consistent with the assumption that some caproate-producing bacteria formed a syntrophic relationship with ethanol-producing bacteria [80]. Another factor supporting the existence of caproate-producing bacteria in the mixed culture is the occurrence of spore-forming *Clostridium kluyveri,* which is known to be resistant to the harsh environments and treatments (e.g., highly acidic conditions) used for selecting HPB [78].

Regarding acetate production, the fermentative pathway of xylose leading to this product can occur spontaneously [27], as shown in Reaction (5). In experimental condition I, acetic acid was the major metabolite produced (Figure 4B), thus implying improved hydrogen yields due to coupled reactions in the metabolic pathway, as demonstrated by Reaction (5). However, stable or increasing hydrogen production was not observed (Figure 3A). On the contrary, the biohydrogen generation profile demonstrated a peak followed by continuous decay, which was possibly caused by the increasing dominance of homoacetogenic bacteria (Reaction 1). A marked decrease in the volumetric biogas production indicated that carbon dioxide and hydrogen were diverted to acetate production, as stoichiometrically demonstrated in Reaction (1). In experimental condition II, a mean concentration of acetate equal to 784.55 mg·L−<sup>1</sup> (HAc Total; Figure 4E) was produced; however, a major part of this production was derived from the hydrolysis of acetyl groups bound to hemicellulosic monomers of SCB. Because the protocol described by Aguilar et al. [46] was followed in this study, the calculations pointed to the physicochemical release of 642 mg·L−<sup>1</sup> of acetate and enabled biologically produced acetic acid to be estimated (HAc Bio; Figure 4E), which was 142.55 mg·L−<sup>1</sup> on average. Whereas in this condition, the acetate production was unstable (HAc Bio; Figure 4E), the generation of this metabolite in condition I showed an increasing profile (Figure 4B). This result indicates the occurrence of homoacetogenic reactions in condition I and the limitations of this biochemical route in experimental condition II. The high concentration of acetate derived from pretreatment could have inhibited the homoacetogenic pathway in the reactor fed with the SCB hydrolysate (condition II) because the accumulation of non-dissociated organic acids is deleterious to acidogenic bacteria [81]. However, a systemic inhibition, i.e., affecting all fermentative groups (including HPB), should be expected in this case. Hence, although continuous, the biogas production in condition II was most likely not high enough to stimulate the development of homoacetogens.

$$\text{AlYolose} + 1.67 \text{H}\_2\text{O} \rightarrow \text{Acetate} + 1.67 \text{CO}\_2 + 3.33 \text{H}\_2 \text{ (}\Delta \text{G}^\circ = -195.5 \text{ kJ}\cdot \text{mol}^{-1}\text{)}\tag{5}$$

Butyrate production was detected at concentrations ranging from 60 to 100 mg·L−<sup>1</sup> in condition I and from 133 to 361 mg·L−<sup>1</sup> in condition II. The reasons for these differences are likely related to the increasing utilization of the substrate in condition II (Figure 3B) and the occurrence of different terminal electron acceptors in condition I, as demonstrated by the production of acetate (Figure 4B) and metabolic trends in the formation of propionate and ethanol (Figure 4C). Similar to acetate production, the generation of butyric acid occurred simultaneously with the production of hydrogen; however, the yield was lower [82], as shown in Reaction (6). In experimental condition II, the lower biohydrogen generation was possibly determined by the biochemical yield demonstrated in Reaction (6), which implied half of the biohydrogen yield compared to that obtained when acetic acid was the by-product (Reaction 5). In both experimental conditions, the predominance of acetate and butyrate as major metabolites suggested the initial occurrence of a butyrate-acetate metabolic pathway, in which *Clostridium* sp. is the specific dominant strain. The same observation was made in previous studies utilizing UAPBR systems and natural fermentation for inoculation procedures [41,42,52]. In the study by Peixoto et al. [41], 16S rRNA sequencing

analysis showed a 91% similarity to *Clostridium* sp, which comprised the butyrate producers *C. acetobutylicum*, *C. tyrobutyricum,* and *C. beijerinckii*.

$$\text{Oxose} \rightarrow 0.83\\\text{Butyrate} + 1.67\\\text{CO}\_2 + 1.67\\\text{H}\_2\text{ (}\Delta \text{G}^\circ = -239.9 \text{ kJ}\cdot \text{mol}^{-1}\text{)}\tag{6}$$

The propionate concentration in condition I increased from 50 to 105 mg·L−1, with a mean production of 66.3 mg·L−1. In condition II, propionate was also detected at an increment of 230 mg·L<sup>−</sup>1, which resulted in an increase from 112 to 342 mg·L<sup>−</sup>1. Although this increase is interesting due to the market value of propionic acid [21], the generation of this metabolite in biohydrogen-producing reactors results in a lower yield. This hydrogenconsuming pathway [13,27,34] is demonstrated by Reaction (7), usually taking place when high hydrogen partial pressures are established in the reactors. One of the factors influencing the instability of biohydrogen production in condition II probably included the peaks of propionate concentration on days 12, 23, and 33 (Figure 4E). This hypothesis was valid because the hydrogen concentration had a simultaneous decrease when these propionate peaks were detected. On the other hand, the decrease in biohydrogen production in condition I could more likely be explained by its conversion into acetate in the homoacetogenic pathway because the average production of propionate was quite low (66 mg·L<sup>−</sup>1) compared to that observed in experimental condition II (228 mg·L<sup>−</sup>1). In spite of the elevated production of propionic acid in condition II, the H2/Propionate ratio (Figure 4F) showed an increase in its trend because hydrogen production did not cease. On the contrary, in condition I, the H2/Propionate ratio was close to 0 by the end of the reactor operation due to the interruption of hydrogen production despite the production of acetate (Figure 4B).

$$\text{Xylene} + 1.67\text{H}\_2 \rightarrow 1.67\text{Propionate} + 1.67\text{H}\_2\text{O} \text{ (}\Delta\text{G}^\circ = -315.1\text{ kJ}\cdot\text{mol}^{-1}\text{)}\tag{7}$$

In addition to the production of hydrogen and volatile organic acids, a relatively low amount of ethanol was generated from the xylose media and SCB-derived pentose liquor. The results in Figure 4A,B show the mean ethanol productions of 43 and 110 mg·L−<sup>1</sup> in conditions I and II, respectively. This decrease in the trend of the H2/Ethanol ratio in condition I occurred due to the progressive termination of biohydrogen production (Figure 3A); however, in experimental condition II, the profile of the H2/Ethanol ratio (Figure 4F) was different due to the opposite dynamics in biohydrogen production.

