1 Introduction

The surface integrity (SI) of bores influences the performance and capability of drilled components. For instance, surface and subsurface characteristics, such as the residual stress state and the microstructure have a decisive impact on e.g. the fatigue behavior of parts and their resistance to stress corrosion cracking [1,2,3]. The SI, in turn, results from subsurface conditioning in drilling, as a consequence of the thermomechanical loads exerted on the bore wall. Figure 1 provides examples of such process-structure–property relationships for the Boring and Trepanning Association (BTA) deep hole drilling process. In light of these complex interrelations, securing favorable properties in the subsurface of bores is of vital importance to guarantee a component's reliability and performance throughout its entire life. However, the accessibility of bores is exceptionally limited. It is thus particularly challenging to conduct measurements and assess SI inside of deep drilled components, especially for relatively small diameters [4, 5]. This calls for developing adapted testing strategies, which are suitable for non-destructively evaluating the SI of bores. Up to this day, however, no method for non-destructively inspecting various aspects of surface integrity on the crucial inner surface of BTA deep-drilled components has been developed. Integrating such concepts into robust process controls for machining processes, enables the reliable generation of subsurface features in machining. This in turn allows for guaranteeing the performance and reliability of components [4]. In this context, this study aims at assessing the interrelations between the design of the drilling process, the resulting thermomechanical loads and the subsurface conditioning in drilling. The findings are used to qualify magnetic Barkhausen noise (MBN) analyses as a holistic method for analyzing SI of the bore hole wall of deep drilled components.

Fig. 1
figure 1

Interplay between the design of the process, the subsurface conditioning, and the performance of components in BTA deep hole drilling

Full size image

2 State of the art

2.1 BTA deep hole drilling

Drilling processes can be classified into two general categories. The first one, conventional drilling with twist drills, is the most common of all drilling methods [6]. The second category of drilling processes is deep hole drilling (DHD). DHD is usually performed using highly specialized tools. In general, a drilling process is considered a DHD process if the length-to-diameter ratio of the bore is larger than ten or if DHD tools and methods are used. One of the most fundamental differences between conventional and DHD tools is the symmetrical and asymmetrical design, respectively [6, 7].

Examples of applications for DHD include safety-relevant components in the aerospace industry and highly loaded parts in the petrochemical industry [6]. Figure 2 shows that the area of application for DHD processes is significantly larger than the one for conventional drilling. Nevertheless, conventional drilling is used more overall, mainly because of its high productivity, low cost, and relative simplicity. However, if the quality of bores is crucial, DHD processes are oftentimes used regardless. In Table 1 some of the key characteristics of conventional drilling and DHD are compared.

Fig. 2
figure 2

Application of deep hole drilling and conventional drilling according to [7]

Full size image
Table 1 Characteristics of conventional and deep hole drilling [6]
Full size table

Figure 3 shows a tool with cutting edges and guide pads for BTA deep hole drilling. The BTA DHD process is usually used for large-diameter bores (D ≥ 40 mm). However, BTA DHD can also be used for machining relatively small diameters, down to D = 6 mm [6]. The asymmetry of the tools leads to a radial force component in BTA DHD. This force would deflect the tool and lead to a straightness deviation of the bore. The drill head is supported by the guide pads, which are pushed onto the bore wall, causing a self-guiding effect and burnishing of the bore surface and deformation of the subsurface. This improves the straightness and diameter deviations of the bore. In contrast to conventional drilling and gun drilling, the chips in BTA deep hole drilling do not come into contact with the bore wall but are evacuated through the chip mouth and the boring bar. Thus, the achievable surface quality of the bore for BTA deep hole drilling is higher than for other DHD processes.

Fig. 3
figure 3

BTA drill head (D = 60 mm) Botek type 12 according to [8]

Full size image

Figure 4 shows the overlapping conditioning mechanisms of the guide pads and cutting edges in BTA deep hole drilling. The influence of the process design on the surface roughness is shown in [9]. The cutting edge and the guide pad condition the subsurface of the bore in different ways, which the authors were able to distinguish [9,10,11].

Fig. 4
figure 4

Conditioning of the bore surface according to [8]

Full size image

When analyzing the influence of the cutting edge the authors found, that without the influence of guide pads, the residual stress state can vary from compressive, for lower cutting speeds and feeds, to tensile, when using higher cutting parameters. Furthermore, white etching layers (WEL) and significantly increased hardness in the subsurface were found in microscopic analyses [9, 10].

The influence of the guide pads is characterized by elastic and plastic deformation of the subsurface, which are distinctive in microscopic analyses. Furthermore, solely compressive residual stresses were measured in BTA deep drilled specimens. In these specimens, larger WEL were found and subsurface hardness was increased even further, compared to specimens that were machined by cutting only [9, 10, 12].

This agrees with the findings of Zhang et al., who examined BTA deep hole drilling of cast iron. They found hardness to be 56% higher after cutting and burnishing, compared to cutting only [13].

Nickel et al. investigated single lip deep hole drilling (SLD) of AISI 4140 + QT with a tool diameter of D = 5 mm. They found that the higher process forces when using high feed rates were accompanied by higher temperatures. In their studies, the increase in the thermomechanical loads led to WEL formation. Based on this, they concluded that a critical parameter combination might exist, which poses a threshold for WEL formation. For identifying these critical combinations, thermo-mechanical loads always need to be regarded in conjunction [14]. Wegert et al. reported highly distorted subsurface zones with large compressive stresses when using a D = 18 mm SLD with indexable inserts and guide pads [15].

