The stakes are considerable. Global energy demand is projected to rise by roughly 48 percent by 2040, and heating and cooling remain stubbornly dependent on fossil fuels. Renewable energy currently supplies only about 23 percent of the heating and cooling sector, a share that lags far behind the rapid gains made in renewable electricity. Heat pumps, which move heat rather than generate it, are widely seen as the most promising bridge technology, but their efficiency depends intimately on the temperature of the reservoir they tap. Air source heat pumps extract thermal energy from ambient air, which is easy to install but fluctuates wildly with the seasons. Ground source systems bury a heat exchanger in the soil, where temperatures remain nearly constant year-round, but they demand drilling, land, and higher upfront capital.

The research team, led by Mustafa Jaradat with Omar Alsotary, Ohoud Aljaloudi, and Madhar Haddad, anchored their analysis in real measurements rather than generic design assumptions. They drilled a ten-meter borehole at the Amman site, lined it with PVC pipe, and lowered a copper-tube heat exchanger into the ground. A UNI-T UT330-USB data logger, rated to ±0.5 °C accuracy and protected against water and dust ingress, recorded ground temperature continuously from December 2021 to December 2022. Over that full annual cycle, the soil temperature at the borehole base stayed within a remarkably narrow band of 21.7 to 22.8 °C, confirming the thermal stability that makes ground source systems attractive in the first place.

To probe how much heat the ground could actually deliver, the team circulated tap water through a two-meter copper coil, manually rolled into a spiral and suspended ten meters below the surface. Water entered the ground loop at roughly 13 to 15 °C and emerged at up to 18.7 °C, a gain of about 3.7 °C that the researchers used in a sensible-heat energy balance to calculate the ground-side heat transfer rate. A short circulation test showed the soil temperature dropping from 22.1 to 21.7 °C as heat was extracted, corresponding to a daily degradation rate of 0.088 °C per day. Crucially, the authors caution that this short-cycle rate describes only the instantaneous response of the coil during active circulation; the full-year record shows the undisturbed ground swings far less than a naive extrapolation would suggest.

On the modeling side, the team built a first-order thermodynamic model of a vapor-compression heat pump cycle, defining four key states: saturated vapor at the compressor inlet, superheated vapor at the compressor outlet, saturated liquid at the condenser outlet, and a saturated mixture after the expansion valve. The model held the indoor setpoint at 24 °C and 38 percent relative humidity, with the compressor outlet fixed at 70 °C and 1866 kPa for the R410A case. A compressor isentropic efficiency of 0.75 was applied uniformly across all three refrigerants to keep the comparison fair. The expansion valve was treated as an isenthalpic throttling device, analyzed through the Joule-Thomson coefficient, which governs how much the refrigerant cools as its pressure drops.

Three refrigerants were put through this framework: R410A, the workhorse of modern residential heat pumps with a global warming potential of 2088; R134A, an older benchmark refrigerant with a GWP of 1430; and R513A, a lower-GWP blend at 631 that is positioned as a near-drop-in alternative to R410A. The inclusion of R513A is what gives the study its contemporary edge, because international agreements such as the Kigali Amendment to the Montreal Protocol and the EU F-Gas Regulation are phasing down high-GWP hydrofluorocarbons. The question is whether a greener refrigerant can match the thermodynamic performance of the incumbent, or whether efficiency and climate impact remain locked in a trade-off.

The answer, at least for the air source configuration, is that R410A still wins on raw efficiency. Modeled against the full October-to-March weather record for Amman, the R410A air source heat pump achieved an average coefficient of performance of 4.97, with hourly values ranging from 4.22 to 6.20. R513A came in at an average COP of 3.49, ranging from 2.97 to 4.50, while R134A trailed at 3.13 on average. Hourly compressor power told the same story: R410A consumed the least, between 0.79 and 1.05 kWh, R513A sat in the middle at 0.82 to 1.27 kWh, and R134A demanded the most at 0.91 to 1.44 kWh. R410A also absorbed the most heat at the evaporator, peaking at 4.14 kWh, compared with 2.86 kWh for R513A and 2.79 kWh for R134A.

The ground source system, by contrast, drew its evaporator heat from a buffer tank of water held at a steady 18 °C, fed by the buried heat exchanger. The model assumed the refrigerant left the evaporator as saturated vapor at 15 °C and 1250.8 kPa, conditions tied directly to the measured buffer-tank temperature rather than an assumed design margin. Scaled so that its delivered heating capacity met the same one-ton target used to size the air source unit, the ground source heat pump achieved a COP of 4.79, delivering 3.51 kW of heat with a compressor power of 0.73 kW and an evaporator load of 2.78 kW. Its annual electricity consumption came to approximately 3203 kWh, lower than any of the three air source configurations, which ranged from 4095 kWh for R410A to 5283 kWh for R134A.

