Heat Pumps in Green Hydrogen Production Systems: A Technical Review

Abstract

Green hydrogen production through water electrolysis is a key pathway to the decarbonization of future energy systems. However, part of the electrical input is transformed into waste heat. In this study, alkaline, proton-exchange membrane, anion-exchange membrane, and solid oxide electrolysis systems are compared in terms of operating temperature, heat generation, heat transfer medium, and integration constraints. Reported COP values for commercial high-temperature vapour-compression heat pumps range from 2.4 to 5.8, depending on operating conditions. The heat-pump technologies reviewed include vapour-compression systems with single-stage, multistage, cascade, and transcritical configurations, together with absorption and adsorption systems, with a focus on suitable working fluids and practical limitations. The review distinguishes between direct heat recovery and heat recovery assisted by heat pumps, and it identifies two main areas of application: external supply for district heating, industrial consumers, and energy communities; and internal support for feedwater preheating, water cycle integration, and steam generation. A selection framework is proposed in which source- and sink-temperature compatibility determines thermodynamic feasibility, COP characterizes heat-pump performance, and LCoH supports techno-economic comparison. Direct heat recovery should be preferred when temperatures are compatible, while heat pumps can operate as enabling technologies when temperature upgrading is required and system-level economic and environmental performance remains advantageous.

1. Introduction

Green hydrogen is increasingly being considered for applications in which direct electrification is technically difficult or does not provide sufficient operational flexibility. These applications include ammonia and methanol production, iron and steel manufacturing, selected high-temperature industrial processes, heavy transportation, and long-term energy storage [1,2,3]. The expansion of wind and photovoltaic generation has further increased interest in water electrolysis because hydrogen can absorb part of the variable electricity production and connect the power sector with industrial, transport, and thermal energy systems [4,5].

Most research on water electrolysis has concentrated on stack efficiency, catalyst activity, membrane durability, current density, and hydrogen production cost [6,7,8,9]. These are essential issues, but they do not describe the performance of the complete production plant. An electrolyser does not convert all supplied electrical energy into the chemical energy of hydrogen. Part of the input is released as heat and must be removed to maintain the required operating temperature [10,11]. In many installations, this thermal energy is transferred to a cooling circuit and rejected into the surroundings. In larger electrolyser systems, however, the rejected heat can represent a relevant thermal stream with potential for useful heat recovery [10,11,12].

The quantity and temperature of the available heat depend on the electrolysis technology, stack design, current density, pressure, and operating strategy [12,13,14]. Alkaline water electrolysers and proton-exchange membrane electrolysers generally release heat at low or moderate temperatures. Anion-exchange membrane electrolysers operate within a similar thermal range, although their commercial development is less mature. Solid-oxide electrolysers require high-temperature steam and therefore have a different thermal balance, with stronger interaction between heat supply, heat recovery, and electrochemical efficiency [15,16,17,18]. These differences prevent the application of a single heat recovery concept to all electrolysis systems.

Low-temperature heat from alkaline, PEM, and AEM electrolysers can be used directly only when a nearby consumer requires heat at a compatible temperature. Possible applications include low-temperature district heating, domestic hot water preheating, greenhouse heating, drying, and selected industrial processes [19,20,21]. Direct use is thermodynamically preferable because it avoids additional electricity consumption. Its practical application, however, is restricted by the temperature level of the cooling circuit, the distance to the heat consumer, seasonal variation of heat demand, and the operating schedule of the electrolyser. A heat source may therefore be technically available but economically unusable when the heat demand is intermittent or remote or requires a higher supply temperature.

Heat pumps can reduce this temperature mismatch by upgrading the heat rejected from the electrolyser to a level suitable for useful consumption [22,23,24]. Conventional vapour-compression systems may be sufficient for space heating, domestic hot water, and low-temperature district heating. High-temperature heat pumps can supply heat for industrial processes at higher temperatures, while cascade arrangements may be used when the temperature lift cannot be achieved efficiently in a single stage [25,26,27]. Heat pump-assisted steam generation has also been investigated for applications requiring hot water or low-pressure steam [28,29]. The choice between these configurations depends on the source temperature, required sink temperature, refrigerant, compressor technology, operating hours, and variations in both hydrogen production and heat demand.

The use of a heat pump does not automatically improve the performance of a hydrogen production system. The recovered heat must have a useful and sufficiently stable destination, while the electricity consumed by the heat pump must be included in the system balance. A high coefficient of performance alone is not enough to demonstrate a benefit. The result also depends on the electricity mix, the displaced heating technology, the value assigned to the recovered heat, and the boundaries selected for the assessment [30,31,32]. A system may show a high overall energy efficiency while providing only a modest exergy benefit because low-temperature heat has limited ability to perform useful work. For this reason, energy analysis should be supported by evaluation of exergy and economic, and environmental factors.

The published literature does not yet provide a consistent basis for comparing these systems. Reviews of green hydrogen have generally examined electrolysis technologies, renewable electricity integration, materials, hydrogen storage, or production costs [6,8,15,33]. Thermal management is often discussed only as a requirement for safe and stable electrolyser operation. Waste heat recovery, when included, is usually treated as a secondary option rather than as a central part of plant design [11,18,34]. Reviews of industrial heat pumps provide detailed information on refrigerants, compressor technologies, and achievable temperature levels, but electrolysers are normally considered as one type of many possible industrial heat sources [22,25,35]. The connection between electrolyser thermal performance and heat pump selection therefore remains insufficiently developed.

Individual studies have examined the use of electrolyser heat in district heating networks [19,20,36], industrial processes [27,29], and integrated energy systems [31,37]. These investigations confirm that useful heat recovery is technically possible, but their results are not directly comparable. Different studies use different definitions of electrolyser efficiency, heat recovery potential, system boundaries, and reference heating systems. Some include stack heat only, whereas others include compression, power electronics, and auxiliary equipment. In several cases, heat pumps have been assessed under fixed operating conditions even though both electrolyser load and heat demand vary over time. These differences can produce substantially different conclusions even for apparently similar systems.

The economic treatment of recovered heat also remains inconsistent. In some studies, thermal energy is assigned a fixed market price, while others calculate the avoided cost of natural gas, electricity, or district heating [30,32,38]. Capital expenditure for heat exchangers, pipelines, heat pump equipment, and backup heat-rejection systems is not always included. The effect of distance between the hydrogen plant and the heat consumer is also frequently simplified, although distribution infrastructure can determine whether a technically feasible concept is economically viable. As a result, promising laboratory or simulation results cannot always be transferred directly to industrial projects.

Environmental assessment presents similar difficulties. Green hydrogen is commonly associated with renewable electricity, but actual electrolyser operation may include electricity from the grid during periods of limited renewable generation [1,39]. The environmental benefit of a heat pump therefore depends on the hourly carbon intensity of electricity and on the fuel or technology replaced by the delivered heat. Recovering electrolyser heat may lead to a clear reduction in emissions when it replaces fossil-fired boilers, but the benefit can become smaller when the reference system is an efficient electric heat pump or a low-carbon district heating network [40,41,42]. Environmental conclusions based only on annual average electricity factors may therefore overlook important operational effects.

A further weakness is the limited treatment of dynamic operation. Electrolysers connected to wind and photovoltaic systems frequently operate at partial load or follow variable electricity availability [4,13,43]. Waste heat production changes accordingly, while heat demand follows its own daily and seasonal pattern. A heat pump selected according to nominal electrolyser capacity may therefore operate inefficiently or remain underused for a considerable part of the year. Thermal storage may reduce this mismatch in some system layouts, but it introduces additional cost, heat losses, and control requirements [44,45,46]. These interactions are central to practical system design, yet they are often omitted from steady-state assessments.

The present review examines heat pumps as part of the thermal design of green hydrogen production systems rather than as an independent heating technology. The analysis covers alkaline, PEM, AEM, and solid-oxide electrolysis, with attention to operating temperature, thermal balance, and recoverable heat. Direct heat use, vapour-compression systems with single-stage, multistage, cascade, and transcritical configurations, thermally driven absorption and adsorption systems, and heat-pump-assisted steam generation are compared according to their technical requirements and suitable applications. Energy, exergy, and economic and environmental aspects are discussed where relevant, with particular attention to the assumptions that limit comparison between published results.

This review addresses four main questions. First, what quantity and temperature level of heat can realistically be recovered from the principal electrolysis technologies? Second, under which conditions is direct heat use preferable to heat pump-assisted temperature increase? Third, which heat pump configurations are suitable for external heat use, internal process support, district heating, and heat and steam generation for industrial processes? Fourth, which technical and methodological gaps must be addressed before these concepts can be consistently evaluated and applied at industrial scale? By focusing on these questions, the paper shifts the discussion from comparing hydrogen and heat pumps as separate decarbonization options, toward understanding heat pumps as enabling technologies within green hydrogen production systems.

Figure 1 presents the conceptual framework adopted in this review, illustrating the link between electrolysis technologies, waste heat generation, heat pump integration, and the principal pathways for useful heat utilisation.

2. Waste Heat and Thermal Management in Green Hydrogen Production Systems

Green hydrogen production by water electrolysis is often discussed in terms of electrical efficiency and hydrogen yield, but its thermal performance is equally important for system integration. A fraction of the electrical input that is not converted into hydrogen chemical energy is released as heat through the electrolyser stack, cooling circuits, electrolyte loops, humid product gases, compression stages, and balance-of-plant components. The quantity, temperature level, and continuity of these heat streams depend strongly on the electrolyser technology, operating point, and plant configuration. With low-temperature electrolysers, such as PEM, alkaline, and AEM systems, the rejected heat is usually available at low- or medium-temperature levels and often requires heat pump upgrading before it can be used in district heating or industrial applications [36,47,48]. In high-temperature electrolysis, especially SOEC systems, heat recovery is more closely connected to process integration, because recovered sensible and latent heat can contribute to feedwater preheating and steam generation [49,50,51]. This section therefore examines the origins of heat losses in electrolysis, the main forms of recoverable heat, the operational constraints of thermal management, and the system-level relevance of waste heat recovery in green hydrogen production plants.

2.1. Heat Losses in Water Electrolysis

In water electrolysis, the supplied electrical energy is not completely converted into the chemical energy of hydrogen. Part of the input is consumed by activation overpotentials at the hydrogen and oxygen electrodes, ohmic resistance in membranes, electrolytes, electrodes, and current collectors, mass-transfer limitations, gas bubble effects, and other irreversible losses. The oxygen evolution reaction is often identified as one of the kinetically limiting steps in water electrolysis, requiring significant overpotential and therefore increasing the electrical energy demand. These electrochemical losses appear as heat that must be removed from the stack to maintain stable operating temperature and durability. At the system level, this heat is released through cooling circuits, electrolyte loops, product gas streams, compression stages, and auxiliary components, creating opportunities for waste heat recovery and heat pump integration.

Among the different irreversible losses, electrode kinetics are particularly important because both half-reactions of water electrolysis require an additional driving force beyond the thermodynamic voltage. The oxygen evolution reaction, occurring at the anode, is commonly regarded as one of the kinetically limiting steps because it involves a complex multi-electron reaction pathway and usually requires a significant overpotential [52]. In parallel, the hydrogen evolution reaction, although generally faster than OER, can also contribute to efficiency losses, especially in alkaline electrolysis, where water dissociation and hydrogen adsorption/desorption processes may limit the reaction rate [53]. These kinetic limitations increase the practical cell voltage and therefore, the electrical energy input required per unit of hydrogen produced. From the perspective of the present review, the relevance of catalyst studies does not apply to the detailed design of the material, but to the fact that HER and OER overpotentials represent a fundamental source of additional energy consumption in electrolysers. This additional input is ultimately dissipated as heat and must be removed by the thermal management system. In PEM electrolysis, the acidic OER environment creates specific challenges related to catalyst activity, stability, and high operating current density, further confirming the importance of electrode losses in the overall energy balance [54]. At industrial current densities, additional effects such as gas bubble formation, mass-transfer limitations, and electrode degradation can further increase voltage losses and affect stable long-term operation [55]. The relative contribution of activation, ohmic, and mass-transfer losses is strongly dependent on electrolyser technology, current density, temperature, electrolyte or membrane properties, and cell design. Therefore, a universal percentage partition of stack heat generation cannot be assigned without specifying the operating conditions.

