Non-invasive early detection of failure modes in total hip replacements (THR) via acoustic emission (AE)

Abstract

Total hip replacements (THR) are becoming an common orthopedic surgucal procedure in the United States (332 K/year in 2017) to relieve pain and improve the mobility of those that are affected by osteoarthritis, ankylosing spondylitis, or injury. However, complications like tribocorrosion, or material degradation due to friction and corrosion, may result in THR failure. Unfortunately, few strategies to non-invasively diagnose early-stage complications are reported in literature, leading to implant complications being detected after irreversible damage. Therefore, the main objective of this study proposes the utilization of acoustic emission (AE) to continuously monitor implant materials, CoCrMo and Ti6Al4V, and identify degradations formed during cycles of sleeping, standing, and walking by correlating them to potential and friction coefficient behavior. AE activity detected from the study correlates with the friction coefficient and open-circuit potential observed during recreated in-vitro standing, walking, and sleeping cycles. It was found that the absolute energy level obtained from AE increased as the friction coefficient increased, potential decreased, and wear volume loss increased. Through the results, higher friction coefficient and AE activity were observed in Ti6Al4V alloys while there was also a significant drop in potential, indicating increased tribocorrosion activity. Therefore, AE can be utilized to predict material degradations as a non-invasive method based on the severity of abnormality of the absolute energy and hits emitted. The correlation between potential, friction coefficient, and AE activity was further confirmed through profilometry which showed more material degradation in Ti6Al4V than CoCrMo. Through these evaluations, it was demonstrated that AE could be utilized to identify the deformations and failure modes of implant materials caused by tribocorrosion.

1. Introduction

Total hip replacements (THR) are becoming increasingly common in the United States to relieve pain and improve the mobility of those who are affected by osteoarthritis, ankylosing spondylitis, or injury¹. Currently, approximately 332K THR surgeries are taking place in the U.S. alone and are expected to increase to 500K in 2030 (Kurtz et al., 2007; Raphel et al., 2016). With the annual increase in the number of total hip replacement (THR) procedures performed, a method to non-invasively and continuously monitor the progression of the implant is necessary to detect and address early-stage complications before higher-risk revision surgery is needed.

These complications may arise in the human body where the hip implant is surrounded by periprosthetic tissue and pseudo synovial fluid, which acts as lubrication and reduces friction coefficient between the ball and socket during sliding motions (Ikeuchi, 1995). Unfortunately, the interaction between the synovial fluid and implant material can lead to corrosion and particle debris (Diomidis et al., 2012; Roques et al., 2004). Implants are thus subjected to tribocorrosion, or material degradation induced by corrosion and friction coefficient, which may accelerate material degradation and lead to THR complications.

Failure modes of an implant should be passively and continuously monitored using non-invasive and non-destructive techniques. These techniques should detect early-stage structural changes without damaging or interfering with the progression of the implant. Revision surgeries may lead to further complications that will ultimately result in implant failure; therefore, complications must be identified prior to the progression of irreversible damage that eventually requires revision surgery. Although several methods to monitor the status and material progression of a THR exist, these methods are sufficient in areas of cost-effectiveness, continuous and non-invasive monitoring, or early-stage detection. Unfortunately, few strategies to non-invasively diagnose early-stage complications have been reported in the literature. For example, although radiography can detect crack propagation or other complications, this technique is dangerous to patients because it can alter molecular structures and increase the risk of cancer (Mavrogordato et al., 2011). In addition, X-rays, another common imaging method, lack the resolution to provide detailed information on the structural degradation of a THR (Karras, Pullin, Grosvenor, Clarke). Therefore, it is difficult to pinpoint the source of degradation directly cost-effectively. With existing methods falling short in evaluation of total hip replacements, a method to non-invasively and continuously monitor early-stage structural degradations is required.

Acoustic emission (AE) can be utilized to monitor both the initiation and development of material degradations. Unlike existing methods that evaluate total hip replacement degradation, AE technique can passively detect both surface and internal failure modes of implants. As shown in Fig. 1, deformations on the material are produced when mechanical loading or force is applied (Cachão et al., 2019). Material degradation is detected through emitted elastic waves that provide information on both local and global integrity. This releases energy as high-frequency stress waves that can compromise the structural integrity of the material. The damaged material then emits more stress waves that can be detected by sensor placed on the surface of the implant (Remya et al., 2020). This sensor is attached to the surface of the material being observed to receive the propagated waves from any arising deformations (Kapur, 2016). The emitted mechanical wave detected by the sensor is then converted into an electrical signal and amplified by a preamplifier (Davies and Harris, 1996). The electrical signal is amplified so that it can be recorded by a data acquisition system (Kapur, 2016). A threshold is defined to filter out undesired noise (Roques et al., 2004) so that signals whose amplitude is beyond the defined threshold are marked and identified as “hits” (Kapur, 2016). Then, the data acquisition board in a computer receives and displays these signals for interpretation (Davies and Harris, 1996).

