Microwave- Versus Conventional Acid-Assisted Thermohydrolysis of Willow and Maize Silage: Methane Yield and the Energy Cost of Pretreatment

Abstract

Hydrothermal pretreatment is widely proposed to improve the anaerobic digestibility of lignocellulosic biomass, yet whether it repays its own energy input is rarely quantified. We compare microwave- and conventional (conductive) acid-assisted thermohydrolysis of willow (Salix viminalis) and maize silage at 110–130 °C, coupling biochemical methane potential (BMP, n = 3) with an incremental energy balance against an untreated control. Substrate, heating mode and temperature were all significant (p < 0.001), with heating mode dominating. Maize silage responded monotonically, reaching 306.0 ± 4.0 NmL CH4 g⁻¹ VS at 130 °C under microwave heating (+31.4%), whereas willow behaved erratically: its best variant reached 310.6 ± 2.5 NmL CH4 g⁻¹ VS (+28.9%), yet two conductively heated variants fell below the control. Furanic by-products stayed at trace levels, well below inhibitory thresholds, excluding toxicity. Critically, no variant recovered its own energy input: every treatment was net-negative, the least unfavourable being microwave heating at 110 °C (−3.57 and −3.48 kJ g⁻¹ DM). Heating mode changed the penalty three- to five-fold, microwave being superior in all six paired comparisons. Under the mild conditions tested, acid-assisted thermohydrolysis was not self-financing in energy terms; where applied for other reasons, microwave heating is the less unfavourable of the two modes tested, although neither returns a net energy gain.

1. Introduction

Anaerobic digestion (AD) is a mature route for converting organic matter into biogas and, after upgrading, biomethane, and it is expected to play a growing role in reaching climate-neutrality targets. It remains one of the most practical routes for recovering energy from low-grade lignocellulosic materials, although its efficiency is constrained by the natural recalcitrance of the plant cell wall [1]. Lignocellulosic feedstocks—both woody short-rotation crops such as willow (Salix spp.) and herbaceous silages such as maize silage—are attractive because of their high organic content and land productivity, but their recalcitrant cell-wall structure limits hydrolysis, which is the rate-limiting step of their anaerobic digestion [2,3]. Reported methane potentials illustrate this constraint: untreated willow typically yields on the order of 198–246 NmL CH₄ g⁻¹ biomass [2], and woody biomass exceeds 0.25 L CH₄ g⁻¹ VS only for the less recalcitrant clones [4], whereas maize silage, being less lignified, reaches roughly 196–402 NmL CH₄ g⁻¹ VS depending on variety, ripeness and ensiling [5,6], with the ensiling regime itself further modulating digestibility and gas yield [7]. Individual lignocellulose components exert distinct effects on methane productivity and on the structure of the methanogenic community, so the contrast between willow and maize silage must be read at the level of structural composition and not merely as an effect of the heating technology [8]; within willow itself, fermentation efficiency is tied closely to woody-tissue traits, particularly lignin content [9]. The chemistry of the pretreatment matters more for willow than its severity alone: aqueous ammonia soaking, an alkaline route, raised the methane yield of willow by 94–162% while solubilising almost no sugars (an 8.9% increase in soluble xylose), against 37–41% for wheat straw and 25–27% for miscanthus [10]. This is consistent with the broader observation that pretreatments directed at hemicellulose removal are poorly suited to biogas production, whereas those that remove lignin deliver the highest methane yields, and that part of the response is mediated by a shift in the microbial community rather than by substrate structure alone [11].

A pretreatment step is, therefore, commonly applied to disrupt the lignocellulosic matrix and accelerate hydrolysis. Current reviews stress that the technological appeal of any pretreatment rests on a trade-off between improved biodegradability on one side and energy cost, reagent consumption and the risk of inhibitor formation on the other [12]. Its effectiveness is, however, strongly substrate-dependent: a recent meta-analysis showed that the methane yield response to pretreatment is governed by the predominant chemical composition of the feedstock, and that mismatched pretreatments can be non-beneficial or even adverse [13]. This makes side-by-side evaluation on chemically contrasting substrates more informative than single-substrate optimisation.

Hydrothermal pretreatment solubilises hemicellulose and increases substrate accessibility, but its benefit is bounded by severity. For willow sawdust specifically, microwave-assisted liquid hot water treatment raised the methane yield by 42.2% at 215 °C and 48.8% at 230 °C, while enhancing hemicellulose solubilisation and leaving lignin and cellulose largely intact—at the cost of rising furan and organic-acid concentrations [14]. The benefit is, thus, bounded from both sides: above a temperature threshold, sugar degradation generates furanic (furfural, 5-hydroxymethylfurfural, HMF) and phenolic by-products [3,15]. For lignocellulosic AD, these compounds inhibit methanogenesis only above comparatively high concentrations—HMF typically becomes noticeably inhibitory above ~0.2 g L⁻¹ and furfural around 1 g L⁻¹, while phenolic model compounds representative of lignin degradation (gallic and tannic acid) show no inhibition up to 2 g L⁻¹ [16,17]—so at low by-product concentrations the main penalty of over-severe treatment is more plausibly the formation of poorly degradable condensation products than direct inhibition. Hydrothermal pretreatment of Napier grass, for example, gave the highest methane yield at 175 °C but formed HMF and furfural at 200 °C that impaired methanogenesis [3].

