Experimental Analysis of Robotic Cortical Bone Specimen Drilling Performance: Effect of Cryogen

The femur, located between the pelvis and knee, is the longest and strongest bone in the human body, playing a vital role in weight-bearing. It comprises dense cortical bone surrounding softer cancellous bone. Cortical bone's high mineral density and resistance to bending and torsional forces are crucial for its load-bearing capacity. However, aging or disease can reduce its density and toughness, increasing the risk of osteoporosis and fractures [1–3]. As a result, orthopedic surgeries often involve bone drilling to effectively treat these fractures [4,5].

In orthopedic cortical bone drilling, it's crucial to manage cutting forces, torque, and temperature to avoid complications. Excessive forces and torque may cause drill breakage, risking bone damage and nerve injury [6–8]. Among these, temperature emerges as one of the pivotal factors in cortical bone drilling, profoundly influencing the outcome of the procedure. During the drilling process, heat is generated due to the plastic deformation of the chip and friction between the bone and the drill. The low thermal conductivity of bone impedes the efficient dissipation of heat, resulting in elevated temperatures that can compromise the blood supply to the bone, leading to osteonecrosis (Fig. 1) [9].

Nevertheless, the exact temperature threshold at which thermal osteonecrosis occurs remains uncertain. Multiple studies have attempted to ascertain the onset temperature for thermal osteonecrosis. Hillary et al. [10] reported that severe bone damage is likely to occur when bone temperature exceeds 55 °C or persists at such elevated levels for 30 min. In contrast, Eriksson et al. [11,12] recommended maintaining temperatures at or below 47 °C within 1 min and continuing to keep them below this threshold to prevent thermal osteonecrosis. Bonfield et al. [13] noted femoral deterioration at temperatures surpassing 50 °C in vivo due to nonequilibrium recovery and irreversible changes in bone structure. Additionally, Lundskog [14] identified that irreversible enzyme perturbation in cortical bone occurred at 50 °C for 30 s. Furthermore, immediate osteocyte death was observed at 70 °C [15–18].

The inconsistency in reported temperature thresholds for the onset of thermal osteonecrosis underscores the ongoing research into preventing this condition. This study specifically aimed to prevent the most severe form of damage, osteonecrosis. Although bone damage initiates at lower temperatures (47–55 °C), the effects at these temperatures are neither immediate nor severe. Therefore, following previous research, we assumed immediate osteocyte death occurs at 70 °C and set this as our threshold.

Feldmann et al. [19] observed temperature fluctuations while employing irrigation tubes with varying water flow rates to reduce cutting forces, torque, and temperature during drilling. They also proposed a novel drill bit design. The results indicated that the likelihood of exceeding the temperature threshold was significantly reduced only when a new drill bit was used in conjunction with a high irrigation rate (30 ml/min) and intermittent drilling (Peck drilling) at intervals of 0.5 mm. Additionally, Tang et al. [20] explored the impact of cooling water temperature and feed rate on bone temperature. They found that a flowrate of 30 ml/min led to a notably faster temperature reduction compared to 10 ml/min, and the temperature increase associated with a feed rate of 1.5 mm/s was significantly lower than that at 0.5 mm/s.

In previous studies, saline solution, serving as a cooling medium, was frequently employed to reduce cortical bone temperature during drilling. However, when cooling water is used, only indirect cooling of a portion of the drill within the bone occurs, resulting in less pronounced cooling at the cutting tip [21]. The blockage of flute channels by bone fragments was identified as a cause of exceeding average temperature peaks, cutting forces, and torque, contributing to an undesired temperature rise [19]. Moreover, indirect cooling is constrained to the bone's outer surface, potentially allowing nonsterile objects to introduce contamination into the sterile surgical area [21,22].