Fig. 1.

Acoustic emission method schematic diagram. Material degradations emit elastic waves that propagate until they reach an acoustic emission sensor. The elastic wave is pre-amplified and output as an electrical signal.

This study proposes using AE to non-invasively and continuously monitor the progression of a THR complications and monitor early-stage abnormalities. In this study, a hip simulator was utilized to recreate the material degradation of Cobalt–Chromium–Molybdenum (CoCrMo) and Titanium alloy (Ti6Al4V) pins taking place during the implementation of sleep, stand, and walk cycles. AE sensors were placed on the hip simulator to detect and correlate the AE activity with the potential and friction coefficient found during these cycles. Through this, the correlation between AE activity and tribocorrosion factors can be evaluated to understand degradation mechanisms and develop AE predictions. These predictions are pertinent in identifying and addressing early-stage complications prior to the formation of irreversible damage.

The main objectives of this study are to understand the AE characteristics of degradation mechanisms in common implant materials, CoCrMo and Ti6Al4V, and identify the progress of degradations formed during cycles of sleeping, standing, and walking by correlating them to potential and friction coefficient behavior. The value of utilizing AE to evaluate material progression directly on a THR implant is unknown. Although previous studies have been conducted to evaluate AE efficacy, these studies have not evaluated AE signals detected directly on an implant when recreating the biomechanical processes that a normal hip undergoes. To our knowledge, existing studies only recreated sliding motions experienced at the hip, so the reliability of AE methods for in vitro studies that more accurately represent the physiological conditions at the hip is still needed to be explored.

2. Materials & methods

2.1. Tribocorrosion hip implant simulator

The model utilized a hip simulator, as shown in Fig. 2A, which can maintain the physiological conditions closer to in vivo (Mathew et al., 2011). Located in this chamber exists a pin-on-ball schematic, shown in Fig. 2B, that mimics the head-cup interface of a hip implant. The 28 mm diameter ball was made of Alumina, and two pin materials, 11 mm diameter by 7 mm thickness, were tested in the experiment: CoCrMo and Ti6Al4V. (CoCrMo is supplied by Carpenters Tech, PA, medical-grade Ti alloy supplied by Supper alloy, CA). These materials are commonly utilized in THR because they demonstrate passivation, or the formation of a protective film, that aid in decreasing wear. In each experiments, a fresh surface of alumina ball is made in contact with pins. As shown in Fig. 3, the ball oscillated ±15° while pressed against the pin at 16 N to complete one cycle. The purpose of applying load between the pin-on-ball schematic was to mimic the load-bearing characteristics conducted at a normal hip joint.

Fig. 2.

(A) Schematic image of tribocorrosion hip simulator with acoustic emission data acquisition system. The working electrode (WE), the counter electrode (CE), and the reference electrode (Ref) placement shown. Acoustic emission (AE) sensor placement shown on hip simulator and attached to AE data acquisition system. (B) Schematic image of pin-on-ball interface. The ball rotates ± 15° with a 16N load from the pin. This can damage the passivation layer at the pin-on-ball interface and produce third-body particles. (C) Tribocorrosion experimental protocol. Cycles of rest/sleep (R) when no load is applied, cycles of standing (S) when 16 N of load is applied, and cycles of walking (W) when 16 N of load and torque at 1 Hz are applied.

Fig. 3.

Open circuit potential experimental data for 7500 s (sec). Indicating friction coefficient (red), potential (black), and absolute energy (blue) for stand, walk, and sleep cycles. CoCrMo sample (A) and Ti6Al4V sample (B). Absolute energy in Ti6Al4V experiments are higher than CoCrMo absolute energy, correlating with the significant drop in potential and increase in friction coefficient demonstrated. Open circuit potential experimental data for one cycle. Indicating friction coefficient coefficients (red) in units of mu (μ), potential (black) versus the saturated calomel electrode (SCE) in units of volts (V), absolute energy (blue), and pin motor angle (green). Ideal behavior (C), CoCrMo sample (D), and Ti6Al4V sample (E). CoCrMo demonstrates high fluctuations in absolute energy, while titanium alloy demonstrates smoother, yet higher absolute energy values. Higher absolute energy values are observed when friction coefficient increases from sliding and potential decreases from increased corrosion.

