Impact–sliding fretting tribocorrosion behavior of 316L stainless steel in solution with different halide concentrations

: Impact–sliding caused by random vibrations between tubes and supports can affect the operation of heat exchangers. In addition, a corrosive environment can cause damage, accelerating the synergism of corrosion and wear. Therefore, the focus of this work was the impact–sliding fretting tribocorrosion behavior of 316L heat exchanger tubes at different halide concentrations. A device system incorporating the in situ electrochemical measurements of impact–sliding fretting corrosion wear was constructed, and experiments on 316L heat exchanger tubes in sodium chloride (NaCl) solution with different concentrations (0.0, 0.1, 0.5, 1.0, 3.5, and 5.0 wt%) were carried out. The synergism between wear and corrosion was also calculated and analyzed. The wear and damage mechanisms were elucidated by correlating the corrosion–wear synergism, morphologies, and material loss rates. The results indicated that the stable wear stage occurred at approximately 9–12 h, after which the corrosion current increased with the expansion of the wear area. As the halide concentration increased, the scale of damage on the wear scars gradually decreased, changing from being dominated by cracks, delaminations, and grooves to being dominated by scratches, microgrooves, and holes. There was an obvious positive synergism between wear and corrosion. The material loss was dominated by pure mechanical wear and wear enhanced by corrosion, but corrosion enhanced by wear contributed more than tangential sliding fretting corrosion. The total mass loss increased gradually in the range of 0.0–0.5 wt% and decreased in the range of 0.5–5.0 wt%. Large-scale damage enhanced by corrosivity and small-scale damage reduced by lubricity dominated the material loss at low and high concentrations, respectively.


Introduction
Turbulent excitation causes heat exchanger tubes to randomly vibrate, resulting in fretting wear in the form of impact-sliding between the tubes and supports [1]. The material loss of heat exchanger tubes operating in corrosive environments is accelerated by the synergism of corrosion and wear [2]. The occurrence of fretting corrosion wear makes it more difficult to predict the life of heat exchanger tubes, which is related to the safe operation of heat exchangers [3,4]. For example, the fretting wear between the U-bend area of heat exchanger tubes and the antivibration bars (AVBs) in a pressurized water reactor (PWR) steam generator (SG) may be accelerated by corrosion [5]. Due to boiling enrichment in occluded areas, enrichment factors as high as 10 6 are observed for nonvolatile impurities with initial ppb-level concentrations, and local highly corrosive environments are formed [5,6]. Therefore, studying the fretting corrosion and wear www.Springer.com/journal/40544 | Friction of heat exchanger tubes is of great significance for equipment safety.
Many studies have been carried out on the fretting wear of heat exchange tubes, mainly focusing on the effects of environmental factors, mechanical parameters, and material properties [7,8] and the analysis of the evolution processes of subsurface structures and material damage mechanisms. Most studies are based on tangential sliding fretting wear, ignoring impact fretting wear, and impact-sliding fretting wear. However, the motion mode of heat exchanger tubes under random vibrations in actual conditions is impact-sliding rather than tangential sliding, which also contributes to the inability to accurately predict wear using the Archard model [9]. Cai et al. [10][11][12] and Guo et al. [9,13,14] conducted studies on impact and impact-sliding fretting wear, discussing the effects of excitation force amplitude, excitation force ratio, support form, and precompressive stress. The results showed that impact-sliding fretting wear exhibits greater wear damage than single-impact fretting wear and single-tangential fretting wear. Guo et al. [9] also developed an energy-based model for predicting impact-sliding wear that is more accurate than the Archard model under impact-sliding conditions.