According to the mass flow rate shown in Table 2, the average missing equivalents in experimental condition I was lower than 2%, implying a 98% correspondence between the calculations and experimental data. One explanation for this unbalance could be the presence of the bacteria *Klebsiella* sp., which was previously identified in acidogenic inocula obtained from carbohydrate natural fermentation [41,83]. This microbial group was involved in the production of ethanol and 2,3-butanediol: a metabolite that was not monitored by the analytical methods used in this study. This metabolite was produced under limited oxygen and low pH conditions [84], similar to those utilized in the experiment presented herein. In experimental condition II, the sum of organic metabolites exceeded the mass balance in 207.24 mg·h<sup>−</sup>1, causing an approximate lack of correspondence at 11%. In this case, it is not likely that the formation of undetected metabolites or the presence of alternative electron sinks accounted for the unbalance. Rather, the other carbon sources that were not monitored, such as glucose (1.58%), galactose (2.81%), rhamnose (6.14%), and arabinose (7.12%), were probably fermented and led to the formation of their own metabolites besides those produced with xylose (82.4%).


**Table 2.** Mass flowrate in experimental conditions I and II. Values in parentheses correspond to the day in which each maximum value was obtained.

<sup>1</sup> Resulting from substrate fermentation. <sup>2</sup> Corresponds to the ratio between the sum of metabolites and the influent xylose. Legend: ND—not detected, NC—not calculated (the sum of metabolites and their correspondence were not calculated for maximum values because they were observed in different days).

According to Table 3, the production of the main volatile organic acids was comparable to other studies reported in the literature, while H2 production was significantly lower. In the studies by Wu et al. [36] and Lin et al. [34], 20 g·L−<sup>1</sup> of xylose was used as the initial substrate concentration, which was more than ten-fold higher than the sugar concentrations used to feed the packed-bed reactor in this study. The effect of the temperature should also be considered because other studies with an equivalent substrate concentration reported hydrogen yields higher than those obtained in this study [26,32,34,36]. Lin et al. [34] demonstrated that increasing temperatures (from 30 to 50 ◦C) promoted higher hydrogen yields and production rates. In their study, temperatures of 30, 35, 40, 45, and 50 ◦C corresponded to 0.4, 0.5, 0.3, 0.8, and 1.3 mol H2·mol−1xylose, respectively. Thus, it is likely that the low biohydrogen production in the UAPBR was greatly influenced by the initial substrate's concentration (or applied OLR, as discussed in Section 3.1) and operation temperature. However, the production of both butyric and propionic acids (Table 3) in the UAPBR was greater than that reported by Wu et al. [36] and Pattra et al. [26].

#### *3.3. Preliminary Analysis of Biohydrogen, VOA and Cogeneration Potential in a Pentose Liquor-Based Biorefinery*

The experimental results provided the main parameters with which to develop and analyze a biorefinery performance in four independent scenarios, as detailed in Section 2.5 and depicted in Figure 5. According to Table 4, a comparison between Scenarios 1 and 2 suggested that exclusively producing 1G and 2G of ethanol (Scenario 2) was more profitable than using 50% of SCB for cogeneration and 50% for 2G ethanol production (Scenario 1) due to fuel [85] and electricity prices [86]. In the abovementioned cases, the economic potential per day of a sugarcane mill operation was 19.6% (Scenario 2) and 26.4% (Scenario 1), which is lower than that reported in Scenario 3. Null environmentally friendly potential was found in Scenarios 1 and 2 because pentose liquor derived from the SCB pretreatment was

not used as a raw material for biohydrogen, VOA, or methane production and could result in the accumulation of 16,800 and 33,600 m3·d−<sup>1</sup> of 2G ethanol byproducts, respectively. In Scenario 3, the greatest economic potential was achieved because the extraction of valueadded VOA was considered. Among these acids, acetate is used as a precursor to industrial polymers, including mainly the vinyl acetate monomer [87], butyrate is used in industrial thermoplastics in the form of cellulose acetate butyrate [23], propionate is utilized as a food preservative [88], and lactate is involved in the production of biodegradable polymers, also known as polylactides [24]. In addition, the separation of these organic acids might have a low cost if performed by an ion exchange resin, as demonstrated elsewhere [16]. In terms of energy production, Scenario 4 provided relevant energetic potential; however, organic acids could not be directly recovered. Rather, they were considered substrates for methane production [89]. In this scenario, methane was considered the raw material for on-site cogeneration because this application did not require purification and also allowed local carbon offset projects with a cogeneration price (64.70 USD·MWh<sup>−</sup>1) that is more attractive than a commodity price of 2.77 USD per 103 ft3 [90]. If the energetic potential depended on burning hydrogen alone, the maximum energy production would be 0.73 MWh, which is not significant for achieving profitability (Scenarios 3 and 4). However, hydrogen must be considered as an available surplus of this process because its production is obligatorily coupled to that of VOA (Reactions 5–6). An on-site application for biohydrogen could be the complete fueling of 56 hydrogen fuel cell vehicles per day with driving ranges of 240 miles [91], which is sufficient for the plant's complete fleet of cars. According to Table 4, the major advantage of Scenarios 3 and 4 is that in the production of organic acids or methane, the environmental impact (assessed as the COD reduction) of the 2G ethanol byproducts could be reduced by 62.7% through the separation/extraction of organic acids (acetate, butyrate, propionate, and lactate) or approximately 75% when employing a process based on sequential hydrogen and methane production [92]. Specifically in Scenario 3, organic matter removal could be further increased through the methanization of the VOA that were not extracted as raw materials for the chemical industry, namely the citric, malic, succinic, formic, valeric, and caproic acids.

**Figure 5.** Second-generation sugarcane biorefinery schemes including potential biotechnological uses for pentose liquor.


**Table 3.** Comparative analysis of systems used for biohydrogen and organic acids production from pentoses.

<sup>1</sup> Terminal soluble metabolite concentrations (measured at the end of the experimental runs). <sup>2</sup> Using pentose liquor. Nomenclature: Temp.—temperature (◦C), HY—hydrogen yield (mol H2·mol<sup>−</sup>1xylose), VHPR—volumetric hydrogen production rate (mL H2·L−1·d−1), VOA—volatile organic acids, HBu—butyric acid, HPr—propionic acid, HAc—acetic acid, CSTR—continuous stirred-tank reactor, AGSB—activated carbon-assisted agitated granular sludge bed, UAPBR—upflow anaerobic packed-bed reactor.

**Table 4.** Energetic, economic and environmental aspects for different second-generation sugarcane biorefinery schemes.