The irreversible losses discussed above are ultimately converted into heat that must be continuously removed from the electrolyser stack and auxiliary systems. This heat removal is necessary to maintain the stack within its allowable operating temperature range, protect membranes, electrodes, and sealing materials, and avoid accelerated degradation. At the system level, the rejected heat is not limited to the electrochemical stack itself, but can also originate from cooling circuits, electrolyte loops, humid hydrogen and oxygen streams, gas drying, compression stages, and power electronics. Studies on PEM electrolysers have shown that part of the stack electricity input can be recovered as useful heat when suitable heat exchangers and heat pump configurations are implemented [36]. Plant-level analyses further indicate that waste heat from green hydrogen production can include both electrolyser cooling heat and heat from hydrogen and oxygen compression stages, making the whole-plant thermal balance more relevant than the stack balance alone [47]. Similar conclusions have been reported for alkaline electrolysis, where waste heat recovery can improve system efficiency and create additional value when the recovered heat is supplied to district heating networks [48]. In high-temperature electrolysis, the role of heat recovery is even more closely connected to the hydrogen production process, since recovered sensible and latent heat from hydrogen–water mixtures can be upgraded and reused for steam generation [49]. Therefore, thermal management should be considered not only as a cooling requirement, but also as a potential route for waste heat recovery, heat pump integration, and improved system-level energy utilisation.

2.2. Heat Release in Low-Temperature Electrolysers

Low-temperature electrolysers, mainly alkaline water electrolysers, proton-exchange membrane electrolysers, and anion-exchange membrane electrolysers, operate at temperature levels where the generated heat is usually removed as low- or medium-temperature thermal energy. Practical alkaline electrolysis systems commonly operate with concentrated KOH or NaOH electrolytes at about 70–110 °C, while comparative reviews of green hydrogen systems often report typical operating temperatures of about 70–90 °C for AWE, 50–80 °C for PEM electrolysis, and 40–60 °C for AEM electrolysis [53]. AEM-specific studies report somewhat higher typical values, around 50–70 °C [56]; therefore, a broader representative range of approximately 40–70 °C is used in the comparative tables in this review. In these systems, the recoverable heat is mainly transferred through stack cooling circuits, circulating electrolyte streams, and humid product gases. This temperature level is often too low for direct integration into conventional district heating or industrial heat networks, but it is well suited for heat pump-assisted upgrading. Studies on PEM electrolysers have shown that a relevant share of the stack electricity input can be recovered as useful heat when the cooling loop is integrated with heat exchangers and heat pumps [36]. Similarly, alkaline electrolysis studies indicate that a considerable part of the electricity input not converted into hydrogen is transformed into heat, which can be recovered and used for low-temperature applications or district heating when appropriate recovery and temperature-upgrading strategies are used [48]. Therefore, low-temperature electrolysers represent the most immediate opportunity for heat pump integration in green hydrogen production systems, particularly when the objective is to convert unavoidable cooling demand into useful heat output.

In PEM electrolysers, heat release is mainly associated with the stack cooling circuit, which removes the thermal energy generated by electrochemical irreversibilities, ohmic losses, and auxiliary operation. Because PEM systems are commonly operated at relatively low temperatures, the recovered heat is usually available as warm water from the cooling loop rather than as high-grade process heat. This makes direct use possible only when a nearby low-temperature heat demand exists, while higher-temperature applications generally require heat pump upgrading. Van der Roest et al. analysed waste heat utilisation from a 2.5 MWel PEM electrolyser and showed that the recovered heat can be used locally, supplied to a district heating network, or upgraded by a heat pump depending on the temperature requirement and distance to the heat user [36]. Their results indicate that heat recovery can increase the total system efficiency, but the economic and energetic benefit depends strongly on the heat sink temperature, transport distance, and local demand profile [36]. At a larger plant scale, Gómez-de-Arteche-Botas et al. considered a 20 MWe green hydrogen plant and showed that PEM electrolyser cooling heat can be combined with heat from hydrogen and oxygen compression stages, increasing the relevance of plant-level heat recovery compared with stack-only analysis [47]. Therefore, PEM electrolysers are particularly suitable for heat pump-assisted waste heat recovery, not because their heat is high-grade, but because it is continuous, water-based, and available at a temperature level that can be efficiently upgraded for district heating or low-temperature industrial applications.

In alkaline water electrolysers, heat release is closely connected to the liquid electrolyte loop, since the concentrated KOH or NaOH solution acts not only as an ionic conductor but also as an important heat carrier. Practical alkaline systems commonly operate at elevated temperatures, often around 70–110 °C, which improves reaction kinetics and electrolyte conductivity but also requires controlled heat removal to maintain stable operation [53]. Compared with PEM systems, alkaline electrolysers usually have larger thermal inertia and a more direct connection between stack temperature, electrolyte circulation, and cooling demand. The rejected heat can therefore be recovered from the electrolyte cooling loop or from associated water-cooling circuits, depending on the system configuration. This heat is generally suitable for low-temperature applications, but integration into district heating or industrial heat networks may still require heat pump upgrading when the required supply temperature is higher than the electrolyser heat rejection temperature. Frassl et al. showed that waste heat recovery from alkaline electrolysis can improve the overall value of green hydrogen production when the recovered heat is supplied to district heating networks, although the feasibility depends on electrolyser size, cooling strategy, heat demand, and economic boundary conditions [48]. From a heat pump integration perspective, alkaline electrolysers are therefore attractive because they provide a relatively stable liquid heat source, but their corrosive electrolyte, heat exchanger material requirements, and operational constraints must be considered carefully in the thermal integration design.

Anion-exchange membrane electrolysers are also relevant to low-temperature heat recovery, although the available system-level waste heat data are still more limited than for PEM and alkaline electrolysers. AEM electrolysis aims to combine advantages of alkaline and PEM systems by using an anion exchange membrane, alkaline charge transport, and potentially lower-cost non-noble catalysts and components [56]. Reported AEM operating temperatures are generally in the same low-temperature range as conventional alkaline and PEM electrolysis, with Vincent and Bessarabov summarizing typical values around 50–70 °C for AEM systems [56]. From a thermal integration perspective, this means that the rejected heat is expected to be low-grade and mainly available through stack cooling, humid product streams, and balance-of-plant components. However, because AEM technology is still developing, with ongoing challenges related to membrane stability, ionic conductivity, catalyst integration, cell durability, and scale-up, there are fewer plant-level studies quantifying recoverable heat streams and heat pump integration potential [56,57]. As a result, AEM electrolysers should be considered promising future candidates for heat pump-assisted waste heat recovery, but their role in green hydrogen plant thermal integration remains less mature than that of PEM and alkaline systems.

The main limitation of low-temperature electrolyser waste heat is the mismatch between the temperature at which heat is rejected and the temperature required by many potential heat users. Although PEM, alkaline, and AEM electrolysers can provide recoverable heat through cooling circuits, electrolyte loops, and humid product streams, this heat is often not sufficiently hot for direct use in conventional district heating networks or industrial processes. Direct heat recovery is therefore most attractive when the heat demand is nearby and compatible with the available temperature level, for example in low-temperature networks, building heating, domestic hot water preheating, or internal plant processes. When higher supply temperatures are required, heat pumps become necessary in order to upgrade the rejected heat to a useful level. This is demonstrated in PEM-based studies, where heat pump integration allows electrolyser cooling heat to be supplied to higher-temperature users instead of being rejected to the environment [36]. At a larger plant scale, combining stack cooling heat with compression heat can increase the amount of recoverable thermal energy, but the usefulness of this heat still depends on the required sink temperature and the achievable temperature lift [47]. For alkaline electrolysis, similar considerations apply; waste heat recovery can improve the overall value of green hydrogen production, but successful integration depends on the cooling strategy, local heat demand, district heating temperature level, and economic conditions [48]. Therefore, the role of heat pumps in low-temperature electrolysis is not only to recover waste heat, but to solve the temperature mismatch between electrolyser thermal management and practical heat use.

The relevance of heat pump-assisted upgrading becomes clearer when the temperature requirements of district heating systems are considered. Many existing second- and third-generation district heating networks in Europe still operate with relatively high supply temperatures, often around 80–100 °C or higher, which limits the direct use of low-temperature waste heat from PEM, AWE, and AEM electrolysers [58]. In contrast, fourth-generation district heating concepts aim to reduce distribution temperatures substantially, with proposed temperature levels in the range of 30–70 °C and representative annual average values around 50 °C for the supply pipe and 20 °C for the return pipe [59]. Other reviews on low-temperature heating systems report target values for 4GDH of 55–60 °C supply and 25–30 °C return, depending on building-side heating systems and domestic hot water requirements [60]. This transition is highly relevant for green hydrogen plants because lower network temperatures increase the share of electrolyser waste heat that can be used directly or upgraded with a smaller temperature lift. Economic analyses further show that lower district heating temperatures improve the competitiveness of heat pumps, geothermal heat, and industrial excess heat, while high distribution temperatures remain an important barrier to renewable and recycled heat integration [61]. Therefore, the feasibility of using electrolyser waste heat in district heating depends not only on the electrolyser temperature level, but also on the generation of the district heating network, its supply/return temperature regime, the required temperature lift, and the presence of nearby heat demand.

2.3. Heat Release in High-Temperature Electrolysis

High-temperature electrolysis, mainly using solid-oxide electrolysis cells, differs fundamentally from PEM, AWE, and AEM electrolysers because water is supplied as steam and the process operates at elevated temperature, commonly in the range of 500 to 850 °C [5,50]. At these temperatures, part of the energy required for water splitting can be supplied as heat rather than electricity, which explains the higher efficiency potential of SOEC compared with low-temperature electrolysis [5,50]. However, the thermal behaviour of SOEC systems is strongly dependent on operating voltage, current density, steam utilisation, gas composition, pressure, heat losses, and heat exchanger effectiveness. Depending on the operating point, the stack may work in endothermic, thermoneutral or exothermic mode. Below thermoneutral operation, external heat is required to maintain the stack temperature, while above it, excess heat must be removed to avoid thermal stress and material degradation [50,51]. Techno-economic studies show that the availability of external or recovered steam can significantly improve SOEC performance. Nami et al. [51] reported that, for a low-pressure SOEC case, supplying saturated steam at the electrolyser operating pressure through external process integration, without additional electricity consumption for steam generation, increased the reported LHV-based efficiency from 79% to 94%. Therefore, heat release in high-temperature electrolysis should not be treated only as a cooling problem, but as part of a wider thermal integration strategy involving steam generation, feedwater preheating, internal recuperation, product gas cooling, and possible external waste heat sources. Several studies have therefore investigated SOEC integration with heat recovery systems, including solar thermal reactors, engine waste heat recovery, ORC-assisted systems and system-level heat integration [62,63,64,65]. From the perspective of the present review, the most relevant point is that part of the recovered heat may remain at a temperature level that is insufficient for direct steam generation after internal recuperation. In this case, high-temperature or cascade heat pumps can upgrade residual low- or medium-temperature heat and return it to the hydrogen production process. Fang et al. demonstrated this concept by recovering latent heat from the hydrogen–water mixture and upgrading it through a cascade heat pump for steam generation in high-temperature water electrolysis [49]. Thus, in SOEC systems, heat pumps can act not only as external waste heat recovery devices, but also as process-support technologies that reduce auxiliary electric heating demand and improve system-level energy utilisation.