Fretting corrosion is one of the fretting modes in corrosive media and is commonly observed in various industries. Fretting corrosion is the result of mechanical, chemical, electrochemical, and other factors acting together, and there is an interaction between wear and corrosion, also known as synergism [15]. The material loss caused by wear and corrosion is often greater than the sum of their individual material losses. However, in some cases, the corrosion products can form a self-lubricating layer to protect the surface, referred to as negative synergism [16]. Therefore, the study of the synergy between corrosion and wear is the key to revealing the mechanism of material damage during fretting corrosion [17]. Many scholars have evaluated the synergism between wear and corrosion and elucidated the damage mechanisms of materials. The effects of various factors (e.g., material type, normal force, displacement amplitude, applied potential, and temperature) on fretting corrosion (tribocorrosion) behavior were investigated in Refs. [18][19][20][21][22][23]. Xu et al. [15,21] found that the corrosion rate of Ti-16Mo alloy increased approximately 100 times under tribocorrosion, and the wear caused by corrosion was approximately 40% higher than that of pure mechanical wear. Chen et al. [22] found that the total volume loss of Monel K500 alloy during tribocorrosion increased with the increasing applied potential. Li et al. [23] found that the wear mechanism depends heavily on the polarization conditions. In addition, Refs. [24][25][26] on fretting corrosion in high-temperature environments have been carried out, and Refs. [24][25][26] are mainly focused on heat exchanger tubes. Studies on the commonly used 316L alloy have shown that factors such as surface treatment [27], applied potential [28][29][30], type of solution [29], and number of cycles [30] significantly affect the tribocorrosion behavior of this alloy. The passivation behavior complicates the tribocorrosion behavior of the passive metal due to the interaction between the deterioration of the base metal and oxidation of the metal [31]. The removegrowth cycles of passive films can significantly accelerate corrosion. On the other hand, the presence of passive films affects plastic deformation and wear behavior [30]. These existing studies are of great significance to understanding fretting corrosion damage mechanisms and corrosion-wear interaction mechanisms. However, the motion mode in most existing studies is tangential sliding. Thus, it is urgent to study fretting corrosion under the impact-sliding motion mode and explore the synergy between corrosion and wear according to the actual working conditions of heat exchanger tubes.
It is not enough to focus on a single solution or particular solution concentration in fretting corrosion studies. Moreover, it is necessary to investigate the effect of halide concentration in addition to the effects of temperature and dissolved oxygen. The tribocorrosion behaviors of Inconel 625 alloy, 304 stainless steel, and nickel-aluminum bronze in artificial seawater with different halide ion concentrations were investigated by Chen et al. [32], Zhang et al. [33], and Zhang et al. [34], respectively. Their results showed a clear positive synergism between wear and corrosion. The good lubricating ability of the high-concentration solution resulted in less material losses. Han et al. [35] investigated the effect of sodium chloride (NaCl) concentration on the tribocorrosion behavior of 2312 Friction 11(12): 2310-2328 (2023) | https://mc03.manuscriptcentral.com/friction SAF 2507 super duplex stainless steel (SDSS) and found that its corrosion-wear state could be divided into three parts according to the NaCl concentration: antagonistic (0.5%-2%), synergistic (3.5%-6.5%) and additive-synergistic (8%). Obviously, the difference in halide concentration made the tribocorrosion behavior more complex. Considering the fretting damage characteristics of heat exchanger tubes, it is necessary to carry out experimental studies with longer durations and more realistic movement modes.
In summary, more experiments are necessary to successfully predict the life of heat exchanger tubes with the risk of fretting corrosion wear. To date, the fretting corrosion-wear behavior of materials, even at room temperature and ordinary pressure, has been insufficiently studied. The synergism of corrosion and wear and the damage mechanism of the fretting corrosion wear of heat exchanger tubes in the impactsliding mode are worth studying. In this work, we focus on the impact-sliding fretting corrosion and wear behavior of heat exchanger tubes and investigate the effects of different halide concentrations. For this purpose, a set of impact-sliding fretting corrosion equipment was obtained, and fretting corrosion experiments with 316L heat exchanger tubes in NaCl solution with different mass percentage concentrations (wt%) were carried out. The synergism between corrosion and wear was quantitatively analyzed, and the wear mechanism was investigated by the optical microscopy (OM), scanning electron microscopy (SEM), and energy dispersion X-ray spectroscopy (EDS) measurements.

Experimental equipment
An impact-sliding fretting corrosion and wear device system incorporating the in situ electrochemical measurements was set up based on the self-designed equipment described in Guo et al. [9,13]. The schematic diagram is shown in Fig. 1(a), and there are three key points for the device system. First, impact-sliding fretting wear was realized. The top of the cantilever beam was excited by two exciters so that the tube specimen fixed to the cantilever beam collides with the plate specimen to cause wear. The two exciters are placed perpendicular to each other to provide turbulent forces in the lift and drag directions, as shown in Fig. 1(b). The excitation signals were controlled by an arbitrary waveform generator (SDG2000X, Siglent), and Second, the vibration and friction data were recorded. Figures 1(a) and 1(c) show the sensor placement. The plate specimen was mounted on the holder, which is fixed by four force sensors to record the contact forces. Two displacement sensors were mounted in two mutually perpendicular directions to measure the displacement of the tube in the normal and tangential directions and to synthesize the motion track. The accuracy of the force sensors is 0.05%, the accuracy of the displacement sensors is 1 μm, and the data collection device (DongHua Testing Technology Co., Ltd.) connected to them has a maximum sampling frequency of 10 kHz.