Simulated joint fluid, buffered newborn calf serum (30 g/L: Protein), was utilized in the electrochemical chamber as an electrolyte and as shown in Fig. 2A, served as a lubricant between the pin-on-ball interface. This basic buffer solution had a pH of 7.6, similar to natural synovial fluid, and was prepared using 0.2 g/L of EDTA, 9 g/L of NaCl, and 27 g/L Tris (hydroxymethyl) aminomethane (Tris) at pH 7.4.

In the electrochemical chamber shown in Fig. 2A, the system used a working, reference, and counter electrode. The tribocorrosion hip simulator was connected to a Gamry Instruments Interface 1000 Potentiostat through these electrodes. The working electrode was the sample that was studied, either CoCrMo or Ti6Al4V, and the reference electrode was a saturated calomel electrode (SCE). In this study, tribocorrosion under free potential mode is selected. Through the connections above, the open circuit potential (OCP) or free potential (voltage) of the electrochemical cell was measured between the working (sample) and the reference electrode. This produced time data in seconds and the measured voltage versus SCE.

Acoustic emission sensors manufactured by MISTRAS Group Inc. were attached with hot glue to the hip simulator in locations shown in Fig. 2A. These sensors were then connected to four 40 dB gain pre-amplifiers and then to PCI-8 data acquisition board manufactured by MISTRAS Group Inc. that recorded electrical signals for evaluation. The major data acquisition variables were the sampling frequency of 3 MHz, the threshold of 35 dB, peak definition time of 200 μs, hit definition time of 800 μs, and the analog filter ranged from 20 kHz to 400 kHz. A micro30 sensor was connected to channel 1, wideband (WD) sensors were connected to channels 2 and 4, and a resonant R6 sensor was connected to channel 3. This connection specifically enabled the detection of the emitted signals at the pin-on-ball interface by the sensors and processed them through an external data acquisition system. However, only the data from micro30 AE sensor directly attached to the electrochemical chamber was considered in this study.

AE data was further analyzed to study both time driven data (TDD), which is independent of threshold specificity, and hit driven data (HDD), which is threshold dependent. HDD features such as AE energy, frequency centroid (i.e., the first frequency spectrum moment), peak frequency and amplitude are extracted from the AE signals once the amplitude of each AE signal is above threshold. Through TDD, AE absolute energy is extracted continuously from transient signal at 10 ms interval. Both TDD and HDD features are used to identify the AE characteristics with tribocorrosion data to indicate material degradation.

2.2. Experimental protocol

The tribocorrosion test protocol under free potential mode is shown in Fig. 2C. The experiment consisted of three series of data collections (i) monitoring open-circuit potential (electrochemical condition of the sample), (ii) monitoring friction coefficient, and (iii) collecting AE data with a specific protocol that consisted of human activities, such as rest/sleep, stand and walk (3 times 900 steps) with a total duration of 7500 s. Sleeping conditions were portrayed by obtaining data from the model when there was no load applied. Then, to recreate standing conditions at a hip joint, a load of 16 N was applied. The normal force is selected based on the tribological contact, pin on ball, that matches approximately with normal contact pressure at head-cup area of a THR, (≤50 MPa). Walking gait cycles were then tested by continuing to apply the load while rotating the ball on the pin at 1 Hz with a 30° oscillating angle. Each CoCrMo and Ti6Al4V protocol was repeated on a new sample for N = 3 per sample. From the experiment, the evolution of potential, friction coefficient, and AE data was collected. Then, scanning electron microscopy (SEM), energy-dispersive X-ray spectroscopy (EDS), and profilometry were utilized to visualize and examine the worn surfaces.

3. Results

3.1. Electrochemical potential, friction coefficient, acoustic emission data for 7500 seconds

Fig. 3A and B displays the potential (black), friction coefficient (red), and absolute energy (blue) data during the entire 7500 s experimental period of CoCrMo and Ti6Al4V samples. The different stand, walk, and sleep cycles are indicated with arrows to observe the relationship between the data in these cycles. The open-circuit potential (OCP) can be observed during the entire 7500 s period. However, the friction coefficient and AE absolute energy are only measured during walking cycles, when the metal interface is under dynamic motion.

Fig. 3A shows the different cycles observed for a CoCrMo sample. The potential remains relatively constant from the beginning to the end of the experiment. However, the potential does drop at the beginning of each walk cycle. The potential at each walk cycle decreases further as the walk cycle progresses but returns to the original potential when the walk cycle ends. The friction coefficient measured increases while AE absolute energy is detected during each walking cycle. In general, the AE absolute energy decreases with cycles. After two walk cycles, approximately 20 percent of AE absolute energy drops.