<sup>1</sup> Medium-sized sugarcane distillery (milling capacity = 500 ton·h−1) considering 1G ethanol yield [93]. <sup>2</sup> According to the anhydrous ethanol market price provided by the Center for Advanced Studies on Applied Economics (CEPEA/ESALQ/USP) [85]. <sup>3</sup> Considering the 2G ethanol yield obtained exclusively from the cellulosic fraction [94]. <sup>4</sup> According to the mean hydrogen yield obtained in experimental condition II. <sup>5</sup> According to the hydrogen market price reported by the U.S. Department of Energy [95]. <sup>6</sup> According to the mean organic acid yield obtained in experimental condition II. <sup>7</sup> According to HLa, HAc, HPr, and HBu reference prices [96]. <sup>8</sup> Calculated as the surplus electricity of a distillery producing only 1G ethanol [93]. <sup>9</sup> According to the market price defined by alternative energy auctions, as released by the Brazilian Energy Research Office (EPE/MME) [86]. <sup>10</sup> Calculated as the COD of all organic matter minus the COD of the extracted acids (HLa, HAc, HPr, HBu). <sup>11</sup> Considering 350 mL CH4·g−1COD removed [89] and 74.7% methanization efficiency of vinasse [92]. <sup>12</sup> Considering CH4 properties [85] and combined cycle turbines [97]. Nomenclature: 1G— first generation, 2G—second generation, HAc—acetic acid, HBu—butyric acid, HLa—lactic acid, HPr—propionic acid, NA—not applicable.

#### **4. Conclusions**

Biohydrogen and volatile organic acids were produced using both a xylose-based media and sugarcane bagasse-derived pentose liquor. The fermentation of the pretreatmentderived pentose liquor generated a 25% higher volatile organic acid concentration compared to that obtained with the xylose-based media. The differences in the composition of both substrates triggered different fermentative metabolic pathways, leading to the stimulus of homoacetogenesis when using xylose and to the occurrence of biohydrogen

production coupled to acetate and butyrate buildup when using pentose liquor. Overall, higher biohydrogen production rates could be observed when modifying the operating conditions applied in the packed-bed reactor, such as applying high organic loading rates and increasing the temperature to achieve thermophilic conditions. Nevertheless, the continuous operation of the reactor yielded both operational and performance parameters that enabled the simulation of the application of dark fermentation in a medium-sized distillery. The best estimate indicated that recovering hydrogen and organic acids from pentose liquor had the potential to enhance profits in an ethanol production plant (1G and 2G) by 24.3% with a simultaneous environmental impact reduction of 62.7%. Therefore, developing the present fermentation technology seems to be crucial for improving the sustainability of conventional 1G and 2G sugarcane biorefineries.

**Author Contributions:** Conceptualization, D.M.F.L.; methodology, M.A.d.L.; formal analysis, F.V.F.; investigation, W.K.M.; resources, M.A.d.L.; writing—original draft preparation, G.P. and G.M.; writing—review and editing, L.T.F.; visualization, I.P.; supervision, M.Z.; project administration, M.Z.; funding acquisition, M.Z. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the FUNDAÇÃO DE AMPARO À PESQUISA DO ESTADO DE SÃO PAULO, grant numbers 2008/56255-9, 2009/17539-4 and 2009/15984-0. The APC was waived by the journal.

**Data Availability Statement:** The data presented in this study are available on request from the corresponding author.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


**Disclaimer/Publisher's Note:** The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

## *Article* **Rosaceae Nut-Shells as Sustainable Aggregate for Potential Use in Non-Structural Lightweight Concrete**

**Veronica D'Eusanio 1,\*, Lucia Bertacchini 2, Andrea Marchetti 1,3,4, Mattia Mariani 1, Stefano Pastorelli 2, Michele Silvestri <sup>2</sup> and Lorenzo Tassi 1,3,4**


**\*** Correspondence: veronica.deusanio@unimore.it

**Abstract:** Apricot (AS), peach (PS), and plum shells (PlS) were examined as sustainable aggregates for non-structural lightweight concrete. The extraction of natural resources has a significant environmental impact and is not in line with the Sustainable Development Goals (SDGs) of Agenda 2030. Recycling agri-food waste, such as fruit shells, fully respects circular economy principles and SDGs. The chemical and physical properties of the shells were investigated using scanning electron microscopy (SEM) for microstructure analysis and TG-MS-EGA for thermal stress behavior. Two binding mixtures were used to prepare the concrete samples, one containing lime only (mixture "a") and one containing both lime and cement (mixture "b"). Lime is a more sustainable building material but it compromises mechanical strength and durability. The performance of lightweight concrete was determined based on the type of aggregate used. PS had a high-water absorption capacity due to numerous micropores, resulting in lower density (1000–1200 kg/m3), compressive strength (1–4 MPa), and thermal conductivity (0.15–0.20 W/mK) of PS concrete. AS concrete showed the opposite trend (1120–1260 kg/m3; 2.8–7.0 MPa; 0.2–0.4 W/mK) due to AS microporosity-free and denser structure. PlS has intermediate characteristics in terms of porosity, density, and water absorption, resulting in concrete with intermediate characteristics (1050–1240 kg/m3; 1.9–5.2 MPa; 0.15–0.3 W/mK).

**Keywords:** sustainability; green building; recycle; food waste; lightweight concrete; lime concrete; fruit shells; coarse aggregate replacement

#### **1. Introduction**

The climate emergency confronts us with the need to minimize the exploitation of our planet's non-renewable resources. The extraction of raw materials has a dramatic impact on the environment, degrading landscapes, polluting soils and waters, irreparably damaging biodiversity, and inefficiently consuming a huge amount of energy [1]. Furthermore, the extraction of natural resources has accelerated exponentially since the 21st century and has grown significantly globally [2]. Indeed, global population growth, unbridled industrialization, and increased consumption have led to an increase in their demand [3]. The indiscriminate exploitation of non-renewable raw materials leads to their depletion: their cost is expected to increase significantly, and many of them may no longer be available in the near future [4,5]. Instead, renewable materials can be produced.

Ah indefinitely with strong environmental benefits, especially if they are waste byproducts from other supply chains. This is the fundamental principle of the circular economy: someone's waste becomes a valuable resource for someone else. The transition to a circular system provides the opportunity to address this problem by reducing the use

**Citation:** D'Eusanio, V.; Bertacchini, L.; Marchetti, A.; Mariani, M.; Pastorelli, S.; Silvestri, M.; Tassi, L. Rosaceae Nut-Shells as Sustainable Aggregate for Potential Use in Non-Structural Lightweight Concrete. *Waste* **2023**, *1*, 549–568. https:// doi.org/10.3390/waste1020033

Academic Editors: Vassilis Athanasiadis, Dimitris P. Makris and Catherine N. Mulligan

Received: 16 March 2023 Revised: 23 April 2023 Accepted: 25 May 2023 Published: 6 June 2023

**Copyright:** © 2023 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

<sup>42124</sup> Reggio Emilia, Italy

of raw materials, protecting material resources, and reducing the carbon footprint [6,7]. It is also expected to bring economic benefits such as an increase in gross domestic product, net savings in raw materials, growth in employment, and reduced volatility in material and supply prices [8,9].