Most hydrothermal studies rely on conventional (conductive) heating. Microwave heating is an alternative that delivers rapid, volumetric energy input and may alter the selectivity of solubilisation. Microwave thermohydrolysis has been shown to release more soluble sugars and to raise biogas yield from maize silage relative to conventional heating at matched temperature and duration [18], and microwave pretreatment combined with acids or other chemical agents has been reported to improve the anaerobic degradability of maize silage [19] and of lignocellulosic waste biomass [20]. An independent comparison of convective and microwave hydrothermal pretreatment, including an energy balance, reached the same qualitative conclusion for Napier grass [21]. Each of these studies, however, examines a single substrate. Whether the advantage of microwave heating is general or substrate-specific, therefore, remains open, and it cannot be settled without a matched comparison across feedstocks of contrasting cell-wall chemistry. Such a comparison—a woody short-rotation crop and a herbaceous silage pretreated under an identical acid-assisted protocol, with full liquid-phase characterisation, biomethane potential and an energy balance evaluated on a common basis—has not been reported.

Accordingly, this study compares microwave and conventional hydrothermal pretreatment of willow and maize silage across matched temperatures (110–130 °C). We hypothesise that microwave heating attains a more favourable energy return than conventional heating at equal temperature through faster, more selective heating, and the response to pretreatment is substrate-specific. The liquid-phase chemistry, biochemical methane potential and the pretreatment energy balance are evaluated jointly to determine whether mild hydrothermal pretreatment is energetically justifiable and, if so, under which heating mode.

2. Materials and Methods

2.1. Feedstock

Two lignocellulosic substrates of contrasting botanical origin were used: chips of common osier (Salix viminalis, willow) and maize silage (Zea mays). Both were characterised for total, mineral and volatile solids. Elemental composition (C, N, H) was determined with a Flash 2000 elemental analyser (Thermo Scientific, Waltham, MA, USA). Fibre fractions were determined by detergent fractionation using a semi-automatic ANKOM220 fibre analyser (ANKOM Technology, Macedon, NY, USA) yielding neutral detergent fibre (NDF), acid detergent fibre (ADF) and acid detergent lignin (ADL); cellulose was calculated as ADF − ADL, hemicellulose as NDF − ADF, and lignin was taken as ADL. The characteristics of both feedstocks are summarised in

Table 1. Characteristics of the raw feedstocks.

Parameter	Willow (Salix viminalis)	Maize Silage (Zea mays)
Total solids [mg/g]	512.33	330.27
Volatile solids [mg/g]	485.27	291.92
Mineral solids [mg/g]	27.06	38.35
Lignin (ADL) [%]	13.22	10.28
Cellulose (ADF − ADL) [%]	43.85	7.69
Hemicellulose (NDF − ADF) [%]	14.17	5.45

. The volatile fraction accounted for 94.7% of the willow dry matter and 88.4% of that of the maize silage. After hydration to 90% moisture, the willow contained 11.71% TS (11.09% VS of fresh mass) and the maize silage 10.0% TS (8.84% VS of fresh mass).

2.2. Hydrothermal Pretreatment

Substrate preparation followed an identical protocol for both feedstocks. Both substrates were first comminuted to a particle size of 5–10 mm and then hydrated with water to 90% moisture content. A 10% hydrochloric acid solution was then added at a fixed dose of 7% relative to the substrate dry matter, and the mixture was shaken for 60 min at 130 rpm (IKA KS 4000 shaker IKA-Werke GmbH & Co. KG, Staufen, Germany) to ensure uniform acid distribution before thermal treatment.

Thermohydrolysis was carried out in a purpose-built HL-P reactor operated at fractional-technical scale. The unit comprises two chambers (Figure 1). Pretreatment took place in the upper, thermally insulated chamber, a 2.4 L PTFE vessel charged through a feed hopper and equipped with a motor-driven stirrer. Because PTFE is transparent to microwaves, the same vessel can be heated either by a magnetron or conventionally; the two modes are used alternately and never simultaneously, which is what allows them to be compared on the same charge and in the same vessel. The charge was stirred throughout heating to keep the temperature uniform, and the temperature was measured by a sensor immersed in the biomass rather than at the chamber wall. Pressure was not regulated: it developed autogenously from the vapour pressure of the hydrated charge at the set temperature, and the chamber is rated to 180 °C and approximately 3–4 bar. The unit is controlled from an HMI panel, which allows the target temperature, the ramp time and the holding time to be set. After the holding period, the charge is decompressed automatically into the larger lower chamber, which is fitted with a safety valve, a pressure gauge and a solenoid drain valve, and is emptied manually.

Treatments were run under two heating regimes—microwave and conventional (resistive) heating—at three process temperatures (110, 120 and 130 °C), with a holding time of 20 min (

Table 2. Experimental design.

Item	Description
Substrates	Willow chips (Salix viminalis); maize silage (Zea mays)
Heating modes	Microwave (MW); conventional resistive heating (Conv.)
Temperatures	110, 120, 130 °C
Hydration	to 90% moisture; shaken 60 min at 130 rpm (IKA KS 4000)
Holding time	20 min
Acid	10% HCl solution, fixed dose of 7% of substrate dry matter
Control	Untreated substrate (no acid, no heating)
Replication	n = 3 (liquid-phase analyses and BMP assays)
Liquid-phase analytics	COD and phenolics: Hach LCK 914/LCK 346; sugars: Megazyme; HMF and furfural: HPLC
BMP assay	AMPTS II, 21-day batch, ISR ≈ 3.5:1 (VS basis)
Variants per substrate	7 (control + 3 temperatures × 2 heating modes)

). The acid dose was held constant across all treated variants, so that heating mode and temperature were the only experimental variables. An untreated sample (no acid, no thermal treatment) served as the control. Following the reactor manual, heating is switched off once the set temperature is reached; during the subsequent holding period, only stirring operates.