2.4. Main Recoverable Heat Streams in Green Hydrogen Plants

At the plant level, recoverable heat in green hydrogen production is not limited to the electrolyser stack. Although stack losses are the primary heat source in many systems, additional thermal streams arise from electrolyte circulation, humid hydrogen and oxygen streams, water vapour condensation, hydrogen and oxygen compression, gas drying, power electronics, pumps, and cooling equipment. This is important because a stack-only analysis can underestimate the practical heat recovery potential of a complete hydrogen production plant. In PEM systems, waste heat is mainly removed through the stack cooling circuit and can be recovered as warm water, while larger plant-level studies show that compression stages can provide additional heat streams that increase the total recoverable heat available for upgrading [36,47]. Vives et al. modelled a 10 MW PEM electrolysis plant with waste heat recovery coupled to an ORC, where recovered heat was used to support hydrogen compression from 30 bar to 200 bar, showing that waste heat may also be used internally for balance-of-plant energy demands rather than only exported as heat [66]. In alkaline systems, the circulating electrolyte acts as both ionic conductor and heat carrier, making the electrolyte loop an important thermal stream for recovery and temperature control [48,53]. Recent district heating studies further confirm that alkaline electrolyser waste heat can be integrated through direct use or heat-pump configurations, depending on the supply temperature of the network and the available heat demand [37]. In high-temperature electrolysis, the dominant recoverable heat streams are different; they are mainly associated with hot hydrogen–steam mixtures, oxygen-side outlet gases, internal recuperation, and condensation heat, rather than only with a conventional liquid cooling loop [49,50,51]. Therefore, the design of heat pump integration should begin with a plant-level heat-source inventory, including temperature level, heat quantity, continuity, heat carrier, pressure level, contamination risk, and material compatibility. The main recoverable heat streams, their heat carriers, typical heat pump relevance, main limitations, and supporting references are summarized in

Table 1. Main recoverable heat streams in green hydrogen production plants.

Recoverable Heat Stream	Main Technology	Heat Transfer Medium/Form	Heat Pump Relevance	Main Limitation	Refs.
Stack cooling circuit	PEM, AWE, AEM	Warm cooling water/glycol	Very high; convenient source for water-source or high-temperature HP	Temperature may be too low for direct district heating use	[36,37,47]
Electrolyte loop	AWE	Heated KOH/NaOH solution	High; stable liquid heat source through heat exchanger	Corrosion and material compatibility	[37,48,53]
Humid hydrogen stream	PEM, AWE, SOEC	Sensible heat and water vapour	Medium to high; useful if gas cooling and condensation heat are recovered	Gas drying, condensation, and safety requirements	[36,49,50,51]
Humid oxygen stream	PEM, AWE, SOEC	Sensible heat and water vapour	Medium; secondary heat recovery source	Lower priority and integration complexity	[47,50,51]
Hydrogen-steam mixture	SOEC	High-temperature gas mixture and latent heat	Very high for internal steam-generation support	Requires condensation, recuperation, and steam-integration strategy	[49,50,51]
Compression heat	Plant-level H2/O2 compression	Compressor cooling heat	High in large plants; useful for district heating, preheating or internal balance-of-plant support	Depends on pressure level and compressor operation	[47,66]
Gas drying and purification	PEM, AWE, plant level	Low-temperature rejected heat	Medium; may be aggregated with other auxiliary heat streams	Often dispersed and lower temperature	[47,66]
Power electronics and rectifiers	All electrolysis systems	Low-temperature air/water cooling	Low to medium; relevant at large scale	Low-grade and dispersed	[47,67]
Pumps, chillers, and cooling equipment	Balance of plant	Low-temperature heat and cooling loads	Medium in large or hot-climate plants; relevant with thermal storage and smart control	Variable operation and parasitic energy use	[67,68]
External hydrogen hub processes	Hydrogen import, ammonia cracking, PtX	Heat and cold streams	High when electrolysis is integrated with ammonia cracking or PtX processes	Site-specific integration complexity	[69]
District heating interface	Green hydrogen plant + DHN	Upgraded hot water	Very high when a nearby heat demand is available	Seasonal demand, network temperature, and distance	[37,58,59,60,61,70]

Note: Heat-pump relevance is assessed qualitatively according to the temperature level, recoverability, continuity, and practical accessibility of each heat stream as a heat-pump source; the categories are comparative rather than quantitative performance ratings.

.

The suitability of each heat stream for direct use or heat pump upgrading depends primarily on temperature level, stability, timing, and connection to heat demand. Low-temperature liquid streams, such as PEM cooling water or alkaline electrolyte cooling loops, are usually the most convenient heat sources for vapour-compression heat pumps because they can be coupled to heat exchangers and provide relatively stable operation. Humid product gases and hydrogen–water mixtures can offer both sensible and latent heat, but their recovery requires additional condensation, gas cooling, separation, and hydrogen safety considerations. Compression heat can be valuable in large plants, especially where hydrogen is compressed for storage, pipeline injection, transport, or downstream power-to-X processes; however, its integration depends on compressor configuration, cooling temperature, and operating schedule [47,66]. Dynamic modelling studies also show that the value of electrolyser excess heat depends strongly on operating strategy; flexible hydrogen production, heat sales, and district heating integration can affect both hydrogen production costs and the economic value of recovered heat [67]. In hydrogen import or port-based hydrogen hubs, heat recovery may extend beyond electrolysers alone; Kramer et al. showed that wind-powered water electrolysis waste heat can be combined with thermal streams from ammonia cracking, creating synergies between hydrogen production, import, and downstream conversion processes [69]. In hot climates, the problem becomes even more complex because electrolyser thermal management interacts with cooling equipment, pumps, chiller operation and, in some layouts, thermal storage. Allan et al. showed that optimized PEM electrolyser waste heat utilisation with thermal storage and smart control can reduce chiller energy consumption, pump energy use, energy costs, and CO₂ emissions under conditions associated with the Persian Gulf region [68]. These results show that plant-level heat recovery should be evaluated as part of the full thermal management system, not only as an additional heat output.

For heat pump design, the most important distinction is whether the heat stream can be recovered continuously and safely without disturbing electrolyser operation. Heat recovery should therefore be designed as an additional layer of thermal management, with bypass cooling, backup heat rejection and, where appropriate, thermal storage to ensure that the electrolyser remains within its required operating temperature range when external heat demand is unavailable. This is particularly relevant for district heating integration, where heat demand is seasonal and network temperatures may not always match the available waste heat temperature. Reviews of fourth-generation district heating show that low-temperature waste heat often requires heat pumps, while distance between the heat source and heat user remains one of the main barriers to practical integration [70]. Consequently, the feasibility of heat pump-assisted recovery depends not only on the heat source itself, but also on network temperature, distance, storage capacity, heat market design, and the flexibility of hydrogen production.

2.5. Thermal Management Requirements and Operational Constraints

Thermal management in green hydrogen production is a primary operating requirement, because electrolysers must remain within a defined temperature range to maintain efficiency, protect membranes, seals, and electrodes, limit degradation, and ensure safe gas separation. The heat balance depends on current density, cell voltage, stack efficiency, water/electrolyte flow rate, and auxiliary operation. Liso et al. showed that PEM cell operation changes with operating temperature and load; at low current densities, the cell may absorb heat from the surroundings, while at higher current densities, irreversible losses dominate and heat rejection becomes necessary [71]. Dynamic modelling of PEM electrolysers under intermittent renewable input further confirms that efficiency, losses, and heat generation vary under fluctuating operation [72]. Therefore, recoverable heat should not be assumed constant, but should be evaluated according to actual operating conditions, part-load operation, and the start-up and shut-down periods.

From a heat pump integration perspective, waste heat recovery must not replace the safety function of the cooling system. It should be added as a heat utilisation layer, while bypass cooling, dry coolers, chillers, buffer tanks, or thermal storage should remain available when heat demand is low, the heat pump is offline, or the required sink temperature is too high. This is especially important for district heating, where heat demand is seasonal and may not coincide with hydrogen production driven by renewable electricity or electricity prices. Studies on PEM and alkaline electrolysers show that the usable share of waste heat depends strongly on heat demand, storage, cooling strategy, plant size, and economic boundary conditions [36,48]. In hot climates, cooling equipment and chiller operation can become significant parasitic loads, making smart control and storage important for reducing energy use and improving total system efficiency [68]. Thus, heat pump-assisted recovery must simultaneously ensure stable electrolyser operation, reliable heat rejection, and useful heat delivery when the temperature level and demand profile are appropriate.

2.6. Plant-Level Waste Heat Availability and System-Level Relevance

At plant level, the relevance of electrolyser waste heat depends not only on the amount of heat generated, but also on its temperature, continuity, recoverability, and match with local heat demand. For low-temperature electrolysers, a significant fraction of the supplied electricity that is not converted into hydrogen chemical energy appears as heat in cooling circuits and auxiliary systems. Van der Roest et al. reported that 14 to 15% of the stack electricity input of a PEM electrolyser could be recovered as heat, depending on the selected integration concept, while Gómez-de-Arteche-Botas et al. showed that, in a 20 MWe green hydrogen plant, combining PEM cooling heat with compression heat could create a substantial plant-level energy stream for heat pump upgrading [36,47]. District-scale simulation studies also indicate that electrolyser waste heat availability follows the operating profile of the electrolyser, which makes load factor, operating strategy, and temporal matching essential for practical utilisation [73]. These results are consistent with recent case studies where heat pump-assisted electrolysis waste heat recovery has been assessed as a route to reduce hydrogen production costs through heat sales and improved district heating integration [37]. Thus, heat recovery should be assessed at plant scale, because the combination of stack cooling heat, compression heat, and auxiliary thermal streams can create a larger integration opportunity than the electrolyser stack alone. These case-study results should not be compared directly because the reported values reflect different system boundaries, operating assumptions, heat-use pathways, and economic conditions.

However, the theoretical amount of available heat should not be confused with the useful heat that can actually be delivered. The usable fraction depends on the required supply temperature, heat pump COP, district heating return temperature, distance to the heat user, thermal storage capacity, and coincidence between hydrogen production and heat demand. Studies on alkaline electrolysis coupled with district heating show that heat recovery can improve the economic value of green hydrogen production, but the benefit depends strongly on plant size, cooling strategy, heat price, and local demand conditions [37,48]. Regarding PEM electrolysis, Massulli et al. showed that waste heat recovery can improve system performance and reduce both specific energy consumption and the levelized cost of hydrogen when the recovered heat is coupled with district heating and compatible end-user heat emission systems [19]. At a wider energy-system level, Moradpoor et al. showed that integrating industrial hydrogen production with district heating in cold climates can support low-carbon heat supply, but that the benefits depend on building renovation level, district heating demand, and hydrogen transport assumptions [74]. Puschnigg et al. quantified electrolysis waste heat potentials across IEA DHC member countries and emphasized that low- to moderate-temperature heat from AWE and PEM electrolysers is suitable for low-temperature district heating networks, while conventional networks require additional temperature upgrading, typically by heat pumps [75]. In multi-energy communities, Decormis et al. showed that waste heat recovery from hydrogen technologies can substantially improve economic performance. In their hydrogen-system scenario, fuel-cell waste heat covered 28% of space-heating demand at a hydrogen price of 6 CHF/kg and 69% at 2 CHF/kg, reducing total operating costs by 16.0% and 75.1%, respectively, compared with the heat-pump-only baseline [76]. They also reported that WHR expands the profitable operating range of hydrogen technologies, allowing fuel cells to remain profitable at hydrogen prices 30% higher than without WHR, while the yearly operational cost savings associated with fuel cells become cost-competitive with those of batteries or photovoltaics when the ratio of the hydrogen price (CHF/kg) to the electricity price (CHF/kWh) falls below 25 [CHF/kg]/[CHF/kWh] [76]. Therefore, plant-level waste heat availability should be evaluated using integrated indicators that include hydrogen output, useful recovered heat, heat pump electricity input, avoided heat production, network temperature level, heat-market value, and the temporal flexibility of both hydrogen and heat production.

The analysis in this section shows that green hydrogen production plants release several recoverable heat streams with different temperature levels, heat carriers, and operational constraints. Low-temperature electrolysers mainly provide liquid-based heat through cooling circuits or electrolyte loops, while high-temperature electrolysis introduces additional opportunities for product-stream heat recovery, condensation heat, and internal steam-generation support. However, the useful application of these heat streams is often limited by temperature mismatch, dynamic operation, seasonal heat demand, and the need to preserve safe electrolyser temperature control. Consequently, heat-pump technologies become key enabling options for upgrading electrolyser waste heat and linking hydrogen production with district heating, industrial heat, and internal process support, while thermal storage may serve as an auxiliary buffer where temporal mismatch is significant.

3. Heat Pump Technologies and Selection Approach for Green Hydrogen Plants

Heat pumps enable the recovery and upgrading of low- and medium-temperature heat streams from green hydrogen production plants when the temperature level of the available heat is insufficient for direct use. The principal heat sources include stack cooling circuits, electrolyte loops, humid product-gas streams, compression stages, and auxiliary equipment, whereas potential heat sinks include district heating networks, industrial processes, thermal storage, feedwater preheating, and steam-generation systems. The appropriate heat pump solution depends primarily on the available source temperature, required sink temperature, and resulting temperature lift, together with the temperature profiles of both streams, operating conditions, and practical system constraints.