2.2 Surface integrity

The SI of components can deeply affect their capability and life. Throughout the past decades, various review papers and a vast amount of research articles have been published, elucidating the complex interrelations between numerous features of SI and the functional performance of components [1,2,3]. In this context, microstructural alterations, such as the formation of WEL in steels, titanium, and nickel alloys play a pivotal role. Due to their detrimental impact on fatigue performance some scientists even consider WEL to be the metallurgical anomaly that is perhaps “the most feared by materials engineers” [16]. In response to the importance of WEL, persistent efforts have been undertaken in the past, to evaluate approaches in non-destructive testing (NDT) for detecting WEL quickly, and reliably. For this, various techniques, such as acoustic emission testing, eddy current testing, optical scattering, and ultrasonic testing have been employed [16]. However, a universally applicable method for diagnosing WEL is not available yet and current strategies oftentimes lack robustness and reliability. Hence, up to this day, metallographic investigations of etched sections remain the gold standard for WEL assessment, despite their practical limitations and numerous disadvantages.

A key obstacle in the task of reliably detecting WEL is that the conditions that cause WEL formation during machining are complex and still not fully understood. Brown et al. published a comprehensive review of the state of the art of research on the mechanisms leading to the formation of WEL [16]. Usually, either mechanical (grain refinement through severe plastic deformation) or thermal effects (phase transformation) dominate WEL formation [16]. This range of thermo-mechanical loads, governing the formation of WEL, results in a vast spectrum of WEL with distinct features and properties, e.g. in terms of residual stresses. Zhang et al. analyzed WEL induced by hard cutting of AISI 52100. They found compressive residual stresses with an intensity of approx. σRes ≈ – 300 MPa when machining by unworn tools, and tensile residual stresses of approx. σRes ≈ 400 MPa when using worn tools [17]. The wide range of distinct properties inside of WEL calls for a flexible inspection technique and demands a comprehensive and holistic evaluation of subsurface properties. Isolated approaches which are limited to certain properties, like residual stress analyses by X-ray diffraction, thus fail to deliver robust results for WEL detection.

One of the most promising techniques for detecting and characterizing WEL in steels is the analysis of magnetic Barkhausen noise (MBN) [18,19,20]. This technique is based on the observation of the movement of domain walls when an alternating magnetic field is applied to ferromagnetic materials. Domain wall movement is affected by various properties, such as microstructural aspects like grain size, and (residual) stresses. As a result, MBN analysis is sensitive to various types of subsurface alterations that are associated with the formation of WEL, such as changes in hardness [21], residual stress state [22], phase composition [23], and grain size [24].

Over the course of the past decades, scientists have qualified MBN analysis for WEL detection for selected experimental setups. Stupakov et al. detected a second peak in the MBN envelope when a WEL was present. In their experiments, increasing tool wear in milling of AISI 52100 corresponded to an increasing WEL thickness and could be detected via MBN analysis [19]. Neslušan et al. supported these findings, providing results for milling of 100Cr6 as well. They found that both subsurface layers, the heat-affected zone, as well as the WEL, contribute to the MBN signal [20].

Analyzing BTA deep-drilled AISI 4140, Strodick et al. found that the maximum MBN amplitude decreases, when WEL are present in the subsurface of specimens [12, 25]. This correlation was later confirmed for specimens that were machined in the previously mentioned analogous experiments by cutting edges only, without the influence of guide pads [10].

Due to the number of factors that affect MBN and the vast spectrum of properties inside of WEL, the results of MBN analysis can appear contradictory and there are severe limitations for transmitting this technique e.g. for altered cutting conditions based on empirical and phenomenological approaches only. Instead, enhancing the robustness of MBN analysis demands mechanism-based and concise studies to identify the factors dominating MBN. This way, understanding of MBN analysis for WEL detection can be enhanced.

2.3 Performance and capability

Baak et al. analyzed the fatigue behavior of components machined by SLD. Based on micromagnetic analyses, they evaluated and extrapolated the performance of specimens, machined with different cutting speeds and feeds. They provided evidence that MBN is an adequate means for non-destructively monitoring material degradation during fatigue [26]. Strodick et al. investigated the impact of cutting parameters on the mechanical properties of BTA deep hole drilled components under quasi-static loading. For this, they performed instrumented compression tests, inspired by the tube-flattening test. Based on this methodology, Strodick et al. analyzed the evolution of residual stresses in BTA deep drilled components under quasi-static and cyclic loading. In this study, they verified that WEL can be detected by MBN analysis. Furthermore, their results indicate that the presence of WEL in the subsurface of BTA deep drilled components is indeed detrimental to performance under quasi-static loading. In specimens with WEL, cracks initiated at substantially lower displacements, compared to specimens free of WEL [27].

3 Materials and methods

3.1 Material

For all experiments, the material was a quenched and tempered AISI 4140 + QT (42CrMo4 + QT, 1.7225) steel in its rolled state. Two batches of the material were employed with relatively similar mechanical properties and chemical compositions. Both batches were austenitized and water quenched. Batch A was tempered at 640 °C, and batch B at 680 °C. The mechanical properties and tempering process details according to material inspection documents can be found in Table 2. The chemical compositions are depicted in Table 3. The specimens initially had an outer diameter of Do = 80 mm. Before drilling they were sawed to a length of l = 250 mm, turned and a chamfer was machined in preparation for BTA deep hole drilling. Segments were extracted at the middle of the bore for light microscopy (LiMi), hardness testing, XRD, EBSD, MFM, and MOKE microscopy. Thus, all specimens were extracted at a similar length of the bore. However, since different segments were used, they were extracted at different angles around the circumference.