Saline irrigation is commonly used for cooling during surgical procedures but presents challenges such as indirect cooling, flute blockage, and potential hygiene compromises [23]. This study suggests using cryogenic liquid nitrogen (LN₂) as an alternative, which offers better cooling efficiency and improves chip evacuation. LN₂ is already utilized in various surgical operations, notably for its benefits in reducing treatment times, sufficient biomechanical strength, and preventing infections [24,25].

However, a notable gap exists in the research regarding the application of LN₂ cooling in cortical bone drilling. As in the experiment by Shakouri et al. [26], bone drilling surgery is currently performed manually using a handpiece, therefore relying on the experience of the surgeon. This approach increases the risk of damage to the bone or surrounding tissue, potentially lowering success rate [27]. Therefore, this study conducted cortical bone drilling experiments using a 6-axis robot, which allows for high precision and consistent feed rates. Additionally, the study aimed to quantitatively analyze the cutting performance under various cooling and processing conditions to identify the optimal parameters.

The overall workflow is depicted in Fig. 2, where internal uncertainties such as measurement precision were rigorously controlled through standardized calibration during the data collection process. Furthermore, experiments were conducted in a controlled environment where temperature and humidity were maintained at constant levels, and consistent cortical bone specimens were selected to ensure uniformity in material properties. Additionally, as drilling tools can experience wear and performance degradation over time, they were regularly inspected and replaced as needed to ensure consistent performance.

During the cortical bone drilling process, various issues such as bone destruction and thermal osteonecrosis can occur. Therefore, understanding the mechanical and thermal properties of cortical bone is essential to reduce the incidence of these problems. In this study, cortical bone specimens from the Bonesim 1800 series, which exhibit mechanical properties similar to actual bovine and human cortical bone, were used. These specimens are utilized in various stages of product evaluation by surgeons and project studies by researchers.

Normally, the thickness of the human cortical bone is 6–6.5 mm [28]. However, it may vary depending on age, sex, and genetic factors. Additionally, the neck and head of the femur may be thicker than the shaft, therefore requiring deeper drilling. In this study, a cortical bone specimen with a diameter of 57 mm and a uniform thickness of 10 mm was used. Consequently, the drilling depth was consistently maintained and penetrated at all locations. This study focuses on cortical bone drilling, excluding cancellous bone, and intensively examines the maximum temperature variations at the hole exit according to different cooling methods. This approach ensures that the collected data primarily reflect the effects of the cooling techniques and drilling parameters rather than variations in depth.

The screw insertion torque shown in Table 1 is the result of testing with a 4.5-mm self-tapping screw. Drilling toughness is the time required to process 1 mm of material at a constant speed and load, and is the result of a test at 1000 RPM and 90 N using an orthopedic drill with a diameter of 6 mm [31].

To use LN₂, a cryogenic storage tank, vaporizer, or pressure and temperature control system are generally required. However, owing to the challenge of implementing the system in the field, a handheld LN₂ spray (Fig. 3) was used in this study, eliminating the need for a separate system. The user can directly spray LN₂ onto bone tissue with concentrated low temperature using LN₂ spray at a constant spray pressure, various types of injection nozzles can be selected to control the spray area, allowing for flexible use depending on the environment. However, the manual application method of the handheld spray could potentially obscure clear trends in the research findings. Therefore, to minimize this variability in our study results, a uniform spray nozzle was adopted to standardize its application. Additionally, multiple experiments were conducted to ensure reproducibility, and by comparing various cutting effects under the same processing conditions, all differences were emphasized, focusing on the cooling effects of the LN₂ spray.

Additionally, since osteonecrosis can accelerate at temperatures below −21 °C when using cryogenic gas for cooling, [32] a contactless infrared thermometer was used to monitor the temperature at the injection site in real-time, preventing it from falling below −21 °C. The detailed measurement specifications and the focus distance ratio of the contactless infrared thermometer are depicted in Fig. 4.