The building materials sector is a major contributor to environmental deterioration as it is one of the largest exploiters of resources, half of which are non-renewable [10,11]. Thus, global government policies are driving the need to use low-energy and renewable building materials for construction, with the aim of combating climate change and minimizing its effects [12]. The UN Agenda 2030 sets 17 Sustainable Development Goals (SDGs) to end poverty, protect the planet, and improve the lives and prospects of everyone and everywhere [13]. The construction sector is closely associated with several SDGs including clean water and sanitation (SDG 6), affordable clean energy (SDG 7), industry, innovation, and infrastructure (SDG 9), and sustainable cities and communities (SDG 11). In addition, also SGD 12 (ensuring sustainable consumption and production patterns) provides interesting insights. This concerns the substantial reduction of waste through prevention, reduction, recycling, and reuse, which are core principles of the circular economy. In fact, the use of waste or by-products as a substitute for fossil raw materials represents an important opportunity, as it allows for plugging the problems related to waste disposal while reducing the exploitation of the resources of our planet. The construction sector can only be considered truly sustainable when it starts using renewable materials or materials recycled from construction waste and demolition residues [14]. The conservation of natural resources must be maximized and the environmental impact during the entire life cycle of the building project must be reduced. The development of environmentally friendly building materials is an inevitable path to sustainable construction and the SDGs of Agenda 2030.

Cement concrete is undoubtedly the most widely used building material [14,15], with a worldwide production of more than 10 billion tons. The enormous demand for concrete has led to the exploitation of a massive quantity of aggregates, causing their depletion or exhaustion in natural basins, as well as significant environmental issues [14,15]. The utilization of recycled or bio-based materials as a substitute for natural aggregates has been identified as a highly effective strategy to enhance sustainability in the construction industry. Several studies have explored the potential of industrial byproducts, agri-food wastes, or demolition wastes as lightweight aggregates (LWAs). LWAs have a lower average density and higher porosity, providing concrete with lower density and thermal and acoustic insulation properties. Lightweight concrete has a dry density of up to 2000 kg/m<sup>3</sup> and a thermal conductivity usually lower than 1 W/(m· ◦C). Therefore, it is used when low weight and insulating properties are relevant. The thermal conductivity of concrete refers to its ability to transfer heat. High thermal conductivity can lead to unwanted energy loss, which can increase energy costs and reduce the comfort levels of indoor spaces. In contrast, low thermal conductivity in concrete can promote energy efficiency and thermal comfort. One way to improve the thermal conductivity of concrete is by adding lightweight aggregates, which typically have good insulation properties. For example, Real et al. [16] reported that the use of lightweight concrete in buildings can reduce heating energy by 15% compared to normal-weight concrete.

Although lime concrete is an old material used in civil engineering [17–19], only a few studies have investigated its potential as an alternative to Portland cement concrete for structural components [20,21]. Lime has some environmental advantages over concretebased materials: it requires less energy for its production, since limestone, the basic raw material, can be burned at lower temperatures (900–1000 ◦C), whereas silicate rocks for concrete require at least 1300 ◦C. Furthermore, part of the CO2 generated during the production process is reabsorbed by hardened lime [22]. Therefore, the main aim of this study was to develop lime-based non-structural lightweight concrete using waste materials from the agri-food chain as coarse aggregates. In particular, *Rosaceae* fruit shells are a widely available waste, as a significant part of the harvested fruit is processed, resulting in a huge amount of waste kernels [23].

Peach (*Prunus persica* L.), apricot (*Prunus armeniaca* L.), and plum (*Prunus domestica* L.), belonging to the same *Prunus* family, are widely cultivated fruits [24] above all for their relevance to human health [25–29] as an important source of phenolic compounds, cyanogenic glucosides, vitamins, mineral salts, and phytoestrogens. According to the "Food and Agriculture Organization (FAO)", in 2020, global peach production was approximately 24 million tons, apricot one 3.7 million tons, and plum one 12 million tons. Their pulp is still the only part that is most appreciated and used by agro-industries, whereas pits are considered low-value agro-industrial residues. In addition to being directly consumed as fresh fruit, most Rosaceae fruits are processed into juices, canned fruits, jams, and sweet snacks. All these productive sectors give rise to a large quantity of kernels as waste, estimated at around 10% of the total mass [30]. Thanks to their high calorific value, the current alternative to landfilling fruit shells is incineration in biomass heating systems. However, this seasonal activity requires temporary kernel storage in large open-air stacks. This leads to some issues such as space availability, environmental hygiene problems, and the development of odorous exhalations due to uncontrolled fermentation of pulp residues and decomposition of organic material [31]. A second and more important factor concerns the serious environmental effects caused by the incineration of these materials. It is estimated that the combustion of agricultural residues, such as wood, leaves, trees, and grass, generates approximately 40% of CO2 emissions, 32% of CO emissions, 20% of particulate matter, and 50% of polycyclic aromatic hydrocarbons (HAPs) [32]. Fruit shells exhibit several characteristics that make them an interesting alternative to common coarse aggregates. For example, their degradation under natural conditions is difficult and slow [33,34], unlike other food waste by-products. Moreover, it is a widely available and low-cost waste material, and its reuse and valorization are perfectly aligned with Agenda 2030 for Sustainable Development [13] and, in particular, with SGD 12.

Several studies have reported the use of agricultural waste materials, such as oil palm shells, palm oil clinkers, wood, mussel shells, date seeds, and coconut shells [35–38], for the production of environmentally friendly concrete. Some studies have used fruit shells to prepare cement-based concrete [39–41], but there is little evidence of lime-based materials. The effects induced on the physical and mechanical properties of lime-based concrete of peach, apricot, and plum shells were investigated, considering density, compressive strength, and thermal conductivity. In this preliminary study, the goal was to evaluate the potential of these materials as LWAs and identify any differences between the different shell types. The compositional and morphological characteristics of the aggregates were evaluated and correlated with the performance of lightweight concrete. Scanning electron microscopy (SEM) was used for the morphological study of the LWAs, and TGA-MS-EGA was used to obtain compositional information and study their behavior under thermal stress.

#### **2. Materials and Methods**

#### *2.1. Raw Materials Properties and Specimens Preparation*

#### 2.1.1. Binder Mixture

The main binder used was hydrated lime (Litokol S.p.A., Rubiera, Italy). To improve the mechanical properties, a few sets of specimens were prepared with the addition of Type I 52.5 grade Portland cement (Litokol S.p.A., Rubiera, Italy). The physical and chemical properties of the binders are listed in Table 1.

**Table 1.** Physical and chemical properties of the binders.



2.1.2. Coarse and Fine Aggregates

Crushed shells were used as alternative coarse aggregates (Figure 1). Peaches, apricots, and plums were obtained from a local orchard in Modena, Italy. The pulp was separated from the pits, cleaned before use, and the residual dried pulp and dust on their surfaces were removed. The pits were preliminarily air dried for 30 days to remove residual moisture. Preliminary coarse grinding allowed the separation of the internal kernel from the external shell. The dried pits were crushed with a crushing machine and sieved with 4.5 and 9.5 mm sieves. Natural alluvial silica sand was used as the fine aggregate (Litokol S.p.A., Rubiera, Italy). The physical properties of the aggregates are listed in Table 2, and the proximate chemical compositions of the fruit shells are listed in Table 3.