In this section, heat pump technologies are classified according to their thermodynamic driving principle and cycle configuration. Vapour-compression systems are therefore distinguished from thermally driven absorption and adsorption systems, while single-stage, multistage, cascade, and transcritical arrangements are treated as configurations within the vapour-compression family. High- and very-high-temperature operation is considered an operating condition rather than a separate heat pump technology. A two-stage technology-selection approach is subsequently proposed. First, an operating map is used to identify thermodynamically feasible heat-recovery and heat-pump options from the source and sink temperatures. Second, where more than one technology is feasible, the alternatives are compared according to temperature-profile matching, thermodynamic performance expressed primarily through COP, techno-economic performance expressed through levelized cost of heat (LCoH), operating flexibility, capacity, and practical implementation constraints. System-level economic and environmental feasibility is considered separately in Section 5.

3.1. Role of Heat Pumps and Thermodynamic Requirements

Heat pumps can be integrated into green hydrogen plants as thermal interface technologies that can minimise unavoidable heat rejection and help to meet useful heat demand. Their role is not to improve the electrochemical conversion inside the electrolyser directly, but to increase the useful value of heat that would otherwise be rejected through cooling systems. This is especially relevant because the main recoverable streams in PEM, alkaline, and AEM electrolysers are usually water- or electrolyte-based heat sources at low or medium temperature, while many useful heat sinks require higher supply temperatures [36,47,48]. In high-temperature electrolysis, heat recovery can also support internal process needs, particularly feedwater preheating and steam generation [49].

From a thermodynamic perspective, direct heat recovery should be considered first whenever the temperature level of the available heat source is compatible with the required heat sink. Under such conditions, introducing a heat pump adds compressor electricity consumption and system complexity without providing a necessary temperature benefit. Heat-pump-assisted recovery becomes relevant when a temperature mismatch exists between the source and sink. In practical heat exchangers, direct recovery also requires a finite temperature difference between the two streams; therefore, temperature equality represents only the theoretical limit for direct heat transfer.

Figure 2 presents an indicative operating map linking the available heat-source temperature and required heat-sink temperature with the heat-recovery and heat-pump configurations considered in this review.

The temperature lift, defined by the relationship between the available source temperature and the required heat-sink temperature, is therefore the principal thermodynamic parameter governing heat pump selection. Moderate lifts can generally be addressed by conventional vapour-compression systems, whereas increasing sink temperature and temperature lift may require multistage compression, cascade, or transcritical configurations. COP remains the principal indicator of the thermodynamic efficiency of a heat-pump cycle and generally decreases as the required temperature lift increases. However, technology selection cannot be based on nominal COP alone because actual performance also depends on heat-exchanger temperature differences, temperature-profile matching, working fluid, component efficiency, part-load operation, and the temporal profiles of both the heat source and heat sink [77,78,79,80].

The direct heat-recovery region represents the first level of the selection hierarchy. Above this region, temperature upgrading is required; the map identifies one or more potentially applicable technologies. The overlap between the operating regions is intentional; temperature levels alone do not necessarily provide a unique technology selection. For example, a given source-to-sink temperature combination may be achievable using multistage, cascade, or transcritical vapour-compression systems. The operating map identifies thermodynamically feasible configurations from the source and sink temperatures but does not quantify their expected efficiency. Reported COP values as a function of temperature lift are therefore synthesized separately in Figure 3. A second selection stage is therefore required to distinguish between these candidates using their actual operating characteristics and techno-economic performance.

3.2. Vapour-Compression Heat Pump Systems

Vapour-compression heat pumps represent the most mature and directly applicable heat-pump technology for upgrading low- and medium-temperature heat from green hydrogen production systems. A basic system comprises an evaporator, compressor, condenser, or gas cooler, an expansion device, and a closed working-fluid circuit [79,80]. Depending on the required temperature lift and sink-temperature profile, the cycle may be configured as single-stage, multistage, cascade, or transcritical.

Commercial high-temperature vapour-compression heat pumps already cover a broad operating range. Arpagaus et al. identified more than 20 commercial products from 13 manufacturers providing sink temperatures of at least 90 °C, with heating capacities from approximately 20 kW to 20 MW and reported COP values from 2.4 to 5.8 depending on source and sink temperatures, temperature lift, and working fluid [77]. Jiang et al. reported industrial applications with heating capacities from 60 kW to 18 MW and output temperatures above 80 °C, while development targets include temperatures above 100 °C, capacities above 1 MW, and COP values above 4 at a temperature lift of 40 K [81]. These data demonstrate that high-temperature operation should be regarded as an achievable operating range of several vapour-compression configurations rather than as an independent cycle category. Recent international work within IEA HPT Annex 58 has further consolidated experimental, demonstrative, and industrial experience on high-temperature heat pumps, with contributions from major research organizations including DTU, Fraunhofer ISE, and TNO [82].

3.2.1. Single-Stage and Multi-Stage Compression Systems

In green hydrogen plants, the evaporator of a vapour-compression heat pump can be coupled to stack cooling water, alkaline electrolyte cooling loops through an intermediate heat exchanger, humid product-gas cooling circuits, compression intercoolers, or auxiliary cooling systems [36,37,47,48,66]. The condenser can supply useful heat to district heating networks, industrial consumers, thermal storage, feedwater preheating, or other plant processes [36,37,47,83].

For PEM and alkaline electrolysis, a single-stage vapour-compression heat pump is generally the first configuration to consider when the required temperature lift is moderate. Industrial heat-pump assessments identify temperature lifts from 30 to 50 K as a representative range for favourable operation, although actual performance depends strongly on the specific source and sink temperatures [26,78]. As the required lift increases, COP decreases while compressor pressure ratio and discharge temperature become progressively more restrictive. Consequently, single-stage applicability should be assessed from the actual operating temperatures rather than from nominal performance values alone.

The importance of this distinction is illustrated by van der Roest et al., who assessed waste-heat recovery from a 2.5 MWel PEM electrolyser through direct local use, district-heating integration, and heat-pump-assisted temperature upgrading [36]. For district-heating integration, the reported levelized cost of electrolyser heat was around EUR 8.4–8.9/MWh_t without a heat pump at about 54 °C, increasing to EUR 36.9/MWh_t when a heat pump was used to raise the supply temperature to 100 °C [36]. This case demonstrates that raising the temperature level of electrolyser waste heat does not automatically improve the economic result and that the additional value of the upgraded heat must be considered together with the energy required for temperature lifting.

When the required temperature lift approaches or exceeds the practical range of a single-stage cycle, multistage compression can reduce the pressure ratio and discharge temperature associated with each individual compression stage. Economizers, vapour injection, intermediate cooling, and two-stage compression can expand the operating range and improve cycle performance at elevated sink temperatures [79,80,81]. These arrangements introduce additional components and control requirements but may remain preferable to more complex cascade architectures when the total temperature lift can still be covered efficiently by a single working-fluid circuit.

3.2.2. Cascade Configurations

Cascade configurations extend the operating range of vapour-compression heat pumps when the required temperature lift becomes too large for efficient single- or multistage operation, or when one working fluid cannot satisfactorily cover the complete source-to-sink temperature range. A cascade system combines two or more refrigerant cycles through an intermediate cascade heat exchanger, which acts simultaneously as the condenser of the lower-temperature stage and the evaporator of the higher-temperature stage [80,81]. Dividing the total lift between the stages reduces individual compressor pressure ratios and discharge temperatures and allows the working fluid of each stage to be matched to its respective temperature range.

Reported cascade performance varies considerably with refrigerant pair, source and sink temperatures, and intermediate temperature. Adamson et al. reported COP values from 2.0 to 4.4 for subcritical cascade cycles below 100 °C and up to 2.5–4.9 for configurations delivering heat above 100 °C [80]. Subcritical–transcritical cascade arrangements have demonstrated COP values of 1.95–5.24 depending on operating conditions [80]. Jiang et al. summarized experimental systems including an R290/R600 cascade achieving an 85 K temperature lift with COP above 2, a 20 kW prototype delivering hot water up to 116 °C, and water-source cascade systems producing hot water up to 170 °C, with COP values from 2.3 to 3.18 [81].

Cascade systems are particularly applicable for high-temperature electrolysis where recovered heat can support steam generation. Fang et al. proposed a two-stage cascade heat pump to recover latent heat from hydrogen–water and oxygen product streams, generating steam at 110 °C [49]. The study reported reductions in electricity consumption of up to 65% relative to electric heating, while lower-latent-heat cases still achieved substantial savings. For a 10 MW electrolyser, the reported results further indicated a strong influence of product-stream water content on recoverable energy and economic performance [49]. These results were obtained for a simulated 10 MW electrolyser producing steam at 110 °C, assuming a compressor isentropic efficiency of 75%, an electric-heater efficiency of 95%, and an electricity price of USD 80.9/MWh; under these assumptions, the reported payback period was 3.43 years.

The main disadvantage of cascade systems is their increased complexity. Additional compressors, refrigerant circuits, heat exchangers and control systems increase capital and maintenance costs, while the cascade heat exchanger introduces an additional finite temperature difference and associated thermodynamic loss [80,81]. Cascade configurations should therefore not be selected from temperature lift alone; their efficiency advantage must be sufficient to compensate for the additional system cost and complexity.

3.2.3. Transcritical Configurations

Transcritical vapour-compression cycles are particularly relevant when the useful heat sink undergoes a substantial sensible temperature increase. Unlike subcritical condensation, heat rejection in a transcritical gas cooler occurs over a continuously varying refrigerant temperature, which can provide improved temperature-profile matching when the sink temperature rises significantly through the heat exchanger [80,84].

CO₂ (R744) is the most established working fluid for transcritical heat-pump operation because its low critical temperature requires heat rejection above the critical point in many heating applications. Kong et al. identified CO₂ heat pumps as promising for high-temperature applications with large sink-side temperature glides, including applications above 100 °C, although high operating pressures and potentially reduced COP under unfavourable gas-cooler conditions remain important constraints [84].

Transcritical operation should therefore not be selected solely because a high sink temperature is required. Its principal advantage relative to conventional condensation or cascade operation is temperature-profile matching. In an operating region where both cascade and transcritical systems are thermodynamically feasible, a transcritical configuration becomes particularly attractive when the sink exhibits a large sensible temperature glide, whereas a cascade configuration may be favoured when a large overall source-to-sink temperature lift must be divided between distinct temperature stages. The final choice requires comparison under the actual source and sink temperature profiles.

3.3. Thermally Driven Heat Pump Systems

Thermally driven heat pumps provide an alternative to mechanically driven vapour-compression systems when suitable thermal energy is available as the driving input. Their electrical consumption is generally limited to pumps, valves, and auxiliary equipment, while their applicability depends strongly on the driving-source temperature, useful heat-sink temperature, and required heat-rejection level [85,86,87,88,89]. For green hydrogen plants, the most relevant concepts are absorption heat pumps, absorption heat transformers, and adsorption heat pumps.

3.3.1. Absorption Heat Pump Systems and Heat Transformers

Absorption systems can operate either as heat pumps, using driving heat to recover and deliver low-grade heat at a useful temperature, or as heat transformers, where part of an intermediate-temperature heat flow is upgraded to a higher temperature while the remaining fraction is rejected at a lower temperature [85,86]. This distinction is particularly relevant to hydrogen production plants because an absorption heat transformer can potentially upgrade part of an electrolyser or plant waste-heat stream with very low mechanical electricity consumption.

Cudok et al. identified 48 absorption heat-transformer installations in 42 industrial plants between 1981 and 2019, representing a total installed capacity of 134 MW [86]. H₂O/LiBr has become the dominant industrial working pair, with reported industrial COP values around 0.47 and upgraded temperatures reaching 165 °C under appropriate operating conditions [86]. These installations represent industrial applications rather than demonstrated electrolyser-integrated systems; their relevance in the current study is therefore to demonstrate the technological maturity and achievable operating range of the concept.