2.3. Liquid-Phase Characterisation

After pretreatment, the liquid fraction was analysed for chemical oxygen demand (COD) and total phenolics using Hach (Hach Lange GmbH, Düsseldorf, Germany) cuvette tests (LCK 914 and LCK 346, respectively), and for total, inorganic and organic carbon (TC, IC, TOC) with a Shimadzu analyser. Five- and six-carbon sugars (glucose and xylose) were determined in the filtrate with Megazyme Assay Kits (Megazyme Ltd., Bray, Co. Wicklow, Ireland).

Furanic by-products (HMF and furfural) were quantified by high-performance liquid chromatography (Sykam system: Sykam GmbH, Eresing, Germany: S 1125 pump, S 5300 sample injector, S 3345 PDA detector, S 4115 column oven) on a Zorbax SB-C18 column (4.6 × 150 mm; Agilent Technologies, Santa Clara, CA, USA). Separation used a methanol/water (1:1) mobile phase, previously filtered through 0.22 µm membrane filters, at a flow rate of 0.300 mL/min with an injection volume of 20 µL; detection was at 280 nm. Calibration curves were constructed from six dilution levels of a 0.5 g/L standard mixture of furfural and HMF (Sigma-Aldrich, St. Louis, MO, USA), each injected in triplicate (R² = 0.9998). Samples were stored frozen at −15 °C, thawed, centrifuged for 10 min at 4500 rpm, and the supernatant analysed. All liquid-phase determinations are reported as the mean of three replicates (n = 3).

2.4. Biochemical Methane Potential

Methane potential was determined in an Automatic Methane Potential Test System (AMPTS II, Bioprocess Control, Lund, Sweden), in which carbon dioxide is absorbed before volumetric measurement, so the recorded volume corresponds to methane; the resolution of the volumetric cells was 10 mL. Digestion was mesophilic: the 500 mL bioreactors were held in a water bath at a constant 36 °C and stirred intermittently (30 s every 10 min at 100 rpm). The inoculum was anaerobic sludge obtained from an agricultural biogas plant, with a total solids content of 20.37 mg g⁻¹, of which 13.14 mg g⁻¹ was volatile (organic) and 7.23 mg g⁻¹ mineral. Each bottle had a working volume of 200 mL, comprising the weighed substrate made up to volume with inoculum (8.06 g of willow with 192 mL of inoculum; 10.1 g of maize silage with 190 mL). Both substrates were loaded at an identical organic loading of 0.894 g VS per bottle (4.47 kg VS m⁻³), corresponding to 8.06 g of hydrated willow (11.09% VS of fresh mass) and approximately 10.1 g of hydrated maize silage (8.84% VS); the difference in weighed mass reflects the different volatile-solids content of the two hydrated substrates. Assays were run against inoculum-only blanks. Each bottle was operated until its daily methane production fell below 5 mL CH₄ d⁻¹, in accordance with the manufacturer’s recommendation, so the duration of the assay differed between bottles (6–21 days); maize silage reached this criterion considerably earlier than willow. Methane yield was calculated as net methane production (sample minus inoculum blank) divided by the substrate VS loaded and is expressed as NmL CH₄ g⁻¹ VS. Biogas composition (CH₄, CO₂) was determined via gas chromatography (Agilent 7890A GC) equipped with a thermal conductivity detector (TCD).

2.5. Energy Balance

The pretreatment energy input was estimated from the rated power of the reactor and the measured time required to reach the set temperature (microwave module 0.9 kW; resistive heater 1.05 kW, confirmed with the manufacturer). Because heating is switched off once the set temperature is reached and only stirring operates during the 20 min holding phase, energy was attributed to the heating phase only. This is a conservative treatment for conventional heating, since any regulator duty during holding would further increase its input. The input was normalised per gram of dry matter (≈200 g DM per 2-L charge). The energy output was calculated from the measured methane yield and expressed per gram of dry matter to match the input. Specifically, the output was obtained as E_out = BMP × (VS/TS) × LHV, where BMP is the methane yield in NL CH4 per gram of volatile solids, VS/TS is the volatile-to-total-solids ratio of the substrate (0.947 for willow and 0.884 for maize silage; Table 1), and LHV is the lower heating value of methane, 35.8 kJ per NL. For example, the untreated willow control (241.1 NmL CH4 g−1 VS) gives 0.2411 × 0.947 × 35.8 = 8.17 kJ g⁻¹ DM. Expressing both input and output per gram of dry matter makes the balance independent of the VS/TS normalisation and directly comparable between substrates. Two balances were then derived for each variant. The gross balance was taken as the methane energy of the treated variant minus the pretreatment input. The incremental balance—the metric that answers whether pretreatment is energetically justified—was taken as the additional methane energy relative to the untreated control minus the pretreatment input, since the control yields methane without any energy input. The incremental balance is used throughout as the primary criterion. These values are estimates based on rated power and measured heating time rather than direct wattmeter measurements and are reported as such.

2.6. Data Analysis

Results are expressed as means ± standard deviation (n = 3). Effects on methane yield were assessed with a three-way factorial analysis of variance (Type II sums of squares) fitted to the treated variants, with substrate, heating mode and temperature as fixed factors and all interactions included. Each treated variant was compared with its untreated control using Dunnett-type two-sided contrasts against the pooled within-variant error, with a conservative Bonferroni-type multiplicity adjustment across the six comparisons per substrate (each nominal p-value multiplied by six); the adjusted p-values are reported in Section 3.3. Cumulative methane curves were described with the modified Gompertz model, from which the maximum methane potential (Pmax), the maximum production rate (Rm) and the lag phase (λ) were estimated by non-linear least squares. Relationships between liquid-phase composition and methane yield were examined by Pearson correlation across the six pretreated variants of each substrate; the untreated control was excluded, since it received no pretreatment, and its liquid-phase composition, therefore, reflects the feedstock rather than the treatment. The assay followed established recommendations for BMP testing regarding inoculum handling, blanks and substrate-to-inoculum ratio [22]. Analyses were performed in Statistica (version 13, StatSoft (Tulsa, Oklahoma)/TIBCO Software (Palo Alto, CA, USA)).