The measurement distance of the contactless infrared thermometer to the LN₂ injection area was set at 250 mm following the measurement ratio of the sensor to measure the temperature of the 3.2 mm hole. Additionally, a real-time output of 0–5 V to the display upon reaching −21 °C was set to prevent osteonecrosis.

A spindle (FME, 4212BS) was attached to the end of the robot arm to drill the cortical bone specimens. To rotate the spindle, a brushless motor drive powered by a 24 volts, direct current supply was used. In the robotic drilling process, parameters such as cutting speed and feed rate can influence drilling thrust and torque. Particularly, excessive drilling forces and torque generated during the process can lead to bone destruction or drill bit breakage. To prevent these continuous issues, it is crucial and necessary to study the relationship between drilling force and torque and all possible influencing factors [6]. Therefore, a tool dynamometer (Kistler, 9139AA) was attached between the spindle and the end of the robot arm to acquire cutting force and torque data. These data were collected using a commercially available data acquisition hardware (Dewesoft DAQ, SIRIUS-8xSTG). The detailed data collection process is shown in Fig. 5. A steel use stainless drill tool measuring 3.2 mm × 150 mm was used for drilling to prevent interference between the spindle and the material. In addition, a dispensing system was established to control the flowrate for quantitative analysis with saline solution. For saline solution irrigation, compressed air was delivered to the controller through an air compressor. Compressed air was injected at 0.2 MPa into the barrel and subsequently discharged. A needle size of ∅ 0.15 was selected for accurate irrigation in the hole processing area, and the barrel was fixed with a magnetic base jig to prevent both interference with the tool and backflow.

An UR10e collaborative robot with a maximum radius of 1300 mm and a payload of 12.5 kg was used in this study. The 6-axis robot used in this study provides essential flexibility and precision for the drilling tasks performed, which is crucial for accurately targeting drilling sites on bone specimens. Additionally, each axis can be independently controlled, allowing for precise adjustments in drilling angles and depths, and ensuring a consistent feed rate. The operation and programing of the 6-axis robot are conducted via a controller (Teach pendant), which is designed to manage and execute complex command sequences necessary for precise robotic movements. This controller features a user-friendly interface for programming and real-time adjustments, essential for the accuracy and repeatability of the experiments.

The drilling process programming involves the use of nonlinear movements (MoveJ) and linear movements (MoveL) executed through the teach pendant. To achieve precise horizontal alignment of the tool tip, the rotation vector [rad] indicating the angle of Wrist2 was converted to RPY [°], and the RY axis, which is the tool axis direction, was adjusted to −180 deg. Subsequently, waypoint designation was employed to program the drilling of multiple holes through the bone specimen with a 20-mm depth and a safety distance of 5 mm between adjacent holes.

Cortical bone drilling experiments are typically conducted with vertical drilling [6,7]. However, as the drilling depth in the bone increases, greater heat is generated inside and at the exit of the hole compared to the entrance. Furthermore, since the entrance of the hole is the area where LN₂ is directly sprayed, temperature reduction is more effective, and it can be controlled within the temperature range to prevent thermal osteonecrosis using a contactless infrared thermometer. However, the temperature generated inside the hole is difficult to predict and measure. Consequently, measuring the maximum temperature at the hole exit is essential. The generation of heat during drilling can significantly impact bone tissue, making accurate monitoring essential. Typically, two approaches are considered for investigating thermal properties: thermocouples and thermal imaging cameras. However, due to physical limitations, thermocouples can only be positioned at a certain distance from the hole walls, making it difficult to directly measure temperatures at the tool-bone interface [33]. In contrast, thermal imaging cameras provide real-time, noncontact temperature data, offering greater flexibility. Therefore, in this study, a thermal imaging camera was positioned at the exit of the hole.

Additionally, when drilling a specimen vertically, saline water accumulates on the specimen and cannot penetrate the hole, complicating accurate temperature measurement. Therefore, to ensure the smooth discharge of the saline solution and accurately measure the temperature at the hole outlet, a jig for horizontal drilling (Fig. 6) was designed and attached to the test bed. The temperature at the hole exit was measured using a thermal-imaging camera (FLIR, A700 Series).