The Ti6Al4V sample is shown in Fig. 3B, where after an initial drop in the first stand cycle, the potential remains relatively constant. However, at the beginning of each walk cycle, the potential drops then increase slightly as the cycle progresses. The friction coefficient steadily decreases with each walking period, while AE absolute energy is also detected during each walking cycle. Similar as the observation in CoCrMo sample, the AE absolute energy decreases with cycles. The absolute energy detected, however, is much larger in Ti6Al4V than CoCrMo samples.

3.2. Electrochemical potential, friction coefficient, acoustic emission data for one cycle

To understand the evolution of the collected data, the expected data trend is plotted for one cycle in Fig. 3C. The expected behavior as the motor angle of the ball (green) slides a total of 30° in one cycle is for the potential (black) to decrease while the friction coefficient (red) and absolute energy (blue) increase. In addition, the expected change in velocity of the ball is represented with the brown line. During periods of increasing potential, the motor angle of the ball will return to its original center position of 0°. Likewise, when the potential decreases, then the motor angle of the ball will increase to 15°, representing the ball being fully twisted either left or right.

Fig. 3D and E shows one cycle of friction coefficient, potential, and AE absolute energy data for CoCrMo and Ti6Al4V samples respectively between the time domain of 2750–2751 s. As expected from the ideal trend graph, all cycles show a trend where the detected potential decreases as the friction coefficient and AE absolute energy increase.

3.3. Acoustic emission hit-driven data

3.3.1. Amplitude histogram

From HDD, the amplitude data of AE hits in decibels were obtained. Fig. 4A and B provide a cumulative amplitude histogram from the three walk cycles of CoCrMo and Ti6Al4V samples, respectively. The amplitude of these detected counts is shown in the x-axis between 0 and 70 dB (dB). The CoCrMo sample had the lowest number of cumulative HDD counts, whereas Ti6Al4V samples had a far greater number of counts, indicating the detection of more AE hits. It can also indicate more AE activities due to corrosion happened in Ti6Al4V samples.

Fig. 4.

Amplitude histogram for number of hit-driven data counts. Provided for CoCrMo sample (A) and Ti6Al4V sample (B). CoCrMo displayed less hits than Ti6Al4V, indicating the less structural degradation in CoCrMo. Amplitude (dB) versus frequency centroid (kHz) scatter plot. Provided for CoCrMo sample (C) and Ti6Al4V sample (D). CoCrMo demonstrated less points at lower amplitudes and frequency centroids, whereas Ti6Al4V demonstrated more points at higher amplitudes and frequency centroids.

3.3.2. Amplitude versus frequency centroid

In addition to the amplitude of AE hits, the corresponding frequency centroid (kHz) was also obtained. This data for CoCrMo and Ti6Al4V samples were plotted in a scatter plot shown in Fig. 4C and D respectively, and the highest amplitudes and frequency centroids are shown in the Ti6Al4V sample. Not only are more hits detected, but the frequency centroid goes as high as 550 kHz with an amplitude of over 64 dB.

3.4. Surface characterization analysis

3.4.1. Scanning electron microscopy

The scanning electron microscopy (SEM) images in Fig. 5 show the wear scar of a CoCrMo sample after the tribocorrosion simulation at a magnification of 100x 500x, 1000x, and 3000x. The wear scar of a titanium alloy sample is shown in Fig. 6 at a magnification of 40x, 100x, 500x, 1000x, and 3000x. The purpose of SEM was to visualize the surface conditions of the sample after significant damage. In Fig. 5A, an SEM image at 100x magnification is obtained for an overview of the entire circular wear scar, while Fig. 6A provides a 40x magnification overview of the larger titanium wear scar.

Fig. 5.

SEM of CoCrMo sample. Magnification at (A) 100xx, (B) 500x, (C) 1000x, and (D) 3000x. Fig. 5A point 1 indicates the overall circular wear scar on the CoCrMo sample. Fig. 9B point 2 indicates the exposed CoCrMo alloy due to the disruption of the passivation layer during sliding cycles. Point 3 and 4 from Fig. 5B and C respectively indicate the formation of protective organic layers on the surface. Fig. 5D point 5 indicates grooving mechanisms formed due to the loading and sliding actions on the material surface.

Fig. 6.