3. Results and Discussion

3.1. Solubilisation and Sugar Release

Hydrothermal treatment markedly increased the release of soluble carbohydrates from both substrates, and the effect intensified with temperature (

Table 3. Liquid-phase composition after pretreatment and the resulting biochemical methane potential (means, n = 3).

Substrate	Variant	COD [g/L]	Phenolics [mg/L]	HMF [mg/L]	Furfural [mg/L]	Glucose [mg/L]	Xylose [mg/L]	BMP [NmL/g VS]
Willow	Control	10.2 ± 0.2	190.0 ± 3.7	0.04	0.26	14.0 ± 0.3	194.0 ± 3.7	241.1 ± 5.7
Willow	Conv. 110 °C	11.9 ± 0.2	130.0 ± 2.5	0.24	0.09	63.0 ± 1.2	208.0 ± 4.0	198.4 ± 3.3
Willow	Conv. 120 °C	12.9 ± 0.4	100.0 ± 2.9	0.28	0.02	78.0 ± 2.2	247.0 ± 7.1	284.9 ± 2.5
Willow	Conv. 130 °C	13.0 ± 0.3	80.0 ± 1.8	0.44	0.05	122.0 ± 2.7	321.0 ± 7.2	222.8 ± 1.3
Willow	MW 110 °C	13.3 ± 0.3	129.0 ± 2.8	0.26	<LOQ	71.0 ± 1.5	361.0 ± 7.8	263.1 ± 2.2
Willow	MW 120 °C	14.7 ± 0.4	112.0 ± 3.0	0.29	<LOQ	82.0 ± 2.2	398.0 ± 10.6	242.7 ± 1.9
Willow	MW 130 °C	13.4 ± 0.2	99.5 ± 1.6	0.62	<LOQ	124.0 ± 2.0	412.0 ± 6.6	310.6 ± 2.5
Maize silage	Control	19.1 ± 0.4	112.3 ± 2.2	0.19	2.01	14.0 ± 0.3	184.0 ± 3.6	232.9 ± 3.1
Maize silage	Conv. 110 °C	28.4 ± 0.5	124.7 ± 2.4	0.44	2.57	58.0 ± 1.1	248.0 ± 4.8	240.9 ± 3.8
Maize silage	Conv. 120 °C	31.9 ± 0.9	134.9 ± 3.9	0.46	6.10	84.0 ± 2.4	347.0 ± 9.9	251.5 ± 3.4
Maize silage	Conv. 130 °C	32.8 ± 0.7	136.8 ± 3.1	0.48	6.83	146.0 ± 3.3	421.0 ± 9.5	260.0 ± 3.5
Maize silage	MW 110 °C	27.7 ± 0.6	135.8 ± 2.9	0.78	7.03	61.0 ± 1.3	268.0 ± 5.8	259.5 ± 3.7
Maize silage	MW 120 °C	30.5 ± 0.8	134.7 ± 3.6	0.85	7.07	79.0 ± 2.1	369.0 ± 9.8	293.2 ± 4.6
Maize silage	MW 130 °C	37.4 ± 0.6	159.9 ± 2.6	0.92	8.29	174.0 ± 2.8	502.0 ± 8.1	306.0 ± 4.0

, Figure 2). For willow, glucose rose from 14 mg L⁻¹ in the control to 124 mg L⁻¹ after microwave treatment at 130 °C, while xylose—the dominant sugar, reflecting the hemicellulosic origin of the solubilised fraction—increased from 194 to 412 mg L⁻¹ over the same range. Maize silage released more sugar than willow at the highest severity, reaching 174 mg L⁻¹ glucose and 502 mg L⁻¹ xylose at 130 °C under microwave heating, consistent with its less recalcitrant, non-woody matrix.

At matched temperature, the heating mode governed the selectivity of solubilisation. Microwave heating consistently released more hemicellulosic xylose than conventional heating, most clearly at low severity: for willow at 110 °C, xylose reached 361 mg L⁻¹ under microwave versus 208 mg L⁻¹ under conventional heating. Soluble COD followed the same monotonic rise with temperature and was substantially higher for maize silage (19.1 → 37.5 g O₂ L⁻¹) than for willow (10.2 → 14.7 g O₂ L⁻¹), indicating a larger pool of dissolved organic matter released from the silage.

3.2. Formation of By-Products

Furanic by-products remained at trace levels throughout (Table 3, Figure 3a–d). HMF did not exceed 0.92 mg/L in any variant: it rose from 0.04 mg/L in the untreated willow control to 0.62 mg/L after microwave treatment at 130 °C, and from 0.19 to 0.92 mg/L for maize silage over the same range. Furfural behaved differently between the two substrates—in maize silage, it increased steadily with severity, from 2.01 mg/L in the control to 8.29 mg/L at 130 °C under microwave heating, whereas in willow, it was highest in the untreated control (0.26 mg/L) and fell below the limit of quantification in all microwave-treated variants.

These concentrations are two to three orders of magnitude below the levels reported to inhibit methanogenesis (HMF above approximately 0.2 g/L, furfural around 1 g/L, and phenolic model compounds of lignin origin showing no inhibition up to 2 g/L [16]). The highest furfural concentration measured here (8.29 mg/L, i.e., 0.0083 g/L) is about 120-times below its inhibitory threshold, the highest HMF (0.92 mg/L) roughly 200-times below, and the highest phenolics concentration (159.9 mg/L) more than 12-times below. Direct inhibitor toxicity is, therefore, unlikely to be a controlling factor for the methane yields reported here, given that the concentrations remained two to three orders of magnitude below reported inhibitory thresholds.