For green hydrogen plants, absorption heat transformers may become attractive when a stable intermediate-temperature heat source exists, a relatively high-temperature heat demand is present, and electricity consumption is particularly costly or constrained. However, only part of the available heat is upgraded, and their lower dynamic flexibility makes them less suitable than vapour-compression systems when the objective is to maximize recovery from variable low-temperature electrolyser cooling streams.

3.3.2. Adsorption Heat Pump Systems

Adsorption heat pumps employ a solid sorbent such as silica gel, zeolite, activated carbon, metal–organic frameworks, or composite materials together with an adsorbate such as water, ammonia, methanol, or ethanol [87,88,89]. Their advantages include low electrical consumption, potentially environmentally benign working pairs, and the possibility of combining thermal upgrading with heat-storage functionality [87,88].

Their main limitations are intermittent cyclic operation, comparatively low specific heating or cooling power, slow heat and mass transfer within the adsorbent bed, bulky heat exchangers, and lower dynamic controllability than vapour-compression systems [87,88]. These characteristics are important in electrolysis plants because stack thermal management must remain continuous and reliable. Adsorption systems are therefore more appropriately regarded as auxiliary, storage-coupled, or niche heat-recovery technologies than as the primary cooling or temperature-upgrading system for an electrolyser stack.

3.4. Working Fluids, Components and Practical Constraints

The working fluid should be selected only after the heat-pump architecture has been identified from the operating map and the Stage 2 technology comparison.

Table 2. Representative refrigerants and working pairs for the heat-pump technologies considered in this review.

Technology/Configuration	Refrigerant/Working Pair	Main Reason for Suitability	Main Constraints Affecting Selection	Refs.
Single-stage	R717; R1234ze(Z); R1233zd(E); R1336mzz(Z); hydrocarbons (e.g., R600/R601)	Suitable for low- to medium-temperature upgrading and, with high-critical-temperature fluids, for selected high-temperature applications. Natural refrigerants and low-GWP HFO/HCFO fluids provide alternatives to legacy high-GWP refrigerants.	R717 toxicity; hydrocarbon flammability; pressure ratio and compressor discharge-temperature limits; compressor, lubricant and material compatibility at elevated sink temperatures.	[77,79,81,90]
Multi-stage	R717; R1234ze(Z); R1233zd(E); R1336mzz(Z); R718; hydrocarbons	Staging permits larger temperature lifts while reducing the pressure ratio and discharge temperature of individual compression stages. Higher-critical-temperature fluids are particularly relevant at elevated sink temperatures.	More complex compressor staging and control; intermediate cooling requirements; volumetric-capacity limitations for some fluids; specialized compression technology for R718.	[27,77,79,81,90]
Cascade	Stage-specific refrigerant combinations; demonstrated R290/R600 pair; high-temperature-stage candidates include R717, R1233zd(E), R1336mzz(Z) and R718	Allows different working fluids to be matched to the low- and high-temperature stages, reducing the operating burden on each individual cycle and extending the achievable temperature lift.	Two refrigerant circuits; additional refrigerant charge and components; cascade heat-exchanger temperature difference; intermediate-temperature optimization; combined safety constraints of both fluids.	[80,81]
Transcritical	R744 (CO2)	High volumetric heating capacity and favourable matching to heat sinks with a large sensible temperature glide; particularly relevant for high-temperature water and process-heating applications.	High operating pressure; COP strongly dependent on gas-cooler outlet conditions and temperature-profile matching; compressor and component pressure-rating requirements.	[80,84]
Absorption heat pump/heat transformer	H2O/LiBr; NH3/H2O	Thermally driven operation with low mechanical electricity consumption. H2O/LiBr is established for industrial absorption heat transformers and can provide upgraded heat at elevated temperatures.	Crystallization and corrosion risks for LiBr systems; vacuum operation for water-based systems; toxicity of NH3; lower dynamic flexibility and dependence on driving-heat and heat-rejection temperatures.	[85,86,89]
Adsorption heat pump	H2O/silica gel; H2O/zeolite; other adsorbent-adsorbate combinations using activated carbon, ammonia, methanol or ethanol	Can utilise low-grade driving heat with very low mechanical electricity consumption and can potentially be combined with thermal-storage functions.	Cyclic operation; low specific heating power; slow heat and mass transfer; bulky adsorbent beds and heat exchangers; vacuum operation for water-based systems.	[87,88,89]

Note: The listed working fluids and working pairs are representative rather than exhaustive. Final suitability depends on evaporation and heat-rejection temperatures, temperature lift, compressor technology, capacity, material compatibility, safety requirements, and applicable environmental regulations. For thermally driven systems, working pairs are listed rather than single refrigerants. R134a and R245fa have been used historically in high-temperature heat pumps but are not included among the preferred forward-looking candidates because of their high GWP.

summarizes representative refrigerants and working pairs applicable to the technologies considered in this review. The listed fluids are indicative rather than exclusive, since suitability depends on the specific evaporation and heat-rejection temperatures, pressure ratio, compressor technology, safety requirements, and applicable environmental regulations.

After the cycle architecture has been identified from the required temperature lift and source–sink temperature profiles, the selected concept must be checked against working-fluid, compressor, and component limitations. Working-fluid selection should therefore be treated primarily as a feasibility and optimization step rather than as the initial basis for selecting between cycle architectures.

For vapour-compression systems, natural and low-GWP working fluids provide different advantages and limitations. R717 has a critical temperature of 132.3 °C and favourable thermodynamic properties but introduces toxicity constraints. R744 provides high volumetric heating capacity and is particularly suited to transcritical operation and large temperature glides but requires high operating pressures. Hydrocarbons such as R600 and R601 can provide favourable thermodynamic performance but are flammable. R718 has negligible direct environmental impact and a very high critical temperature, but its low volumetric heating capacity at low pressure requires specialized compressor technology [84,90]. Existing high-temperature systems have also employed R134a and R245fa; however, their comparatively high GWP values of 1430 and 1030, respectively, limit their long-term attractiveness. By contrast, lower-GWP alternatives such as R1336mzz(Z) and R1234ze exhibit GWP values of 2.08 and 1.37, respectively, according to Regulation (EU) 2024/573 [77,79,81,91].

The transition to low-GWP alternatives nevertheless introduces additional practical considerations. These include refrigerant cost and commercial availability, leakage management, compatibility with compressors and lubricants, and, for some A2L (Refrigerant Safety Class) working fluids, mild flammability. These factors should be considered together with thermodynamic performance when selecting the working fluid for high-temperature applications [82].

At elevated sink temperatures, compressor and component limitations become increasingly important, particularly when sink temperatures approach or exceed 150 °C. Increasing temperature lift raises pressure ratio and compressor discharge temperature and places stronger requirements on compressor technology, lubrication, materials, and heat exchangers [77,78,79]. Under these conditions, compressor availability, allowable discharge temperature, lubricant stability, and component pressure–temperature limits can become decisive constraints. The practical operating limit of a candidate architecture must therefore be evaluated together with compressor availability at the required capacity and with the pressure, temperature, and material constraints of the selected working fluid.

Green hydrogen plants additionally impose specific process-integration and safety requirements. Vapour-compression heat pumps should normally be connected to electrolyser cooling streams through an intermediate heat exchanger and a clean water or glycol circuit [36,37,48]. Such hydraulic separation reduces the consequences of refrigerant leakage and avoids direct contact between the refrigerant circuit and hydrogen-, oxygen-, or electrolyte-containing streams. The original electrolyser cooling function must remain available independently of heat-pump operation through bypass cooling, backup heat rejection, or other suitable redundancy. Refrigerant flammability, toxicity, and pressure level must also be considered together with hydrogen safety zoning, ventilation, and leak-detection requirements.

These constraints can eliminate technologies that appear thermodynamically feasible in the operating map. Consequently, working-fluid and component compatibility represents a feasibility check within the second selection stage, rather than a substitute for temperature-based technology identification.

3.5. Performance Evaluation and Technology Selection

The selection approach proposed in this review separates thermodynamic feasibility from subsequent performance and techno-economic technology selection. In Stage 1, the operating map in Figure 2 is used to identify the heat-recovery and heat-pump configurations capable of providing the required source-to-sink temperature relationship. Where direct heat recovery is feasible, it represents the preferred thermodynamic baseline. Where temperature upgrading is necessary, the map may identify one or several candidate heat-pump technologies.

Where the Stage 1 operating regions overlap, Stage 2 determines which of the technically feasible candidates is preferable. Temperature-profile matching is considered first because technologies with similar nominal source and sink limits may interact differently with the actual heat-source and heat-sink profiles. For example, transcritical operation may benefit from a large sink-side temperature glide, whereas cascade operation may become advantageous when a large total temperature lift can be divided efficiently between two temperature levels [80,81,84].

Thermodynamic performance provides the next level of comparison between the technically feasible alternatives. Figure 3 summarizes representative reported COP values for vapour-compression heat pumps over temperature lifts from 30 to 100 K, including experimental and market data as well as industrial and demonstration systems [77,82]. The compiled data show the expected decrease in COP as the required temperature lift increases. Based on the average trend of the reported values, COP decreases from 4.6 at a 30 K lift to about 3.6 at 50 K, 3.0 at 70 K, and 2.7 to 2.8 at lifts approaching 90–100 K. The temperature lift in Figure 3 is defined from the external process-fluid temperatures as ΔT = T_sink,out − T_source,in, using the reported operating temperatures where available; it should therefore not be interpreted as the refrigerant condensation-to-evaporation temperature difference.

Figure 3. Representative reported COP values of vapour-compression heat pumps as a function of temperature lift, together with the average COP trend and local literature spread. The compiled data include experimental/market [77], industrial [82], and demonstration systems [82]. The average curve represents an indicative synthesis of the reported values and should not be interpreted as a universal performance correlation, since actual COP depends on cycle configuration, working fluid, source- and sink-temperature profiles, component performance, and operating conditions.

Figure 3. Representative reported COP values of vapour-compression heat pumps as a function of temperature lift, together with the average COP trend and local literature spread. The compiled data include experimental/market [77], industrial [82], and demonstration systems [82]. The average curve represents an indicative synthesis of the reported values and should not be interpreted as a universal performance correlation, since actual COP depends on cycle configuration, working fluid, source- and sink-temperature profiles, component performance, and operating conditions.

COP therefore provides an important first performance indicator for comparing thermodynamically feasible heat-pump configurations, particularly because it directly reflects the electricity required to deliver a given amount of useful heat. However, COP alone does not account for differences in capital investment, system complexity, maintenance requirements, electricity price, annual operating hours, or plant lifetime. A configuration with a higher COP may therefore not necessarily represent the most favourable overall solution if the efficiency improvement is achieved through substantially greater equipment and integration costs.

The dispersion around the average trend is also important, as systems operating at similar temperature lifts can exhibit substantially different COP values due to differences in cycle architecture, working fluid, source- and sink-temperature profiles, component efficiency, and operating conditions. Figure 3 should therefore be interpreted as an indicative synthesis of reported vapour-compression performance rather than as a universal COP correlation.

For this reason, the principal broader techno-economic indicator proposed for comparing the remaining candidates is the levelized cost of heat (LCoH). LCoH expresses the discounted lifetime cost of delivering useful heat relative to the discounted useful heat supplied and therefore combines capital expenditure, operating and maintenance costs, and electricity consumption with the actual thermal output. A generic formulation adapted to heat-pump systems can be expressed as follows:

$$LCoH={\displaystyle \frac{I_0+{\sum }_{t=1}^n\frac{C_{OM,t}+C_{el,t}}{{\left(1+r\right)}^t}}{{\sum }_{t=1}^n\frac{Q_{heat,t}}{{\left(1+r\right)}^t}}}$$

(1)

where I₀ is the initial investment, C_OM,t is the annual operation and maintenance cost, C_el,t is the electricity cost, Q_heat,t is the useful heat delivered, r is the discount rate, and n is the analysis period [92].