**Figure 1.** Crushed peach shells (PS, sx), crushed apricot shells (AS, centre), and crushed plum shells (PlS, sx).


PS = Peach Shells; AS = Apricot Shells; PlS = Plum Shells.


**Table 3.** Proximal chemical composition of nutshells from different *Rosaceae* fruits.

\* Data taken from the literature [42–44]; PS = Peach Shells; AS = Apricot Shells; PlS = Plum Shells.

The methods recommended by the Association of Official Analytical Chemists [45] were used to determine the levels of moisture, ash, crude protein, and residual oil. Moisture content was determined by drying the samples at 105 ◦C to a constant weight. The ash content was determined using a laboratory furnace at 550 ◦C and the temperature was gradually increased. Nitrogen content was determined using the Dumas method and converted to protein content by multiplying by a factor of 6.25. The residual fat fraction was recovered using the Soxhlet method, exhaustively extracting 10 g of each sample using petroleum ether (boiling point range 40–60 ◦C) as the extractant solvent. Each measurement was performed in triplicate, and the results were averaged.

Finally, we emphasize that the chemical composition and physical properties of vegetable matrices are significantly influenced by certain factors, including the geographical origin, degree of ripeness, and cultivar to which they belong [46].

#### 2.1.3. Lime-Concrete Design and Specimen Preparation

Normal tap water was used in this study. The mix proportions of all specimens are listed in Table 4. For each shell, the mix proportion of the related concrete was kept constant (PSC = Peach Shell Concrete; ASC = Apricot Shell Concrete; PlSC = Plum Shell Concrete). Specimens were removed from the mold after 24 h. They were stored in a laboratory room with a relative humidity of 95 ± 5% and a temperature of 20 ± 2 ◦C until the test age. Binder mixture a (PSC\_a, ASC\_a, PlSC\_a) only includes lime, while mixture b (PSC\_b, ASC\_b, PlSC\_b) involves the addition of cement. Three sets of specimens were prepared: one for the compressive strength test, one for the demolded, air-dry, and oven-dry density evaluation, and one for the thermal conductivity test. Each set contained three cubic specimens (100 × <sup>100</sup> × 100 mm3), and the average values were obtained for each test result.


**Table 4.** Mix proportions of concrete (kg/m3).

\* water-binder ratio.

The specimens were prepared as follows: river sand, lime, and cement were poured into a blender and dry-mixed for 1 min. Water was added and the mixture was mixed for 3 min. The lightweight aggregates were finally added to and mixed for 5 min. After mixing, fresh mixtures were then poured into the mold and compacted. The specimens were placed in the laboratory room and were removed from the molds after approximately 24 h.

#### *2.2. Experimental Methods*

#### 2.2.1. Morphological Analysis of the Aggregates

The field emission scanning electron microscope (SEM) instrument (Nova NanoSEM 450, FEI, Hillsboro, OR, USA) was used to evaluate the microscopic morphology of coarse lightweight aggregates.

#### 2.2.2. TGA-MSEGA

A Seiko SSC 5200 thermal analyzer (Seiko Instruments Inc., Chiba, Japan) was used to perform the thermogravimetric analysis (TGA) in an inert atmosphere. A coupled quadrupole mass spectrometer (ESS, GeneSys Quadstar 422) was used to analyze the gases released during the thermal reactions (MS-EGA) (ESS Ltd., Cheshire, UK). Sampling was performed using an inert and fused silicon capillary system, which was heated to prevent condensation. The intensity of the signal of selected target gases was collected in multiple ion detection mode (MID); a secondary electron multiplier operating at 900 V was collected in multiple ion detection mode (MID), the intensity of the signal of selected target gases. The signal intensities of *m*/*z* ratios of 18 for H2O, 44 for CO2, 60 for C2H4O2 (acetic acid), 94 for C6H6O (phenol), 39 for C3H3 <sup>+</sup> (furfural fragment), 96 for C5H4O2 (furfural), and 151 for C8H8O3 (vanillin) were measured, where *m*/*z* is the ratio between the mass number and charge of the ion. The heating conditions were 20 ◦C/min in the thermal range of 25–1000 ◦C using ultrapure He at a flow rate of 100 μL/min as the purging gas.

#### 2.2.3. Demolded, Air-Dry, and Oven-Dry Densities

Demodeld, air-dry, and oven-dry densities were determined following ASTM C567 [47]. The demolded mass was measured after demolding (after 24 h of curing), and the air-dry mass was measured after 28 days of curing. The test method for oven-dry density is more complex. The specimens were immersed in water (at about 20 ◦C) for 48 h, then the surface water was removed, and the saturated surface-dry mass was measured. Then, it was suspended in water with a wire, and the apparent mass of the suspended-immersed specimens was determined. The samples were then oven-dried at 110 ◦C for 72 h. The oven-dry density was calculated from Equation (1):

$$O\_m = \frac{D \times 997}{F - G} \tag{1}$$

where *Om* is the measured oven-dry density (kg/m3); *D* is the specimen mass (kg); *F* is the mass of saturated surface-dry specimen (kg); *G* is the apparent mass of suspendedimmersed specimen (kg).

#### 2.2.4. Mechanical Test

The compressive strength test was performed after 28 and 56 days using a Technotest compression test machine (Technotest, Modena, Italy). The average value of at least three specimens was used as the test result. It was performed in conformity with the European standard for structural concrete (EN 12390-3:2009), although our concrete had no structural purpose. Lime mortar (EN 1015-11:1999) would be more suitable for the intended use, but the presence of coarse aggregates prevents its application.

#### 2.2.5. Thermal Conductivity of Lime-Concrete Specimens

A KD2 Pro thermal properties analyzer (Decagon Inc., Pullman, WA 99163, USA) was used for thermal conductivity measurements. It is a portable device fully compliant with ASTM D5334-08 and is used to measure the thermal properties of materials based on probe/sensor methods (transient line heat source), as confirmed by Decagon Devices Inc. Operator Manual version 11. It consists of a portable controller and sensors probe to

be inserted into the medium to be measured. The measurement consists of heating the probe for a certain time and monitoring the temperature during heating and cooling. The influence of the ambient temperature on the samples must be minimized to obtain more accurate values. The measurement range of thermal conductivity of KD2 Pro is 0.02 to 2.00 W/(mK). In this study, three cubic specimens for each sample (100 × <sup>100</sup> × 100 mm3), at 28 days of curing were selected to measure thermal conductivity at dry conditions. The samples were oven-dried for 24 h at 100 ◦C prior to testing.