SEM of Ti6Al4V samples. Magnification at (A) 40x, (B) 100x, (C) 500x, (D) 1000x, (E) 3000x. Fig. 6A indicates the overall circular wear scar on the Ti6Al4V sample. Fig. 10C point 2 indicates grooving mechanisms formed due to the loading and sliding actions on the material surface. Point 3 in Fig. 6D indicates the formation of a protective organic layer on the surface, while point 4 in Fig. 6E indicates debris or third-body particles formed in the grooves.

3.4.2. Energy Dispersive X-Ray spectroscopy

In Fig. 7A and C and Table 1, the Energy Dispersive X-Ray Spectroscopy (EDS) data is provided for a CoCrMo sample within the wear tracks where there is a high carbon weight percentage found with a lower weight percentage of Co, Cr, and Mo. Sample group 1 and group 2 spectrum scales are also provided in Fig. 7B and D, respectively. This high carbon content reveals that the proteins from the new born calf serum (NCS) within the electrochemical chamber adhered to the wear scar to form a protective passive layer. This indicates the presence of a passive film within the wear tracks of the sample. The EDS data for the same CoCrMo sample was obtained at a lighter gray region near, but not inside, the wear tracks. The carbon weight percentage dropped while the chromium and cobalt weight percentage increased.

Fig. 7.

Energy-dispersive X-ray spectroscopy for CoCrMo. Provided for sample (A) group one and (C) group two. Higher levels of organic particles are found closer to the grooving in sample group one, indicating the formation of a protective passivation layer. CoCrMo EDS spectrum provided for sample (B) group 1 and (D) group 2.

Table 1.

CoCrMo EDS.

Element	Weight Percentage (%)
	Sample Group One	Sample Group Two
Carbon	19.41	3.16
Silicon	0.73	0.65
Chlorine	0.42	–
Chromium	23.15	26.96
Cobalt	51.56	64.26
Molybdenum	4.73	4.96

The EDS data is shown in Fig. 8A and C and Table 2 for a Ti6Al4V sample within the wear tracks but in the lighter gray region indicated by circular, third-body particles. Sample group 1 and group 2 spectrum scales are also provided in Fig. 8B and D respectively. Fig. 8C provides EDS data within the same titanium alloy wear track but in a deeper, darker region than Fig. 8A. There was a slight increase in chlorine and titanium weight percentage compared to Fig. 8A, but a decrease in aluminum and vanadium. This represents that within the wear tracks, the passivation layer was disrupted, exposing the titanium alloy material. Therefore, EDS analysis confirms the disruption of the protective passive layer within the wear tracks, exposing the Ti6Al4V alloy to corrosion and destructive tribology induced by third-body particles. These factors contributed to the higher wear volume loss determined through profilometry for Ti6Al4V samples.

Fig. 8.

Energy-dispersive X-ray spectroscopy for Ti6Al4V sample (A) group one and (C) group two. Sample group one indicates debris formation either wear- or corrosion-induced. Higher levels of titanium are found within the grooves of sample group two, indicating fewer organic particles. Ti6Al4V EDS spectrum provided for sample (B) group 1 and (D) group 2.

Table 2.

Ti6Al4V EDS.

Element	Weight Percentage (%)
	Sample Group One	Sample Group Two
Chlorine	3.36	2.56
Aluminum	4.99	4.43
Titanium	87.52	89.17
Vanadium	4.13	3.84

3.4.3. Profilometry analysis

Fig. 9A and B provide a three-dimensional observation of the wear scar on the surface of a CoCrMo and Ti6Al4V sample, respectively. The wear scar in the CoCrMo samples is smaller and less deep than the Ti6Al4V samples. With the profilometry data, the roughness average and total average material loss of the wear scar were obtained, and values are provided in Tables 3 and 4, respectively. The roughness average of the standard baseline outside of the wear scar was obtained for each CoCrMo and Ti6Al4V sample with profilometry then averaged to use as control variables. The control roughness average for CoCrMo and Ti6Al4V respectively was found to be ~0.063 μm (μm) and ~0.018 μm respectively, as shown in Fig. 9C. Initially, the CoCrMo sample had a higher roughness average than the Ti alloy sample due to the heterogeneous alloy mixture and element properties. The roughness average of each sample, regardless of the material, increased in the wear scar. The average Ra value of the CoCrMo samples were calculated as ~0.24 μm, and Ti alloy samples were estimated aŝ0.84 μm. In addition, the total average material loss in micrograms (μg) per sample found through profilometry analysis is shown in Fig. 9D. The CoCrMo samples lost an average of 9.60 × 10⁻⁶ ± 1.39 × 10⁻⁵ μg, while the Ti6Al4V samples lost a greater amount with an average of 8.03 × 10⁻⁵ ± 8.73 × 10⁻⁵ μg.