Finally, the in situ electrochemical corrosion testing was carried out. Figure 1(a) shows an electrochemical workstation (CHI660E, CH Instruments) with a three-electrode system. The RE is Ag/AgCl, the CE is platinum gauze, and the WE is a tube specimen. To achieve a liquid environment, a solution tank was set up, which is divided into two parts, upper and lower, using bolted connections. The upper and lower solution tanks were pressed against the upper and lower ends of the plate specimen holder, respectively, and sealed using O-rings. The lower tank was connected to the cantilever beam using a thin-walled hose to achieve a dynamic seal and ensure the accuracy of the contact force measurement. In addition, the components were made of insulating materials [36].

Materials
The tube specimens are shown in Figs. 2(a) and 2(b).
The material is 316L stainless steel, which is a common material for equipment processing and manufacturing. 316L has also been widely used in the manufacturing of heat exchanger tubes and is an early material for SG heat exchanger tubes. The plate specimen material is zirconia (ZrO 2 ), which is insulated and has a high hardness; this material was used to study the fretting corrosion behavior of the tube. The length and diameter of the tube specimen are 32 and 18 mm, respectively; the thickness of the plate specimen in the vertical direction is 10 mm. The roughness of the contact surface was controlled below 0.2 Ra.
The tube specimens were encapsulated by a selfdeveloped device, and the encapsulated tube specimen is shown in Fig. 2(c). First, the tube was filled with epoxy resin inside and at both ends, and the through-hole was reserved for installation. Then, the wire was soldered. The solder joint was covered with light-curing resin, and the wire was fixed. Finally, the tube was covered with nail polish (main components: ethyl acetate, butyl acetate, and acrylate copolymer), and a 150 mm 2 surface was exposed as the working surface. The working surface was aligned with the plate specimen before experiments to ensure that wear occurs on it.

Experimental settings
The excitation force signals were obtained based on the boundary-normalized power spectral density (PSD) of the turbulence-induced random excitation measured by Oeng and Ziada [37]. The normalized PSDs of lift (Φ L ) and drag (Φ D ) are shown in Eqs. (1) and (2), respectively. These PSD curves based on the  reduced frequency ( R f ) can be transformed by an algorithm into mutually independent time-series force signals of lift (normal) and drag (tangential) [9]. 0. 53 Specifically, the flow velocity and density were first determined from the flow field data [4], and then the normalized Φ L and Φ D described by Eqs. (1) and (2), respectively, are transformed into lift and drag the PSDs by Eq. (3) and finally into force-time records using the harmonic superposition method [4,37].
where F S is the PSD of the fluid force, Φ is the normalized PSD,  is the fluid density, V is the gap velocity, l is the tube length, and d is the tube diameter.
The transformed results are shown in Figs. 3(a) and 3(b), and the root-mean-square (RMS) values of lift and drag are 1.6 and 0.66 N, respectively. The actual PSDs of the force-time records are in good agreement with the original PSDs, as shown in Fig. 3(c). The parameter settings for the experiments are shown in Table 1. The smaller initial clearance implies a higher contact rate, which facilitates the stability of the electrochemical signal. A longer duration is beneficial for revealing the material damage mechanism. The experiments were performed at room temperature to focus on the effects of halide concentration and impact-sliding. The mass percentage concentrations of the NaCl solution were set with both a large overall range and good local refinement. Among these concentrations, 3.5 wt% is close to the NaCl concentration of seawater, and 0.0 wt% corresponds to deionized water.