Total phenolics followed opposite trends in the two feedstocks (Figure 3e,f). In maize silage, they rose with temperature, from 112.3 mg/L in the control to 159.9 mg/L at 130 °C under microwave heating, as expected from progressive lignin depolymerisation. In willow, they did the reverse: the untreated control carried the highest phenolic load of the entire study (190.0 mg/L), and every treated variant fell below it, down to 80.0 mg/L after conductive heating at 130 °C.

Taken together with the furfural data, this pattern is best explained by condensation chemistry rather than by a failure of the pretreatment to mobilise phenolics. Willow is naturally rich in water-extractable phenolics—Salix species accumulate salicylates and phenolic glycosides—which accounts for the high concentration leached by the untreated control without any thermal or acid treatment [23]. Under acidic thermal conditions, such phenolics are known to repolymerise and to condense with furanic aldehydes, yielding insoluble humins and pseudo-lignin that deposit on the solid residue [3,15,17]. Such a sink would remove phenolics and furfural from the liquid phase simultaneously, which is precisely what is observed: willow phenolics fell by 32–58% relative to the control while furfural fell from 0.26 mg/L to below the limit of quantification, even though 361–412 mg/L of xylose had been released and was available for dehydration. Maize silage, whose untreated control carried a much lower native phenolic load, shows the opposite behaviour, with both phenolics and furfural accumulating as severity increases. The two substrates, therefore, differ not only in how much they solubilise but in the fate of what is solubilised.

3.3. Biochemical Methane Potential

Because both substrates were digested at an identical organic loading, their methane yields are directly comparable (Figure 4, Table 3). The factorial analysis resolved all three main effects (p < 0.001 for substrate, heating mode and temperature; R² = 0.993, residual SD = 3.21 NmL CH₄ g⁻¹ VS). The heating mode carried by far the largest effect (F = 1137), and the three-way interaction was also significant (F = 396, p < 0.001), indicating that the pattern of response to heating mode and temperature differs fundamentally between the two feedstocks.

Maize silage responded in an orderly manner. Every treated variant equalled or exceeded the untreated control (232.9 ± 3.1 NmL CH₄ g⁻¹ VS), microwave heating surpassed conductive heating at every temperature, and yields rose monotonically with temperature under both modes, reaching 306.0 ± 4.0 NmL CH₄ g⁻¹ VS at 130 °C under microwave heating (+31.4%, p < 0.001). All contrasts against the control were significant except conductive heating at 110 °C (+3.5%, not significant).

Willow behaved erratically. Its highest yield, 310.6 ± 2.5 NmL CH₄ g⁻¹ VS, was obtained by microwave heating at 130 °C (+28.9%, p < 0.001), and conductive heating at 120 °C also improved the yield markedly (284.9 ± 2.5; +18.2%, p < 0.001). Microwave heating at 110 °C also raised the yield significantly (263.1 ± 2.2; +9.1%, p < 0.001). However, two variants fell significantly below the untreated control (241.1 ± 5.7): conductive heating at 110 °C (198.4 ± 3.3; −17.7%, p < 0.001) and at 130 °C (222.8 ± 1.3; −7.6%, p < 0.001), while microwave heating at 120 °C was indistinguishable from the control. Mild acid-assisted thermohydrolysis of this recalcitrant woody substrate can, therefore, reduce the methane yield as readily as it increases it, and the outcome is not a monotonic function of severity. Yield losses under acid-assisted thermal treatment have been reported before: in wheat straw, thermal-sulfuric acid pretreatment lowered the methane yield by 29–44% relative to the untreated control, an effect attributed there to furfural and HMF accumulation [24]. The present case differs mechanistically, since the furanic concentrations measured here remain two to three orders of magnitude below inhibitory levels (Section 3.2), which points to the loss of degradable carbon into condensation products rather than to inhibitor toxicity.

The measured values correspond to 48–75% of the theoretical maximum for carbohydrate substrates (415 NmL CH₄ g⁻¹ VS by the Buswell relationship), i.e., within the range physically attainable for partially lignified biomass. The methane content of the biogas, expressed as CH₄/(CH₄ + CO₂), was comparable across all variants (67–71%), so the differences in potential reflect the extent of substrate conversion rather than gas quality. The yields are consistent with reported potentials for willow (approximately 198–246 NmL CH₄ g⁻¹ biomass) [2] and maize silage (approximately 196–402 NmL CH₄ g⁻¹ VS) [5,6].

An unexpected outcome is that the two substrates delivered comparable methane potentials from their untreated controls despite markedly different composition—willow contained 13.2% lignin and 43.9% cellulose against 10.3% and 7.7% recorded for maize silage—the controls differing by only 3.5% (241.1 versus 232.9 NmL CH₄ g⁻¹ VS). What separated the substrates was not the ultimate yield but the rate at which it was reached (Section 3.5).