The morphology of chips generated during the drilling process can vary depending on process parameters or cooling conditions. The heat generated inside the hole dissipates through thermal conduction to the bone tissue and is expelled by the flow of chips, making it necessary to investigate the chip evacuation process [34]. In this study, a high-speed camera (PHOTRON, mini AX series) was positioned at the entrance of the hole to examine the chip evacuation morphology under different torque and cooling conditions. The frame rate of the high-speed camera was set to 4000 fps, and each condition was recorded for 5.4 s.

The hole-machined cortical bone specimens were cut using abrasive cutter (RandB, RB203 Abcut-M), as shown in Fig. 7. Subsequently, the surface roughness was measured using a surface-measuring instrument (MITUTOYO, SV-2100), and the image of bone surface was measured using scanning electron microscope (SEM) (HITACHI).

Representative independent variables that can be controlled during the robot drilling process are cutting speed and feed rate. Because cutting speed and feed rate affect machining performance, machining conditions were diversified. Owing to the high strength and density of cortical bone, the feed rate of the tool must be lowered to drill the bone at low cutting speeds. If the feed rate exceeds the cutting speed, a risk of tool breakage exists because the feed occurs without simultaneous bone cutting. In the context of robot drilling, the force acting on the tool tip could be detected as interference, potentially causing a program halt. In particular, when drilling with a robot, relatively higher cutting speeds may be required due to its structurally lower rigidity compared to conventional computer numerical control or handpiece drills. Additionally, drilling at low cutting speeds can lead to increased drilling time due to the demand for low feed rates, and the prolonged contact time between bone tissue and the tool may render it more susceptible to high temperatures. Therefore, in this study, the minimum and maximum cutting speeds were set to 5000 and 8000 RPM, respectively, and the tool feed rate was set to a minimum of 0.5 mm/s and a maximum of 1.1 mm/s. The cutting speed was incremented at 1000 RPM intervals, the feed speed at 0.2 mm/s intervals, and the cutting force, torque, and temperature were measured under each condition. Measurements were conducted under the same processing conditions for both cooling methods (using saline solution and LN₂), totaling 32 different processing conditions (Fig. 8).

Previous studies have reported that the cutting force and torque increase as the feed rate rises [7,35]. On the contrary, Lee et al. [4,36] reported that cutting force and torque decrease as cutting speed increases. In this way, in previous experiments, the relationships between the cutting speed, force, and torque have exhibited predominantly conflicting or indistinct patterns. In addition, when saline solution was used, small bone chips discharged through bone drilling combined with the liquid saline solution (Fig. 9) to form a sludge at the hole entrance and caused clogging inside the hole. This prevented smooth chip discharge, resulting in an excessive temperature increase and high cutting force and torque [14].

The cutting force and torque results obtained from the dynamometer located between the spindle and tool are shown in Fig. 10. According to the thrust force results, under both cooling conditions, a cutting speed of 7000 RPM at a feed rate of 1.1 mm/s or lower maintained relatively low thrust force. The observed results are anticipated to be attributed to the mechanical properties associated with the transition from brittleness to ductility induced by heat activation of the bone. The microstructure of cortical bone is similar to that of metallic superalloys [37], with approximately 50% volume fraction of brittle and fracture-prone intermetallic compounds dispersed within a ductile matrix [38]. Therefore, higher cutting speeds and lower feed rates result in elevated temperatures near the perforated surface [39], and it is anticipated that the rate of ductile transition has increased in the heat and drilling forces generated, particularly in the 7000 RPM range.

But in the case of LN₂ cooling conditions, there was a sharp decrease in thrust force at a specific feed rate condition of 0.9 mm/s. For 5000 RPM, the thrust force decreased by a maximum of 37.77%, and for 6000 RPM, it decreased by 47.21%. These results contradict the hypothesis stated earlier.