#### **3. Results and Discussion**

#### *3.1. Lightweight Aggregates*

LWAs, their nature, and compositional characteristics critically define the physical and chemical properties of concrete [48–50]. Depending on the type of aggregate used, the uses and functions of the final product change drastically [51]. Therefore, defining the compositional characteristics of our fruit shells is fundamental to understanding their potential as LWAs.

Bulk density is one of the most important characteristics [17] because it significantly affects the final density of concrete, which, in turn, determines its mechanical and insulation properties. This depends on the size and shape of the aggregates, the moisture content, and the porosity. PS had the lowest bulk density, whereas AS had the highest density. The SEM analysis (Section 3.2) highlights the marked morphological differences, which fully explains these density trends.

The water absorption of lightweight aggregates is generally significantly higher than that of conventional coarse aggregates. This is certainly due to the greater porosity but also to the different chemical compositions. In particular, fruit shells, lignocellulosic biomass, are composed mainly of the three main natural polymers, cellulose, lignin, and hemicellulose [52,53], and the content of these components in the examined matrices are collected in Table 2. Unlike lignin, which is highly hydrophobic, cellulose has a marked ability to absorb water. The latter acts as the glue that connects cellulose and hemicellulose [54]. In fact, in the cell wall of certain biomasses, especially wood species, lignin functions to cement cellulose fiber [55]. It is a three-dimensional strongly cross-linked macromolecule. Cellulose, on the other hand, differs markedly, as it is a linearly structured homopolymer, and in plants, it plays a fundamental role as a supporting matrix for the cell membrane. Hemicellulose is a heterogeneous, completely amorphous, weak polymer. Hemicellulose is decidedly more soluble and labile [53]. From Table 3, it can be seen that PS showed greater water absorption than AS and PlS. This phenomenon can certainly be attributed to the increased porosity (as will be explained later in Section 3.2). However, it is possible to draw conclusions by analyzing the chemical composition of the shells. In fact, the lignin content of PS was the lowest compared to that of AS and PlS, while that of cellulose and hemicellulose was higher. Considering the greater affinity of the latter towards the water and the hydrophobicity of lignin, the greater water absorption can be easily explained. Water absorption affects some important properties of concrete such as its strength, density, and time-dependent deformation [17,48].

#### *3.2. Morphological Analysis of the Aggregates*

The surfaces of peach shells (PS), apricot shells (AS), and plum shells (PlS) are shown in Figures 2–4, respectively. We reported only images relating to the external surface of the shells, in which we identified the most significant differences between PS, AS, and PlS. The internal one, in fact, was extremely smooth and compact for all three types of shells. Therefore, it seemed more important to pay attention only to the outer part of the shells, since the outer surface layers probably contribute more to the properties of the aggregates and, consequently, to the behavior of the specimens.

**Figure 2.** SEM images of crushed peach shells (PS) at different enlargements.

**Figure 3.** SEM images of crushed apricot shells (AS) at ×500 (**left**) and ×1000 magnification (**right**).

The surfaces of all shells appear rough, irregular, and have many cavities. The PS and AS cavities are ovoidal shaped, where the grater diameter is approximately 50 μm and the smaller one is about half, 25 μm. The PlS cavities, on the other hand, are much more irregular and lack a specific shape. The most evident observation concerns the presence of microporosity inside the cavities, which are only present in PS and PlS. The size of the microporosities was approximately 2.0 μm, which can be better viewed in Figure 5.

**Figure 4.** SEM images of crushed plum shells (PlS) at ×500 (**left**) and ×1000 magnification (**right**).

**Figure 5.** SEM images of the micropores of PS (**left**) and PlS (**right**), 4000× magnification.

The PlS micropores were denser, although some were not completely empty. It is likely that a small fraction of the fruit pulp remained trapped inside the micropores, which made it difficult to remove by simple mechanical separation. The SEM observations can be correlated with the data shown in Tables 2 and 3. The greater AS bulk density was justified by the absence of porosity and the consequent greater compactness. The PlS bulk density was higher than that of PS, which could be due to the more superficial and shallower porosities. The water absorption (24 h, %) data were also in line. PS has the highest value, indicating greater trapping of water inside the pores, AS has the lowest value, and PlS has an intermediate value.

#### *3.3. TG-MS-EGA Analysis*

Fruit shells are primarily composed of lignocellulosic material. TG-MS-EGA analysis allows us to obtain information about the different degradation processes involving all constituents, which occur in defined thermal ranges identifiable in the thermogram. Materials with complex compositions give rise to different degradative reactions that can occur simultaneously, and the thermogram profile is the sum of the various contributions. In these cases, deconvolution and interpretation of the signals are not particularly easy, especially if different processes lead to the formation of the same reaction products, such as H2O and CO2. For effective interpretation of thermograms, the entire temperature range is usually divided into thermal regions of different sizes and characteristics, as shown in Table 5. Furthermore, Table 5 shows some thermal windows or subdomains of the regions where particular deformations of the TG/DTG profiles are observed, corresponding to specific behaviors due to some degradation processes of the studied samples.

**Table 5.** Representative values of TG/DTG profiles of Figures 6–8 obtained in inert atmosphere (He).


**Figure 6.** TG (black line) and DTG (grey line) curves of AS sample at heating rate of 20 ◦C/min in He atmosphere. Vertical dashed lines delimit the five thermal regions (I–V) described in the text. For the meaning of the numbers in parentheses, see Table 6.


**Table 6.** Representative values of TG/DTG profiles of Figure 6 (AS sample), obtained in inert atmosphere (He).

To = onset temperature (beginning of thermal step processes); Tm = maximum temperature for the largest mass loss rate; Tc = conclusion temperature (end of thermal step processes).

**Figure 7.** TG (black line) and DTG (grey line) curves of PlS sample at heating rate of 20 ◦C/min in He atmosphere. Vertical dashed lines delimit the five thermal regions (I–V) described in the text. For the meaning of the numbers in parentheses, see Table 7.

**Table 7.** Representative values of TG/DTG profiles of Figure 7 (PlS sample) obtained in inert atmosphere (He).


To = onset temperature (beginning of thermal step processes); Tm = maximum temperature for the largest mass loss rate; Tc = conclusion temperature (end of thermal step processes).

**Figure 8.** TG (black line) and DTG (grey line) curves of PS sample at heating rate of 20 ◦C/min in He atmosphere. Vertical dashed lines delimit the five thermal regions (I–V) described in the text. For the meaning of the numbers in parentheses, see Table 8.

**Table 8.** Representative values of TG/DTG profiles of Figure 8 (PS sample) obtained in inert atmosphere (He).


To = onset temperature (beginning of thermal step processes); Tm = maximum temperature for the largest mass loss rate; Tc = conclusion temperature (end of thermal step processes).

