Thermophysical Properties of Low GWP Refrigerants: An Update

In the last decades, the industry of HVAC&R has faced continuous changes trying to identify environmentally friendly refrigerants for the numerous applications of the sector. However, the amount of low GWP fluids still available as potential refrigerants is limited to some natural fluids and, among synthetic chemicals, to hydrofluoroolefins (HFO). The knowledge of the thermophysical properties of these compounds and the evaluation of their energy efficiency in experimental apparatuses is essential to properly address the selection of the most suitable fluids. However, regarding the wide majority of HFOs, the information on the thermophysical properties, especially for the blends, are still scarce and require further research. In this work, an analysis of the possible substitutes and the available experimental data sets on their thermophysical properties was carried out to find out for which fluids further studies are needed to obtain an accurate representation of their thermophysical properties. Specifically, for 21 pure refrigerants, an overview of the thermodynamic (critical point, psat, PVT, heat capacity and speed of sound) and transport properties (λ, μ, σ) data published in the peer reviewed literature was provided. In addition, a more comprehensive analysis was carried out for four fluids (R1243zf, R1233zd(E), R1336mzz(Z), and R1224yd(Z)), for which major efforts have been made in the last 4 years to investigate the above thermophysical properties. Although an increasing amount of data sets on thermophysical properties have been compiled in recent years, the present study indicates that research efforts are still needed, especially on transport properties, as only 4 of the fluids of interest for the present research have been fully investigated (R1234yf, R1234ze(E), R1233zd(E), R1243zf), while other 4 (R1234ze(Z), R1336mzz(Z), R1224yd(Z), R1336mzz(E)) have been almost completely characterised.


Introduction
In the last decades, the refrigeration industry has undergone several changes due to the need to replace the employed refrigerants with more environmentally friendly fluids. As described by Calm (2008) [1], nowadays we are facing the search of the fourth generation of refrigerants, driven by the necessity of addressing the Global Warming issues. Large restrictions have indeed been imposed to the largely used hydrofluorocarbons (HFC) by the EU F-gas Regulation (REGU-LATION (EU) No.517/2014) [2] due to their high GWP (Global Warming Potential), giving a big impulse to the search of environmentally benign compounds, at least as efficient and safe as the current generation. The process of high GWP refrigerants substitution, at least in Europe, has been clearly programmed by the definition of the deadlines to limit, control and eventually block the emissions of harmful refrigerants, depending on the specific HVAC&R applications [2]. In the medium-long term, all the present high GWP refrigerants should be substituted by low GWP (< 150) refrigerants, either natural (ammonia, CO 2 , hydrocarbons) or synthetic. However, natural refrigerants cannot always satisfy all process requirements, due to toxicity and/or flammability issues, beside thermodynamic properties not suitable for all the applications. As refer the group of synthetic compounds, an interesting work by Mc Linden et al. (2014) [3] restricted the number of possible fluids to be used in refrigeration at a very low number of compounds. They considered a set of about 1200 candidate fluids, identified from more than 56 000 small molecules examined by applying screening criteria to be considered for GWP, flammability, stability, toxicity, and critical temperature. At the end, the number of fluids was reduced to 62. They found that no fluid is ideal in all regards: all of them have one or more negative attributes, such as poor thermodynamic properties, toxicity, chemical instability, low to moderate flammability, or very high operating pressures. Among all the considered fluids, the potential solutions were practically restricted to chemicals belonging to few groups, in particular hydro-fluoro-olefins (HFOs), hydro-chloro-fluoro-olefins (HCFOs) or hydro-fluoro-ethers (HFEs). HFOs and HCFOs are fluorinated hydrocarbons characterised by the presence of at least one carbon-carbon double bond in the molecules. The reactivity of the double bond to the atmospheric hydroxyl radical is higher than that of single bond and as a consequence the double bond compound is more unstable in atmosphere (McLinden et al., 2014 [3]); this determines a lower atmospheric lifetime and consequently a much lower GWP. However, if their double bond represents an advantage as refers to GWP and ODP, on the other hand it might pose a problem if related to their compatibility with materials, environmental safety and toxicity. Moreover, HFOs are mildly flammable and several contradictory studies on HFOs in the last years have risen some doubts on their compatibility with the environment.
To date, only few applications can employ definitive working fluids. In automotive, R1234yf has completely substituted R134a for mobile air conditioning. Few car manufacturers are now trying to produce air conditioning systems for cars basing on CO 2 , but R1234yf is surely the most used. R134a has been substituted Bold indicates the fluids which need to be substituted two fluids, out of the 17 analysed, were already well studied, namely R1234yf and R1234ze(E). Other two fluids (R1234ze(Z) and R1233zd(E)) had at least one set of data available for all except one of the seven thermophysical properties considered in the analysis. For all the other fluids, only few sets of data or no data at all were found. In this paper, the situation will be updated by analysing the open literature in the period 2018 to 2022, to identify the new sets of experimental data produced for the main thermophysical properties regarding potentially alternative synthetic low GWP refrigerants. After a description of the methodology applied, the sets of data found for each fluid will be synthesised in tables and briefly discussed.
2 Literature Analysis for the Period 2018 to 2022