Fig. 9.

Profilometry analysis. Obtained for CoCrMo sample (A) and Ti6Al4V sample (B). The wear scar of Ti6Al4V samples are larger than CoCrMo wear scars. Less structural degradation is portrayed in the CoCrMo sample compared to Ti6Al4V. (C) Average roughness average (μm) for CoCrMo and Ti6Al4V samples. A larger increase in roughness average is observed in Ti6Al4V samples. (D) Total average material loss (μm) for CoCrMo and Ti6Al4V samples. Ti6Al4V demonstrated a higher total material loss due to wear and corrosion (K_WC) in micrograms (μg) than CoCrMo.

Table 3.

Average roughness average (μm) for CoCrMo and Ti6Al4V samples.

CoCrMo Control	CoCrMo Ra	CoCrMo SD	Ti6A14V Control	Ti6A14V Ra	Ti6A14V SD
0.063 μm	0.24 μm	0.21 μm	0.018 μm	0.84 μm	0.72 μm

Roughness average (Ra), Standard deviation (SD), Micrometer (μm).

Table 4.

Average material loss (μg) for CoCrMo and Ti6Al4V samples.

CoCrMo Material Loss	CoCrMo SD	Ti6A14V Material Loss	Ti6A14V SD
9.60 × 10−6 μg	1.39 × 10−5 μg	8.03 × 10−5 μg	8.73 × 10−5 μg

Standard deviation (SD), Microgram (μg).

4. Discussion

4.1. Correlation between friction coefficient, potential, and AE absolute energy

Proteins found in the NCS could have contributed to the lubrication of the friction coefficient surface, attaching to the surface of the material and forming a thin, organic film (Liao et al., 2014). In addition, the presence of carbon detected through EDS analysis on the surface of the material provides lubrication during tribology and corrosion resistance (Mahapatro, 2015). However, the presence of third-body particles, such as organic particles or wear particles, can also abrasively scratch the material surface during walking periods, removing the passive film and exposing the material (Liao et al., 2014). By exposing the scratched material to the NCS during periods of rest, the passive layer forms again and improves corrosion resistance once more (Liao et al., 2014). The molybdenum in CoCrMo has a high affinity for the protein that is present in the NCS (Kamakoti et al., 2016). Therefore, the proteins from the NCS are more likely to adhere to the CoCrMo surface, forming a tribolayer. This tribolayer can aid in the reduction of friction coefficient and wear, resulting in a decrease in corrosion. However, third-body particles can also form from material degradation and make abrasive contact with the surface. During walking cycles, the rubbing of these third-body particles between the pin-on-ball interface can destroy the protective passivation layer. This leads to the exposure of the CoCrMo pin surface, which in turn leads to an increase in corrosion of the alloy. The drop in potential as the walking cycle progress is a result of the corrosion increase. Therefore, as friction coefficient increases in these sliding periods from the presence of organic or third-body particles, corrosion also increases and leads to a decrease in potential. At the end of each walking cycle, an increase in potential is observed with each sample as the surface repassivates, promoting corrosion resistance of the material (Mathew et al., 2011).

In addition, when two different materials of different moduli are placed together and interact without an intervening material to reduce friction coefficient, an increase in stress will occur. Therefore, the titanium alloy samples could have demonstrated more initial damage and drops in potential compared to CoCrMo due to the roughness of the surface. More specifically, Ti6Al4V has a modulus of elasticity of 110 Gpa which is significantly lower than CoCrMo with a value of 210 GPa (Mahapatro, 2015). This demonstrates that Ti6Al4V is less resistant to stress and more prone to deformation. Therefore, during the running-in period when the samples first interact as sliding initiates, more damage is observed in the less rough CoCrMo samples. Profilometry and SEM analysis provide the support that the wear scar of the Ti6Al4V samples are significantly larger in size than the CoCrMo samples.

It is also observed that as the potential decreases, the AE absolute energy values recorded from AE sensor either increase or demonstrate more fluctuations. The AE absolute energy also drops as the samples increase in potential and become more electrochemically stable. In addition, the potential of the samples decreases as the friction coefficient increases in the sliding periods. This shows that the presence of third-body wear particles can interact with the pin-on-ball interface and cause higher friction coefficient and wear at the surface. Therefore, AE data positively correlates with the increase in friction coefficient and a decrease in potential. This implies that AE can be utilized to predict tribocorrosion behavior in CoCrMo and Ti6Al4V samples.