3.4. Linking Liquid-Phase Chemistry to Methane Yield

Rising soluble sugar concentrations with temperature did not translate into proportional methane gains, and for willow, the highest-severity variants did not maximise yield (Figure 5). A comparison of several physicochemical pretreatments applied to a single corn stover feedstock under identical digestion conditions likewise found that the greatest organic-matter solubilisation did not correspond to the greatest methane yield [25]. Because furanic and phenolic by-products remained two to three orders of magnitude below inhibitory thresholds (Section 3.2), this ceiling cannot be attributed to toxicity. It is instead consistent with the limited additional bioaccessibility gained under mild conditions and with the formation of poorly degradable condensation products. For willow, the liquid-phase data are consistent with the latter: phenolics and furfural both disappeared from solution as severity increased (Section 3.2), which is the expected signature of repolymerisation into humins and pseudo-lignin on the solid residue. That the two willow variants with the lowest methane yields (conductive heating at 110 and 130 °C, both significantly below the untreated control) were also those with the lowest residual phenolic concentrations is consistent with this interpretation, although the present design cannot establish the link causally. Microwave irradiation has been shown to disrupt the lignocellulosic structure of corn straw and release cellulose, hemicellulose and their derivatives, with the effect confirmed by SEM, FTIR and XRD and a methane gain of up to 73.1% [26]. The more selective hemicellulose solubilisation achieved under microwave heating—higher xylose release at comparable HMF formation—is consistent with its generally higher or comparable methane yields relative to conventional heating. We emphasise that this interpretation, although consistent with the liquid-phase data, remains a hypothesis: it cannot be confirmed without characterisation of the solid residues (for example by SEM, FTIR, XRD or a mass balance on the solid fraction), which was outside the scope of this study.

3.5. Methane Production Kinetics

The modified Gompertz model described the cumulative curves well for every variant (Figure 6; R² = 0.979–0.996;

Table 4. Modified Gompertz parameters fitted to the cumulative methane curves.

Substrate	Variant	Pmax [NmL/g VS]	Rm [NmL/g VS/d]	Lag λ [d]	R2
Willow	Control	300.9	13.0	0.00	0.993
Willow	Conv. 110 °C	254.9	10.9	0.00	0.988
Willow	Conv. 120 °C	329.0	17.2	0.00	0.991
Willow	Conv. 130 °C	254.6	15.3	0.00	0.990
Willow	MW 110 °C	306.3	16.7	0.37	0.995
Willow	MW 120 °C	272.3	16.2	0.00	0.993
Willow	MW 130 °C	335.7	22.4	0.00	0.996
Maize silage	Control	226.6	112.4	0.11	0.995
Maize silage	Conv. 110 °C	235.8	103.7	0.08	0.993
Maize silage	Conv. 120 °C	240.5	96.5	0.04	0.989
Maize silage	Conv. 130 °C	248.5	91.1	0.00	0.983
Maize silage	MW 110 °C	251.4	80.9	0.00	0.989
Maize silage	MW 120 °C	286.3	100.1	0.00	0.985
Maize silage	MW 130 °C	292.2	75.1	0.00	0.979

). The most striking difference between the substrates lies in the production rate rather than in the potential: the maximum rate reached 75–112 NmL CH₄ g⁻¹ VS d⁻¹ for maize silage against only 10.9–22.4 NmL CH₄ g⁻¹ VS d⁻¹ for willow, a five- to eight-fold difference. Maize silage consequently approached its plateau within about a week, whereas willow was still producing methane at the end of the assay. This distinction is of direct practical relevance, because the retention time—and, hence, the digester volume—is governed by the rate, not by the ultimate potential.

Within each substrate, microwave heating raised the maximum rate for willow (up to 22.4 against 10.9–17.2 NmL CH₄ g⁻¹ VS d⁻¹ under conductive heating), consistent with its higher yields. For maize silage, the rate was already high in the untreated control (112.4 NmL CH₄ g⁻¹ VS d⁻¹), and pretreatment did not accelerate it further; the benefit for this substrate lay in the extent of conversion rather than its rate. The lag phase was effectively absent in all variants (λ ≤ 0.4 d), which independently corroborates the absence of inhibition inferred from the by-product concentrations (Section 3.2).

One consequence of the differing kinetics deserves emphasis. For willow, the Gompertz Pmax exceeded the measured 21-day yield by roughly 20–25% (for example 300.9 against 241.1 NmL CH₄ g⁻¹ VS for the control), indicating that the assay was terminated before this substrate had fully plateaued and that its methane potential is, therefore, somewhat underestimated. For maize silage, the two agreed closely (226.6 against 232.9), confirming that the assay duration was sufficient. Since the truncation applies equally to all willow variants, the comparisons between heating modes and temperatures remain valid, but the absolute willow values should be read as conservative.

3.6. Net Energy Balance of the Pretreatment

The energy assessment is the central result of this study (

Table 5. Energy balance of the pretreatment step (estimates based on rated power and measured heating time).

Substrate	Heating	T [°C]	E_in [kJ/g DM]	E_out [kJ/g DM]	ΔE_out VS Control	Gross Net	Incremental Net
Willow	Control	—	0	8.17	—	—	—
Willow	Microwave	110	4.32	8.92	+0.75	+4.60	−3.57
Willow	Microwave	120	5.40	8.23	+0.06	+2.83	−5.34
Willow	Microwave	130	7.56	10.53	+2.36	+2.97	−5.20
Willow	Conventional	110	10.71	6.73	−1.45	−3.98	−12.16
Willow	Conventional	120	17.64	9.66	+1.49	−7.98	−16.15
Willow	Conventional	130	18.59	7.56	−0.62	−11.03	−19.21
Maize silage	Control	—	0	7.37	—	—	—
Maize silage	Microwave	110	4.32	8.22	+0.84	+3.90	−3.48
Maize silage	Microwave	120	5.40	9.28	+1.91	+3.88	−3.49
Maize silage	Microwave	130	7.56	9.69	+2.32	+2.13	−5.24
Maize silage	Conventional	110	10.71	7.63	+0.25	−3.08	−10.46
Maize silage	Conventional	120	17.64	7.96	+0.59	−9.68	−17.05
Maize silage	Conventional	130	18.59	8.23	+0.86	−10.36	−17.73

, Figure 7). The criterion applied here follows the principle that a pretreatment is justified only when its energy balance is positive [27]. This threshold is not arbitrary: an energy assessment of thermal and microwave pretreatments concluded that treatments below 100 °C tend to return a positive energy balance, whereas those at or above 100 °C generally do not [28]. The present temperature window (110–130 °C) lies entirely above that threshold. Microwave heating reached the set temperatures in roughly half the time of conventional heating and consequently required 2–3-times less input energy; at 120 °C, for instance, the input was 5.40 kJ g⁻¹ DM under microwave heating versus 17.64 kJ g⁻¹ DM under conductive heating.

