Wear 304 (2013) 191–201

Contents lists available at SciVerse ScienceDirect

Wear

0043-16http://d

n CorrUnivers

E-m

journal homepage: www.elsevier.com/locate/wear

Tribological performance of environmentally friendly refrigerantHFO-1234 yf under starved lubricated conditions

M. Wasim Akram a, Kyriaki Polychronopoulou a,b, Christopher Seeton c,Andreas A. Polycarpou a,d,n

a Department of Mechanical Science and Engineering, University of Illinois at Urbana-Champaign, 1206W Green Street, Urbana, IL 61801, USAb Department of Mechanical Engineering, Khalifa University of Science, Technology and Research, Abu Dhabi, P.O. Box 127788, UAEc Honeywell International Inc., Buffalo, NY, USAd Department of Mechanical Engineering, Texas A & M University, College Station, TX, USA

a r t i c l e i n f o

Article history:Received 7 September 2012Received in revised form27 April 2013Accepted 30 April 2013Available online 15 May 2013

Keywords:Green tribologyRefrigerantCompressorXPS

48/$ - see front matter & 2013 Elsevier B.V. Ax.doi.org/10.1016/j.wear.2013.04.035

esponding author at: Department of Mechaniity, College Station, TX, USA. Tel.: +1 979 458ail address: apolycarpou@tamu.edu (A.A. Poly

a b s t r a c t

Concerted research efforts have been made in the field of alternative refrigerants due to environmentalconcerns. Newly developed, HFO-1234 yf (2,3,3,3-Tetrafluoropropene) refrigerant has been proposed as adirect replacement solution for substituting the existing harmful hydrofluocarbon refrigerants, such asR134a (1,1,1,2-Tetrafluoroethane). This paper presents a comparative tribological study of R134a and HFO-1234 yf refrigerants. Specifically, we present scuffing resistance, friction and wear, using gray cast ironinterface under starved lubricated conditions. Better tribological performance was observed with HFO-1234 yf, compared to R134a. Optical microscopy and profilometry techniques were employed to analyzewear, and scanning electron microscopy was carried out to reveal the morphological changes along withthe mode of wear. Chemical composition of the substrate was studied using energy dispersive X-rayspectroscopy, which showed the presence of fluorine element in the case of HFO-1234 yf refrigerant. Thisresult in conjunction with the X-ray photoelectron spectroscopy results (disks and pins analyses),support the presence of fluorine-containing tribolayers preventing metal–metal contact and sustainingprolonged interface functionality.

& 2013 Elsevier B.V. All rights reserved.

1. Introduction

During the last several years, the research field of alternativerefrigerants has been intensified due to environmental concerns.The introduction of chlorofluorocarbon (CFC), in the 1930s, asrefrigerants solved the initial problems related with flammabilityand toxicity of earlier refrigerants [1]. However, in the 1970s and1980s it was proven that the existence of chlorine in CFCs causedthe depletion of the ozone layer [2,3], which led to the replace-ment of CFC with hydrofluorcarbon (HFC) refrigerants. Since then,HFC refrigerants have been almost exclusively used in refrigerationand air-conditioning compressor applications. Among HFC-basedrefrigerants, 1,1,1,2-Tetrafluoroethane (R134a) is commonly used,especially in automotive air-conditioning systems due to itseffective thermodynamic performance [4–6]. However, due to itsGlobal Warming Potential (GWP) index of above 1100 it hasadverse effects on the environment in terms of global warming.According to the Montreal and Kyoto Protocols the detrimental

ll rights reserved.

cal Engineering, Texas A & M4061; fax: +1 979 845 3081.carpou).

effects on the environment of the high GWP producing refrigerantmust be reduced [7,8]. Therefore since the early 2000s, industrieshave concentrated their efforts in identifying natural and/orenvironmental friendly refrigerants.

Among different natural refrigerants such as water, air, isobu-tene, and ammonia, CO2 is an attractive candidate that has receivedrecent attention [9], and the tribological performance of compressorsurfaces in CO2 atmosphere has also been studied. Demas et al. [10]and Nunez et al. [11,12] found that CO2 exhibits the same or bettertribological performance compared to R134a, especially at higheroperating chamber pressures [13]. Therefore, existing systems willrequire modification with high pressure equipment, if CO2 is to beused to replace conventional refrigerants.

Newly developed, 2,3,3,3-Tetrafluoropropene (HFO-1234 yf)refrigerant is considered as a “drop-in” solution for possessingsimilar thermodynamic properties compared to R134a [14–19].Spatz et al. also demonstrated low toxicity of HFO-1234 yf similarto R134a along with manageable flammability [14] indicating analternative refrigerant for automotive air-conditioning compres-sors. However, this proposition needs to be supported withinterfacial characterization of compressor material surfaces. Sincethere is a difference in chemical nature, namely carbon–carbonbonding, between the two refrigerants, this could contribute to

Sliding Face

Fig. 1. Photographs of typical gray cast iron (a) disk and (b) pin.

M.W. Akram et al. / Wear 304 (2013) 191–201192

different tribological behavior. Note that, the chemical formulasfor HFO-1234 yf and R134a are H2C¼CFCF3 and H2FC–CF3, respec-tively. This means that HFO-1234 yf consists of an unsaturatedcarbon–carbon bond (–C¼C–), whereas R134a has only saturatedcarbon–carbon (–C–C–) bonds. These bonding could react differ-ently at the material interface which was observed in an earlierstudy [20] under unlubricated conditions. Under such conditions,better tribological performance was found in HFO-1234 yf envir-onment and gray cast iron interface. In current compressorsystems, lubricants are used and could play an important role indetermining the tribological properties, especially under starvedlubricated conditions, which was excluded in Ref. [20]. Under suchstarved lubricated conditions, the maximum critical load at theinterface, termed as scuffing load/resistance, along with thefrictional resistance, wear, and sub-surface contact temperatureare very important and could lead to sudden catastrophic failure oflubricated sliding contacts [21–25]. This severe phenomenon ofwear is accompanied by abrupt rise in friction, and sometimesassociated with noise and vibration. Delaying or eliminatingscuffing failures in the presence of lubricant/refrigerant mixturemeans better performance, i.e., the interface can withstand higherload in that particular refrigerant/lubricant mixture. In addition toscuffing performance, we also investigate the wear and frictionperformance using constant load experiments.