Table 2 Material properties of AISI 4140 + QT
Full size table
Table 3 Chemical composition of AISI 4140 + QT
Full size table

3.2 Experimental setup

3.2.1 BTA deep hole drilling

BTA deep hole drilling was carried out on a horizontal BTA machine (Giuseppe Giana GGB 560). A schematic sketch of the BTA machine, along with the sensors used in the different series of experiments, is depicted in Fig. 5. The sketch shows that the tool consists of two parts, the boring bar, and the drill head, which is mounted onto the bar. Due to the slender design of the tool system, the process is susceptible to vibrations while machining. The BTA experiments were carried out using a drill head of D = 60 mm (Botek type 12) (Fig. 3). The deep drilling oil Berucut RMO TC 22 was supplied through the oil pressure head (BOZA). For this, a pressure of p = 11 bar and a flow rate of \(\dot{V}\) = 300 l/min was used. Cutting speeds in the range of vc = 60–120 m/min were used in combination with three feeds (f = 0.150; 0.225; 0.300 mm). Depending on the measurement setup, it was necessary to either work with a rotating workpiece or with a rotating tool. When using strain gauges to measure the bore moment, and/or the force sensors, to measure the normal force perpendicular to the guide pad and the bore surface, the workpiece was rotating to create the cutting motion. To use the force sensors under the guide pad, it was necessary to modify the drill head to integrate the sensors. A detailed account of this can be found in [28]. When using the thermal sensors, the workpiece was stationary and the tool was rotating. Only the microphone was not dependent on the mode of operation. The microphone was used to record the audible chatter vibrations, analyze the frequencies, and compare them to the data of the force sensors, and strain gauges.

Fig. 5
figure 5

Different experimental setups on the BTA-DHD-machine Giuseppe Giana GGB 560

Full size image

3.2.2 Surface integrity

For the non-destructive analysis of SI, various complementary techniques were employed. Micromagnetic investigations based on MBN analysis were conducted for a holistic and comprehensive assessment of multiple areas of SI. To back up the findings of this technique and identify the underlying mechanisms, that govern MBN analysis, additional mechanism-based investigations were performed for selected specimens. Scanning electron microscopy (SEM), magnetic force microscopy (MFM), magneto-optical Kerr effect (MOKE) microscopy, electron backscatter diffraction (EBSD), and X-ray diffraction (XRD) were employed for an in-depth analysis of selected specimens. The general strategy along with the equipment employed is depicted in Fig. 6.

Fig. 6
figure 6

Strategy and equipment employed for the holistic assessment of SI. Scales for the pictograms can be found in the results section

Full size image

For the assessment of the thickness of WEL in the subsurface, sections were extracted from the bores. After embedding, polishing, and etching with Nital, the specimens were analyzed by optical microscopy with the microscopes VHX-7000 (Keyence Corporation) and Axio Imager.M1m (Carl Zeiss AG).

MBN analyses were performed, using the devices FracDim (Fraunhofer IKTS) and Rollscan 350 (Stresstech Ltd.). Multipurpose sensors were used along with custom-designed positioning devices. In addition to this, a sensor was developed in close collaboration with Stresstech Ltd. which can be mounted into a drill head [29]. A detailed account of the procedure and parameters, employed for MBN analyses is given in [10, 25]. The equipment used is depicted in Fig. 7a) for FracDim-based analysis and Fig. 7b) for Rollscan measurements.

Fig. 7
figure 7

Equipment for MBN analysis. a Setup for FracDim analyses, previously published in [25]; b Setup for Rollscan 350 analyses

Full size image

The device HMV-G21 was used for analyzing microhardness mappings (HV 0.01). The distances specified in DIN EN ISO 6507–1 were transgressed. This practice is well established and employed frequently to assess microhardness gradients at very close distances to surfaces. [30].

XRD analyses were undertaken using the X-ray diffractometer D8 Discover (Bruker Corporation) and the portable system Pulstec µ X360s. In the Bruker device, a copper tube was employed, with a current of 40 mA and voltage of 40 kV. A collimator with a diameter of d = 1 mm and a Ni-filter were used. Scans were performed from 2θ = 35° to 155° with a stepsize of approx. 0.0016°. For a rough estimation of crystallite size, the Scherrer equation was used (1), estimating that crystallites were spherical and uniformly sized:

$$D = \frac{k\lambda }{\beta cos\theta }$$
(1)

where D is the crystallite size, k is a constant (approximated with k = 0.94 for spherical particles [31]), λ is the wavelength of X-rays (Cr-Kα = 2.291 Å), β is the full width at half maximum (FWHM) of the peak at 2θ = 156.396° and θ is the Bragg angle (θ = 78.198°). The dislocation density δ was calculated, using the formula:

$$\delta = \frac{1}{{D}^{2}}$$
(2)

where D is the crystallite size, calculated by the Scherrer Eq. (1) [31].

The Pulstec system was employed with a chromium tube to obtain full width at half maximum (FWHM) and residual stresses. A Bragg angle of 2θ = 156° (Fe-α, {211}) was analyzed and residual stresses were obtained by the cosα-method. In contrast to the sin2 ψ-method, which requires various ψ-inclinations, one single diffracted Debye–Scherrer cone is analyzed. As a result, this method tends to be faster than the sin2 ψ-method, while leading to similar results [32].