The decisive comparison, however, is not the gross methane energy of a treated variant but the additional methane it delivers relative to an untreated control, which requires no energy input at all. On this incremental basis, no variant tested here recovered its own energy input. The additional methane energy gained through pretreatment was modest—at most +2.36 kJ g⁻¹ DM (willow, microwave 130 °C) and +2.32 kJ g⁻¹ DM (maize silage, microwave 130 °C)—while the lowest input required was 4.32 kJ g⁻¹ DM. Every variant was, therefore, net-negative, the least unfavourable being microwave heating at 110 °C for both willow (−3.57 kJ g⁻¹ DM) and maize silage (−3.48 kJ g⁻¹ DM). For two willow variants treated conductively, the additional methane energy was itself negative (−1.45 and −0.62 kJ g⁻¹ DM at 110 and 130 °C), i.e., pretreatment reduced rather than increased the recoverable energy.

The heating mode nonetheless remains decisive, because it changes the magnitude of the penalty three- to five-fold. Microwave variants clustered between −3.5 and −5.3 kJ g⁻¹ DM, whereas conductive heating fell between −10.5 and −19.2 kJ g⁻¹ DM; microwave heating was superior in all six paired comparisons. Since the methane gain achieved by the two heating modes was of comparable magnitude, this difference is driven almost entirely by the input side. Importantly, this comparison is insensitive to the normalisation of the methane data: the incremental methane term is identical for both heating modes at a given temperature, so the ranking of heating modes depends only on the measured heating energy.

Because the energy input was estimated from rated power and the measured time to set-point rather than from a wattmeter, it carries a nominal uncertainty. This does not affect the comparison between heating modes: the two- to three-fold gap in input energy greatly exceeds any plausible estimation error, so even a 20–30% error in either rated power would preserve the ranking. A direct wattmeter measurement would remove this uncertainty and is planned for future work. A further caveat concerns the comparison itself: the two heating modes differ in heating rate and in spatial temperature uniformity, and localised superheating (hot spots) under microwave irradiation cannot be excluded. Such effects may contribute to the observed differences and cannot be resolved by the present set-point-based design, so the microwave advantage reported here reflects the performance of this specific reactor rather than a demonstrated intrinsic microwave mechanism. Continuous temperature–time and pressure–time profiles were not logged during these runs; per-run logging is identified as a priority for ongoing work, since it would allow the process severity of the two modes to be compared directly rather than through the set temperature alone.

The dissociation between yield and energy return is instructive. For both substrates, the highest-yielding variant was microwave heating at 130 °C, yet the least unfavourable energy balance was obtained at 110 °C, because the additional methane gained between 110 and 130 °C did not cover the additional heating energy. Conductive heating at 120 °C illustrates the point more sharply still: for willow, it delivered the second-highest yield of the study (284.9 NmL CH₄ g⁻¹ VS, +18.2%) at an incremental cost of −16.15 kJ g⁻¹ DM, four- to five-times worse than any microwave variant. Optimising pretreatment on methane yield alone would, therefore, select the wrong operating point on both counts—the wrong temperature and, more importantly, the wrong heating technology. The practical implication is that mild acid-assisted thermohydrolysis cannot be justified on energy grounds alone; if it is applied for other reasons—most plausibly to exploit the five- to eight-fold faster kinetics of maize silage or to shorten the retention time of willow—microwave heating is the only mode for which the penalty remains within a plausible trade-off.

One benefit does, however, fall outside the energy balance and favours pretreatment for the woody substrate: a shorter retention time. Fitting the modified Gompertz model and estimating the time to reach 90% of the maximum methane potential, untreated willow required about 28 days, whereas microwave heating at 130 °C reduced this to about 18 days—a 35% reduction (27% at 120 °C and 21% at 110 °C). Because, at a fixed organic loading, the hydraulic retention time scales with the digester volume, this corresponds to a proportionally smaller reactor for the same throughput. These figures are order-of-magnitude estimates derived from batch kinetics rather than continuous-reactor measurements, but the direction is clear and rests on the maximum production rate, which is well constrained even where the willow potential itself is conservative. For maize silage, the argument does not apply: the untreated control already reached 90% of its potential within about 2.5 days, so pretreatment offered no kinetic benefit and, if anything, slightly slowed the approach to plateau. The practical case for mild acid-assisted thermohydrolysis is, therefore, confined to slow, recalcitrant substrates such as willow, and, even there, it rests on reactor volume rather than on energy return.

Two qualifications are important and strengthen rather than weaken this conclusion. First, the input estimate is conservative for conventional heating, since the energy consumed by the temperature regulator during the holding phase was excluded; the real balance of conductive heating is, therefore, likely to be worse than reported. Second, on a strict net energy basis, the untreated control, which requires no energy input, returned more energy than any pretreated variant, because the methane gain achievable under mild conditions was too small to offset even the low microwave input. The practical message is, therefore, not that pretreatment maximises net energy under these conditions but that if hydrothermal pretreatment is applied, microwave heating reduces the penalty to a plausible margin, whereas conductive heating multiplies it, so microwave heating incurs a smaller penalty than conductive heating, although, under these conditions, neither mode returns a net energy gain.