During the lifetime of a compressor, different stages of wearmay occur including running-in and steady-state wear. The initialstages of running-in wear, especially under boundary/mixed orstarved lubricated conditions are critical and strongly influencethe overall tribological performance [26]. Starved lubricated con-ditions in a compressor (i.e., in the absence of full film conditions)represent aggressive operating conditions, which are eitherencountered by tribopairs away from flooded areas, or duringtransient operation. In this work, a small predetermined amountof lubricant is directly applied at the interface, which is alsoflooded with vapor refrigerant. Polyalkylene Glycol (PAG) lubricantwas used for this purpose, as it is commonly used in automotiveair-conditioning compressors and has good miscibility with therefrigerant and good compatibility with compressor materials,including elastomers [27]. PAG is a linear polymer of ethyleneand propylene oxides having the general formula RO–[CH2CHOCH3]n[CH2CH2O]m–H. Note that, mineral oil was notappropriate for R134a refrigerant due to the immiscibility issuewhich, therefore, prompted the development of synthetic lubri-cants such as PAG, POE and others [28,29]. Since the focus of thisstudy was to compare the tribological performance in the presenceof R134a, and HFO-1234 yf and for keeping the same conditions forcomparison, PAG was chosen as the lubricant.

Noteworthy, tribological performance of compressor interfacesis greatly influenced by the development of tribo-layers during thetribo-experiments. Researchers have reported the beneficial actionof tribolayers on the topmost surface, such as in the case of CO2

environment, where an oxide-rich anti-wear layer was identifiedusing Augur Electron Microscopy (AES) [30]. The chemical com-position of the surface can also be analyzed by employing X-rayPhotoelectron Spectroscopy (XPS). In our previous study, we usedXPS to report fluorine-passivated (FeF3) tribo-layer in the presenceof HFO-1234 yf originating from tribochemical reactions [20]. Notethat, the fluorine-enriched layer exhibited better tribologicalperformance [36]. Development of fluorine-enriched preventivelayer is the underlying reason to achieve lower friction and lesswear for PTFE coating [30,31,37]. Theiler et al. found FeF3 layerwhile working with PTFE coatings against steel surfaces [37]. Thusit was believed that facilitating fluorinated reaction at the contactinterface would be beneficial. However, the intact nature of thepreventive layer is also important to maintain better tribologicalperformance during sliding action. Based on the fact that XPS is

capable of acquiring data over a larger area, compared to AES [31]and since tribo-layers are developed within few nanometers of thesurface, XPS is a powerful diagnostic tool for probing the forma-tion of tribolayers.

In this paper, we compare the tribological performance of HFO-1234 yf with R134a using PAG lubricant under starved lubricatedconditions. We elucidate the influence of HFO-1234 yf for com-pressor material surfaces with the aid of detailed tribo-chemicalanalysis over the worn surfaces. Specifically morphologicalchanges of the surface were analyzed under Scanning ElectronMicroscopy (SEM), and constituent elements of the bulk materialwere identified using Energy Dispersive System (EDS). XPS wasemployed for identifying any chemical changes on the surface ofthe worn gray cast iron surfaces (disks and pins). Three differentsets of experiments were carried out for comparing the scuffingand wear performance in the presence of HFO-1234 yf and R134arefrigerants. In the first set, the interfacial maximum critical loadsi.e. scuffing resistance was evaluated under step loading condi-tions. In the second and third sets of experiments, wear experi-ments were conducted under constant load conditions.

2. Experimental procedure

2.1. Controlled tribological experiments

The exact replication of the conditions found inside a com-pressor is complicated due to issues related to variable speeds,complex loading conditions, contact geometries, impurities, sur-face finish and treatments among other factors. Using the samematerials, contact geometry, and controlled temperature andpressure in conjunction with a specified normal load, experimen-tal simulation results correlate with compressor system results. Inthis work, a specialized high pressure tribometer (HPT) was usedfor performing three sets of controlled tribological experiments.The tribo-pairs were gray cast iron disks against nominally flatgray cast iron pins. This combination was chosen to best simulate ascroll-type air-conditioning compressor [23].

Fig. 1 shows photographs of a typical disk and pin used in thiswork. The contact geometry used in this study was a pin-on-diskgeometry, where the disk was the upper rotating body, and the pinwas the lower stationary body. The pins were oriented to create anominally flat contact as illustrated in Fig. 1b and are 6.7 mm indiameter and 8 mm long, with a 1 mm diameter hole for miniaturethermocouple insertion. The disks were 76 mm diameter and6.4 mm thick. The thermocouple was positioned 2 mm beneaththe contact interface. The pin was fixed within a special holderwhich was then placed in a fixture such that it was allowed to

M.W. Akram et al. / Wear 304 (2013) 191–201 193

self-align to the disk surface ensuring a uniform contact. Thetribometer has an upper rotating spindle to which the disk isattached and a stationary lower fixture that holds the pin. Acontrolled normal load ranging from 45 N to 4450 N can beapplied with this tribometer. The lower fixture is mounted on a6-axis force transducer which measures the forces in the x, y, and zlinear directions to obtain the coefficient of friction.

The first set of experiments was scuffing type experimentsunder step loading at room temperature of 23 1C. Two differentrefrigerants were used namely, HFO-1234 yf and R134a. Details ofthe refrigerants can be found in the manufacturer's data sheet[34,35]. The chamber pressure was kept constant at 0.2 MPa,which was found at the inlet of automotive air-conditioningcompressor system. A small amount (about 23 mg) of PAG lubri-cant (Idemitsu Kosan Co., Ltd., Grade: PZ68S) was applied directlyat the interface as shown in Fig. 2, before the initiation of the test.This lubrication method precluded the influence of lubricantsupply rate and variability of pressure/miscibility effect on thetribological performance. A rotational speed of 1000 rev/minwhich corresponds to a linear speed of 2.4 m/s was used. Theinterfacial normal load was increased in steps of 89 N every 1 minuntil scuffing occurred (as characterized by an abrupt change infriction). A preload of 50 N was applied to ensure complete andsmooth running-in.