Available Experimental Data
A thorough analysis of the open literature was conducted in a previous review paper by Bobbo et al. 2018 [4], which listed experimental data on the key thermodynamic (critical point, saturation pressure, PVT properties, speed of sound) and transport (thermal conductivity, viscosity and surface tension) properties of fluids considered as potential working fluids for HVAC&R applications. Fluids characterised by low global warming potential (< 150) and belonging to the groups of hydrofluoroolefins (HFO), hydrochlorofluoroolefins (HCFO) or hydrofluoroethers (HFE) were considered. The review revealed that only for 2 fluids (i.e., R1234yf, R1234ze(E)) at least one set of experimental data was available for each of the main thermophysical properties considered in the analysis, while for the other 2 fluids (R1234ze(Z) and R1233zd(E)) no data sets were available for only one property (isobaric heat capacity and thermal conductivity, respectively). Here, the available literature data for all fluids included in the Bobbo et al. study are updated with the data sets reported in the literature since 2018. In addition, this study also considers four new fluids to be used for HVAC&R and ORC applications that were recently added to ANSI / ASHRAE Standard 34 (R1224yd(Z), R13I1, R1132a, R1130(E)), along with HFE RE356mmz, which is suitable for organic Rankine cycle and high temperature heat pump applications. The ASHRAE designations, IUPAC names and basic properties of the HFOs, HCFOs and HFEs considered in this article are listed in Table 2.
In particular, this paper provides a more in-depth analysis for those fluids that meet the following three criteria: (1) Fluids for which at least 7 of the 8 thermophysical properties considered in this study have been experimentally investigated in at least one available data set: as shown in Table 3, which lists the number of papers providing experimental data for each refrigerant, only 8 fluids are fully or almost fully characterised (i.e., R1234yf, R1243zf, R1234ze(E), R1234ze(Z), R1233zd(E), R1336mzz(E), R1336mzz(Z), R1224yd(Z)), with at least 1 data set available for each thermophysical property, except for R1234ze(Z), R1336mzz(Z), and R1224yd(Z), for which specific heat capacity data are still missing).  [14] (2) Fluids for which a fundamental equation of state (FES), explicit in Helmholtz energy, is implemented in the open-source software REFPROP 10.0 [5]. Equations of state are necessary to calculate the properties of the working fluids in an HVAC&R system: without them, the achievement of an optimal system design is far-fetched. In the following, the reference values for thermodynamic properties are calculated using REFPROP 10.0 and thus for each fluid the respective EoS available in the software is applied.  Fig. 1 shows, for each thermophysical property, the number of peerreviewed references available in 2018 (Bobbo et al. [4]) and in 2022, displaying only the six fluids that can be considered fully characterised (point 1) and for which an EoS is already implemented in REFPROP 10.0 (point 2) (R1234yf,     R1243zf, R1234ze(E), R1234ze(Z), R1233zd(E), R1336mzz(Z); R1224yd(Z) is not considered as it was not included in the previous review by Bobbo et al.). As can be seen from the figure, the 3 fluids R1243zf, R1233zd(E) and R1336mzz(Z) displayed a higher percentage increase in the number of published articles for all the considered properties, thus demonstrating a growing interest in these fluids from the scientific community; for this reason, particular attention is paid in this article towards R1243zf, R1233zd(E) and R1336mzz(Z), plus the fourth refrigerant R1224yd(Z), which was not included in the previous study (Bobbo et al., 2018 [4]).
For these four fluids, selected as described above, the following text proposes an extended analysis only for newly published articles (2018 to 2022) and for articles published before 2018 and not included in the review by Bobbo et al. [4]. On the other hand, in the following figures and tables, all data sets available so far for each of the above-mentioned refrigerants are listed, unless otherwise stated.