Experimental process
Before the experiments, the mounted specimens were kept under tangential fretting conditions for more  This can preclude the effect of crevice corrosion during long-term static storage and can also mitigate the changes in surface electrochemical properties caused by fretting-driven solution-stirring rather than wear [38]. To simultaneously observe vibration and friction behaviors and obtain thermodynamic and kinetic information of corrosion in situ, the following three parts of the main process of the experiment are carried out simultaneously, as shown in Fig. 4: (i) The impact-sliding fretting wear started 600 s after the beginning of the experiment and stopped after 24 h, and the OCP continued to be recorded for 1,200 s.
(ii) During this period, displacement and contact force were measured every 1 h for 80 s to observe the change in friction behavior with time. Intermittent high-sampling-frequency measurements reduced the total data volume while ensuring accuracy and representativity. (iii) A potentiodynamic polarization test was performed every 4 h to obtain Tafel curves, and the OCP was monitored for the remainder of the time. The longer scan interval time and smaller scan time ratio minimized the influence on the mass loss results. The scanning range is −0.8 to 0.3 V (vs. Ag/AgCl), which can preclude pitting, and the scanning rate is 1 mV/s. It should be clarified that the time periods of the potentiodynamic polarization test and the displacement force measurement are staggered to prevent the effect of polarization on the coefficient of friction (COF).

Posttest procedure
The COF for each collision and the mean values were calculated with a self-written program. It is necessary to subtract the baseline of the contact force and select the data for each collision before executing the calculation. An optical three-dimensional (3D) profiler (ST400, Nanovea) was used to measure the wear volume and calculate the mass loss. In addition, the Tafel curve data were processed using the CHI660E companion software (CHI Version 17.06, CH Instruments) to calculate the corrosion current and corrosion rate [22]. The synergy between corrosion and wear was quantitatively analyzed with reference to American Society for Testing Material (ASTM) G119-09, as described in detail in Section 4.1. The morphologies and elemental distributions of the wear surface were analyzed using the OM (VHX-900, KEYENCE), SEM (Apreo S LoVac, FEI), and EDS (EDAX Octane Elect Super, AMETEK).

Vibration and wear
The motion trajectory is measured to confirm the rationality of the excitation method. The results indicate that the motion is disordered, showing the characteristics of random vibration. Normal and tangential displacements are located mostly within the 200 μm range, which is typical of fretting. The COF can reflect the characteristics of the friction and the evolution of the wear process on the 316L/ZrO 2 surface. As shown in Fig. 6, the COF exhibits the same trendline change for different concentrations.
The COF values are low at the beginning of wear, all located between 0.37 and 0.38, and then rise rapidly into the fluctuation stage and finally stabilize gradually. This indicates that the initial state of the material is the same, and there is also a break-in period for the impact-sliding fretting wear. In addition, the higher   the concentration is, the lower the COF. The reason for this is that solution with a high concentration has high corrosivity [22] and high lubricity [33,34], which may be related to the dissolution of the metal material, the stripping of corrosion products, and the lubrication of the solution.
The total mass loss reflects the wear performance of the 316L tube at different halide concentrations, as shown in Fig. 7. As the concentration increases, the mass loss of 316L tubes increases, and then decreases, peaking at 0.5 wt%. This same trend is also observed for both the maximum wear depth and wear area. This is not entirely consistent with the results of Refs. [33,34], which found that the lubricity of highly concentrated solution reduces material loss. This mechanism could explain the mass loss trend from 0.5 to 5.0 wt%, as shown in Fig. 7, but is not fully applicable to this work. This is because of the differences in the motion modes and normal loads. The impact-sliding fretting between the tube and plate caused by random vibration is obviously different from rotational or reciprocal sliding and has a lower load level. This affects the contributions of mechanical wear and electrochemical corrosion to material loss and also changes the mechanism of material damage. Therefore, the effect of wear on corrosion is further studied, and the synergism of corrosion and wear is analyzed with the help of electrochemical techniques.

Electrochemical corrosion
The in situ information on the surface states of the tubes under fretting corrosion is obtained by the OCP measurements, as shown in Fig. 8. Figure 8(a) shows the results of splicing directly without treatment.