Table 3 shows that ~90% of AS, PS, and PlS consisted of cellulose, hemicellulose, and lignin. Therefore, the TG-MS-EGA analysis showed the thermal steps leading to the degradation of these three fractions. For this reason, the evolution of some analytes characteristic of the degradation of hemicellulose and cellulose (i.e., acetic acid, *m*/*z* = 60; furfural, *m*/*z* = 96 and 39 for its fragment C3H3 +) and lignin (i.e., phenol, *m*/*z* = 98; and vanillin *m*/*z* = 151), was evaluated [56]. This analysis allows for better differentiation of the thermal processes and better identification of the process temperature range that involves every component of the matrix. Unfortunately, no significant results have been obtained regarding the evolution of the emitted gases phenol and vanillin, which will not be reported below.

The result of the TG, together with its first derivative (DTG), runs in inert atmosphere (He) as shown in Figures 6–8, and the related quantitative considerations are summarized in Table 5. For each thermogram, a summary table is provided (Tables 6–8), that collects the representative values of the TG/DTG profile.

The thermograms are divided into five regions (I, II, III, IV, and V), each representing the behavior of the samples following specific processes. Before examining the TG/DTG curves relating to the three nut-shell samples, it must be emphasized that the separation

limits of the various regions are not rigidly identified with values of T/◦C in an absolute way. Conversely, these are limits with a mobile index because the material fractions that undergo thermally activated processes can produce experimental evidence typical of the sample rather than of the type of matrix. Therefore, any attempt to generalize the thermal intervals of each region could lead to a speculative investigation, which is incompatible with the inspiring criteria of this study.

Region I, which covers a temperature range up to ~120 ◦C, is attributed to the removal of moisture and particularly volatile organic compounds (VOCs). The mass loss in this thermal step was greater for PS (Δm% = −4.2%), followed by PlS (Δm% = −3.8%) and AS (Δm% = −2.8%). Within this region, other thermally activated processes occur without mass loss, such as protein denaturation by unfolding [57,58].

Region II, which covers the temperature range from ~120 ◦C to ~215 ◦C, concerns the mass loss related to bound water, i.e., water typically retained by the inorganic fraction, such as the crystallization of mineral salt water. In this region, semi-volatile compounds with medium-low vapor pressure (SVOCs), which are present in the initial matrix or formed during the heating phase, are completely removed. The removal of structural water begins at 160 ◦C, following condensation reactions of the -OH groups mainly present in simple non-cellulosic carbohydrates [59]. The formation and removal of the reaction water passes through the entire thermogram up to and including region IV. Therefore, near the upper-temperature limit (~180 ◦C), free amino acids begin to undergo thermal degradation [60], while proteins persist up to ~200–220 ◦C. Thus, the processes occurring in this region suggest that the chemical structure of the biomass begins to destabilize, partly depolymerize, and plasticize.

Region III, in the temperature range from ~215 ◦C to ~423 ◦C, is the main pyrolysis window, where structural decay reactions of proteins (~240 ◦C), hemicellulose (~300 ◦C) [61,62], and cellulose (~370 ◦C) [57,61,63] take place. This was confirmed by the emission of acetic acid and furfural in this thermal window, as shown in Figure 9. These analytes are formed by the thermal degradation of cellulose and hemicellulose, as reported in several studies [56,64,65].

**Figure 9.** Evolution trend of H2O, CO2, furfural fragment C3H3 <sup>+</sup> (*m*/*z* = 39), and acetic acid (*m*/*z* = 60) during the heating of PS sample; for ease of comparison the DTG curve is also shown. Intensity of *m*/*z* is in arbitrary units.

Region IV begins at ~413 ◦C and extends up to ~695 ◦C. In this thermal window, the gradual mass decrease is mainly due to the slow pyrolysis of the lignin fraction [66], which is associated with the sample vitrification and volatilization of carbon microparticles. The additional small mass loss can be attributed to the thermal decomposition of carbonaceous matter (biochar), which is mostly related to the hemicellulose fraction [67], although lignin components can also degrade [68].

In region V, above ~700 ◦C, up to the final temperature (1000 ◦C), the last residues of biomass degradation can be observed. This is the typical carbon pyrolysis window with the thermal decomposition of low volatile matter such as carbon fragments C20–C40 in the presence of mineral ash. This thermal process was confirmed by the evolution profile of CO2 (Figure 9), where an increase in the signal occurred between 800 ◦C and 1000 ◦C. The TG/DTG profiles of PS, AS, and PlS are typical of lignocellulosic raw materials, highlighting the contents of hemicellulose, cellulose, and lignin. This observation was confirmed by the proximate composition analysis (Section 2.1.2, Table 3).

As an example, the evolution of gases is reported below (Figure 9) for only one sample (PS), as the results were almost similar for the three matrices.

#### *3.4. Density and Compressive Strength of the Lime-Based Concrete*

Lightweight concrete can be classified according to its density, which normally ranges from 320 to 1920 kg/m3 according to the ACI Committee 213 Guide for Structural Lightweight Aggregate Concrete [69]. The classification of concrete according to density provides three groups of materials: (i) low-density concretes (300–800 kg/m3); (ii) moderate-strength concretes (800–1350 kg/m3); (iii) structural concretes (1350–1920 kg/m3) [70]. These three classes are also associated with specific strength range: 0.7–2.0 MPa, 7–14 MPa, and 17–63 MPa, respectively [70]. Density is one of the most important variables in concrete design, as compressive strength depends on it [71]. The reduction of concrete density implies the increment of its porosity, which is achieved by the direct replacement of normal-weight aggregates with LWAs.

The density data obtained are collected in Table 9.



\* Data are expressed as mean ± SD.

All the samples are in the density range relating to moderate-strength lightweight concretes. As expected, the PSC has lower density values, as PS has a lower bulk density and specific gravity than AS and PlS. As observed by the SEM analysis, PS has a high porosity, AS low, and PlS intermediate between the two. The porosity of an aggregate significantly defines concrete density, as it affects both the porosity of the concrete itself and allows it to trap more air inside. Furthermore, some studies reported that the aggregate shape affects concrete density. In fact, the flaky shape easily traps air inside the concrete, increasing its porosity and consequently reducing its density. This phenomenon has been observed in concrete containing seashells [72], peach shells [38], and recycled polyolefin waste [73]. In particular, irregular shapes hinder the complete compaction of concrete, thus contributing to higher air content. In addition to this, there is also trapped air due to the high porosity of LWAs. Furthermore, in these studies, it is reported that the organic matter content is also able to increase the air inclusion in the concrete. Moreover, the extremely

irregular shape of the aggregates leads to a difficult compaction of concrete, which leads inevitably to an increase in the occluded air [35]. This decrease in density, however, involves the reduction of the compressive strength [74], as explained below.

It is also important to underline that the use of lime as a binder allows it to obtain lower density values when compared with specimens prepared with similar lightweight aggregates [36–39], because of its lower specific gravity and bulk density. This is confirmed by the higher density values observed in concrete-containing specimens (Table 5).

The results of the compressive strength tests at 28 and 56 days for the concrete specimens are shown in Table 10.