Critical Point
Experimental data for the critical parameters are available for 14 of the selected refrigerants reported in Table 2; values for critical temperature T c , critical density ρ c , and critical pressure p c are listed in Table 5 and graphically displayed in Fig. 2a and b as they are distributed in the P-T and ρ-T planes. For all the reported data sets critical temperatures and densities were directly measured, while the critical pressure p c is usually either extrapolated from T c or calculated as an adjustable parameter during vapor pressure curve regression, as indicated in Table 5.

Vapor Pressure
Experimental vapour pressure data are available for 16 of the selected refrigerants and for each fluid they are listed in Table 6 together with the respective deviations from REFPROP 10.0 [5].      Fig. 3b.

PVT Properties
As shown in Table 3, experimental PVT data are currently available for 17 of the selected refrigerants; for each fluid, these data are listed in Table 7, divided into single-phase and saturated data points, accompanied by the percent absolute average deviation AAD% of each data set. Looking at the number of papers available, R1234yf and R1234ze(E) remain the most studied fluids, although, as can be seen from Fig. 1, much effort has been put into the experimental characterisation of R1336mzz(Z), R1233zd(E) and R1224yd(Z) over the last 4 years; for this reason, a more comprehensive analysis of the peer-reviewed literature published after 2018 is carried out for these 3 fluids and R1243zf, exactly as in the previous paragraph.  Figure 4a shows the distribution of the data on the P-T plane, with the pressure shown in logarithmic scale for clarity. On the other hand, Fig. 5a shows the percentage deviations of the experimental data for the fluid from REFPROP.
Compressed liquid: Higashi and Sakoda (2018) [39] reported 28 points in the limited temperature range T r = 0.87 to 0.998 and a much wider pressure range P r = 0.64 to 1.96 (P max = 6.73 MPa) with AAD% = 0.29%, MAD% = 1.12% and higher deviations at lower temperatures.  [54] reported 128 density data with T r = 0.67 to 0.98 and P r = 0.03 to 0.82 (P max = 2.88 MPa); the AAD% is 0.92% with MAD% = 1.78% and deviations evenly distributed over the temperature range.
Supercritical region: Higashi and Sakoda (2018) [39] also reported 25 data points in the near-critical region, between T r = 1.003 and 1.128 and P r = 1.040 to 1.962. Deviations from REFPROP 10.0 are AAD% = 0.42% and MAD% = 1.23%, slightly higher at pressures closer to the critical region.
Saturation: Higashi and Sakoda (2018) [39] provided 6 liquid density data and 7 vapour density data in the near critical region T r = 0.96 to 0.999 for experimental R1224yd(Z): 3 data sets with a total of 245 data points were identified in the literature: all four sets contain compressed liquid data, 1 superheated vapour data, 1 saturated vapour data and 2 saturated liquid data. Figures 4b and 5b show the distribution of the data on the P-T plane and the percentage deviations of the experimental data for the fluid from REFPROP, respectively.