The OCP values are stable when only tangentially fretting before wear, while they undulate after wear. Magnification of the data near 20,000 s shows that the  undulations are superimposed by high-frequency and low-frequency components. The high-frequency undulations reflect the damage to the surface with each impact-sliding. The low-frequency undulations are due to the uneven distribution of the excitation force level over time, and higher excitation and contact force levels in a short period of time cause the OCP to decrease, indicating that the increased load causes a higher corrosion tendency [38]. Figure 8(b) shows the results of fitting the relatively stable part of each time period to remove the effect caused by potentiodynamic polarization. After the beginning of wear, the OCP decreases rapidly, stabilizes, and gradually recovers after the end. This is attributed to the destruction and regrowth of the passive film. Notably, the higher the concentration is, the lower the OCP before and after wear, indicating an increased corrosion sensitivity of the material [34]. According to the galvanic coupling model developed by Vieira et al. [39], corrosion enhancement at the OCP is related to the galvanic coupling corrosion between worn/unworn surfaces, which promotes the cathodic reduction of oxygen and accelerates the anodic dissolution of alloying elements [33]. In addition, the magnitude of the OCP reduction is not as large as that in Refs. [33,34,40] due to differences in the material type and load level.
Corrosion kinetic information is obtained from the Tafel curves. Figures 9(a)-9(d) show the Tafel curves for no wear and the 1st, 9th, and 21st hour of the wear period, respectively. The difference between the corrosion current and the self-corrosion potential before and after wear is not obvious, and there are passivation intervals in the anodic region. However, the reference line helps to demonstrate that the anodic polarization current increases significantly after wear, especially in the 1st hour, indicating that the wear breaks or even removes the passive film, and thus accelerates corrosion. According to the wear-accelerated corrosion model developed by Mischler et al. [41], wear damage to the passivation layer exposes more corrosion-sensitive substrate to solution, leading to metal oxidation and dissolution. Repeated depassivation/repassivation cycles can cause wearaccelerated corrosion. References [42,43] have shown that the donor and acceptor densities in 316L passive films increase with the increasing Cl − concentration. Defects at the metal/film interface and within the film reduce the adhesion of the passive film to the metal substrate and the stability of the passive film [44]. Therefore, with the increasing Cl − concentration, the passive film more easily breaks and peels off, leading to wear-enhanced corrosion. In addition, plastic deformation within the wear scars under tribocorrosion can lead to many defective spots, such as cracks and dislocations, which can act as corrosion sites and accelerate corrosion [15]. The difference between the 9th and 21st hours is small, and the wear may have entered a stable stage. In addition, Figs. 9(b)-9(d) show that the higher the concentration is, the lower the self-corrosion potential (to the left) and the higher the anodic polarization current (to the up). Further analysis is performed by calculating the corrosion current. The calculated corrosion currents without wear and during wear are shown in Fig. 10. The results indicate that the higher the concentration is, the higher the value of the corrosion current, indicating a higher corrosion rate. The corrosion kinetics reflected by the Tafel curve are consistent with the corrosion thermodynamics reflected by the OCP. In addition, the corrosion current increases in the 1st hour after the start of wear, decreases rapidly afterward, and then gradually increases again. This is because the change in corrosion current is dominated by the contact stress and area of the wear scar with the development of wear. The high stress at the contact location at the beginning of wear results in a high corrosion current. Subsequently, the expansion of the wear scar reduces the contact stress, resulting in a lower corrosion current. Finally, the corrosion current starts to rise gradually and slowly due to the increasing area of the wear scar. A comparison of the corrosion current values indicates that the change before and after wear is small compared to the results of Refs. [22,33,34]. This is because the effect of wear on the polarization curve depends on the ratio of the worn area to the unworn area [45]. The lower average load of the impact-sliding motion mode in this work results in a smaller area ratio. In addition, the contact collisions of the tube and plate specimens are intermittent, and the wear damage is also intermittent. Therefore, the increase in corrosion current caused by wear is smaller. The impact-sliding mode is more consistent with the actual damage pattern of the heat exchanger tubes, which accumulates slowly rather than rapidly, leading to failure.