But the authors are careful not to declare an unconditional winner. The ground source system’s lower absolute consumption partly reflects its smaller delivered capacity, 3.51 kW versus roughly 4.65 kW for the air source unit, rather than efficiency alone. Per unit of heat actually delivered, the air source heat pump running R410A retains a marginally higher COP of 4.97 compared with the ground source system’s 4.79. The ground source advantage lies elsewhere: its hour-to-hour energy consumption is remarkably steady because the underground temperature barely moves, whereas the air source system’s output fluctuates with every cold snap. The researchers also quantified their measurement uncertainty, noting that the ±0.5 °C sensor accuracy propagates to roughly ±19 percent on the 3.7 °C temperature difference used in the ground-side heat calculation, a band too narrow to explain the 28 percent gap in annual electricity between the two systems.

The study’s broader message is that refrigerant selection and heat source selection are not independent choices. R410A delivers the strongest thermodynamic performance in both architectures but carries the heaviest climate burden, while R513A offers a meaningful reduction in global warming potential at a measurable cost in efficiency. R134A, the legacy option, performs worst on both counts in this comparison. For homeowners and policymakers, the practical takeaway is that the optimal configuration depends on priorities: if minimizing electricity per unit of heat is the goal, the air source heat pump with R410A leads among the modeled options; if steady output and lower absolute consumption at a smaller capacity are preferred, and if drilling costs and land availability permit, the ground source system with R410A remains a defensible choice. As heat pump adoption accelerates worldwide, studies like this one, grounded in a full year of site-specific measurements rather than generic assumptions, will be essential for guiding that transition in the climates where it matters most.

Subject of Research: Comparative experimental and modeling analysis of air source and ground source heat pumps using R410A, R134A, and R513A refrigerants

Article Title: Experimental and comparative analysis of air source and ground source heat pumps: refrigerant performance and energy efficiency

Article References: Jaradat, M., Alsotary, O., Aljaloudi, O., & Haddad, M. (2026). Experimental and comparative analysis of air source and ground source heat pumps: refrigerant performance and energy efficiency. Results in Engineering, 32, Article 113582. https://doi.org/10.1016/j.rineng.2026.113582

Image Credits: AI Generated

DOI: 10.1016/j.rineng.2026.113582

Keywords: heat pumps, ground source heat pump, air source heat pump, R410A, R513A, R134A, coefficient of performance, refrigerants, geothermal energy, energy efficiency, Amman, heating and cooling

The study, led by Ihsan M. Khudhur of Northern Technical University with colleagues Omer K. Ahmed, Omar Rafae Alomar and Sameer Algburi, was conducted outdoors in Kirkuk, a hot, semi-arid city at roughly 35.5 degrees north latitude that experiences punishing summer irradiance and distinct seasonal swings. The team built a direct-expansion photovoltaic/thermal heat pump prototype in which a one-square-meter, 150-watt photovoltaic module doubles as the evaporator of a small vapor-compression heat pump. A roll-bond aluminium plate bonded to the back of the module allows refrigerant to evaporate directly at the panel surface, absorbing heat that would otherwise accumulate in the silicon cells. On the condenser side, a water-cooled heat exchanger upgrades that recovered heat into useful hot water, so the same incident sunlight yields both electricity and thermal energy from a single footprint.

The passive half of the cooling strategy is unusually low-tech. The researchers dispersed titanium dioxide nanoparticles into linseed oil, a renewable and biodegradable drying oil, at a ratio of 15 percent nanoparticles to 85 percent oil, homogenized the mixture with magnetic stirring and ultrasonic dispersion, and applied it to the glass cover of the panel at three nominal thicknesses: 100, 200 and 300 nanometers. Screening with a solar cell analyzer under 1000 watts per square meter of halogen illumination showed the thinnest layer performed best, raising the conversion efficiency of coated mini-cells from 3.44 percent uncoated to 4.41 percent at 100 nanometers, with improved short-circuit current density and fill factor. Thicker coatings actually degraded performance, likely by scattering or absorbing useful light, so the 100-nanometer layer was carried forward into the field trials.