4.2. Acoustic emission hit-driven data

Hit-driven data provides the correlation between AE characteristics and material degradation. For example, hits are AE points that are detected above a specific threshold, indicating the presence of an abnormal material flaw. Therefore, the higher the number of hit-driven data counts detected, the more damage there is on the material. Fig. 4A and B shows that fewer HDD counts were detected in CoCrMo samples, while Ti6Al4V samples had more counts. This correlates with the average material loss calculated for Ti6Al4V and a lesser value for CoCrMo. HDD also provides insight into the amplitude and frequency centroids of the hits, which can be used for further characterizing the AE signals with the degradation mechanism. The amplitude versus frequency centroid scatter plot in Fig. 4C and D shows that the Ti6Al4V samples with more material loss emitted more hits that occurred at a higher amplitude and frequency. These samples are also observed a greater increase in roughness average that may have resulted from the formation of third-body particles in the wear scar (Boness and McBride, 1991; Karras, Pullin, Grosvenor, Clarke). More significant amounts of third-body particles were generated in correlation with the drastic drop in potential observed in Fig. 3A and B. This indicates that with the drop in potential, corrosion was promoted and resulted in the production of additional third-body wear particles. These additional particles further contributed to the increase in tribocorrosion of the Ti6Al4V material.

4.3. AE evolution and tribocorrosion mechanisms

SEM images show what appears to be a passive protective layer, which is further confirmed with EDS analysis. In Table 1, the weight percentage of carbon is higher inside the wear scar of sample group one than in sample group two. This suggests that the proteins from the NCS within the electrochemical chamber adhered to the wear scar to form a passive protective layer. CoCrMo has a high affinity for protein, so the destruction of the passive layer during sliding periods exposed the alloy, allowing denatured proteins to adhere and repassivate. This repassivation accounts for the lower amount of wear volume calculated with profilometry in CoCrMo samples than Ti6Al4V samples.

SEM image reveals the presence of debris or organic particles within the Ti6Al4V wear scars. This is further verified with energy-dispersive X-ray spectroscopy in the analysis of sample group one in Table 2. Sample group one analyzes outside the wear grooves on what appears to be a third-body particle. A higher weight percentage of chlorine and lower weight percentage of aluminum, titanium, and vanadium is present in this sample. This reveals the presence of either third-body particles or a passivation layer. EDS analysis for sample group two in Table 2 examines within the wear tracks and shows a decrease in chlorine and an increase in titanium weight percentage. This demonstrates that within the wear tracks, the passivation layer was disrupted, exposing the titanium alloy material. Therefore, EDS analysis confirms the disruption of the passive protective layer within the wear tracks, exposing the Ti6Al4V alloy to corrosion and destructive tribology induced by third-body particles. These factors contributed to the higher wear volume loss determined through profilometry for Ti6Al4V samples.

The schematic diagram in Fig. 10 shows mechanical and corrosion synergistic interactions, known as tribocorrosion, that are linked to AE data. During the walking cycles in the experiment and disruption of the passive layer, the potential for both CoCrMo and Ti6Al4V samples dropped as corrosion increased, indicating that mechanical exposure (sliding) negatively impacted the tribocorrosion behavior (Mathew et al., 2011). In addition, in both samples, the AE absolute energy displayed higher fluctuations during applications of load and friction coefficient. This infers the correlation between worsening structural degradations of a sample with increasing AE stress waves. This implies that surface friction coefficient and wear correlate with increasing energy values. Besides, correlating with the larger wear volume loss found in Ti6Al4V samples, the disruption of the protective passive layer found with EDS data leads to higher AE activity. Therefore, AE can be utilized to identify the deformations and failure modes of implant materials by non-intrusive means.

Fig. 10.

Schematic of mechanical and corrosion synergistic interactions, known as tribocorrosion, that are linked to acoustic emission data. Acoustic emission data can be analyzed to establish a parameter for early prediction and identify implant failure.

4.4. Clinical implication of the AE data with patient physical data

This study proposes an innovative method to continuously monitor the performance of a hip implant and non-invasively detect complications at early stages through various recreated cycles of sleeping, standing, and walking. With the progression of this continuous method, total hip replacement patients can undergo acoustic emission evaluation immediately following surgery, as baseline data and then the periodic assessment could assist in monitoring the hip implant performance. The AE data obtained when structural degradations do not exist can be referenced each time the patient periodically returns for check-ups. Following each visit, AE TDD and HDD can be utilized to monitor for abnormalities or AE spikes, which would indicate instability/loosening within the THR interfaces. With this information, clinicians can utilize AE to detect early failure modes prior to irreversible, extreme damage that would require a secondary revision surgery. Hence, the prevention of revision surgeries, which normally lead to worsening complications and higher failure rates, can ensure the optimization of the THR lifetime. By periodically monitoring the conditions of the implant and addressing complications early on, there will be a reduction in patient health complications as well as procedure costs. Therefore, the AE method is favorable to reduce complications and improve patient satisfaction.