SEM analyses were performed, using the Mira 3 XMU system by Tescan and the focused-ion beam (FIB) SEM system Crossbeam 550 by Carl Zeiss. EBSD analyses were performed using the SEM Mira 3XMU by Tescan equipped with an EDAX Velocity Plus detector. For pattern acquisition, an acceleration voltage of 25 kV with a stepsize of 25 nm was used. Visualization and post-processing were done in OIM Analysis using Neighbor pattern averaging and reindexing algorithm.

For an in-depth assessment of the structure of magnetic domains in the micrometer range, MFM investigations were performed for selected specimens, using the piezo-based AFM LiteScope by Nenovision. MFM investigations were done with magnetic Akiyama probes as two-pass measurements. Topography was acquired in the first pass with setpoints of around 5 Hz in tapping and the change of phase was mapped in the second pass contactless with offsets around 150 nm. To find the region of interest and to ensure a stable oscillation of probes, the AFM investigations were performed under a vacuum inside the Zeiss Crossbeam 550.

In addition to MFM analyses, MOKE microscopy was performed, to enhance the understanding of the structure of magnetic domains and domain wall movement when an alternating magnetic field is applied. The MOKE is based on the analysis of small rotations of the polarization plane of reflected light from a magnetized specimen [33]. The differences in the grayscale of the images show the domain structure depending on the magnetization conditions. The MOKE investigation was carried out using a high-resolution Kerr Microscope equipped with a magnetometer. This device is composed of a Light Microscope Axio Imager manufactured by Carl Zeiss Microscopy GmbH for the optics. A LED lamp and an ORCA-CMOS camera manufactured by Evico magnetics GmbH were installed to perform high-resolution microscopy. A stage with a piezo module for drift correction and a rotatable magnet to induce the magnetic field in the specimens were used during the experiments. Specimens for MOKE investigation underwent metallographic preparation including sectioning, embedding using a phenolic conductive resin, mechanical grinding, and polishing with conventional liquid suspensions (6, 3, and 1 μm). The final step includes polishing with colloidal silica (SiO2) using the corresponding cloth and suspension oxide polishing suspension (OPS).

4 Results

4.1 Thermomechanical loads in BTA deep hole drilling

In Fig. 8 the results from previous works of the authors regarding the bore moment and guide pad normal force measurements are summarized. The mechanical loads revealed an influence of the cutting speed and the feed on the bore moment and the normal force. At higher cutting speeds, the measured bore moment would slightly decrease, probably due to thermal consolidation. An increase in the feed rate resulted in an increased drilling torque and increasing normal forces.

Fig. 8
figure 8

Measured bore moments and normal forces under the guide pad [28]

Full size image

A detailed discussion of these experiments can be found in [28]. Figure 9 gives a short review of the experiments regarding the thermal measurements in the workpiece while BTA drilling. The authors were able to measure the temperatures in the contact zone between the cutting edge and bore surface using a fiber optic ratio pyrometer (FIRE-3 by en2Aix with a Pigtail WF 300 dF = 300 µm optical fiber). The temperatures in the subsurface zone were measured with type K thermocouples with a diameter of dTC = 250 µm at three distances from the bore surface.

Fig. 9
figure 9

Measured temperatures in the contact zone and the subsurface zone [34]

Full size image

An increase in feed leads to a higher temperature in the subsurface zone of the bore. An increase in the cutting speed resulted in higher temperatures in the contact zone between the cutting edge and bore surface, indicating the effect of thermal consolidation in BTA deep hole drilling in this material [34].

Figure 10 shows an example of the recorded chatter vibrations using the microphone and strain gauges. Both signals were processed using the Fast-Fourier-Transformation and displayed in a spectrogram. In both spectrograms the dominant frequency FD ≈ 1210 Hz shows the highest amplitude. This shows the possibility to analyze the dynamic state of the BTA process using a microphone instead of more complex strain gauges.

Fig. 10
figure 10

Comparison of the bore moment and the audio signal using spectrograms

Full size image

4.2 Modelling and simulation of the BTA deep hole drilling process

To complement the quite complex and effortful real experiments by numerical simulations several approaches towards the numerical simulation of the process were developed and partly validated.

The major results expected from the process simulations run based on the developed models are the cutting forces, the temperatures close to the cutting edge, and the remaining residual stresses introduced into the subsurface of the bore wall. These values are intended to be used to supplement the physical BTA deep hole drilling experiments by additional data, to set up the fast regression models for the process control. The major idea here is to carry out virtual drilling processes with different parameter combinations for the cutting speed and the feed rate and to simulate the mentioned values to establish additional relations between the process parameters and the resulting surface properties. In the case of the residual stresses, the simulated values can be used directly for the control system. In contrast, for the WEL, the underlying hypothesis is, that there is a relationship between the occurring cutting forces, the introduced temperatures close to the cutting edge, and the stress fields generated by the cutting edge and the guide pad and the occurrence of WEL. To calculate the aforementioned process values, two different 2D and 3D model were developed, representing an abstraction of the complex BTA process. The investigations started by setting up a 3D model based on the commercial chip formation simulation system AdvantEdge® [35]. Figure 11 shows the model and some results.