Experiments in Set 2 and Set 3 were carried out by keeping thenormal load constant to measure the wear under starved lubrica-tion. Set 2 experiments were performed at 222 N for 30 min in thepresence of both refrigerants, while Set 3 experiments wereconducted at 445 N. Besides the loading schemes, the remainingexperimental parameters were identical in all three Sets ofexperiments, as detailed in Table 1. Note that, wear experimentswere also conducted at 222 N under dry condition in the presence

One drop of PAG

Pin

Self-aligning pin holder

Fig. 2. Photograph of a pin in the self-aligning holder showing application oflubricant.

Table 1Summary of experimental conditions.

Conditions Set 1 Set 2 Set 3

Type of experiment Scuffing Wear WearDirection Unidirectional Unidirectional UnidirectionalLoading 89 N/min 222 N 445 NTemperature, 1C 23 23 23Chamber pressure, MPa 0.2 0.2 0.2Lubricant PAG PAG PAGAmount of lubricant, mg 23 23 23Linear speed, m/s 2.4 2.4 2.4

of HFO-1234 yf only, to understand the influence of lubricant inthe tribological performance.

Prior to and following testing, all specimens were ultrasonicallycleaned in acetone for 10 min and subsequently rinsed with2-propanol. Each experiment was performed at least twice toensure repeatability. After each experiment, surface topographywas analyzed using a standard low magnification optical micro-scope, and wear was quantitatively analyzed using a Tencor P-15contact prfilometer.

2.2. SEM/EDS analysis

Two scuffed and two worn samples that were tested understarved lubricated conditions were analyzed using SEM to inves-tigate morphological changes that occurred. In some cases, EDSwas also employed to obtain the chemical composition of thesubstrates. A JEOL6060LV SEM operated at variable excitingvoltage coupled with an EDS detector, which measures the relativeabundance of emitted X-rays against their energy, was utilized forthis purpose.

2.3. XPS analysis

XPS analysis was conducted on a Perkin Elmer PHI 5400spectrometer equipped with a hemispherical electron analyzerand a non-monochromatic Mg Kα X-ray source (1253.6 eV). Allreported photoelectron binding energies are referenced to the C1 s feature of adventitious carbon at 285 eV (internal standard) totake into account charging effects. XPS studies were performedinside and outside the wear tracks of the gray cast iron disksurfaces and on the gray cast iron pins. The survey spectra wereacquired at pass-energy of 178.95 eV. A certain region of thespectrum was scanned a number of times to obtain a goodsignal-to-noise ratio (the detailed scan was performed at a pass-energy of 35.75 eV). The measurements were performed in threedifferent areas inside the wear track of each individual sample forrepeatability purposes. The peak fitting was performed using CasaXPS (version 2.3.14) software.

3. Results and discussion

3.1. Friction and scuffing performance

Fig. 3 depicts typical experimental results for Set 1 scuffingexperiments, comparing the performance of HFO-1234 yf withR134a under identical conditions. Fig. 3a shows the maximumscuffing load and Fig. 3b presents the frictional behavior in thepresence of the two refrigerants. As seen in Fig. 3a, the scuffingresistance of the cast iron-cast iron interface in the presence ofR134a environment was approximately 600 N, whereas for HFO-1234 yf was twice as much at about 1250 N.

The steady-state friction coefficient (before scuffing occurred)in the presence of R134a was slightly higher (0.09) compared toHFO-1234 yf (0.075). These tests indicated the maximum inter-facial load that a tribo-pair can withstand, and their lubricity(measured via their steady-state friction coefficient). Clearly, theinterface could sustain a significantly higher load in the presenceof HFO-1234 yf refrigerant, compared to R134a under starvedlubrication. Note that these observations, which were understarved lubrication, were in accord with our previous study underdry conditions [20].

In-situ friction coefficient and near-contact temperature underconstant load wear experiments (Sets 2 and 3, Table 1) are presentedin Fig. 4. The friction behavior under the two different loadings isshown in Fig. 4a, comparing the tribological performance of the two

00.0

0.1

0.2

0.3

0.4

0.5

CO

F

Sliding Distance (m)

R134a at 220 N R134a at 445 N HFO-1234yf at 445N HFO-1234yf at 220 N HFO-1234yf at 220 N (dry)

HFO-1234yf

0

20

40

60

80

100

Nea

r Con

tact

Tem

p. (C

)

R134a HFO-1234yf

R134a

HFO-1234yf

700 1400 2100 2800 3500 4200 0Sliding Distance (m)

700 1400 2100 2800 3500 4200

Fig. 4. Experimental results under constant load wear experiments in the presence of HFO-1234yf and R-134a: (a) friction coefficient, Set 2 (222 N), Set 3 (445 N), (b) near-contact temperature Set 2 (222 N).

00

200

400

600

800

1000

1200

1400

1600

Load

(N)

Time (min)

HFO-1234yf R134a

Scuffing Load (HFO-1234yf)

Scuffing Load (R134a)

0.0

0.1

0.2

0.3

0.4

CO

F

HFO-1234yf R134a

Scuffed

2 4 6 8 10 12 14 16 18 20 0Time (min)

2 4 6 8 10 12 14 16 18 20

Fig. 3. Comparison of (a) scuffing load, and (b) friction coefficient in the presence of HFO-1234yf and R-134a refrigerants (Set 1, Table 1).

M.W. Akram et al. / Wear 304 (2013) 191–201194

refrigerants. The friction characteristic of HFO-1234 yf refrigerantunder dry condition is also presented.

In all cases the initial transient friction coefficient behavior wassomewhat erratic, showing fluctuating values around 0.1 due thestarved lubricated conditions and until the surfaces were run-in.Interestingly steady-state behavior was delayed until about2800 m sliding distance, even though the near-contact tempera-ture reached steady-state after 1400 m. The steady-state valueswere almost the same for both refrigerants. At the low load (Set 2,222 N), the average friction coefficient for HFO-1234 yf was 0.06compared to 0.065 for R134a. At the higher load (Set 3, 445 N), theinterface failed in the presence of R-134a, before the completion ofthe experiment. In contrast to the R134a environment, the tribo-samples were able to operate successfully without failure for theentire duration of 4320 m sliding distance in the presence of HFO-1234 yf refrigerant. Thus, it was envisaged that as the testprogressed, tribo-beneficial layers developed at the interface inthe presence of HFO-1234 yf refrigerant, which protect the sub-strate from adhesive wear. It is worthy to note that, even under drycondition the experiment did not fail and the average frictioncoefficient was about 0.16.