Compressed liquid: the 16 data from Sakoda and Higashi (2019) [47] exhibit small deviations over the entire temperature and pressure range (T r = 0.83 to 0.94 and P r = 0.59 to 1.92, P max = 6.41 MPa), with AAD% = 0.08% and  [94] provided 93 density data in the medium temperature range T r = 0.66 to 0.85 and the much wider pressure range P r = 0.075 to 10.507; the deviations are really small: AAD% = 0.046% and MAD% = 0.12%. However, it must be considered that 4 data points were excluded because they are outside the validity range of the EoS used in REFPROP. Finally, Lago et al. (2022) [95] presented 80 data points, 9 of which exceed the validity range of the EoS implemented in REFPROP, in the range T r = 0.64 to 0.94 and P r = 0.29 to 10.55 (P max = 35 MPa); the agreement with the reference data from REFPROP is excellent, with AAD% = 0.03% and MAD% = 0.098%.  [47] present 30 data in the range T r = 0.77 to 0.98 for pressures P r = 0.11 to 0.55 (P max = 35 MPa); deviations are higher than those of the data for compressed liquid, with AAD% 0.19% and MAD% = 0.57%.
Saturation: the 17 data from Sakoda and Higashi (2019) [47] includes data points for saturated liquid and 8 for saturated vapour in the range T r = 0.82 to 1.00, with high deviations increasing with temperatures: AAD% = 1.804% and MAD% = 5.57%.   [94] presented 9 points for saturated liquid from T r = 0.66 to T r = 0.85; the deviations are low in the whole temperature range (AAD% = 0.04%, MAD% = 0.07%).
R1233zd(E): a total of 532 data points and 7 papers are reported for this fluid; compressed liquid data, as well as superheated vapour data are contained in 4 of the 7 papers. 2 papers contain data for the supercritical region and 2 contain data for the saturated region. With reference to   [4], 2 papers are added to this review and are discussed in the following lines. Figure 4c shows the distribution of the data on the P-T plane, while Fig. 5c shows the percentage deviations of the experimental data for the fluid from REFPROP.  [4], which are described in more detail below. They come to a total of 601 data points and three papers: all of them report compressed liquid data, 2 report superheated vapour data, one reports near critical point data and one reports supercritical data. As with the previous fluids, Figs. 4d and 5d show the data distribution on the P-T plane and the percentage deviations of the experimental data for the fluid from REFPROP, respectively.
Compressed liquid: the Sun et al. (2019) [97] paper contains 137 density data distributed in the medium temperature range T r = 0.64 to 0.84 and a much wider pressure range P r = 0.035 to 34.55; the maximum reported pressure is 100 MPa, exceeding the validity range of the EoS used in REFPROP (P max = 46 MPa), so 40 points were excluded from the analysis. The data showed good agreement with REFPROP, with AAD% = 0.03% and MAD% = 0.11%. Then McLinden and Akasaka (2020) [32] provided 97 data points in the wider temperature range T r = 0.52 to 0.99 for pressures P r = 0.22 to 12.33 (P max = 35.70 MPa); again, the agreement with the reference data was excellent (AAD% = 0.01% and MAD% = 0.07%). Finally, supercritical: McLinden and Akasaka (2020) [32] identified 7 data near the critical point with T r = 1.01 to 1.03 and P r = 1.14 to 1.64, showing higher deviations compared to the compressed liquid and superheated vapour regions, i.e., AAD% = 0.13% and MAD% = 0.26%.