Fig. 11:
figure 11

3D model of the BTA process and simulation results

Full size image

The material model used throughout all the models, was the well-established Johnson–Cook model, and the parameter set for the AISI 4140 steel was adopted by own Split-Hopkinson-Pressure-Bar material characterization tests [36]. As a friction model, Coulomb friction with a constant factor of µ = 0.65 is used and all mentioned FE-calculations are explicit.

Although some results with respect to cutting forces and the forces acting on the guide pad could be calculated, it was not possible to validate the 3D model, which can be attributed to the too low resolution of the mesh. The residual stresses could not be extracted from the model, because there was only one element covering the first 500 µm below the surface, where all the residual stresses and their development are measured. The values for the cutting force and the force acting on the guide pad deviated by a factor of three to five. Due to the already very long computational times, the use of a finer mesh was not possible.

To overcome these problems, a reduction from 3 to 2D was chosen, which was reflected in two different modelling approaches, both using the commercial general purpose Finite Element software Abaqus [37] as a solver. The simplification of the BTA deep hole drilling process from 3 to 2D is shown in Fig. 12.

Fig. 12:
figure 12

3D to 2D simplification of the BTA deep hole drilling process

Full size image

Here, a virtual plane orthogonal to the middle axis of the drilling head is assumed and the model only reflects the process mechanics in this plane, following the theory of orthogonal cutting. In addition, the width of the cut has to be defined, in order to calculate real force values.

Due to the excessive, mesh distortion occurring in the modelling of cutting operations, it is necessary to apply a continuous remeshing procedure, assuring a high mesh quality throughout the simulation. Therefore, an adaptive remeshing method procedure based on the supernodal patch recovery method according to Zinkiewicz and Zhu was implemented by Python scripts [38, 39]. The mesh shown in the upper part of Fig. 13, is the result of the local and error controlled mesh adaption [40].

Fig. 13
figure 13

Course of the 2D simulation of the BTA deep hole drilling process using the adaptive remeshing procedure.

Full size image

The Finite Element Model was split into two parts. First, the cutting edge removes the chip and in the second step the guide pad slides over the same surface, already involved in the chip formation below the cutting edge, Fig. 13.

In Fig. 14 some results from this approach are shown. The process forces partly are in good agreement to the measured values, whereas the temperatures still show larger deviations. The computed residual stresses already appear in the typical hook-shaped manner, here single values match to measured ones in an acceptable way but further validation is necessary.

Fig. 14
figure 14

Results of the 2D simulation of the BTA deep hole drilling process using the adaptive remeshing procedure

Full size image

Another option to avoid the already mentioned massive mesh distortion is to switch from a Lagrangian meshing scheme to an Eulerian meshing scheme, where the mesh is fixed in space and the material is shifted through it. For this purpose Abaqus offers the CEL method. The model implementation is shown in Fig. 15. The light green area in Fig. 15 shows a magnified extract of the mesh in the Eulerian space, where it can be seen, that the resolution close to the surface of the bore hole’s wall is set. To support the further development of the control system by adding additional BTA deep hole drilling tests, it is most likely that the CEL approach will be utilized, whereas an extension back to 3D might be necessary, due to the three-dimensional nature of the contact situation between the guide pad and the bore hole’s wall. This contact situation is elucidated in more detail in [41] where an additional analytical- numerical model is shown as well.

Fig. 15:
figure 15

2D model of the BTA deep hole drilling process using CEL

Full size image

4.3 Surface integrity resulting from BTA deep hole drilling

4.3.1 Microstructure

In previous works of the authors, the occurrence of WEL in the subsurface of BTA deep drilled bores was analyzed. Evidence was provided that WEL can form in BTA deep hole drilling of quenched and tempered AISI 4140 when using relatively high cutting speeds and feed rates [12, 28]. Exemplary results for the metallographic analyses of cross-sections extracted in the direction of cut are given in Fig. 16. The occurrence of dynamic process disturbances was found to be a reinforcing factor for WEL formation. When chattering occurred in deep hole drilling, WEL were observed for lower cutting speeds and feeds, compared to a process free of dynamic disturbances [25]. In analogous experiments to BTA deep hole drilling, it was found that WEL can form during cutting and without the effect of guide pads [10].

Fig. 16
figure 16

Exemplary results of light microscopy, for specimens of batch A, cross sections extracted in the direction of cut, according to [12]

Full size image

In metallographic analyses, specimens with WEL were found to consist of three layers. The WEL is followed by a transitional layer and the bulk material. The transitional layer is marked by severe material drag in the subsurface. For some specimens, it appears darker than the bulk material (Fig. 16) [12]. In the direction of the feed, the thickness of the WEL can exhibit a periodicity, which corresponds to the feed employed. This is displayed in Fig. 17. On a larger scale, the thickness of WEL in the direction of the cut seems to change periodically as well. Between the spots marked with A and B in Fig. 17, the thickness of the WEL increases and then drops again. This might result from the combined effects of feed and cutting speed. Based on these results, it can be assumed, that the WEL can exhibit a helical structure along the bore.