In HFO-1234 yf environment, an interesting trend in thefrictional behavior is observed after the initial run-in period forboth dry and starved lubricated conditions. As indicated above, theinitial friction coefficient was low and then increasing and fluctu-ating, and subsequently becomes stable. This characteristic wasbelieved to be due to the chemical reactions at the interface whicheventually lead to the formation of the beneficial tribo-layers. Thusit was hypothesized that after this stage which can also be termedas “transitional phase” preventive tribo-layers were formed on the

surface. This conjecture will be verified with the aid of SEM andXPS analysis. No such behavior was observed in R134a environ-ment, and thus the tribo-layers formed should be different.

Comparing the overall tribological behavior under unlubricated[20] and starved lubricated conditions, clearly starved lubricatedconditions, under both HFO-1234 yf and R134a refrigerants, offersignificant tribological benefits, as far as maximum load carryingcapacity, and low friction are concerned. Having said this, underboth unlubricated and starved lubricated conditions HFO-1234 yfoutperforms R134a refrigerant.

Fig. 4b shows the near-contact temperature from the Set 2 wearexperiments (222 N) under starved lubrication. Even though theminiature thermocouple was positioned 2 mm underneath thecontacting interface and could not measure the exact interfacialtemperature, it provided useful information of the near-contacttemperature and frictional heating that the interface experienced.Fig. 4b shows that in the case of the 222 N load, the near-contacttemperature increases during the running-in period for bothrefrigerants and reached a steady value after about 10 min or1400 m sliding distance. The initial rate of temperature increase ishigher in HFO-1234 yf environment. The steady-state temperaturevalues were 65 1C and 40 1C for HFO-1234 yf and R-134a, respec-tively. The rise in temperature in the presence of HFO-1234 yfcould be partly corroborated to the aggressive chemical reactivityof this refrigerant which is expected on the basis of fluro-olefins.Thus, higher frictional heating is evident in the presence of HFO-1234 yf. The elevated sub-surface temperature indicates thatdifferent events could be taking place on the surface such as bondbreaking and concomitant reactivity in the case of HFO-1234 yfrefrigerant. Therefore, sufficient amount of elemental components

M.W. Akram et al. / Wear 304 (2013) 191–201 195

of HFO-1234 yf would be available at the interface for theformation of different surface species allowing different chemistryto happen. Components of the surface chemistry will be presentedand discussed in the XPS section. The tribological propertiesobtained from the experiments are summarized in Table 2, clearlyshowing that HFO-1234 yf exhibits better performance comparedto R134a.

3.2. Wear analysis

Wear analysis was first carried out qualitatively using lowmagnification optical microscopy. The images were taken insideand near the wear tracks of the disk surfaces as shown in Fig. 5.The sliding direction is indicated by the solid yellow arrow. Initialmachining marks, especially outside the wear tracks (shown withred arrows) are also visible. Fig. 5a and b shows the wear tracks inthe presence of HFO-1234 yf and R134a refrigerants (Set 2), andFig. 5c and d for Set 3.

In the case of Set 2 (222 N), there was no measurable wear onthe disk surface under HFO-1234 yf environment, just some minorburnishing. On the other hand, significant wear was seen in theR134a environment as shown in Fig. 5b. At 445 N (Set 3), in thepresence of HFO-1234 yf, burnishing is seen (Fig. 5c). Since theinterface failed at 445 N in R134a environment, severe wear andscuffing is seen in Fig. 5d. Clearly, the presence of HFO-1234 yf

Table 2Comparative tribological performance of HFO-1234yf and R-134a.

Properties HFO-1234yf R134a

Set 1: Scuffing load, N 1200 600Set 1: Average friction coefficient (before scuffing) 0.09 0.085Set 2: Average friction coefficient (222 N) 0.06 0.065Set 2: Near-contact temperature (222 N), 1C 65.0 40.0Set 2: Maximum wear depth (222 N), mm o1.0 4.0

3 mm

3 mm

Wear track

Wear track

Sliding direction

Sliding direction

Fig. 5. Low magnification optical microscopy images after wear experiments in the pres(d) Set 3. (For interpretation of the references to colour in this figure, the reader is refe

refrigerant was beneficial leading to improved tribological per-formance.

To quantify the amount of wear, long profilometric line scansacross the disk wear tracks were performed, as shown in Fig. 6.Fig. 6a and c shows the profilometric analysis of the worn disks inthe presence of HFO-1234 yf at 222 N and at 445 N, respectively.As seen in the optical microscopy images, with 222 N and HFO-1234 yf environment, no wear was seen, whereas a moderateamount of wear (about 4 mm deep) was found in R134a environ-ment, Fig. 6b. At 445 N, burnishing and very light wear of 1 mmdepth is seen in Fig. 6c, in the presence of HFO-1234yf. In contrastto the HFO-1234yf environment, severe wear of about 12 mm alongwith deep wear scratches (about 16 mm) were seen in the presenceof R134a which is depicted in Fig. 6d.

The influence of starved lubrication on the wear behavior of graycast iron in HFO-1234yf environment appeared to be negligible, asit is similar to the unlubricated case. Fig. 7 depicts a low magnifica-tion image and a profilometric scan of the worn disk tested underdry condition in HFO-1234yf environment. Referring to Fig. 7, nosignificant wear is seen on the disk surface. Thus, in the presence ofHFO-1234yf under a constant load of 222 N, mild burnishing isobserved of the order of 1–2 mm, irrespective of using either starvedlubrication or dry condition. Therefore, it is believed that protectivelayers are formed during the wear experiments irrespective of usingthe lubricant or not in HFO-1234 yf environment. These protectivelayers were believed to form during the friction transitional phaseas discussed in the previous section. This was in contrast withR134a refrigerant, where wear behavior was significantly improvedwith the application of lubricant. Details on the mode of wear areanalyzed via SEM in the following section.