For heat pumps, electricity consumption links thermodynamic efficiency directly to LCoH. For annual assessment,

$$E_{el,ann}={\displaystyle \frac{Q_{heat,ann}}{SPF}}$$

(2)

Higher COP generally reduces the electricity-related component of LCoH, but the technology with the highest COP does not necessarily provide the lowest LCoH. For example, a cascade system may improve thermodynamic efficiency by dividing a large temperature lift between two cycles, but the additional compressors, refrigerant circuits, and cascade heat exchanger increase capital and maintenance costs. Conversely, a technically simpler configuration may operate at a slightly lower COP while achieving a lower lifetime cost of useful heat. COP and LCoH should therefore be regarded as complementary indicators. COP characterizes thermodynamic efficiency, whereas LCoH provides a broader techno-economic basis for comparing feasible technologies. For all candidate technologies, LCoH should be calculated using identical system boundaries, electricity-price assumptions, lifetime, discount rate, operating hours, and useful heat demand. It should also preferably be based on annual or time-dependent heat-pump performance rather than a single nominal COP, since source temperature, sink temperature, part-load operation, and demand profiles influence real electricity consumption [77,78].

Although Figure 3 provides indicative COP levels for different temperature lifts, a universal minimum acceptable COP cannot be defined because economic feasibility depends on electricity price, operating hours, capital and maintenance costs, system lifetime, and the value of the alternative heat supply. COP should therefore be used as a thermodynamic performance indicator, while LCoH provides a broader economic comparison. After the LCoH assessment, the preferred candidate must also satisfy practical criteria, including operating flexibility and part-load response, heating capacity and scalability, component availability, working-fluid compatibility, system complexity, maintainability, hydraulic integration, and safety requirements. These criteria are summarized in

Table 3. Stage 2 technology-selection criteria for heat-pump technologies identified as thermodynamically feasible by the operating map (Stage 1).

Technology/Configuration	Main Selection Feature	COP and LCoH Considerations	Operational and Scalability Considerations	Refrigerant/Working Pair and Main Feasibility Constraints	Refs.
Single-stage	Best suited to moderate temperature lifts and conventional source-sink temperature profiles.	High COP and favourable LCoH are generally achievable at moderate temperature lifts, but both deteriorate as temperature lift and electricity consumption increase.	Mature technology; good controllability; broad commercial capacity range.	R717, hydrocarbons, HFO/HCFO and other suitable refrigerants; limited primarily by pressure ratio, compressor discharge temperature, and refrigerant safety requirements.	[77,78,79,90]
Multi-stage	Suitable for moderate-to-large lifts when the required temperatures remain achievable within a single working-fluid architecture.	Improved COP at larger temperature lifts can reduce electricity-related LCoH, but the additional staging and components increase CAPEX.	Good load controllability; additional compressors, intercooling/economizer components, and more complex control.	Working fluid selected according to the required operating temperatures; compressor staging, discharge temperature, lubrication, and component availability become important constraints.	[79,80,81]
Cascade	Particularly suitable for large overall temperature lifts and for dividing the source-to-sink lift between distinct temperature levels.	Higher COP may be achieved at large temperature lifts, but additional compressors, refrigerant circuits and heat exchangers increase CAPEX and may offset the LCoH benefit.	Suitable for high-temperature and large-capacity applications; greater system and control complexity than single- or multi-stage systems.	Different refrigerants may be selected for the low- and high-temperature stages; feasibility depends on refrigerant pairing, intermediate-temperature optimization, and additional heat-exchanger losses.	[49,80,81,84]
Transcritical	Particularly advantageous when the heat sink exhibits a large sensible temperature glide, allowing improved temperature-profile matching in the gas cooler.	COP depends strongly on gas-cooler temperature matching, while LCoH is additionally influenced by the cost of high-pressure components.	Potentially simpler cycle architecture than cascade systems, but performance is highly sensitive to operating conditions and control.	R744 is the principal representative working fluid; high operating pressure, compressor availability and sensitivity to gas-cooler outlet conditions are key constraints.	[80,84]
Absorption heat pump/heat transformer	Relevant where suitable driving heat or stable intermediate-temperature waste heat is available and low electrical consumption is advantageous.	Low electrical consumption can reduce LCoH, but relatively low thermal COP and larger equipment may offset this advantage.	Best suited to relatively stable operation; generally slower dynamic response than vapour-compression systems.	H2O/LiBr and NH3/H2O are representative working pairs; crystallization, corrosion, vacuum operation, toxicity, and heat-rejection requirements may limit applicability.	[85,86,89]
Adsorption heat pump	Suitable for low-grade thermally driven heat recovery, auxiliary operation and storage-coupled applications.	Very low electrical consumption can reduce LCoH, but low thermal COP and large equipment size may limit the overall benefit.	Cyclic operation and slow thermal response limit suitability for variable electrolyser loads but can be advantageous with thermal storage.	H2O/silica gel, H2O/zeolite, methanol/activated carbon and related pairs; limitations include slow heat and mass transfer, bulky beds and, for water-based systems, vacuum operation.	[87,88,89]

Note: Source- and sink-temperature screening is provided by the operating map in Figure 2. Table 3 is applied when two or more heat-pump technologies remain thermodynamically feasible; the preferred option is then selected based on temperature-profile matching, COP, and LCoH, followed by operational, component, integration, and safety feasibility checks.

and serve as technology-selection and feasibility constraints for the alternatives identified in Stage 1.

This two-stage procedure distinguishes technology selection from the subsequent decision on whether the selected heat-pump solution should actually be implemented. A third, system-level assessment is therefore required after the preferred technology has been identified. That assessment should compare the selected heat-pump solution with direct heat use and relevant alternative heat-supply options while considering annual heat availability, temporal demand matching, electricity price, value of displaced heat, storage requirements, economic performance, and avoided emissions. The resulting selection hierarchy is therefore as follows: source- and sink-temperature compatibility for thermodynamic feasibility, COP for heat-pump performance assessment, and LCoH for the broader techno-economic comparison of feasible alternatives. Accordingly, LCoH is used as the principal overall comparison criterion, while COP remains essential for evaluating the thermodynamic performance of each candidate technology. These system-level feasibility criteria are addressed in Section 5.

4. Heat Pump Integration in Green Hydrogen Production Systems

The integration of heat pumps in green hydrogen plants depends mainly on the relationship between the available waste heat temperature and the required heat sink temperature. Two principal pathways can be distinguished: external heat use, where upgraded electrolyser heat is supplied to district heating, industrial users, or energy communities; and internal process support, where recovered heat is returned to the hydrogen production process for feedwater preheating, steam generation, or auxiliary thermal management. The appropriate integration concept must preserve electrolyser cooling reliability while improving the useful value of otherwise rejected heat.

To support the selection of suitable integration pathways,

Table 4. Waste heat temperature levels and waste heat fractions for hydrogen-related technologies.

Hydrogen-Related Source/Technology	Waste Heat Temperature	Waste Heat/ Input Energy (%)	Type of Heat	Refs.
PEM electrolyser stack cooling	~50–80 °C	~14–20%; useful recovered heat often reported around ~14–16%	Stack cooling water/glycol loop	[36,71,72]
PEM electrolyser plant, including stack and product-gas cooling	~50–80 °C; plant-level heat may be recovered as warm water	~20–30%, depending on plant boundary and operating point	Stack cooling + humid H2/O2 gas cooling	[36,47]
PEM plant with hydrogen/oxygen compression heat included	Cooling heat from electrolysis plus compressor cooling; temperature is site- and pressure-dependent	Plant-level recoverable heat can approach ~30% in modelled systems	Electrolyser cooling heat + compressor cooling heat	[47,66]
Alkaline electrolyser cooling/electrolyte loop	~70–110 °C; often ~70–90 °C	Not usually reported as one fixed value; theoretical thermal losses often ~20–40% if 60–80% efficiency is assumed	Heated KOH/NaOH electrolyte or secondary cooling loop	[48,53]
AEM electrolyser	~40–70 °C; often ~50–70 °C	Not well quantified at plant level; theoretical losses may be around ~25–35% depending on cell efficiency	Stack cooling loop and humid gas streams	[56,57]
SOEC stack/high-temperature electrolysis	Stack operation ~500–850 °C	No fixed waste-heat fraction; system may be endothermic, thermoneutral, or exothermic depending on operating point	High-temperature steam, H2-steam mixture, oxygen-side gas	[49,50,51]
SOEC hydrogen-water mixture/condensation heat	High-temperature product stream; recovered heat can support steam generation around ~110 °C	Not separately reported as a fixed % of input	Sensible and latent heat from H2-water mixture	[49]
Hydrogen compression stages	Site-, pressure- and compressor-dependent	Usually not reported separately; may add a few percentage points to plant-level recoverable heat	Compressor cooling heat	[47,66]
Gas drying and purification in hydrogen plant	Low-temperature rejected heat; site-dependent	Not separately reported	Auxiliary rejected heat	[47,66]
Power electronics, rectifiers and auxiliary systems	Low-temperature air/water cooling; site-dependent	Not separately reported	Cooling air or cooling water	[47,67]
Hydrogen-enabled integrated energy system	Minimum waste heat temperature ~50 °C	30% of hydrogen-device input energy assumed as waste heat	Mixed waste heat from hydrogen-related devices	[93]

Note: The reported waste-heat fractions are indicative of values stated in the literature and should not be interpreted as directly comparable across rows. The cited studies apply different system boundaries, including stack-only cooling and plant-level heat recovery that may additionally include product-gas and compression heat, as well as different operating conditions and efficiency definitions. Moreover, recoverable heat is load-dependent; for PEM electrolysers, net stack heat generation may decrease substantially and the cell may even absorb heat from the surroundings at low current densities [71,72].

summarizes the representative waste heat temperature levels, approximate waste heat fractions relative to input energy, heat type, and source references for the main hydrogen-related technologies and plant components. This source-side classification provides the basis for deciding whether direct recovery is sufficient or whether heat pump-assisted temperature upgrading is required.

4.1. Direct and Heat Pump-Assisted Waste Heat Recovery

Direct heat recovery should be considered first when the available waste heat temperature is compatible with the heat demand. In this configuration, heat from the electrolyser cooling circuit, electrolyte loop, product gas cooling, or auxiliary systems is transferred through heat exchangers to a nearby heat user without additional compressor electricity consumption. This pathway is most suitable for low-temperature district heating, return-line preheating, domestic hot water preheating, greenhouse heating, drying processes, or internal low-temperature plant uses [36,37,47,48,94,95].

Based on the temperature match between the recovered heat and the heat demand, two main integration pathways can be distinguished: direct heat recovery and heat pump-assisted temperature upgrading. These pathways are schematically illustrated in Figure 4.

Studies on low-grade waste heat show that streams at 60 to 95 °C may be recovered by heat exchangers for local heating, while lower-temperature streams around 30–60 °C generally require heat pump upgrading when higher supply temperatures are needed [94,96]. A district heating demonstration using multiple low-grade industrial waste heat sources between 20 and 90 °C recovered 390,000 GJ of heat, reduced CO₂ emissions by 35,000 t, and saved more than 150,000 t of water, demonstrating that low-grade heat can become useful when the network temperature and hydraulic design are adapted to the source [94].

The direct low-temperature pathway is selected when the recovered heat can meet the required sink temperature with a sufficient heat-exchanger temperature difference; otherwise, heat-pump-assisted upgrading is required. When the heat sink requires a higher temperature than the available waste heat, heat pump-assisted recovery becomes necessary. In this case, the heat pump acts as a temperature-upgrading interface between the electrolyser thermal management system and the final heat user [93,96,97,98]. This is particularly relevant for waste heat in the range of around 45–60 °C, where heat pumps have been shown to deliver 2.5–11 times more energy value than power cycles at equal waste heat input, while heat engines become more competitive mainly when waste heat temperatures reach about 100 °C or above [96]. The same study indicates that waste heat water streams at 45–60 °C can be partly cooled to about 5 °C and partly upgraded to 90–130 °C, with COP values of about 4–7 depending on operating conditions [96]. This temperature interval is relevant for PEM and alkaline electrolyser cooling heat, because such heat is often too low for direct use in conventional heat networks but sufficiently high for efficient heat pump upgrading [36,47,48,96].