Specific Heat Capacity
Specific heat capacity measurements are only available for seven of the selected fluids, which turns out to be the least studied property among those considered in this work. Table 8 summarises the available data sets and distinguishes between data related to isochoric heat capacity in the single-phase region (c v ), ideal gas capacity (c p 0 ), isobaric heat capacity under saturated conditions (c p sat ) and in the single-phase region (c p ), the latter being the most frequently measured among the heating properties; as can be seen from the Table, R1234yf and R1234ze(E) are also the most studied fluids for this property, while for two of the four fluids on which this study focuses, i.e., R1243zf and R1224yd(Z), no data are available at all. On the other hand, three new papers have recently been published for R1233zd(E) and R1243zf: R1243zf: a total of two papers and 150 data points are now available, both regarding directly measured isochoric heat capacity in the compressed liquid region.

Speed of Sound
As far as the speed of sound is concerned, the number of analysed fluids has more than doubled compared to the last review, for a total of eleven fluids, namely R1234ze (E), R1234yf, R1233zd(E), R1234ze(Z), R1336mzz(Z), R1224yd(Z), R1243zf, R1336mzz(E), R13I1, R1123 and RE356mmz. For all these fluids speed of sound data are listed in Table 9.
For R1243zf andR1224yd(Z), one data set each is available in the peer-reviewed literature, both for the superheated vapour phase. Kano   R1233zd(E) is now the most studied fluid, with a total of 461 points and 5 data sets, three of which were not included in the previous review, one for the compressed liquid and two for the superheated vapour: Lago et al. (2018) [143] reported 43 data for compressed liquid in the ranges T r = 0.62 to 0.80, P r = 0.08 to 9.28; the deviations from REFPROP are AAD% = 0.18 and MAD% = 0.35, slightly higher for lower temperatures. On the other hand, Kano (2021) [144] and Peng et al. (2022) [145] have both recently presented data for the compressed liquid, 36 and 92 data points, respectively: Kano (2021) [144] data range from 0.71 to 0.92 for T r and 0.08 to 0.24 for P r and show higher deviations from REFPROP for temperatures  Finally, two speed of sound data sets are available for R1336mzz(Z) so far, one published by McLinden and Akasaka (2020) [32] published after the 2018 review: the study reports 140 speed of sound data in the liquid phase in the ranges T r = 0.72 to 1.08 and P r = 0.01 to 0.70; the data are quite scattered between Table 9 Available experimental data for the speed of sound of selected refrigerants AAD relative to the EOS cited in Table 4 Fig. 8 shows the distribution of the available data on the P-T plane, while Fig. 9 gives the deviations from REFPROP.

Thermal Conductivity
Experimental measurements of thermal conductivity have increased by 300% in the last 4 years: while only 3 working fluids were studied in 2018, this number has now increased to 9, including R1234ze(E), R1234yf, R1233zd(E), R1234ze(Z), R1336mzz(Z), R1224yd(Z), R1243zf, R1336mzz(E) and RE356mmz. For all these  Table 10, divided into single-phase data and saturated data. As for the four fluids R1336mzz(Z), R1233zd(E), R1243zf and R1224yd(Z), 2 publications are available for each of R1336mzz(Z) and R1233zd(E), while only one thermal conductivity data set is available for each of the other fluids. As no thermal conductivity data were available for R1243zf and R1224yd(Z) prior to this paper, reference values calculated with REFPROP 10.0 are based on an extended corresponding states model in predictive mode [152], with an estimated uncertainty of 20% for both liquid and vapor phases.  [156] measured the thermal conductivity both in the compressed liquid region (74 data, temperatures from 314 K to 435 K) and superheated vapor region (92 data, temperatures from 321 K to 96 K) for pressures lower than 4 MPa; based on these data the correlation implemented in REFPROP 10.0 was made, with an estimated uncertainty of 3% along saturation and in gas phase, and higher at higher pressures and near critical point. Perkins and Huber (2020) [157] gathered a huge dataset of 3294 data points, 1167 for the compressed liquid phase (T = 192 K to 428 K, declared uncertainty: 1%), 1323 for the superheated vapor phase (T = 303 K to 444 K, declared uncertainty: 1.5%), and 804 for the supercritical region (T = 457 K to 498 K, declared uncertainty: 1%). Deviations from REFPROP, shown in Fig. 11d, are AAD% = 5.93 and MAD% = 16.71. Measurements were performed at pressures up to 69 MPa: since the correlations implemented in REFPROP for this fluid is only  Figure 10a, b, c, d displays the distribution of thermal conductivity data on the P-T plane, while Fig. 11a, b, c, d shows the deviations from REFPROP 10.0.