With both cooling mechanisms in place, the team ran a full seasonal comparison across four representative months: January, March, August and October. For each season they tested three refrigerants in the same hardware: R134a, the legacy hydrofluorocarbon baseline with high global warming potential; R600a, an isobutane hydrocarbon with negligible climate impact; and R1234yf, a low-GWP hydrofluoroolefin increasingly used in automotive and stationary systems. The same refrigerant charge mass was used for all three fluids to keep the comparison consistent, an intentional design choice the authors note may understate each fluid’s optimal potential. Measurements were taken on three clear-sky days per condition between 07:00 and 17:00 using calibrated irradiance meters, K-type thermocouples, an infrared thermometer, and current-voltage tracing, with standard deviations reported for repeatability.

The temperature results are the headline. In August, when an uncoated baseline panel reached a scorching 69 degrees Celsius at 14:00, the coating alone brought the surface down to 62 degrees, the heat pump alone to 39 degrees, and the combined active-passive system to 35 degrees, a 34-degree mitigation. In winter the combined approach reduced peak surface temperature from 46 to 29 degrees Celsius; in spring the drop was from 46 to 27 degrees; and in autumn from 48 to 24 degrees. Intriguingly, the combination outperformed the sum of its parts: the passive coating delivered 6 to 10 degrees of relief while the heat pump alone provided 19 to 22 degrees, yet together they produced nonlinear gains that exceed what either mechanism achieves independently.

Those temperature reductions translated directly into electrical performance. Compared with the uncoated baseline panel, the heat pump alone raised electrical efficiency by roughly 19 to 24 percent across the seasons, while the heat pump plus coating pushed gains to 24 to 29 percent. The best seasonal configurations show the pattern clearly: electrical efficiency climbed from 10.21 to 10.60 percent in winter with R1234yf, from 10.48 to 10.75 percent in spring with R600a, from 8.86 to 9.25 percent in summer with R600a, and from 9.87 to 10.18 percent in autumn with R600a. The authors argue that the coating acts as a thermal buffer that stabilizes the glass-to-cell interface and dampens morning transients, allowing the semiconductor junction to operate closer to its ideal fill factor and open-circuit voltage, while the heat pump removes heat from behind the panel and eliminates localized hot spots.

The heat pump’s coefficient of performance followed a daily rhythm, rising through the morning, peaking near solar noon when irradiance was greatest, and falling toward evening. Winter favored R1234yf, which posted a maximum COP of 2.65 at 13:00 under 760 watts per square meter of irradiance, with a daytime average of about 1.73. But from spring onward the hydrocarbon R600a dominated: a peak COP of 3.78 in March, 4.27 in August under 882 watts per square meter, and 3.85 in October. This seasonal flip is the study’s most practically consequential finding. It suggests that no single refrigerant is optimal year-round, and that system designers in hot, high-irradiance climates should match refrigerant selection to the season, favoring R1234yf in cold months and R600a for the bulk of the year.

Overall efficiency, defined as the sum of electrical and thermal efficiencies, reached its highest values in the transitional seasons with R600a: 73.44 percent uncoated rising to 76.59 percent coated in spring, and 75.27 percent rising to 77.32 percent coated in autumn. Even in winter, the best case with R1234yf improved from 62.97 to 64.93 percent with the coating. The authors frame this as more than an incremental gain: the coating does not merely assist the heat pump, it enables it, presenting a thermally stable evaporator surface that reduces the compressor’s workload per unit of heat extracted, so electrical and thermal outputs improve together rather than trading off against each other.

The team is candid about limitations. The campaign covered a single location with three test days per season, refrigerant charges were not optimized per fluid, and the long-term durability of an organic linseed oil matrix under ultraviolet exposure, oxidation, dust accumulation and thermal cycling remains unproven. Still, the engineering appeal is clear: the system uses commercially available components and a cheap, easily applied bio-nano coating requiring no manufacturing changes, offering a practical template for solar heating and cooling in the sweltering regions where photovoltaic losses bite hardest.

Subject of Research: Experimental enhancement of a direct-expansion photovoltaic/thermal heat pump system using a TiO2-linseed oil nanocoating and alternative refrigerants across four seasons

Article Title: Performance enhancement of photovoltaic/thermal systems with heat pump units using alternative refrigerants and nanoparticle TiO 2 with linseed oil cooling coating

Article References: Khudhur, I., Ahmed, O. K., Alomar, O. R., & Algburi, S. (2026). Performance enhancement of photovoltaic/thermal systems with heat pump units using alternative refrigerants and nanoparticle TiO2 with linseed oil cooling coating. Results in Engineering, 32, Article 112728. https://doi.org/10.1016/j.rineng.2026.112728

Image Credits: AI Generated

DOI: 10.1016/j.rineng.2026.112728

Keywords: photovoltaic thermal, heat pump, TiO2 nanocoating, linseed oil, R600a, R1234yf, R134a, coefficient of performance, solar cooling, renewable energy, seasonal performance, passive cooling