In addition to patient benefits, hip implant manufacturers may utilize AE to analyze the performance of new implant treatments and designs before experimenting on human subjects, reducing experimentation costs while still obtaining information that portrays the condition of an implant in the human body. Therefore, monitoring the hip implants through acoustic emission techniques can provide patient, hospital, and industrial benefits that reduce procedure costs, time, and complications.

4.5. Limitations and future scope

Although AE can non-destructively evaluate the structural integrity of materials, this study does fall short in some areas. For example, limitations arise from utilizing AE on biologically complex structures such as the THR. In addition to the varying anatomical structures of THR candidates, the biomechanics and mobility of each patient also differ. For example, the load and friction coefficient applied to an implant will be greater for an obese patient versus an underweight patient. Not only will the biomechanics vary, but the overweight patient will have denser tissues that may compromise the emitted waves that are detected by the AE sensors. More precisely, the variations in synovial fluid and tissue thickness of each patient must be accounted for to deploy AE as a non-invasive early structural monitoring tool in orthopedics. However, to combat the signal attenuation, more AE sensors can be utilized to receive signals. Therefore, to clinically utilize AE for implant monitoring, it must be tailored to the needs of each patient, and sensor location must be carefully chosen to optimize the capturing of the signals (Remya et al., 2020).

In addition, although in vitro AE method provides information on the structural degradation of total hip replacements, further testing must be conducted to bring this technology into a clinical setting. In- vitro testing cannot recreate the exact biological and biomechanical structures of the hip; therefore data obtained cannot be utilized as a concrete representation of AE data in vivo (Mavrogordato et al., 2011). It is also worth to state that, the study is a proof of concept investigation on the applicability of AE for the hip implant monitoring; hence the Ti and CoCrMo alloys were tested at same tribological conditions. More specific sample selection will be made based on the hip implant interface and loading conditions in the further study. In this study, the weight loss is calculated on the profilometry estimation. The weight loss estimation solution analysis may be more significant in the case of biomedical application, which will considered in future. Future studies will address other advanced factors including multi-directionality of the loading patterns.

A better understanding is necessary regarding the correlation between detected friction coefficient, potential, and AE absolute energy during normal hip movements as material degradation progresses. Understanding the relationship between these three factors will allow the AE method to be utilized to single-handedly predict tribocorrosion failure modes. Therefore, in a clinical setting, AE can predict the progression of tribocorrosion mechanisms to monitor a patient’s THR. With this information, failure modes can be detected early-on to address complications before irreversible damage that requires a revision surgery takes place.

Besides, the attenuation of signals caused by the presence of biological layers must be better understood. This knowledge can be applied to create a standard evaluation curve that can clinically predict the AE signal emitted from specific failure modes. To elaborate, estimated signal attenuation can be predicted based on the patient’s biological layers at the implant-to-sensor interface. By better understanding the specific signals emitted from different failure modes in correlation with the impact of the tissue interface, AE can be utilized in a clinical setting to evaluate the material progression of an implant despite attenuation or noise produced.

Current findings were directly obtained from the in-vitro hip model. However, there are many other aspects to be considered for clinical applications, such as the tissues and muscles around the implants. The research is in progress to address those limitations, and to develop a reliable AE based diagnostic system for orthopedic patients with THR.

5. Conclusions

The main findings of this study are as follows: acoustic emission technique was shown to be capable to predict the presence and severity of in vitro total hip replacement material degradation, and surface characterization analysis verified that the samples that observed more acoustic emission activity underwent more surface damage. An increase in detected surface degradation would lead to an increase in AE activity. In addition, the study showed a higher increase in AE activity in the Ti6Al4V material, which demonstrated higher levels of wear than CoCrMo material. Therefore, acoustic emission data can correlate with the damage of a total hip replacement material.

By recreating the biomechanics at the hip joint during periods of standing, walking, and sleeping, the relationship between material degradation and friction coefficient, potential, and absolute energy was observed. Although in vitro AE method provides information on the structural degradation of total hip replacements, further testing must be conducted to bring this technology into a clinical setting.