Morphologies and elemental distributions of wear scars
The morphological characteristics of the wear scar surfaces are observed by the OM and SEM to analyze the damage mechanisms. The OM images of the wear scars at different concentrations are shown in Fig. 11. Figures 11(a)-11(f) demonstrate that the boundaries of the wear scars are triangular. This is because there is an initial clearance between the tube and the plate, and the cantilever beam has a certain deflection after being excited. Thus, wear initially occurs due to the point contact between the tube and the edge of the plate, after which the contact mode develops from point contact to line contact and surface contact, and eventually wedge-shaped wear scars form. This is also the reason for the trend of the corrosion current with time, as shown in Fig. 10. The depth of the triangular wear scar is not uniform. The clearance is strictly controlled before the experiment with the help of a displacement sensor, so the trends of the maximum wear depth, wear area, and mass loss (Fig. 7) are the same, but the depth is also important for wear prediction. In the case of a low area and high depth, the ratio of wear depth to wear volume may be high, and the risk of tube damage may increase. In addition, there are many wear debris and corrosion products accumulating around the wear scars, which requires further composition analysis. As seen in Figs. 11(g)-11(l), there are many scratches on the wear scars. The higher the concentration is, the fewer and shallower the scratches are, and the smoother the surfaces of the wear marks. The difference in wear surfaces at different concentrations is further analyzed by the SEM, as shown in Fig. 12. Overall, the wear scars are relatively smooth, with no serious surface deterioration and no obvious accumulation of wear debris. Clearly, the wear damage in the room-temperature liquid environment is relatively slight, and the wear debris is more easily discharged. The surfaces of the wear scars at all concentrations have grooves, which are long black-gray areas, indicating that abrasive wear is the main wear mechanism. Notably, the interiors of the black and gray regions are still covered with microscopic scratches and grooves. Therefore, we refer to the former as macroscopic large-scale grooves and the latter as microscopic small-scale grooves based on the significant size difference. This phenomenon is further observed and analyzed, as shown in Fig. 13. A comparison of the images at different concentrations shows that the higher the concentration is, the fewer, narrower, and shallower the macroscopic large-scale grooves, which is consistent with the OM results, as shown in Figs. 11(g)-11(l). In addition, there are some irregularly shaped black areas ( Fig. 12(a)), which are caused by the re-extrusion of undischarged wear debris onto the surface. This indicates that an     The higher-magnification SEM images are obtained to observe the details of the minor damage, as shown in Fig. 13. At all concentrations, there are many microscopic small-scale grooves, scratches, and holes on the surfaces of the wear scars. This is quite different from the wear-scar morphology of tangential sliding wear, attributed to the randomness of the motion and the minor damage caused by the impact. It is noteworthy that there are some delamination and spalling pits (Fig. 13(a)) and some cracks and spalling pits (Figs. 13(b) and 13(c)). With a further increase in concentration, this damage is almost absent, as shown in Figs. 13(d)-13(f). This indicates that the lubricating and corrosive effects of the solution change the wear and damage mechanisms. In addition, the sizes of the microscopic small-scale grooves, scratches, and holes (Figs. 13(d)(d)-13(f)) decrease sequentially, and in combination with the characteristics of the morphology (Figs. 13(a)-13(c)), it can be found that the higher the concentration is, the smaller the scale of the damage.
To observe the products around the wear scars, the SEM images of the edges of the wear scars are obtained, as shown in Fig. 14. Large amounts of wear debris and corrosion products accumulate on the outsides of the wear scars, while there are almost none on the insides of the wear scars. By combining these results with those shown in Fig. 11, it can be inferred that most of the wear debris and corrosion products are discharged and attach around the wear scars to be further oxidized. To verify this, elemental analysis is performed. The results show that the oxygen content at the edges of the wear scar is obviously higher than those inside the wear scar and on the unworn surfaces, as shown in Figs. 15(a) and 15(b), confirming the accumulation of wear debris and corrosion products around the wear scar. The same elemental distributions inside the wear scar and on the unworn surfaces indicate that there is no formation of a wear debris layer (WDL) on the wear surface under the combined effect of the liquid environment and impact-sliding motion. In addition, a comparison of the elemental distributions at 0.0 and 5.0 wt% reveals that the oxygen content is higher at 5.0 wt%, with little difference among the other elements, which is due to the greater mass loss causing more accumulation. Figure 15(c) shows the results for the insides of the wear scar at 5.0 wt%, demonstrating that the oxygen content in the black groove is slightly higher, indicating that a small amount of wear debris has entered the groove and participated in the subsequent wear.