The results clearly illustrate that the force applied to the material increased in tandem with the feed rate, as shown in Figs. 10(b) and 10(e) and the maximum RMS was recorded at the lowest cutting speed of 5000 RPM and the highest feed rate of 1.1 mm/s. Likewise, the lowest thrust forces are maintained at low feed rates (0.5 mm/s), where high heat occurs.

As for torque, except for the conditions of 8000 RPM cutting speed and 1.1 mm/s feed rate, LN₂ cooling conditions maintained lower torque compared to saline solution cooling conditions. This is also attributed to the bone sludge generated inside the hole under saline solution cooling conditions. Therefore, for LN₂ cooling conditions, it is generally judged to maintain low torque, and it is recommended to choose appropriately within the range of feed rates below 0.9 mm/s. particularly, based on the RMS results, maintaining the segment with the lowest thrust force, which is the feed rate of 0.5 mm/s, is advised.

Most of the heat generated during cortical bone drilling is discharged by chips, rendering smooth chip evacuation essential for drilling. Additionally, the size and shape of the chip may vary depending on variables such as feed rate and cutting speed. According to the torque results shown in Figs. 10(c) and 10(f), the processing conditions that recorded the highest torque value were a cutting speed of 8000 RPM and feed rate of 0.5 mm/s, whereas the processing conditions that recorded the lowest torque value were a cutting speed of 5000 RPM and feed rate of 0.9 mm/s. To assess the chip discharge status under both conditions, images were captured using a high-speed camera under dry conditions without cooling.

Bone chips were collected immediately after the drilling experiment under LN₂ cooling conditions, and the results indicate that, in conditions with the highest torque, bone chips are fragmented into fine powders during evacuation. Conversely, under the lowest torque condition, bone chips were thicker and exhibited irregular shapes [Fig. 11(a)]. Chips were collected under all processing conditions, and their morphology was examined under a microscope. As torque increased, the chips exhibited a powdery form, and as torque decreased, the chips were larger. This is illustrated in Fig. 12.

Under LN₂ cooling conditions, at 8000 RPM and a feed rate of 1.1 mm/s (Lowest torque in 8000 RPM), the chips exhibited an irregular area shape and the maximum temperature measured under this condition was 57.61 °C. On the other hand, at 8000 RPM and a feed rate of 0.5 mm/s (Highest torque condition), the chips were finely segmented [Fig. 11(b)], and the maximum temperature measured under this condition was 45.52 °C. In other words, high torque significantly contributes to the fine segmentation of chips, and finely segmented chips contribute significantly to temperature reduction [34,40]. Therefore, it can be observed that temperature can vary based on the shape of the chips, and the shape of the chips is determined by the torque.

As mentioned earlier, the use of saline solution for external cooling hampered smooth chip evacuation, as the bone chips combined with the liquid inside the hole (Fig. 13), causing clogging.

Additionally, bone sludge formed at the hole entrance, contacting the drill bit and adhering to the flute, thereby reducing cutting performance and causing bit corrosion. However, with LN₂, the cooling gas infiltrated the hole, aiding in its interior cooling [41,42] and facilitating smooth chip evacuation through the gas injection pressure (Fig. 14) [26,43].

Real-time temperature monitoring was conducted by checking the temperature displayed on the thermometer to prevent the temperature from dropping below −21 °C in the cortical bone hole entrance spray zone. In the actual drilling process, the high temperatures generated by the friction between the tool and bone tissue, and the low temperatures resulting from the penetration of LN₂ into the hole interior, offset each other, posing little difficulty in maintaining the temperature within the permissible range.