Viscosity
To date, experimental viscosity data for nine of the selected fluids have been published in the literature and are listed here in Table 11. As shown in Fig. 1, R1234yf is still the most studied working fluid, followed by R1233zd(E), for which four new sets of viscosity data have been published since 2018. The distribution of available data on the P-T plane for the four fluids R1243zf, R1224yd(Z), R1233zd(E)  Fig. 12a, b, c, d, while Fig. 13a, b, c, d shows the deviations from the reference data from REFPROP. It must be underlined that since before 2018 no viscosity data were available for these fluids, except from the data set from Hulse et al. (2012) [50], values provided in REFPROP 10.0 are calculated with an ECS and different estimated uncertainties depending on the fluid [152].
R1243zf: only the data set by   [163] is available today, which reports 12 data for saturated liquid kinematic viscosity and presents deviations higher than 10% at higher temperatures (MAD% = 13.25 at 372.84 K).

Surface Tension
Compared to 2018, the number of fluids whose saturated liquid surface tension has been studied has more than doubled, from five in   [4] to eleven at the present time. The publicly available datasets are listed in Table 12, along with their respective temperature range. R1233zd(E) has the highest number of available datasets, with four publications and 39 measurements in the range from 273 K to 432 K, while R1336mzz(Z) is the least studied working fluid, with only one dataset and 20 data in the range from 303 K to 399 K. For R1243zf, compared to the review by   [4], 12 additional data were measured in the range of 303 K to 432 K by  [163], while two different data sets were reported for R1224yd(Z) (Kondou and Higashi (2018) [181], Kondou (2019) [182]), totalling 35 data, both in the temperature range between 266 K and 372 K. The experimental data for the fluids are graphically displayed in Fig. 14. The deviations from REFPROP 10.0 are shown in Fig. 15. To calculate the reference values for R1243zf and R1233zd(E), the empirical correlation presented by klghfkljKondou et al. (2015) [183] and implemented in REFPROP was used. For R1224yd(Z) and R1336mzz(Z), on the other hand, no data were correlated at the time the latest version of the software was developed, so the values for these fluids were calculated using an ECS in predictive mode, with an estimated uncertainty of 10% [152].

Conclusions
It is a fact that for several HVAC&R applications there is still no definitive operating fluid able to simultaneously fulfil the requirements given by thermodynamic and new regulations addressing the Global Warming issue. As things stand today, the problem is often related to the lack of data on the thermophysical properties of new fluids. To this end, a literature search was conducted to identify new experimental data sets on the thermophysical properties of low global warming potential refrigerants published in the open literature during 2018 to 2022 and to update a previous review published in 2018. The amount of experimental data available for pure compounds, including refrigerants not considered in the previous review, has increased significantly in recent years, but research efforts are still needed to provide a complete description of all fluids of interest and to find the right substitutes for different applications. Currently, only 4 fluids have been fully investigated (R1234yf, R1234ze(E), R1233zd(E), R1243zf), while other 4 (R1234ze(Z), R1336mzz(Z), R1224yd(Z), R1336mzz(E)) have been almost completely characterised. For all fluids, several sets and a large amount of data are available for p sat and PVT, while fewer sets are available for specific heat capacity and speed of sound. Especially for the transport properties, more information is needed. It must also be stressed that when searching for new refrigerants, it is not only thermodynamic properties that need to be considered, but also ODP, GWP, flammability, toxicity, and material compatibility; in particular, further clarifications are needed regards the possible environmental pollution created by HFOs.