3.3. SEM analysis

Fig. 8a and b shows SEM images inside the wear tracks of thedisk and the pin in the presence of R134a (Set 2). Abrasive wear

Sliding direction

Wear track

3 mm

Scuffed region

3 mmSliding direction

Machining marks

ence of HFO-1234yf (a) Set 2, (b) Set 2, (c) Set 3 and in the presence of R-134a and,rred to the web version of this article.)

0-5

-4

-3

-2

-1

0

1

2

Wea

r Dep

th (u

m)

Wear Scan (mm)

Burnishing

-5

-4

-3

-2

-1

0

1

2

Wea

r Dep

th (u

m) Mild Wear

-5

-4

-3

-2

-1

0

1

2

Wea

r Dep

th (u

m)

-20

-15

-10

-5

0

5

Wea

r Dep

th (u

m)

Wear

Severe wear (Scuffed)

1 2 3 4 5 6 7 0Wear Scan (mm)

1 2 3 4 5 6 7

0Wear Scan (mm)

1 2 3 4 5 6 7 0Wear Scan (mm)

1 2 3 4 5 6 7

Fig. 6. Wear profiles after Set 2 (222 N) experiments: (a) HFO-1234yf; (b) R134a, after Set 3 (440 N) experiments: (c) HFO-1234 yf, and (d) R134a.

3 mm

Wear track

Sliding Wear

0-5

-4

-3

-2

-1

0

1

2

Wea

r Dep

th (u

m)

Wear Scan (mm)

Burnishing

1 2 3 4 5 6 7

Fig. 7. (a) Optical microscopy image and (b) wear profile of the worn disk tested at 222 N under unlubricated conditions in HFO-1234 yf environment.

M.W. Akram et al. / Wear 304 (2013) 191–201196

was visible on both samples resulting in deep grooves (Fig. 8b).These findings inferred the absence of any significant protectivetribo-protecting layer in the presence of R134a or if any wasformed it was prematurely expelled out from the contact zone,providing no surface protection. The material removed from theinterfaces in the form of wear debris amalgamated inside thecontact region and eventually caused cold junction, namelyadhesive wear, which is clearly seen at higher loads, Fig. 8c (disk)and d (pin) (Set 3).

In contrast to the effect of R134a, under HFO-1234 yf, smootherwear is observed, as shown in Fig. 9 for both Set 2 and Set3 experiments. In both cases, no dominant wear mode is observedexcept some minor burnishing. There was also less evidence ofabrasive wear over any of the samples. This implied the presenceof some sort of protective layers on the surface. Remarkably, someperforated like surface was visible over the pin surface. The

formation of such kind of surface layer might be originated fromthe chemical reaction of the refrigerant at the interface.

As discussed above, in the presence of HFO-1234yf some sort oftransitional friction behavior is observed. To investigate thisbehavior a separate experiment in the presence of this refrigerantand under Set 3 (445 N) load conditions was conducted andstopped after 5 min or 700 m of sliding distance. Fig. 10a depictsSEM images of the evolutionary changes of the pin surface in thepresence of HFO-1234 yf refrigerant at this transitional stage.A lower exciting voltage of 5 kV was used in order to avoidcharging effects coming from the nonconductive “patch-like” filmon the surface and hence for achieving a better capturing of thesurface changes. Observing Fig. 10a, there were some particles,aligned with the sliding direction. These bundles of material werebelieved to be the initiation of the tribo-layers originating fromthe chemical reaction involving HFO-1234 yf refrigerant and

20 kV 20 kV

20 kV 20 kV

Sliding direction

Sliding directionSliding direction

Sliding directionAbrasive wear

Amalgamation of materials

x350 50µm x350 50µm

x350 50µmx350 50µm

Fig. 8. SEM images in the presence of R134a: (a) disk, Set 2; (b) pin, Set 2; (c) disk, Set 3; and (d) pin, Set 3.

20 kV x350 50µm 20 kV x350 50µm

20 kV x350 50µm 20 kV x350 50µm

Sliding direction

Sliding direction

Sliding direction

Sliding direction

Smoother surface

Mild burnishing

Fig. 9. SEM images in the presence of HFO-1234 yf: (a) disk, Set 2; (b) pin, Set 2; (c) disk, Set 3; and (d) pin, Set 3.

M.W. Akram et al. / Wear 304 (2013) 191–201 197

surface iron. Since these particles could act as “bumps” or obstaclesat the interface, the friction behavior was not steady, as discussedin the preceding section. However, as the test progresses, theparticles grew continuously over the surface and eventuallydeveloped a fluorine-enriched layer. This layer had a tribo-beneficial effect which reduced the friction coefficient and wear.Note that, no such particles were seen on the disk surface asshown in Fig. 10b. This could be explained from the contact pointof view. Since the pin is engaged constantly, tribo-layer can be

formed easily over the pin surface compared to the disk surface. Inaddition, during the experiments, some fluorine-enriched materialmight transfer to the pin surface from the disk surface which will beverified later via EDS and XPS. There is no evidence of fluorine-basedlayer when scuffing occurred. In other words, interfacial failureoccurred when the last protective layer, which might be consistingof fluorine-containing species, is completely expelled from thecontact interface. Therefore, it might be envisaged that the presenceof fluorine layer prolongs the scuffing behavior, which is the case in

5kV 5kV

Fluorine enriched particles

Direction of sliding

x350 50µm x350 50µm

Fig. 10. SEM images in the presence of HFO-1234f during the transitional stage (700 m sliding distance): (a) pin, 445 N; and (b) disk, 445 N.

Fe

FeO

C

O

F Si

Fe

Fe

R134a environment (inside wear track)

HFO-1234yf environment (inside wear track)

C

2

200

400

600

800

1000

100

200

300

400

500

Energy (keV)

Cou

nts

Cou

nts

4 6 8

2Energy (keV)

4 6

Si

Fig. 11. EDS analysis after Set 2 wear experiments: (a) R134a; and (b) HFO-1234 yf.

M.W. Akram et al. / Wear 304 (2013) 191–201198

HFO-1234yf environment. This is in accordance with our previousfindings, where we demonstrated the existence of fluorine underdry condition for non-scuffed samples [20].