Fig. 17
figure 17

Exemplary results of light microscopy, for a specimen of batch B machined with a feed of f = 0.300 mm and a cutting speed of vc = 100 m/min

Full size image

4.3.2 Magnetic Barkhausen noise analysis

It was shown in previous works that WEL can be detected by MBN analyses, as they cause significantly lower values for maximum Barkhausen noise Mmax and root mean square RMS of MBN [10, 12, 25]. Figure 18 presents exemplary results of MBN analyses using FracDim and Rollscan 350. The specimen machined with a cutting speed of vc = 120 m/min reveals significantly lower Mmax and RMS compared to the specimen machined with a cutting speed of vc = 60 m/min. In light microscopy, relatively thick WEL were observed for the bore machined with vc = 120 m/min. Based on these studies, the focus of this study is to evaluate and separate the underlying mechanisms, governing MBN analyses. The properties of WEL which affect MBN the most need to be identified, to make MBN analysis a more robust method for the assessment of SI resulting from BTA deep hole drilling.

Fig. 18
figure 18

Exemplary results of MBN analysis and light microscopy for specimens of batch B, machined with a feed of f = 0.225 mm and two cutting speeds, using uncoated cutting edges and guide pads. The results of MBN previously have been published previously in [42], and the micrographs in [25]

Full size image

4.3.3 Holistic evaluation of SI

Figure 19 shows the results of the extensive investigations on surface integrity for the BTA deep drilled bores, depicted in Fig. 16, following the strategy presented in Fig. 6. For the three specimens, which had no signs of WEL in light microscopy, higher Mmax values were reported in MBN analyses, compared to the specimens which showed WEL. The standard deviation was significantly higher as well. A more extensive analysis of these MBN values can be found in [12]. The observation of low MBN amplitudes when WEL are present agrees to the findings of Neslušan et al., who observed poorer MBN for specimens which had WEL after milling. They identify the austenite volume content, compressive residual stresses and very fine grains inside the WEL as the decisive factors, governing MBN analyses [20].

Fig. 19
figure 19

Results of the holistic assessment of SI of specimens of batch A, by magnetic Barkhausen noise (MBN), X-ray diffraction (XRD), Microhardness testing (Hm) and light microscopy. Results for WEL thickness, Hm and Mmax have been previously published in [12]

Full size image

Besides higher Mmax values, specimens without WEL had significantly lower FWHM in X-ray diffraction, indicating finer grains inside the WEL. Subsequently, the calculated crystallite size was lower for all of the specimens that had WEL. Following the direct correlation of Eq. (2) between crystallite size calculated by the Scherrer equation and dislocation density, the dislocation density was higher for all of the specimens that had WEL. The observations agree with the findings of Brown et al., who analyzed machining-induced WEL in aeroengine materials. They found that WEL can be detected reliably and rapidly by significant peak broadening in X-ray diffraction. They identify the ultrafine grain size and high lattice strains as the intrinsic properties of WEL, which cause peak broadening [43].

No clear correlation was observed between the micromagnetic and diffractometric results and the thickness of the WEL, reported based on the micrographs given in Fig. 16. A potential reason for this might be found in the periodically changing thickness of the WEL along the bore. In light of the inconstant thickness of the WEL, the micrographs analyzed might not be fully representative of the expansion of the WEL in the subsurface of the specimens. In addition to this, the assessment of the thickness of WEL based on micrographs can be challenging and oftentimes includes a certain bias. This is particularly the case if boundaries between WEL and bulk are not completely clear in micrographs (e.g. Fig. 16, vc = 60 m/min and f = 0.300 mm).

Microhardness in a distance of dsurf = 5 µm significantly increased when WEL were present in the subsurface. The hardness inside of the WEL was found to exceed the hardness of the bulk material by more than three times. Additional microhardness mappings revealed an increase in hardness after BTA deep hole drilling for all cutting speeds and feed rates [12].

X-ray diffraction according to the cos-α method revealed that all specimens had compressive residual stresses in axial direction close to the surface of the bore. Specimens free of WEL had residual stresses in the range of σaxial = – 459 to – 556 MPa. When WEL were present at the surface, the intensity of residual stresses was significantly higher, compared to specimens free of WEL. The highest intensity of compressive residual stresses of σaxial = – 1425 MPa was reported for the specimen, drilled with the highest feed rate and the highest cutting speed. At the same time, this specimen had the highest FWHM and thus led to the lowest size of crystallites. This observation corresponds with the findings of Brown et al., who state that WEL show extreme residual stresses [16].

In Fig. 20 the results of depth profiles of FWHM and residual stresses are displayed for a specimen expected to be free of WEL (Fig. 20a) and a specimen expected to have WEL (Fig. 20b). The FWHM values measured in axial and tangential directions for the specimen which was free of WEL were identical. For the specimen with a WEL the values of FWHM measured in axial and tangential direction are slightly distinct. For both specimens, the values of FWHM decrease with increasing depth. The reason for this might be found in increasing grain size with increasing depth below the surface. For the specimen, expected to have a WEL, there is a particularly strong reduction of the value of FWHM after a distance of d = 10 µm (Fig. 20b). The reason for this sudden drop might be found in the removal of the WEL by this point, since the layer had a thickness of presumably tWEL ≈ 11 µm (Fig. 16). The reduction of the FWHM values of both specimens becomes less pronounced below a depth of approx. tWEL ≈ 20 µm. At a distance of dsurf = 100 µm, the FWHM values almost stay constant. This might indicate, that thermomechanical loads in drilling have an impact on grain size until this approximate depth. However, the FWHM of the specimen with the WEL (Fig. 20b) is still slightly higher at this depth, compared to the FWHM value of specimen free of WEL (Fig. 20a). Thus, a very slight reduction in FWHM can still be expected for the specimen with the WEL below a depth of dsurf = 100 µm.