Synergism between wear and corrosion
Referring to ASTM G119-O9 (Standard guide for determining synergism between wear and corrosion) [46], the following parameters are calculated with the help of Eqs. (4)-(6). | https://mc03.manuscriptcentral.com/friction where T is the total material loss after a corrosionwear test; S is the total synergism; C and W are the material losses caused by electrochemical corrosion and mechanical wear, respectively; W C , 0 C , and W C  are the total corrosion, pure corrosion, and wearenhanced corrosion, respectively; C W , 0 W , and C W  are the total wear, pure wear, and corrosion-enhanced wear, respectively; and W C and 0 C are calculated from the corrosion currents ( corr I ) measured with and without mechanical wear, respectively, by Faraday's law [22], as shown in Eq. (7). In this paper, 0 W is obtained from the material loss of the wear test in pure water. Because the oxidation rate is low in this case, it can be ignored [40].
where F (= 96,500 C/mol) is the Faraday's constant, M is the atomic mass, and Z is the number of electrons. M/Z calculated based on the atomic percentage composition of 316L is 25.504. Regarding the units of the above variables, and considering the curved profile of the tube specimen and the initial installation clearance, the depth of the wedge-shaped wear scar is not uniformly distributed over the entire surface; thus, weight/time rather than depth/time is used as a uniform unit. Therefore, the variables in Eqs. (4)- (6) and C in Eq. (7) have units of "g/h". In addition, we need to correct the corrosion rates of the unworn and worn areas. The measured corrosion current represents the corrosion rate over an area of 150 mm 2 . In fact, the ratio between the unworn and worn final areas is 50-100, and the difference is even larger during wear. This leads to a small change in the corrosion current and to a higher corrosion rate derived from the Tafel curve than the total material loss rate in some cases, which is clearly not reasonable. Therefore, we first assume that the corrosion current density on the unworn surface ( unworn i ) is constant. Then, the areas of the worn and unworn surfaces ( worn S and unworn S ) at any moment are calculated based on the assumption that the total mass loss rate is approximately constant. After measuring the total corrosion current ( total I ), the corrosion current density on the worn surface ( worn i ) can be calculated by Eq. (8). Eventually, the corrosion rate on the worn surface at any moment can be calculated by Eq. (7). This treatment may introduce a bias but reduces the effect of having a small wear area. total worn worn unworn unworn Three factors are also defined in the standard to describe the synergistic effect of corrosion and wear. The total synergistic factor is T/(T − S), the wear  Figure 16(a) shows the calculated results of the factors. All factors are higher than 1, indicating a significant positive synergism between corrosion and wear. The corrosion enhancement factor exhibits little difference between 100 and 200 at all concentrations, except for the higher value at 0.1 wt%. The wear enhancement factors and the total synergistic factors are in the range of 1.5-5, demonstrating a high degree of accordance with the concentration trend, indicating that wear enhancement dominates the synergism of corrosion and wear. Figure 16(b) shows the change in corrosion enhancement with concentration. 0 C does not increase monotonically with the increasing concentration, which is due to having the same exposed area but different wear areas. The area becomes an influencing factor because "g/h" is used as a unit. The smaller wear area at 5.0 wt% leads to a lower W C on the wear surface. For this reason, the small graph (the inset of Fig. 16(b)) is plotted, showing the corrosion current density at different concentrations. The results show that the higher the concentration is, the higher the corrosion current density, which indicates that the corrosion rate per unit area monotonically increases with the concentration. In addition, the W C generally increases with the increasing concentration but has an exceptional value at 0.5 wt%. This is attributed to the high wear surface area at this concentration, as very clearly shown in Figs. 7 and Fig. 11(c). Figures 16(c) and 16(d) show the C W and S, respectively. Notably, there is almost no difference between 0 W and 0 0 W C  (represented by the red parts of bars), indicating that the pure corrosion rate is clearly lower than the pure wear rate. There is a discrepancy between (represented by the blue parts of bars in Figs. 6(c) and 16(d)), indicating that corrosion enhancement due to wear accounts for a higher percentage of the total mass loss compared to the results of Refs. [15,33,34,40]. This is attributed to the lower load level in this work. The lower total material loss rate and the longer experimental duration allow the enhanced corrosion to play a more important role. Nevertheless, the trends of C W  and T with concentration are in perfect agreement, and it is clear that dominates the material loss. The frequency of serious collisions caused by random vibrations is lower than the total contact frequency because some of the collisions are   [22]. Further analysis is needed, as well as surface morphological characteristic studies, since these characteristics are related to the wear and damage mechanisms of the material.