Several integration layouts can be distinguished. The simplest option is direct heat exchange between the electrolyser cooling circuit and a compatible heat sink. A second option is return-line integration, where electrolyser heat preheats a district heating return line or a secondary circuit before final heating [37,94,95]. A third option is heat pump-assisted upgrading, where an intermediate water or glycol loop supplies the evaporator of the heat pump, while the condenser supplies district heating, industrial heat, internal process heat, or an intermediate thermal buffer [93,97,98,99]. In district heating applications, high-temperature heat pump integration has achieved COP values of 3.2–5.4 when the network acted as the heat sink, while using the district heating network as the heat source for 110 °C heat delivery resulted in COP values of 2.8–5.7 [98]. A field study of a medium-temperature heat pump in an Italian district network showed that return flows of 30–50 °C could be upgraded to 65–75 °C over 5256 operating hours, with a weighted average COP of 3.96, a thermal output of 134.5 kW, a levelized cost of heat of EUR 0.0245/kWh, a payback period of about six years, and CO₂ emission reductions of about 120 t compared with gas-fired boilers [100].

For hydrogen-based integrated energy systems, the benefit of heat pump-assisted recovery depends strongly on the temporal profile of waste heat. Hydrogen-related devices can release heat at different temperature levels, and one recent study estimated that about 30% of the input energy of hydrogen-related devices is converted into waste heat, with a minimum temperature of about 50 °C [93]. In the same study, a cascade heat pump-assisted recovery system achieved an average COP of 3.3 in an extremely cold region, reduced energy consumption by up to 36% compared with a conventional single-stage air-source heat pump, achieved a payback period of about 2–6 years depending on electricity price, and reduced CO₂ emissions by 600 t/year [93]. These values are consistent with wider district-heating and large-scale heat-pump studies, which show that heat pumps can enable the use of low-temperature heat sources in thermal networks when the temperature lift, operating hours, and heat demand are favourable [97,98,99,100].

The choice between direct and heat pump-assisted recovery should therefore be based on source temperature, required sink temperature, temperature lift, heat-flow stability, distance to the heat user, annual operating hours, backup cooling requirements, and the economic value of the delivered heat [36,37,47,48,93,96]. Direct recovery is thermodynamically preferable when the temperature match is sufficient, whereas heat pump-assisted recovery is preferable when a moderate electricity input allows a larger fraction of the waste heat to be used at a higher-value temperature level [96,97,98,100]. Practical implementation also depends on the heat-user distance, ownership structure and business model. For industrial excess heat collaborations, economic feasibility has been shown to depend strongly on distance and energy prices; one study found that under baseline prices, a greenhouse-heating case was profitable at a source-user distance below about 2 km, while higher gas prices could make distances of approximately 8 km viable [101]. Other studies emphasize that cooperation models, clear interfaces, heat pricing, risk allocation, and trust between partners are decisive for implementation, especially when the heat user and heat supplier are different organizations [102,103]. Therefore, for green hydrogen plants, direct and heat pump-assisted waste heat recovery should be treated as a site-specific integration problem rather than a universal efficiency improvement.

4.2. External Heat Use: District Heating, Industry and Energy Communities

External use of recovered heat from green hydrogen production is most attractive when the hydrogen plant is located close to a continuous heat demand. The main receiving systems are district heating networks, industrial users, greenhouses, public buildings, and local energy communities. Among these options, district heating is usually the most suitable large-scale heat sink because it can aggregate many heat consumers and absorb low- or medium-temperature heat either directly or through heat pump upgrading [70,104,105,106,107].

The suitability of district heating depends strongly on network temperature. Fourth-generation district heating and fifth-generation district heating and cooling systems are particularly suitable because they operate at lower temperatures and can integrate renewable heat, industrial excess heat, and distributed heat pumps more easily than conventional high-temperature networks [70,108,109,110]. Therefore, electrolyser waste heat is more valuable in low-temperature networks, where the required temperature lift is smaller and the heat pump COP is higher. In existing high-temperature networks, the same waste heat may still be useful, but central or decentralized heat pump upgrading is usually required before injection.

Industrial users and energy communities represent a second external integration route. Industrial parks, food-processing facilities, greenhouses, and mixed-use communities may demand more stable year-round heat than residential district heating alone. However, these applications require careful matching of temperature level, heat quantity, operating hours, and distance between the hydrogen plant and the heat user [104,105,106,107]. Spatial studies show that excess heat integration is limited not only by the amount of heat available, but also by the proximity between heat source and heat sink, the network investment cost, and the achievable supply temperature [105,106,107].

The external heat-use pathways discussed above differ mainly with regard to the required heat supply temperature and in the temperature lift that must be provided by the heat pump. Low-temperature district heating and energy communities can often be supplied by single-stage water-source heat pumps, while conventional district heating and industrial heat users generally require high-temperature, two-stage, cascade, or transcritical configurations.

Table 5. Heat pump configurations and representative refrigerants for external use of recovered heat from green hydrogen production.

Recovered Heat Used for External Applications	Typical Heat Source	Required Heat Output Temperature	Suitable Heat Pump Configuration	Refrigerant	Main Application	Refs.
Low-temperature district heating and energy communities	PEM/AWE/AEM cooling water, electrolyte loop, auxiliary cooling streams	~50–80 °C	Single-stage vapour-compression heat pump; water-source heat pump	R717, R1234ze(E), R290, R600a	Low-temperature district heating, public buildings, energy communities	[36,37,70,77,78,81,83,104,105,106,107,108,109,110]
Conventional district heating	PEM/AWE cooling heat, compression heat, aggregated plant waste heat	~80–120 °C	High-temperature vapour-compression heat pump; two-stage heat pump	R717, R1234ze(Z), R1233zd(E), R1336mzz(Z)	District heating supply-temperature upgrading	[77,78,79,80,81,83,93,94,96,97,98,100]
Industrial hot water and low-/medium-temperature process heat	Electrolyser cooling heat, compression heat, plant-level waste heat	~80–150 °C	High-temperature heat pump; two-stage or cascade heat pump	R717, R1233zd(E), R1336mzz(Z), R1234ze(Z), hydrocarbons	Food industry, drying, process water, industrial symbiosis	[26,77,78,79,80,81,83,93,96,104,105,106,107]
District heating return-line or process-water upgrading	Warm return water, low-temperature network flow, industrial return streams	~70–120 °C	Transcritical heat pump	R744	Return-line boosting, process-water heating, large temperature-glide applications	[80,84,98,100]

summarizes the most relevant heat pump configurations and representative refrigerants for these external heat-use pathways. Refrigerant selection, including environmental characteristics such as GWP, is treated separately in Section 3.4, because these properties are associated with the selected working fluid rather than with the heat-use pathway itself.

The economic implementation of external heat use also depends on ownership and market design. When the hydrogen producer and heat user are different organizations, the heat price, connection responsibility, backup supply, metering point, and risk allocation must be clearly defined. Liberalized or open district-heating market concepts can support the participation of smaller excess-heat producers, while price-signal and market-participation approaches may be needed when many distributed heat sources are connected to the same network [111,112]. Consequently, external use of electrolyser waste heat should be evaluated as a local heat-market problem rather than only as a thermodynamic recovery option.

4.3. Internal Process Support: Feedwater Preheating and Steam Generation

Internal use of recovered heat is most relevant when the heat can reduce auxiliary energy demand inside the hydrogen production plant itself. Unlike external heat use, where the value of recovered heat depends on a nearby consumer, internal process support can improve the energy balance of the hydrogen plant by preheating feedwater, supporting steam generation, assisting thermal management, or supplying heat to auxiliary water-treatment processes. This pathway is particularly important for SOEC systems, because high-temperature electrolysis requires steam and involves strong coupling between electrical input, heat supply, recuperation, and product-gas cooling [49,50,51,65,113,114,115].

In low-temperature electrolysis systems, internal heat use is mainly limited to feedwater preheating, balance-of-plant heating, and water-treatment support. PEM, alkaline, and AEM electrolysers reject heat at temperatures that are generally too low for direct steam generation, but this heat can still be useful for preheating demineralized water, reducing cooling demand or supporting thermally driven desalination [36,47,48,116,117]. Experimental studies have shown that waste heat from water electrolysis can be used to drive membrane distillation and produce desalinated water for subsequent use in hydrogen-production systems. In one integrated system, the membrane distillation unit produced water with a conductivity of 2.8 μS/cm. This value can satisfy the less stringent requirements reported for some electrolyser systems (<5 μS/cm), but it remains above the <1 μS/cm level commonly associated with high-purity electrolyser feedwater and the <0.25 μS/cm requirement reported for PEM electrolysers; additional processing may therefore be required depending on the electrolyser technology [116,118]. A later study integrated a membrane distillation heat exchanger with a 2.2 kW commercial electrolyser and showed that the system could simultaneously remove heat from the electrolyser and produce pure water for hydrogen production [117]. Therefore, for low-temperature electrolysers, internal heat recovery is less related to high-temperature process heat and more related to cooling-load reduction and water-cycle integration.

For SOEC systems, the role of internal heat recovery is more central. Because steam must be supplied to the stack, recovered heat from hot outlet streams, product-gas cooling, and condensation can reduce the need for electric heaters or external boilers. System-level SOEC studies show that heat integration can improve hydrogen production efficiency by recovering exhaust heat for feedwater preheating, steam generation, and internal recuperation [65,114,115,119]. External heat sources can also be beneficial for high-temperature electrolyser systems because part of the energy required for water splitting can be supplied as heat instead of electricity [113]. Recent studies have therefore examined SOEC coupling with engine waste heat, industrial waste heat, high-temperature process heat, and downstream synthesis processes, showing that the value of heat recovery in SOEC systems lies not only in avoiding heat rejection but also in reducing auxiliary electrical energy needed to prepare steam and maintain thermal balance [63,120,121,122].

Heat pump-assisted steam generation becomes important when the recovered heat is available at a useful temperature level but is still insufficient for direct steam supply. Fang et al. proposed a cascade heat pump to recover sensible and latent heat from the hydrogen–water mixture leaving high-temperature water electrolysis and to generate steam for the process [49]. Zhang and Zhang further integrated SOEC operation with H₂–O₂ combustion and a heat pump for low-emission high-temperature heating in the chemical industry, showing that heat pump-based recovery of condensation and process heat can reduce the demand for electricity associated with steam generation and cooling-medium preparation [123]. These studies indicate that heat pumps can act as internal process-support devices, especially where condensation heat, medium-temperature heat, or product-gas cooling heat must be upgraded to steam-generation conditions.

For internal process support, heat pump selection is governed by the function of the recovered heat inside the hydrogen plant. Low-temperature electrolyser heat is mainly suitable for feedwater preheating, water-cycle integration, and auxiliary thermal loads, while SOEC-related heat recovery may require high-temperature or cascade heat pumps for steam-generation support.

Table 6. Heat pump configurations and representative refrigerants and working pairs for internal use of recovered heat in green hydrogen production.

Recovered Heat Used for Internal Processes Support	Typical Heat Source	Required Heat Output Temperature	Suitable Heat Pump Configuration	Refrigerant/Working Pair	Main Application	Refs.
Feedwater preheating	PEM/AWE/AEM cooling heat, humid gas cooling, auxiliary cooling streams	~40–90 °C	Direct heat exchanger with auxiliary vapour-compression heat pump; single-stage water-source HP	R717, R1234ze(E), R290, R600a	Demineralized water preheating, feedwater temperature control	[36,47,48,77,78,81,83,116,117]
Water-treatment support	Low-temperature electrolyser cooling heat	~40–80 °C	Vapour-compression heat pump or heat exchanger-assisted recovery	R717, R1234ze(E), R290, R600a	Membrane distillation, pure water production for electrolysis	[116,117]
SOEC steam-generation support	H2-steam mixture condensation heat, product-gas cooling, residual sensible heat	>100 °C; steam-related temperature levels	Cascade heat pump; high-temperature heat pump	R717, R1233zd(E), R1336mzz(Z), R718	Steam generation, reduction of auxiliary electric heating, SOEC thermal support	[49,63,65,79,80,81,84,113,114,115,116,117,119,120,121,122,123,124]
Internal high-temperature process support	SOEC exhaust heat, industrial waste heat coupled to electrolysis, high-temperature plant streams	~100–160 °C or higher, case-specific	High-temperature or very-high-temperature heat pump; cascade system	R1233zd(E), R1336mzz(Z), R718, R717	Feedwater heating, steam-related processes, high-temperature electrolysis support	[49,65,77,79,80,81,84,113,114,115,122,123,124]
Auxiliary thermally driven upgrading	Intermediate-temperature plant heat or external waste heat	Medium-temperature useful heat	Absorption heat pump/heat transformer; adsorption heat pump	H2O/LiBr, NH3/H2O, water-silica gel, water-zeolite	Auxiliary heat upgrading, niche internal heat recovery	[85,86,87,88,89]

summarizes suitable heat pump configurations and representative refrigerants and working pairs for internal use of recovered heat in green hydrogen production systems.