Fig. 20
figure 20

Depth profiles of residual stresses and full width half maximum (FWHM) for a specimen expected to be free of WEL (a) and a specimen expected to have WEL (b), both batch A. “Distance to surface” indicates the depth of electrolytical polishing

Full size image

The results for residual stresses close to the surface in the axial direction correspond well to the findings presented in Fig. 19. For the specimen free of WEL, the residual stresses in axial and tangential directions are compressive (Fig. 20a). Their intensity decreases constantly until a depth of approx. dsurf = 20 µm, where they reach approximately σ ≈ 0 MPa.

The intensity of axial residual stresses for the specimen with a WEL, only falls slightly within the first 10 µm. After this point, a relatively strong drop in their intensity can be observed. Just like for the drop in FWHM at this point, the reason might be found in the complete removal of the WEL at this point.

The axial residual stresses for this specimen then decrease steadily, until a depth of dsurf = 50 µm. Between dsurf = 50 to 100 µm the axial residual stresses for this specimen remain constant at a level of approx. σ ≈ -300 MPa. Tangential residual stresses for the specimen with WEL are in the tensile range with a maximum intensity of approx. σ ≈ 250 MPa at a depth of dsurf = 5 µm. After this point, they decrease and reach approx. zero at dsurf ≈ 90 µm. The reason for the tensile stresses in the tangential direction will be analyzed in future studies.

Figure 21 displays two exemplary diffractograms obtained by Bruker D8 Discover with a copper tube.

Fig. 21
figure 21

Diffractograms for a specimen expected to be free of WEL (a) and a specimen expected to have WEL (b), both batch A. Magnified regions in the red boxes were recorded with an increased scanning time of t = 10 s/step. The measured peak positions are indicated. They show a slight shift to lower two theta angles compared to stress free steel (α-Fe{110}: 44.670°; α-Fe{200}: 65.019°; α-Fe{211}:82.329°; α-Fe{220}: 98.937°; α-Fe{310}: 116.372°; α-Fe{222}: 137.141° [44])

Full size image

Figure 21a) depicts a specimen expected to be free of WEL (vc = 60 m/min, f = 0.150 mm).The regular α-Fe peaks were observed and no traces of γ-Fe peaks were found. The specimen expected to have a WEL (vc = 80 m/min, f = 0.300 mm) exhibits the same α-Fe peaks, which, however, have slightly less intensity and appear broader. In addition to this, they are shifted to slightly lower 2 theta angles (Fig. 21). This corresponds well to the trends for FWHM and residual stresses, displayed in Fig. 19, since higher compressive residual stresses are a potential reason for peak shifting to slightly lower 2 theta angles. In contrast to the specimen free of WEL, the specimen with WEL revealed traces of γ-Fe peaks, e.g. the γ-Fe peak for {200} (Fig. 21b).

Based on this observation, it can be assumed that austenite was present inside the WEL, indicating austenitization of the subsurface during drilling. Thus, it can be hypothesized that the formation of WEL occurred mainly due to the phase transformation mechanism [16]. This assumption is supported by the temperatures observed in drilling experiments, which exceeded austenitization temperature when using high cutting speeds and feed rates (Fig. 9).

Figure 22 displays the results of the EBSD analysis. Both specimens exhibit significantly refined grains at the very surface. The specimen with the WEL (Fig. 22b) has very small grains up to a distance of approx. dsurf ≈ 12 µm, which corresponds to the expected thickness of the WEL in these specimens (Fig. 16). Beneath this layer, a severely distorted grain structure can be observed, resulting from the material drag during cutting by the cutting edges and burnishing by the guide pads. The specimen free of WEL (Fig. 22a) reveals a similar structure. However, the initial layer of significantly refined grains is smaller, and material drag in the direction of the cut appears less strong.

Fig. 22
figure 22

Results of electron backscatter diffraction (EBSD) for a specimen expected to be free of WEL (a) and a specimen expected to have WEL (b), both batch A. Background correction disabled for the specimen in (a)

Full size image

Figure 23 shows the results of the MFM analysis for the same area depicted in Fig. 22. MFM results reveal that for the specimen with a WEL, no magnetic domain structure can be observed at the very edge of subsurface area until a depth of approximately dsurf ≈ 5 µm.

Fig. 23
figure 23

Results of Magnetic Force Microscopy (MFM) for a specimen expected to be free of WEL (a) and a specimen expected to have WEL (b), both batch A

Full size image

The closer to the surface of the bore, the less phase shift due to magnetic force acting on the Akiyama probe could be measured by MFM. Besides the actual nonexistence of domains, which would be surprising for body-centered cubic α-Fe as detected by EBSD and XRD, the physical reason could also be that they are too small or too weak in the unmagnetized state to be detected by MFM. This is in contrast to measurements in bulk area, but most importantly also to the subsurface area for the specimen expected to be free of WEL. These results support the decrease of Mmax for every single specimen with a WEL compared to specimens free from WEL.