The balance is, moreover, a gate-to-gate estimate of the electrical energy of the heating step alone. The embodied energy of the acid, the manufacture of the reactor vessel, and any biomass drying or handling were not included; accounting for these terms would render the balance more negative and would, therefore, reinforce, not weaken, the conclusion.

Since all treatment variants showed a negative net energy balance, it is necessary to determine which changes could enable recovery of the energy used for pretreatment. At a moisture content of 90%, a 2 L charge contains 1.8 kg of water and only 0.2 kg of dry matter; therefore, most of the energy is used to heat the water. The minimum heat required to raise the charge from ambient temperature to 110, 120, and 130 °C is 3.52, 3.91, and 4.30 kJ·g⁻¹ DM, respectively (

Table 6. Energy input compared with the thermodynamic minimum for heating the charge.

T [°C]	Theoretical Minimum [kJ/g DM]	MW Input [kJ/g DM]	MW Efficiency	Conventional Input [kJ/g DM]	Conv. Efficiency
110	3.52	4.32	82%	10.71	33%
120	3.91	5.40	72%	17.64	22%
130	4.30	7.56	57%	18.59	23%

).

Comparison of these values with the measured energy consumption confirms the advantage of microwave heating. The ratio of the thermodynamic minimum to the actual energy input was 82%, 72%, and 57% at the respective temperatures, which is consistent with the device specification of a 0.9 kW power draw and 800 W microwave output. For conventional heating, this ratio was only 22–33%, mainly because of heat losses through the reactor walls. The decrease in microwave heating efficiency from 82% to 57% with increasing temperature resulted from the longer heating time and greater energy losses. This indicates that the energy consumption of the microwave treatment was already close to the thermodynamic minimum.

A second option for improving the balance is to increase methane yield, although this is limited by stoichiometry. Assuming the Buswell value of 415 NmL CH₄·g⁻¹ VS as the maximum theoretical limit, the greatest possible increase in energy relative to the untreated control is 5.90 kJ·g⁻¹ DM for willow and 5.76 kJ·g⁻¹ DM for maize silage. At 130 °C, however, the microwave energy input alone was 7.56 kJ·g⁻¹ DM. Therefore, treatment at this temperature could not achieve a positive energy balance even with complete substrate conversion. At 110 and 120 °C, a positive balance would theoretically be possible, but it would require conversion close to the stoichiometric maximum, which is unlikely for partially lignified biomass.

The greatest potential, therefore, lies in increasing the dry matter content of the charge. Since the energy input per gram of dry matter decreases as the solids concentration increases, energy break-even would be reached at approximately 28% DM for maize silage treated with microwaves at 120 °C, approximately 33% DM at 130 °C, and approximately 32% DM for willow treated at 130 °C. This is particularly relevant because the raw maize silage contained 33% DM. The negative balance, therefore, resulted primarily from diluting the charge to a moisture content of 90%, rather than from the heating process itself.

A similar effect could be achieved by recovering 65–80% of the thermal energy, for example, from the steam released during decompression. Another option is to recover the energy contained in the liquid phase. Anaerobic digestion of the hydrolysate generated during microwave-assisted hydrothermal pretreatment as a co-substrate, instead of discarding it, may increase the overall energy gain of the process [29]. In the present balance, the soluble sugars, organic acids, and phenolic compounds released into the liquid phase (Section 3.2) were included only to the extent that they were digested together with the solid fraction.

4. Conclusions

Microwave- and conventional acid-assisted thermohydrolysis treatments of willow and maize silage were compared under matched conditions and at an identical organic loading, combining liquid-phase characterisation and biochemical methane potential with an energy balance of the pretreatment step. A three-way factorial analysis resolved substrate, heating mode and temperature, with the heating mode carrying the largest effect and a significant three-way interaction: maize silage responded in an orderly manner, rising monotonically to +31.4% over the untreated control, whereas the more recalcitrant willow responded erratically, gaining up to +28.9% in its best variant yet falling significantly below the control in two conductively heated ones. Gompertz fits showed the substrates to differ five- to eight-fold in maximum methane rate rather than in ultimate potential, with no lag phase in any variant. Furanic and phenolic by-products remained two to three orders of magnitude below reported inhibitory thresholds, so the yield ceiling is attributed to limited additional bioaccessibility and to condensation products rather than to inhibitor toxicity.

Expressed incrementally against an untreated control—the only basis on which the question can properly be answered—none of the treatments repaid their own energy input. The additional methane energy never exceeded 2.4 kJ g⁻¹ DM, while the cheapest pretreatment cost 4.3 kJ g⁻¹ DM. Under the mild conditions studied, acid-assisted thermohydrolysis is, therefore, not self-financing in energy terms, and reporting only the gross energy of the treated stream would have obscured this.

The heating mode nonetheless governs the magnitude of the penalty. Microwave heating reached the target temperatures with 2–3-times less energy and confined the incremental deficit to between −3.5 and −5.3 kJ g⁻¹ DM, against −10.5 to −19.2 kJ g⁻¹ DM for conductive heating, and was superior in all six paired comparisons; for both substrates, the highest-yielding variant (130 °C) was not the one with the best energy return (110 °C). Where pretreatment is applied for reasons other than energy return—shorter retention time, smaller digester volume, or co-production of fermentable sugars—microwave heating incurs a markedly smaller energy penalty than conductive heating; under the conditions tested, however, neither mode returns a net energy gain, so pretreatment is not warranted on energy grounds alone for either substrate.