**Table 10.** Compressive strength at 28-day and 56-day.

\* Data are expressed as mean ± SD.

The compressive strength of the specimens prepared with the cement-free mixture is less than 3 MPa. This value is too low to allow the material to fall into the category of moderate-strength concretes. At the same time, the density value of these specimens is too high for them to be considered "low-densities concretes". However, not falling into a specific category does not preclude possible applications. For example, non-structural mortar beds for wooden floors have larger density allowances (1400–1600 kg/m3) and low strength requirements [17]. Compressive strength between 1 and 2 MPa is recommended in this case. This is a clear example of how depending on the specific application to be assigned to a material, specific ranges of density, and compressive strength are required. Specimens prepared with the cement-containing mixture "b" fall perfectly into the category "moderate strength concrete", as they have a density between 800 and 1350 kg/m3 and a compressive strength exceeding 3.4 MPa. The addition of cement, even if in a small percentage, entails a significant improvement in mechanical properties and little compromises the density value, slightly greater than lime-based concrete. Moderate strength lightweight concrete is a versatile material that can be used for various purposes in construction. One of its most useful applications is as a non-structural filler for thermal and acoustic insulation. Non-structural infills are materials that do not bear any significant weight or load in a building but provide important functions such as insulation. One benefit of lightweight moderate-strength concrete is its relative ease of installation and transport, owing to its relatively low weight. This can be particularly advantageous in situations where access to the construction site is limited or there are restrictions on the use of heavy machinery.

Several studies in the literature demonstrated that the compressive strength of concrete is mainly affected by the properties and volume content of aggregate [75,76]. LWAs are also relatively weak if compared with normal-weight coarse aggregates, and their strength is an additional limiting factor affecting concrete strength [77]. As previously mentioned, the most important characteristic of LWAs is its internal porosity, which results in a lower density and higher water absorption. These factors adversely affect the compressive strength and making concrete less compact and porous. In particular, the greater water absorption by aggregates leads to greater porosity of concrete [78]. This results in lower density and lower compressive strength. PS showed increased water absorption, as explained in Section 3.1, due to the increased cellulose content and higher porosity. AS, on the other hand, having greater lignin content, a hydrophobic polymer, and free of surface porosity showed lower water absorption. These observations are in line with the values given in

Table 6: PSC specimens have a lower strength, given the greater porosity of the aggregates. On the contrary, those ASC show the highest values, while PlSC intermediate ones.

#### *3.5. Thermal Insulating Properties*

Thermal conductivity is a fundamental parameter in the design and application of thermal-insulating lightweight concrete. These materials are becoming important in the context of the climate crisis. The development of energy efficiency strategies is increasingly being studied since the design of energy-efficient buildings is crucial for the realization of a sustainable future [79]. Room air conditioning, ventilation, and occupant comfort account for 29 of CO2 emissions from the building sector. By increasing the energy efficiency of buildings, also using thermal insulation materials, it is possible to reduce consumption and reduce the environmental impact and CO2 emissions significantly [80,81].

Several factors affect the thermal properties of concrete: type and content of aggregates, air voids content, pore size distribution and geometry, moisture content, w/b ratio, and types of admixtures [82]. In particular, the thermal conductivity of conventional building materials is inversely proportional to the porosity ratio. This trend is due to the relatively low thermal conduction of air (0.025 W/mK at room temperature and free of convection) and the interfaces promoted by the pores. Microstructural characteristics are thus critical factors for the consequent thermal conductivity of concrete.

The results are collected in Table 11.

**Table 11.** Thermal conductivity.


\* Data are expressed as mean of three replicates ± SD.

PSC has a lower thermal conductivity, due to the highly porous structure of the lightweight aggregate and the consequent high porosity of the concrete. Generally, lowcompaction concrete has better thermal insulation properties because more air bubbles are carried into the concrete during the mixing. Consequently, the thermal insulation properties improve with increasing porosity of both the lightweight aggregate and the related concrete. For the same reason, there is a significative correlation between concrete density and its thermal conductivity. Lightweight aggregates change density by forming voids, incorporating more air inside the concrete. AS, being practically porosity-free, is an aggregate that does not involve a significant inclusion of air, and therefore, does not provide a significant decrease in thermal conductivity. The addition of cement (mixture b) results in better compactness of the samples and a consequent worsening of the thermal insulation properties. This trend is in agreement with the observation reported in the literature, which suggests that the thermal conductivity of concrete decreases as its density decreases [83].

Moderate-strength lightweight concrete is known to have a thermal conductivity ranging from 0.2 to 0.6 W/mK [84]. The values in Table 8 indicate that all the concretes obtained fall within this range. Notably, some of these materials, specifically those produced using mixture "a" consisting of PS and PlS, exhibit even lower thermal conductivities, indicating enhanced thermal insulation properties. In general, all the materials obtained, having reduced thermal conductivity, have a marked potential for application as non-structural fillers to improve energy savings in buildings, thus improving environmental sustainability.

#### **4. Conclusions**

The potential of peach, plum, and apricot shells as lightweight aggregates was evaluated. The TG/DTA profile is typical of lignocellulosic material, confirming the proximal analysis. The use of lime as the main binder allowed it to obtain more eco-friendly building materials, both because it is more ecological than cement, and because it gives particularly favorable thermal insulation properties due to its greater porosity. PS, AS and PlS prepared several specimens of non-structural lightweight concrete. The specimens containing only lime as binder had an oven-dry density between 1000 and 1200 kg/m3, a low 28-day compressive strength (<3 MPa), and low thermal conductivity values. PSC had lower conductivity and density values, and this is mainly due to the high porosity of PS highlighted by SEM analysis. ASC instead showed the highest values and is practically free of porosity. PlSC showed intermediate characteristics, which reflects the reduced porosity content of PlS. In addition, PS showed greater water absorption, probably due to the higher content of cellulose. This parameter greatly affects the chemical–physical characteristics of concrete, leading to a worsening of compactness, the reduction of density, the formation of greater voids, and consequently the lowering of thermal conductivity. The addition of cement greatly improves the mechanical properties but negatively affects thermal conductivity.

This study showed that there is a feasibility of application of these agro-industrial wastes, which can, therefore, be reused and valorized, reducing dependence on natural raw materials.

**Author Contributions:** Conceptualization, V.D. and A.M.; methodology, V.D. and M.M.; software, V.D., S.P. and M.S.; validation, L.T., S.P., V.D. and A.M.; formal analysis, V.D.; investigation, V.D., L.B. and M.M.; resources, L.T. and S.P.; data curation, V.D., M.M. and M.S.; writing—original draft preparation, V.D. and L.B.; writing—review and editing, A.M. and L.T.; visualization, V.D.; supervision, L.T. and S.P.; project administration, V.D.; funding acquisition, V.D. and L.T. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Not applicable.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


**Disclaimer/Publisher's Note:** The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.