Figure 24 depicts the results of MOKE investigations. The greyscale of the images illustrates the magnetic domains observed under different levels of external magnetization. Under low levels of magnetization, the pictures look light-gray with small dark points, which means that the external magnetic field is not high enough to move the domain walls and evidence the magnetic microstructure of the specimens. With increasing magnetic fields, the magnetic domains can be seen clearly as the contrast between light and dark regions. In both specimens, the shape and size of the highly magnetized regions correspond to areas where the microstructure is formed by α-Fe. Around 5 µm depth from the surface, no dark regions were observed on specimens with WEL. In this region, the microstructure is characterized by nano-sized grains without visible domain structure. An analysis of the whole picture of both specimens at the fully magnetized state leads to the conclusion that the specimen without WEL reaches a higher magnetization level compared to the specimen with WEL. This is evidenced by the fact that the pictures look darker, due to the higher density of magnetic domains associated with the α-Fe phase. The MOKE results correlate well with the MBN investigations, where higher Mmax measurements were reported for the specimen without WEL. The MOKE investigation reveals details of the magnetic structure in a broader area compared to MFM. The MFM results could be interpreted as a magnification of the MOKE pictures without an external magnetic field but with higher resolution and sensitivity. A comparison of MOKE and MFM results concerning the measurements on the 5 µm depth from the surface shows that in both cases there is a low level of magnetization on the nano-sized grains.

Fig. 24
figure 24

Results of the magneto-optical Kerr effect (MOKE) investigations on the subsurface for a specimen expected to be free of WEL (a) and a specimen expected to have WEL (b), both batch A

Full size image

5 Outlook

As shown in Sect. 4.3, it was possible to detect WEL in BTA drilled specimens with the MBN method. This motivated the integration of an MBN sensor into a BTA drill head, to analyze the subsurface in-process. In preparation for the in-process integration, some first results on the influence of workpiece rotation on MBN analysis have been published in [42]. In future studies, such an in-process sensor will be used in a robust process control for BTA deep hole drilling. This will enable the reliable generation of subsurface features in drilling. This in turn will allow for guaranteeing the performance and reliability of components. Some first remarks on this developed sensor will be presented here.

Due to the size of conventional MBN sensors, a significantly smaller sensor was developed in close collaboration with Stresstech Ltd., which was adapted to the geometry of the drill head used. The most important development steps are summarized in Fig. 25. First, the size of the sensor was reduced by dividing the sensor into smaller sub-systems. The second step was the modification of the drill head, to install one sub-system of the sensor. The last step was the mounting of the BTA sensor head on the BTA machine and tests in the laboratory. A detailed account on the design and implementation of this novel sensor for an in-process MBN analysis is given in [29].

Fig. 25
figure 25

Development of a BTA MBN sensorhead [29]

Full size image

6 Conclusion

Various complementary in-process and post-process techniques were employed for a mechanism-based and multi-scale assessment of the interplay between the design of the BTA deep hole drilling process and the resulting surface integrity of bores. Results for normal forces, bore moments, and temperatures in drilling are given. To complement the quite complex and effortful experiments, different concepts for the 2D and 3D numerical simulation of BTA deep hole drilling are presented. SI in the deep drilled bores was assessed extensively using light microscopy, microhardness testing, magnetic Barkhausen noise (MBN) analysis, X-ray diffraction (XRD), electron backscatter diffraction (EBSD), magnetic force microscopy (MFM) and magneto-optical Kerr effect (MOKE) microscopy. Based on the results presented, the main conclusions can be summarized as follows:

  • An increase in feed rate resulted in higher drilling torques and higher normal forces. Subsequently, the temperatures in the interface between the workpiece and cutting edge and the subsurface increased as well with higher feed rates.

  • WEL were observed after BTA deep hole drilling when using high feed rates and cutting speeds. The thickness of WEL alters periodically and in accordance with the feed and cutting speed employed in drilling. The WEL thus appears to have a helical shape along the length of the bore, corresponding to the movement of the tool while drilling.

  • Evidence was provided, that WEL can form in BTA deep hole drilling due to a combination of elevated temperatures and severe plastic deformation. When using high feeds and cutting speeds, temperatures during drilling exceeded austenitization temperature and traces of austenite peaks were found in XRD when a WEL was present at the surface of the bore. This supports WEL formation according to the phase transformation mechanism.

  • In the 3D numerical simulations, the necessary mesh resolution was not achievable, due to the high computational times. This problem could be negligible in the future, when higher computing performance is available.

  • With the 2D numerical simulations, good agreements regarding the cutting forces were achieved. Furthermore, it was possible to extract residual stresses in the subsurface zone. Lastly, the CEL method was introduced for future investigations.

  • WEL can be detected non-destructively by MBN analysis (by lower Mmax and RMS) and by X-ray diffraction (by higher FWHM).

  • In the axial direction, all specimens had compressive residual stresses. Higher intensity of residual stresses was reported for specimens that had WEL. The reasons for tensile residual stresses in tangential direction in a specimen with WEL will be further investigated in the future.

  • Both MOKE and MFM results of the magnetic domain structure support in a correct way the micromagnetic characterization of the specimens, performed using Mmax measurements. Using these techniques, the WEL can be detected due to the lack of magnetization of the nano-sized grains.

The results presented for dislocation densities might have to be taken with a grain of salt, following the simplicity of equations and the limited data available. In future investigations, these results will thus be confronted with scanning transmission electron microscopy (STEM). Electron channeling contrast imaging (ECCI) and transmission Kikuchi diffraction (TKD) will be performed to further elucidate the underlying characteristics of WEL and acquire experimental results for properties such as dislocation densities inside the WEL.

The results form the basis for the integration of an MBN sensor, already shown in this paper, into the drilling process. This MBN sensor will allow for evaluating SI holistically in deep hole drilling. Based on this sensor, a closed-loop control will be developed. In addition to this, the novel miniaturized sensor concept can be transmitted to other applications, e.g. when the accessibility of surfaces is similarly limited as it is in drilling.