3.4. EDS analysis

The chemical composition of the pin surface is shown in Fig. 11with the spectra extracted from EDS analysis after the wearexperiments at 222 N (Set 2). An area map of 200 mm�200 mmwas used for this purpose. No EDS analysis was performed for thewear experiments at 445 N and for the scuffing experiments (Set1). Only successful (no catastrophic failure) wear experimentswere analyzed for a fair comparison. Through this analysis theinfluence of the two different refrigerants on the bulk chemistry ofthe cast iron was investigated. Fig. 11a presents the spectrum inthe presence of R-134a, while Fig. 11b presents the spectrum in thepresence of HFO-1234 yf. Part of oxygen that was detected, mightbe originated from the PAG lubricant, since in the chemicalstructure of PAG there is terminal –OH group. Similar oxygen peakwas detectable in the HFO-1234 yf spectrum. One of the significantdifferences between the two spectra was the existence of fluorineelement on the substrate. The enrichment of the surface influorine was only found in the HFO-1234yf spectrum. Since theexperiment was performed in a closed/controlled chamber, theappearance of the fluorine peak in EDS originated from therefrigerant. However, no such peak was identified in the case ofR134a refrigerant. This result suggested fluorine-containingadsorbed layers in HFO-1234yf atmosphere, which providedanti-wear protective properties to the gray cast iron surface.

3.5. XPS analysis

3.5.1. Disk surface analysisXPS analysis was performed in the cases of wear tracks formed

under scuffing and wear experiments conditions as described inTable 1. Fig. 12 presents the F1s core level spectra obtained overthe wear tracks in the presence of PAG/R134a (Fig. 12a) and PAG/HFO-1234yf (Fig. 12b) under Set 3 conditions (445 N). The mainpeak was observed at 684.3 eV corresponded to the inorganicmetal fluoride species, most likely FeF3, due to the predominantpresence of Fe in the gray cast iron disk surface. No fluorinespecies were found in the case of scuffing experiments either inthe case of PAG/HFO-1234yf or in the case of PAG/R134a. Thiscould be understood based on the severity of the contact leadingto collapsing and expelling of any formed tribo-protective layer.

Among the starved lubricated wear experiments with HFO-1234yf Set 2 (222 N), HFO-1234 yf Set 3 (445 N) and R134a Set 2,fluorine species were found only in the case of HFO-1234yf Set 3.Note that the gray cast iron disk did not survive under the R134aSet 3 conditions and scuffed. The fluorine species in the case of

HFO-1234yf demonstrate the superiority of HFO-1234yf refriger-ant compared to R134a. The fact that fluorine-containing specieswere found only in the case of Set 3 (445 N) and not in the case ofSet 2 (222 N) is indicative that under higher load (and contactpressure) conditions, the tribo-protective layer that is formed inthe contact zone is more well-adhered. It appears that in the caseof Set 2, the tribo-layer which is adhered due to chemical affinitybetween the surfaces and mechanical pressure is loosely bondedto the gray cast iron surface.

3.5.2. Pin surface analysisThe pins surface was also characterized using XPS following the

scuffing and wear experiments in the presence of HFO-1234yf andR134a refrigerants. Fig. 13 presents the F1s XPS spectra for thescuffing Set 1 and wear Set 2 (222 N) experiments. As mentionedabove, no fluorine species on the gray cast iron disk surface werefound in the case of scuffing experiments either in the case ofHFO-1234yf or in the case of R134a. This was not the case for thepins, where in both the HFO-1234yf and R134a cases, fluorinespecies were found. Though, a major difference was noticed; in thecase of HFO-1234yf refrigerant, the F1s XPS spectrum was

Fig. 12. F1s core level spectra obtained over the gray cast iron disks following the PAG starved lubricated experiments (Set 3) in the presence of (a) R134a and (b) PAG/HFO-1234 yf.

Fig. 13. F1s core level spectra obtained over the gray cast iron pins following the PAG starved lubricated experiments in the presence of (a) HFO-1234 yf, Set 1; (b) R134a, Set 1;(c) R134a, Set 2; and (d) HFO-1234 yf, Set 2.

M.W. Akram et al. / Wear 304 (2013) 191–201 199

dominated by two peaks, (∼685 and ∼688.7 eV), while in the caseof R134a only one F1s peak was observed (689.2 eV).

The same trend as above, regarding the fluorine species wasnoticed in the case of pins after Set 2 (222 N) wear experiments. Inthe case of R134a only one peak (∼685 eV) was observed, whereasin the case of HFO-1234yf two distinct peaks (∼684.4 and∼688.4 eV) were identified. As mentioned before, on the disksurface, fluorine-containing species were found only in the caseof Set 3 (445 N) and not in the case of Set 2 (222 N) experiments.This happened since, as the pin analysis herein proves, the fluorinetribolayer in the particular cases has migrated onto the pin surface.

The peak at ∼685 eV corresponds to metallic fluorine-containingspecies, whereas the peak at ∼688 eV corresponds to organic

fluorine-containing species as has been reported in our previouswork [20]. The metallic component (∼685 eV) is mainly coming froma reaction between the refrigerant and the gray cast iron surface(tribo-chemical reaction), whereas the organic fluorine species(∼688 eV) are formed due to refrigerant fragmentation. From thisperspective it is important to notice that in the case where HFO-1234yf refrigerant is present at the interface, the formation ofmetallic F-containing species was facilitated. Ultimately, due to themechanical forces applied, these species were found on the pinsurface (Fig. 13a and d). This was true in both the scuffing and wearexperiments. The organic fluorine-containing species play the role ofthe third body ensuring the sustainability of the interface in terms offriction and wear; whereas the metallic component provides better

M.W. Akram et al. / Wear 304 (2013) 191–201200

adhesion due to its nature (refrigerant/lubricant fragment reactedwith elements/species coming from gray cast iron surface).