Damage mechanism
It is clear that the damage mechanism in this work is dominated by abrasive wear and corrosive wear. This can be supported by the following phenomenon. The OM images (Fig. 11) have obvious scratches parallel to the sliding direction. The SEM images (Fig. 12) have obvious macroscopic large-scale grooves. There are many microscopic small-scale scratches, grooves, and holes (Fig. 13). In addition, Fig. 16 shows that corrosion significantly increases wear and that corrosion exacerbated by wear is not negligible. Interestingly, with the increasing concentration, the T and S all have the same trend as C W : first increasing, and then decreasing, reaching a maximum at 0.5 wt%. We have learned that both the corrosivity and lubricity increase as the concentration increases, as shown in Figs. 8, 10, and 16(b) and Fig. 6, 11, and 12, respectively. Higher corrosion promotes microcrack initiation and propagation [40], which increases the material loss rate. Higher lubricity reduces deformation and dislocation [33], which decreases the material loss rate. Clearly, these two mechanisms dominate the material loss in this work, resulting in material loss that increases, and then decreases with concentration. Therefore, it is speculated that there is a competition between the solution corrosivity and lubricity, corresponding to different damage mechanisms. From the high-magnification SEM images, as shown in Fig. 13, we can observe the details of the damage at different concentrations. Delamination and spalling pits are present in Figs. 13(a) (0.0 wt%), 13(b) (0.1 wt%), and 13(c) (0.5 wt%), and spalling pits and cracks are present. These results imply large pieces of material spalling and serious material damage. However, in Figs. 13(d) (1.0 wt%), 13(e) (3.5 wt%), and 13(f) (5.0 wt%), there are almost no cracks and large spalling pits and only a large number of microscopic small-scale grooves, scratches, and holes. It is coincidental that the damage details are also differentiated with respect to 0.5 wt%.
In summary, the damage mechanism of impactsliding fretting corrosion and wear is schematically shown in Fig. 17. At lower halide concentrations (0.0-0.5 wt%), the solution lubricity is not sufficient to inhibit serious material damage. As the halide concentration increases, corrosion promotes crack initiation and propagation and accelerates the spalling of large pieces of material, although collisions and contacts are lubricated. At higher halide concentrations (1.0-5.0 wt%), the solution lubricity is high enough that the damage to the material is dominated by microscopic scratches, grooves, and spalling. The higher the concentration is, the better the lubricity and the lighter the deformation and dislocation. The resulting microdamage formed is smaller in size, which reduces the material loss rate. In addition, some experimental settings may be the cause of this phenomenon. For example, impact-sliding fretting corrosion can be caused by random vibrations, lower load levels, longer experimental durations, etc. The turbulent excitation of heat exchanger tubes is broadband, and its load level is relatively low. The resulting collision damage accumulates with the operating time, and corrosion plays a more important role. The fretting mode of impact-sliding results in the coexistence of different scales of damage. Thus, the material losses due to fretting corrosion wear at different halide concentrations are dominated by corrosion and lubricity. This suggests that the local damage acceleration due to the difference in halide concentration should be considered in the wear prediction of heat exchanger tubes.

Conclusions
In this paper, an impact-sliding fretting corrosionwear device system incorporating the in situ electrochemical measurements is built to perform fretting corrosion-wear experiments on 316L heat exchanger tubes. The purpose is to investigate the effect of the halide concentration and to analyze the synergism of corrosion/wear and the damage mechanism. The electrochemical analysis shows that the corrosion current increases rapidly at the beginning of wear, and then decreases gradually. The COF results indicate that the wear process enters a stable stage at approximately 9-12 h, after which the corrosion current increases continuously with the expansion of the wear scars. There is a clear positive synergism in the impact-sliding fretting corrosion-wear behavior. 0 W and C W  dominate the material loss, but W C  is not negligible in impact-sliding mode. As the halide concentration increases, the total mass loss increases gradually in the range of 0.0-0.5 wt% and decreases in the range of 0.5-5.0 wt%. This is attributed to the competition between the corrosivity and lubricity of the NaCl solution, which influences the damage mechanism. Large-scale damage enhanced by corrosivity and small-scale damage reduced by lubricity dominate the material loss at low and high concentrations, respectively. Therefore, the local damage acceleration due to the difference in halide concentration should be considered in the wear prediction of heat exchanger tubes.