The suitability of internal heat recovery depends strongly on the electrolysis technology. For PEM, alkaline, and AEM systems, internal recovery is mainly useful for low-temperature preheating, water treatment, and auxiliary thermal loads. For SOEC systems, the integration potential is more extensive because heat recovery directly affects steam supply, stack thermal balance, operating voltage, and system efficiency [65,113,114,115,119,122]. However, internal heat use must not compromise safe operation of the electrolyser. The cooling system must still maintain stack temperature control, and any heat recovery loop should include bypass cooling, backup heat rejection, or thermal storage when heat availability and process demand are not synchronized [124]. Therefore, internal process support should be designed as a controlled thermal integration layer that improves the hydrogen plant energy balance while preserving the primary cooling and thermal management requirements of the electrolyser system.

5. Discussion

Despite the increasing number of studies on electrolyser waste heat recovery and industrial heat pump integration, the use of heat pumps in green hydrogen production systems remains at an early stage of system-level development. The reviewed literature shows that heat pump integration should not be interpreted as a universal efficiency improvement, but as a site-specific thermal integration option whose feasibility depends on electrolyser technology, waste heat temperature, recoverable heat fraction, required sink temperature, operating profile, heat-user proximity, electricity price, and displaced heat-production technology [36,37,47,48,93,94,95,96,97,98,99,100,101,102,103]. Therefore, the main research challenge is not only to demonstrate that electrolyser waste heat can be recovered, but to define the operating conditions under which direct heat use, heat pump-assisted external heat supply, or internal process support provide measurable thermodynamic, economic, and environmental benefits.

Accordingly, the position adopted in this review is that direct heat recovery should be preferred whenever the source and sink temperature levels are compatible, whereas heat pumps should be regarded as enabling technologies rather than default solutions. Their implementation is justified when a meaningful temperature mismatch exists and when the resulting thermodynamic benefit is supported by favourable LCoH and system-level economic and environmental performance.

The selection framework developed in Section 3 provides indicative quantitative guidance rather than universal acceptance thresholds. For vapour-compression systems, the literature synthesis in Figure 3 indicates average COP values decreasing from 4.6 at a 30 K temperature lift to 3.6 at 50 K, 3.0 at 70 K and about 2.7–2.8 at 90–100 K. These values can be used to identify increasingly demanding operating conditions, but they should not be interpreted as minimum acceptable COP values. The economically acceptable COP is application-specific and depends on electricity price, annual operating hours, capital and maintenance costs, and the value of the displaced heat supply. Consequently, Figure 2 should first be used for temperature-based feasibility screening, Figure 3 for indicative thermodynamic performance, and LCoH for the subsequent techno-economic comparison of feasible alternatives.

A first critical gap, already identified in the Introduction and highlighted by the source-side comparison in Table 4, is the lack of a consistent system boundary for defining recoverable waste heat [36,47,48,66]. Consequently, reported waste-heat fractions cannot be compared directly across studies without distinguishing stack-level from plant-level heat recovery. For PEM electrolysers, useful heat recovery has been reported at both stack and plant level, while alkaline systems are often evaluated through the electrolyte cooling loop and district heating integration [36,47,48]. For AEM systems, plant-level waste heat data remain limited because the technology is still less mature [56,57]. For SOEC systems, the heat balance is even more complex, since the stack may operate in endothermic, thermoneutral or exothermic mode depending on operating voltage, steam utilisation, current density, and heat exchanger effectiveness [49,50,51,65,113,114,115]. Future studies should therefore clearly distinguish between theoretical thermal losses, technically recoverable heat, useful delivered heat, and heat pump-upgraded heat.

A second gap, also identified in the Introduction and discussed in Section 2.5, is the limited availability of dynamic and part-load assessments. Because electrolyser operation and heat demand may follow different temporal profiles, nominal heat-pump sizing does not necessarily represent annual operating conditions [4,13,43,67,73]. Current steady-state studies are useful for identifying theoretical integration options, but they are insufficient for predicting annual performance, seasonal COP, useful heat delivery, cooling reliability, and economic value. Future work should include time-resolved simulations and, where possible, measured operating profiles for both hydrogen production and heat demand. To make such datasets useful for heat-pump design, future studies should also report the electrolyser load or current density, operating pressure, cooling-loop inlet and outlet temperatures, heat-transfer-medium flow rate, and corresponding recoverable thermal power over the operating range. This would allow the dynamic quality and quantity of the available heat source to be linked directly to heat-pump evaporator conditions rather than represented by a single nominal temperature and heat-flow value. Thermal storage may be considered as an auxiliary buffer in such systems, but it should be evaluated together with its cost, heat losses, and control complexity rather than assumed to solve the temporal mismatch automatically [44,45,46,68].

A third important research gap concerns the coupling between electrolyser thermal management and heat pump operation. Waste heat recovery must not compromise the primary cooling function of the electrolyser. PEM, alkaline, and AEM stacks require stable temperature control to protect membranes, electrodes, separators, and sealing materials, while alkaline systems also introduce material compatibility issues due to the corrosive electrolyte [36,48,53]. In SOEC systems, thermal gradients, stack temperature control, and steam supply are even more critical because they directly affect degradation, thermal stress, and electrochemical efficiency [49,50,51,65,113,114,115,122,123,124]. This coupling creates an important design trade-off. Increasing the cooling-loop outlet temperature within the allowable electrolyser operating range raises the heat-pump source temperature and reduces the required temperature lift, which can improve COP and expand the potential for direct heat use. However, the cooling set point cannot be selected solely to improve heat-recovery performance, because stack temperature must remain within limits imposed by efficiency, material stability, degradation, and safe operation. Similarly, coolant or electrolyte flow-rate control affects both the temperature profile and the quantity of heat available to the heat pump. Cooling temperature and flow rate should therefore be treated as coupled electrolyser–heat-pump design variables rather than as fixed boundary conditions. Therefore, heat recovery should be designed as an additional thermal layer with hydraulic separation, bypass cooling, backup heat rejection, and safe control logic. More experimental studies are needed to validate heat pump operation under realistic electrolyser cooling constraints, rather than treating the electrolyser as a simple constant-temperature waste heat source.

A fourth gap concerns the validation of the technology-selection framework developed in Section 3 under realistic hydrogen-plant operating conditions. Although Figure 2 and Table 3 provide thermodynamic and practical screening criteria, comparative studies simultaneously considering temperature lift, COP, LCoH, compressor discharge temperature, working-fluid constraints, safety requirements, and part-load operation remain limited [77,78,79,80,81,82,83,84,90]. Future work should therefore validate these criteria against measured electrolyser heat-source profiles and representative heat-demand profiles, particularly for high-temperature and cascade configurations where compressor and refrigerant limitations may become decisive.

A fifth gap concerns the system-level conditions governing external heat recovery. As discussed in Section 4.2, technical feasibility alone does not ensure useful heat recovery, because the value of exported heat depends strongly on heat-user proximity, required network temperature, temporal demand, connection infrastructure, and heat-market conditions [70,94,95,99,100,101,102,103,104,105,106,107,108,109,110,111,112]. Lower-temperature district-heating networks are generally more favourable because they reduce the required temperature lift and therefore improve heat-pump performance, whereas higher-temperature networks impose greater electricity consumption and may require more complex configurations. Future assessments should therefore evaluate external heat use using the combined criteria of temperature compatibility, annual heat-demand matching, transport distance, connection cost, heat price, and backup-supply responsibility rather than COP alone.

A sixth gap is associated with internal process support. For low-temperature electrolysers, internal heat use is mainly limited to feedwater preheating, water-cycle integration, water-treatment support, and auxiliary thermal loads [36,47,48,116,117]. For SOEC systems, the integration potential is deeper because recovered heat can support feedwater preheating, steam generation, internal recuperation, and reduction of auxiliary electric heating [49,50,51,63,65,113,114,115,116,117,119,120,121,122,123,124]. However, the boundary between direct heat recovery, heat pump-assisted steam generation and general SOEC heat integration is not always clearly defined. Future studies should quantify how much electric heating can be avoided by heat pump-assisted recovery, how the heat pump affects total system efficiency, and whether the additional complexity is justified compared with direct recuperation or external heat supply. Importantly, internal process support and external heat use should not be interpreted as mutually exclusive pathways. In an integrated plant, recovered heat may first be allocated to internal demands such as feedwater preheating or steam generation, particularly in SOEC systems, while any thermally and temporally available surplus can subsequently be supplied to district heating or other external users. The preferred allocation therefore depends on the temperature hierarchy, temporal coincidence, and economic value of competing internal and external heat sinks.

A seventh methodological gap concerns the consistent system-level assessment of heat-pump-assisted recovery after technology selection. As established in Section 3, source- and sink-temperature compatibility should first determine thermodynamic feasibility, COP should characterize heat-pump performance, and LCoH should provide the principal broader techno-economic comparison of feasible alternatives. The subsequent implementation assessment should then compare the selected heat-pump solution with direct heat recovery and relevant alternative heat-supply options using consistent system boundaries and annual operating profiles. At minimum, this assessment should consider electrolyser input energy, recoverable and useful heat, heat-pump electricity consumption, seasonal performance, electricity price, the value and carbon intensity of displaced heat production, storage requirements, LCoH and avoided CO₂ emissions [30,31,32,36,37,47,48,93,100].

As a simple illustrative example, consider the delivery of 1 MWh of useful heat. A heat pump operating at COP = 3 requires 0.333 MWh of electricity, whereas at COP = 5, the electricity demand decreases to 0.200 MWh. Assuming an electricity carbon intensity of 250 kg CO₂/MWh, the corresponding operational emissions are 83.25 and 50 kg CO₂/MWh of useful heat, respectively. If the displaced heat supply has a carbon intensity of 80 kg CO₂/MWh, the COP = 3 case provides no operational CO₂ benefit, whereas the COP = 5 case avoids 30 kg CO₂/MWh. These values are illustrative rather than case-specific, but they demonstrate why COP should be evaluated together with the electricity mix and the displaced heat source.

On this basis, the final system-level decision can be formulated as a third stage following the technology screening and selection procedure described in Section 3. If direct heat recovery can satisfy the heat demand, it should remain the reference option. Where temperature upgrading is necessary, the heat-pump configuration identified through the temperature map, COP assessment, and LCoH comparison should be implemented only after confirming sufficient coincidence of annual heat availability and demand, acceptable economic performance relative to the relevant alternative heat supply, and a net environmental benefit under the applicable electricity mix. The selected solution must additionally preserve electrolyser thermal-management and safety requirements. Thus, the technical feasibility of a heat pump does not by itself constitute justification for its implementation.

6. Conclusions

As discussed in Section 5, the reviewed literature still lacks sufficient pilot- and industrial-scale demonstrations that combine actual electrolyser operation with heat-pump-assisted waste heat recovery under representative operating conditions. Most existing studies rely on simulations, conceptual system diagrams, or case-specific technical and economic assumptions. Future research should focus on demonstration facilities in which the electrolyser load, cooling temperature, heat pump capacity, amount of recovered heat, heat demand from consumers, backup cooling operation, electricity consumption, and economic indicators are measured over long periods. In the short term, priority should be given to standardized definitions of recoverable and useful heat, dynamic performance models, and safety-oriented integration schemes. In the medium term, research should focus on the validated selection of heat pump cycles, working fluids with low GWP or even natural refrigerants, the integration of district heating, and internal heat-recovery and steam-generation concepts in SOEC systems. In the long term, heat pump-assisted waste heat recovery should be evaluated as part of comprehensive green hydrogen production systems, where its role is compared to direct heat use, industrial symbiosis, low-temperature district heating, and other low-carbon heat supply options. Accordingly, heat pumps should not be regarded as the default option for electrolyser waste heat recovery, but as enabling technologies that are justified when direct heat recovery cannot satisfy the required sink temperature and when the resulting temperature lift, LCoH, annual utilisation, and environmental performance remain favourable.