Another interesting point of the current findings is that thepeak corresponding to the organic fluorine-containing speciesappeared at different binding energies (eV). In particular, for thecases of HFO-1234yf (Set 1), R134a (Set 1) and HFO-1234yf (Set 3)the binding energies are 688.8 eV, 689.5 eV, and 688.3 eV, respec-tively. It is observed that the binding energies in the case of R134ais far away (689.5 eV) from the other two values (688.8 eV and688.3 eV for HFO-1234yf). This can be due to the differentchemical identity of the organic patch formed (e.g., differenthydrogen (H-) or oxygen (O-) content in the fragments) [33]. Thiscan be caused due to different interaction between PAG and R134aor HFO-1234yf under the surrounding mechanical forces.

3.5.3. Tribochemistry considerationsAs stated above, the organic fluorine-containing species (∼688 eV)

are coming from the unsaturated refrigerant (HFO-1234yf) fragmenta-tion and were not identified in the case of the saturated refrigerant(R134a). In our previous study [20], where aggressive unlubricatedconditions were applied, thus allowing for direct comparison of howthe chemical nature of refrigerants affects their ultimate performance,it was reported that the unsaturated refrigerant facilitates the forma-tion of a well-adhered tribologically protective layer. Beyond this,some chemical considerations should be taken into account in order toexplain the observed behavior. In principle, the C¼C double chemicalbonds are more reactive than the single bonds [38]. During thetribological performance evaluation, material removal off the surfaceof gray cast iron takes place, leading to the exposure of new “fresh”bonds on the surface (dangling bonds). These new bonds function asactive sites and react with the surroundings, which in this case is thePAG/R134a and PAG/HFO-1234yf systems. Since PAG lubricant is acommon component in both cases, themain difference is coming fromthe interaction of PAG with the saturated and unsaturated refrigerant(miscibility factor) and from the refrigerants itself (chemical naturefactor). It is anticipated that C¼C double bond in the HFO-1234yf, dueto their more delocalized and labile electronic cloud, lead the surfaceto a more energetically stable state (reduction of surface energy due tothe dangling bonds) through increased interaction with the surface.Ultimately, it is believed that the increased reactivity in the case ofunsaturated refrigerant likely leads to an increased coverage ofadsorbed refrigerant (most likely refrigerant fragments which wereseen only in this case, peak at 688 eV) which give rise to a morehomogeneous protective layer (confirmed by the morphological SEMfindings). In the case of the saturated R134a refrigerant the metalliccomponent of fluorine species (FeF3 observed at 685 eV) is originatedfrom the reaction between the Fe (gray cast iron as source) andfluorine (F) likely coming from the rapture of the highly polarizedCδ+–Fδ– bond (pool of F-species) [38].

4. Conclusions

An experimental investigation simulating realistic air-conditioningand refrigeration compressors was carried out to compare thetribological performance under PAG starved lubricated conditionsand two different refrigerants, HFO-1234yf and R134a. Three differentsets of tribological experiments were conducted: time-to-failure viastep loading known as scuffing experiments, and wear experimentsunder two different loading conditions. The findings, based onthe scuffing experiments, show that under HFO-1234yf atmosphere,the gray cast iron interface can sustain as much as twice the loadsbefore failure compared to R134a. The wear experiments also con-firmed the superiority of the HFO-1234yf refrigerant, with less wear.These findings are in accord with earlier studies under unlubricatedconditions.

SEM images revealed the initiation and formation of tribo-protective layers on the gray cast iron surface in the presence HFO-1234yf. During the formation of this tribo-layer, the friction coeffi-cient exhibited some instability due to formation and nucleation ofbump like particles. Once the tribo-layers were formed, the frictioncoefficient was stable and low. Also from SEM images, abrasive wearwas evident in R134a environment. The abrasive particles amalga-mated perpendicular to the sliding direction and eventually causedfailure (adhesive mode). This mode of wear was not seen in the caseof HFO-1234yf. This can be corroborated to the formation ofpreventive layers which impede abrasive wear. However, eventualfailure was occurred when the tribo-layers completely removed fromthe contact interface. That's why no fluorine-enriched substance isidentified after scuffing experiments. Note that, fluorine-based sub-stance was found for the non-scuffed samples which is in agreementour previous findings. In the presence of HFO-1234yf, fluorine basedelements were detected from EDS analysis which was furtherverified using XPS analysis. The mechano-chemical environment(applied load, refrigerant/lubricant mixture) seems to have an impacton the formation and migration of the F-tribo-protective layers fromthe disk surface to the pin.

Acknowledgments

This research was supported by the 25 member companies ofthe Air Conditioning and Refrigeration Center (ACRC), an Industry-University Cooperative Research Center at the University of Illinoisat Urbana-Champaign. Micromechanical and chemical analyseswere carried out at the Center for Microanalysis of Materials at theUniversity of Illinois, which is supported by the U.S. Department ofEnergy under Grant DEFG02-96-366ER45439.

References

[1] R.U. Ayres, L.W. Ayres, V. Pokrovsky, On the efficiency of US electricity usagesince 1900, Energy 30 (2005) 1092–1145.

[2] M.J. Molina, F.S. Rowland, Stratospheric sinks for chlorofluoromethanes:chlorine atom-catalyzed destruction of ozone, Nature 249 (1974) 810–814.

[3] R. Basu, D.P. Wilson, Thermophysical properties of 1,1,1,2- tetrafluoroethane(R134a), International Journal of Thermophysics 10 (1989) 591–603.

[4] S. Papasavva, O. Anderson, Green-MAC-LCCP: life-cycle climate performancemetric for mobile air conditioning technology choice, Environmental Progressand Sustainable Energy 30 (2011) 234–247.

[5] G. Morrison, D.K. Ward, Thermodynamic properties of two alternative refrig-erants: 1,1-dichloro-2,2,2- trifluoroethane (R123) and 1,1,1,2-tetrafluroethane(R134a), Fluid Phase Equilibria 62 (1991) 65–86.

[6] S. Safari, M. Hadfield, Wear behavior of the piston/gudgeon pin in a hermeticcompressor with replacement CFC refrigerants, Wear 219 (1998) 8–15.

[7] IEA, Energy technologies for reducing emissions of greenhouse gases, in:Proceedings Seminar Paris OECD/IEA (2nd ed.), Paris, 1990.

[8] J.M. Calm, Emissions and environmental impacts from air-conditioning andrefrigeration systems, International Journal of Refrigeration 25 (2002)293–305.