When using saline solution for cooling (Fig. 9), the interior of the hole was covered with saline solution, hindering accurate temperature measurement. In addition, as drilling depth increased, heat accumulation led to elevated temperatures. This was caused by a reduction in convective heat transfer owing to a diminished exposure of the cutting area to surrounding air, [44] and by longer periods of friction between the material and the tool, which resulted in higher temperatures at the hole exit.

In accordance with previous results, high cutting speeds resulted in significant increases in temperature [45]. Conversely, a lower temperature within the cortical bone was recorded with an increased feed rate [7,44,46–49]. The primary factor leading to an increase in bone temperature is frictional heat [50], indicating that reducing the friction duration between the drill and bone tissue is more effective in temperature reduction.

As illustrated in Figs. 15(c) and 15(d), when using LN₂, a significant reduction in temperature was observed at a cutting speed of 7000 RPM with the highest feed rate of 1.1 mm/s. Furthermore, at a cutting speed of 8000 RPM, clear reductions in temperature were also evident at lower feed rates of 0.5 mm/s and 0.7 mm/s. At the highest feed speed (1.1 mm/s) and a cutting speed of 7000 RPM, the highest thrust force (42.09 N) and the second highest torque value were recorded under LN₂ cooling conditions. The highest torque value was observed at the lowest feed speed (0.5 mm/s, 0.7 mm/s) with a cutting speed of 8000 RPM, resulting in the discharge of finely a divided chip. Consequently, the increase in cutting force and torque contributed to temperature reduction through the discharge of a finely divided chip. Although a high cutting force and torque are effective in reducing the temperature, they can cause tool damage and shorten tool life.

Furthermore, at 8000 RPM, both cooling conditions exhibit a trend of temperature increase with the rise in feed rate. This is because, especially at high speeds like 8000 RPM, increasing the feed rate results in a higher material removal rate [51]. As material removal increases, more energy must be converted to heat in a shorter time. Despite the use of coolant to mitigate this heat generation, the rapid rate of energy conversion at high feed rates overwhelms the coolant's ability to effectively absorb and dissipate the heat. Additionally, as the material removal rate per unit time increases, friction actually increases, leading to a higher accumulation of thermal energy than the coolant can efficiently dissipate [52].

However, considering the cooling efficiency of LN₂, it can prevent thermal osteonecrosis even at high cutting speeds or low feed rates that can prevent excessive force and torque. Therefore, it provides significant advantages in the selection of processing parameters. LN₂ cooling comparatively decreased temperatures by an average of 33.96% and a maximum of 49.59%, reaching values below 70 °C with the all processing conditions.

In this study, “through” holes facilitate efficient cooling and chip removal as the heat and chip generated during drilling can exit through the opposite side. However, in practical applications where holes do not fully penetrate, the release of heat and chip may become more challenging. In scenarios where there is no exit for the hole, particularly with saline solution cooling, there is a greater likelihood of heat becoming trapped within the bone, which can lead to higher localized temperatures. Additionally, saline solution can obstruct chip evacuation at the hole entrance and interfere with the drill bit, thereby increasing friction and heat production [21]. However, because LN₂ cooling is supplied in a high-pressure gaseous state through the drill flutes, it can prevent chips generated inside the hole from adhering to the drill flutes [32]. This facilitates more effective chip evacuation at the hole entrance, unlike saline solution cooling conditions, thereby preventing the accumulation of heat within the hole. Consequently, the advantages of LN₂ cooling can be further maximized in practical applications. Additionally, when applied to the drilling process of cortical bone with average thickness, it is anticipated that the drilling time will be reduced, leading to a significant decrease in heat transfer to the bone [22]. Therefore, it is expected that temperatures will remain below the lower threshold under all processing conditions.

After performing cortical bone drilling procedures and applying screw fixation in osteoporotic patients, complications such as screw loosening or displacement may occur [53]. In patients without osteoporosis, various screw loosening failure rates range from 1% to 15%, while in osteoporotic patients, it has been observed to reach up to 60% [54]. Therefore, achieving more robust internal fixation is a crucial concern for surgeons, and the surface of cortical bone [55], which promotes bone growth and facilitates secure screw insertion, is a critical factor.