[9] S.B. Riffat, C.F. Afonso, A.C. Oliverira, D.A. Reay, Natural refrigerants forrefrigeration and air-conditioning systems, Applied Thermal Engineering 17(1997) 33–42.

[10] N.G. Demas, A.A. Polycarpou, T.F. Conry, Tribological studies on scuffing due tothe influence of carbon dioxide used as a refrigerant in compressors, TribologyTransactions 48 (2005) 336–342.

[11] E.E. Nunez, K. Polychronopoulou, A.A. Polycarpou, Lubricity effect of carbondioxide used as an environmentally friendly refrigerant in air-conditioningand refrigeration compressors, Wear 270 (2010) 46–56.

[12] E.E. Nunez, N.G. Demas, K. Polychronopoulou, A.A. Polycarpou, Comparativescuffing performance and chemical analysis of metallic surfaces forair-conditioning compressors in the presence of environmentally friendlyCO2 refrigerant, Wear 268 (2010) 668–676.

[13] N.G. Demas, A.A. Polycarpou, Ultra high pressure tribometer for testing CO2

refrigerant at chamber pressures up to 2000 psi to simulate compressorconditions, Tribolology Transactions 49 (2006) 291–296.

[14] M. Spatz, B. Minor, HFO-1234yf, A Low GWP Refrigeration for MAC, SAE WorldCongress, Detroit, Michigan, April 2008.

[15] M. Spatz, B. Minor, HFO-1234yf: low refrigeration update, in: Proceedings ofInternational Refrigeration and Air-conditioning Conference, Purdue, Indiana,July 2008.

M.W. Akram et al. / Wear 304 (2013) 191–201 201

[16] M. Koban, Automative Material Investigation with Low GWP Refrigerant HFO-1234yf, Vehicle Thermal Management System, 2010.

[17] J. Benouali, K. Iljima, HFO-1234yf refrigerant development, in: Proceedings ofthe International Symposium on New Refrigerants and Environmental Tech-nology, 2008, pp. 225–233.

[18] R. Akasaka, K. Tanaka, Y. Higashi, Thermodynamic property modeling for2,3,3,3-tetrafluoropropene (HFO-1234yf), International Journal of Refrigera-tion 33 (2010) 52–60.

[19] O.J. Nielsen, M.S. Javadi, Atmospheric chemistry of CF3CF¼CH2: Kinetics andmechanisms of gas-phase reactions with Cl atoms, OH radicals and O3,Chemical Physics Letters 439 (2007) 18–22.

[20] M.W. Akram, K. Polychronopoulou, A.A. Polycarpou, Lubricity of environmen-tal friendly HFO-1234yf refrigerant, Tribology International 57 (2013) 92–100.

[21] O.O. Ajayi, C. Lorenzo-Martin, R.A. Erck, G.R. Fenske, Scuffing mechanism ofnear-surface material during lubricated severe sliding contact, Wear 271(2011) 1750–1753.

[22] J.M. Han, R. Zhang, O.O. Ajayi, G.C. Barber, Q. Zou, L. Guessous, D. Schall,S. Alnabulsi, Scuffing behavior of gray iron and 1080 steel in reciprocating androtational sliding, Wear 271 (2011) 1854–1861.

[23] A. Suh, J.J. Patel, A.A. Polycarpou, T.F. Conry, Scuffing of cast iron and Al390-T6materials used in compressor applications, Wear 260 (2006) 735–744.

[24] A. Dyson, Scuffing—a review, Tribology International 8 (1975) 77–87.[25] K.C. Ludema, A review of scuffing and running-in of lubricated surfaces with

asperities and oxides in perspective, Wear 100 (1984) 315–331.[26] K. Mizuhara, M. Tomimoto, The effect of refrigerants in the mixed lubrication

regime, G.E. Totten et al. (Eds.), Proceedings of the 1995 Symposium inTribology of Hydraulic Pump Testing, ASTM STP 13101996) 38-48.

[27] E. Preisegger, R. Henrici, Refrigerant R134a: the first step into a new age ofrefrigerants, International Journal of Refrigeration 15 (1992) 326–331.

[28] K.N. Marsh, M. Kandil, Review of thermodynamic properties of refrigerants+lubricant oils, Fluid Phase Equilibria 199 (2002) 319–334.

[29] E.E. Nunez, N. Demas, K. Polychronopoulou, A.A. Polycarpou, Tribologicalstudy comparing PAG and POE lubricants used in air-conditioning compres-sors under the presence of CO2, Tribology Transactions 31 (2008) 790–797.

[30] E.E. Nunez, S.-M. Yeo, K. Polychronopoulou, A.A. Polycarpou, Tribological studyof high bearing blended polymer-based coatings for air-conditioning andrefrigeration compressors, Surface and Coatings Technology 205 (2011)2994–3005.

[31] D. Dascalescu, K. Polychronopoulou, A.A. Polycarpou, The significance oftribochemistry on the performance of PTFE-based coating CO2 refrigerantenvironment, Surface and Coatings Technology 204 (2009) 319–332.

[33] A.M. Ferraria, J.D. Lopes da Silva, A.M. Botelho do Rego, XPS studies of directlyfluorinated HDPE: problems and solutions, Polymer 44 (2003) 7241–7249.

[34] ⟨http://www.nationalref.com/PDF's/MSDS/Refrigerants/MSD134A.pdf⟩(accessed 13.03.13).

[35] ⟨http://www51.honeywell.com/sm/lgwp-uk/common/documents/msds-documents/FP_LGWP_UK_HFO-1234yf_uk_MSDS.pdf⟩ (accessed 13.03.13).

[36] M. Nosaka, M. Kikichi, H. Kikuyama, Effects of iron fluoride on durability ofcryogenic high-speed ball bearing rocket turbopumps, Tribology Transactions43 (2000) 163–174.

[37] G. Theiler, W. Hubner, T. Gradt, P. Klein, K. Friedrich, Friction and wear of PTFEcomposites at cryogenic temperatures, Tribology Transactions 35 (2002) 449–458.

[38] J.E. McMurry, Organic Chemistry, 8th ed., Thomson, Australia, ISBN 978-0-8400-5444-9, 2012.