Surface roughness plays a critical role in the success of orthopedic surgeries involving screw fixation, and a high surface roughness is necessary to enhance the internal growth rate of bone tissue [56]. Koistinen et al. [57] reported that the growth and attachment of bone could be enhanced by the rough surface or texture of implants, and in the study by Wennerberg et al. [58], the surface was intentionally grit-blasted to create a rough texture. Thus, researchers in previous literature have concluded that a rough surface is required for the more secure attachment of screws to the bone and to promote bone growth in the body portion [59]. Therefore, surface roughness is a key parameter in determining osseointegration and mechanical stability.

To measure the surface roughness of the hole, the center was cut with an abrasive cutter while maintaining the processing conditions, and the roughness of the inner surface was measured using a Mitutoyo surface-measuring instrument. The trace length was set as 8 mm and measured based on the arithmetic average roughness (Ra). The Ra value is the primary metric used to evaluate surface quality, providing a quantitative measure of the deviations in the surface profile, reflecting the effects of various cooling methods and drilling parameters.

The hole's surface, processed with saline solution irrigation, displayed an Ra value below 10 μm, and there was no noticeable trend in Ra concerning variations in cutting speed or feed. Contrastingly, holes machined using LN₂ cooling exhibited relatively high Ra values, which, like thrust and torque, decreased with increasing feed rate. Although Ra increased at a feed rate of 1.1 mm/s, it tended to decrease at 8000 RPM. Therefore, it is determined that a feed rate of 0.5 mm/s is the most suitable for generating a rough surface while maintaining low cutting forces.

Figure 16 illustrates the difference between the roughness peak (Rp) and roughness valley (Rv) at 0.5 mm/s with an 8000 RPM cutting speed, which recorded the highest torque value under all cooling conditions. The maximum height (Rz), which is Rp + Rv, is 70.7733 μm for the sample processed with saline solution cooling, whereas it increased by 86.08% to 131.6961 μm with LN₂ cooling (Tables 2 and 3).

The saline solution has a dynamic viscosity of about 1.02 mPas at 20 °C, which is higher than that of water, and appears to have minimized friction between the hole surface and the tool during hole processing. However, in LN₂ cooling, the administration of a gaseous coolant may cause friction on the hole surface, similar to a noncooling process.

In conclusion, the choice of cooling method significantly influences the surface roughness of drilled holes in cortical bone. LN2 cooling ensures cooling performance while simultaneously creating a rougher surface, thereby enhancing the bonding and fixation strength between cortical bone screws and the cortical bone in orthopedic procedures. These findings underscore the importance of selecting appropriate drilling parameters and cooling techniques to achieve the desired surface characteristics and enhance clinical outcomes.

Scanning electron microscope measurements were conducted to detect microcracks inside the hole. The samples for these measurements were obtained under a total of four conditions, combining both cooling conditions and the highest and lowest torque conditions. As a result, the sample corresponding to the saline solution cooling condition and the lowest torque condition (5000 RPM, 0.9 mm/s) exhibited a cleaner surface compared to that processed under the lowest torque and LN₂ (Dry) cooling condition [Fig. 17(a)]. This was verified through surface roughness measurements; with LN₂ cooling, a somewhat rougher surface was observed [Fig. 17(b)]. Under LN₂ cooling conditions, despite measuring the lowest torque range with a Ra value similar to the saline solution cooling condition, a rough surface was still generated.

On the other hand, in the condition where the highest torque was recorded (8000 RPM, 0.5 mm/s), there is a clear difference between both cooling conditions. In saline solution cooling conditions, the surface conditions remained relatively unchanged compared to the lowest torque counterpart, except for the formation of tool marks in the form of regular thin stripes [Fig. 18(a)]. This can be attributed to the bone sludge adhering to the drill bit inside the hole, as a low feed rate was used in contrast to the high cutting speed of 8000 RPM, leading to the formation of tool marks. On the contrary, under LN₂ cooling conditions, despite recording the highest torque, the efficient evacuation of chips within the hole prevented the formation of tool marks [Fig. 18(b)].

In the case of LN₂ cooling conditions, the thrust force maintained the lowest thrust at a cutting speed of 7000 RPM, a clear correlation between cutting force and feed rate can be observed in the RMS analysis, where an increase in feed rate results in an increase in cutting force. and in terms of torque, except for specific processing conditions (8000 RPM and 1.1 mm/s feed rate), it maintained lower torque compared to saline cooling conditions. However, the low torque resulted in the formation of chips with a larger area, hindering smooth discharge. Therefore, lowest feed rate (0.5 mm/s), which allows for the lowest thrust force, and the cutting speed of 7000 RPM for smooth chip evacuation while maintaining low thrust, were deemed ideal for the cortical bone drilling process under LN₂ cooling conditions.

The shape of the chips generated during the drilling process significantly influenced the temperature. Under high torque conditions, the chips were fragmented into finer pieces, contributing to a decrease in temperature. Additionally, the use of saline cooling impeded chip evacuation, whereas LN₂ cooling facilitated smooth evacuation through pressure injection.

Temperature tended to decrease at low cutting speeds and high feed rates. When using saline solution cooling, the maximum temperature at the hole exit exceeded the critical value with the exception of low cutting speed (5000 RPM and 6000 RPM) and high feed rate in 6000 RPM (0.9 mm/s and 1.1 mm/s) conditions. Conversely, with LN₂, temperatures below the critical value were observed under all processing conditions. Consequently, this approach allows for the exclusion of thermal osteonecrosis issues and the selection of processing conditions that ensure low cutting forces, smooth chip evacuation, and the generation of a rough surface.

Through surface roughness and SEM measurements, we found higher surface roughness under LN₂ cooling conditions. Furthermore, it appears that friction between the drill and the surface increases under dry conditions, proving effective in creating a rough surface. However, it's worth noting that in actual surgery, blood with similar physical properties to saline is produced. Therefore, the results in actual field conditions may be inconclusive compared to the extremely dry processing conditions evaluated in this study.

In this experiment, efforts were made to identify the optimal LN₂ cooling conditions across various processing parameters. Currently, there is a need for real-time temperature monitoring due to concerns about thermal osteonecrosis when using LN₂ as a coolant, and the variability in spray methods may compromise reliability. However, the LN₂ cooling method maintains lower thrust and torque by facilitating smooth chip evacuation, offering significant advantages in terms of temperature control.

Despite advances in modern medical technology, achieving the high accuracy and efficiency required in precision surgery necessitates continual technical improvements. Particularly in complex medical procedures such as robotic bone drilling, the accuracy of real-time data analysis and prediction plays a critical role. Recently, various medical fields have been utilizing data mining or multidimensional self-supervised learning algorithms to handle complex datasets, and in complex medical procedures such as robotic bone drilling, more sophisticated data processing can be anticipated [60,61]. Future efforts will consider the application of these methodologies and aim for a more precise analysis of the various cutting effects that occur during the robotic cortical bone drilling process.

• National Research Foundation (NRF) of Korea (Award ID: 2020R1C1C1008113; Funder ID: 10.13039/501100003725).

• Ministry of Science and ICT of Korea and “Development of a Metalearning Framework for Intelligent Predictive Maintenance in Robotic Arm” Korea Institute of Industrial Technology (Award ID: KITECH-JE-240026; Funder ID: 10.13039/501100003695).

The datasets generated and supporting the findings of this article are obtainable from the corresponding authors upon reasonable request.

$n$ =

number of observations

$yi$ =

observed values

$ŷi$ =

predicted values