In-situ evidence for the existence of surface films in electrochemical machining of copper in nitrate electrolytes

This contribution analyzes the electrochemical etching of copper in a sodium nitrate electrolyte using in-situ Raman spectroscopy. Experiments were conducted in a specially designed process chamber at voltages below, at, and above the limiting current plateau. Below the plateau, a Raman peak at 1050 cm – 1 is detected, which is attributed to the symmetrical nitrate ion. However, the line shape changes at and above the plateau. Detailed line shape analysis indicates the presence of a copper nitrate surface film by ex-situ comparison with copper nitrate solutions. In addition, the conductivity and pH of copper nitrate solutions were determined. A 5 M copper nitrate solution exhibits a low conductivity of 55 mS cm – 1 and a pH close to 0. A significant influence, especially of the low pH, on the surface structure and therefore on the etching mechanism is assumed.


Introduction
Since the beginning of the last century, electrochemical machining (ECM) has been utilized as a non-traditional technique for shaping and structuring of metals [1] in industries such as aerospace [2] and automotive [3].As the understanding of the process deepens, the range of applications is expected to expand.ECM offers several advantages over traditional metal shaping techniques: Very high etching rates can be achieved, it is contactless, there is no tool wear, harmless and inexpensive chemicals are used, and it is applicable to many different metals.In addition, the good surface quality, which is obtained with certain electrolyte-metal combinations, is one of the main advantages of ECM [4].This polishing effect is observed especially with nitrate-based electrolytes at high current densities and is attributed to the presence of aso far only postulatedsurface film [5].The polishing effect is explained by a random dissolution of metal ions into the surface film.It is dependent only on the presence of "vacancies" in the surface film and is therefore independent of grain boundaries or orientation [6,7].It is not clear from the literature when the surface film formation sets in.However, it is likely related to the onset of mass-transport-limitation.
Many different surface film models have been proposed in the literature.They can generally be divided into solid and supersaturated/ viscous films or a combination of both.The solid layers can be either compact or porous and can be oxide-or salt based.The latter implies a precipitation of a salt consisting of electrolyte anions and metal cations [8,9].Liquid, viscous or supersaturated surface films, consisting of highly concentrated metal-electrolyte solutions, have been introduced already in the past century, mainly for phosphoric acid electrolytes [10].Lohrengel et al. postulated the formation of such "melt-like" films in nitrate electrolytes [11].In addition, they provided some evidence against solid product films in nitrate electrolytes by investigating the melting and crystallization behavior of metal nitrates.They showed that molten metal nitrates are metastable and can take minutes to weeks to recrystallize.Due to the short dwell time of the metal ions at the surface, they rule out the presence of solid product films [12].The postulated "melt-like" films could also be described as solvate ionic liquids, which by definition consist of a solvent/ligand complexed metal ion and a counterion [13].This fits well with the low melting points of less than 100 • C of the hydrated metal nitrates [14], and the fact that the first published ionic liquid was nitrate based [15].
To our knowledge, no in-situ evidence for the presence of any of the postulated surface films under ECM conditions at high current densities has been published.Some authors have found ex-situ evidence for the presence of solid CuCl surface films using low-concentration NaCl or HCl electrolytes at low current densities [16,17], which was possible due to the low solubility of CuCl [17,18].Ex-situ studies of liquid nitrate surface films are very difficult, because they are washed away immediately after the etching process is stopped due to the high solubility of the product salts (> 800 g L -1 [14]).
We present in-situ evidence for product surface films during ECM of copper in sodium nitrate electrolyte.The main challenges that have been addressed are (1) the fast moving surfaces during ECM at high voltages, (2) the need for an incident beam perpendicular to the copper surface, which requires a specially designed process chamber, and (3) the presence of sodium nitrate electrolyte, which requires an analytical technique that enables to distinguish between copper nitrate and sodium nitrate.
The third challenge was solved using Raman spectroscopy: An undistorted nitrate ion, due to its symmetric structure (D 3h ), has only three vibrational modes that can be distinguished by Raman spectroscopy.Since sodium nitrate is completely dissociated, the nitrate ion is unaffected in its symmetry.Therefore, Raman investigations of such solutions reveal only the three signals corresponding to the three vibrational stretching modes, up to saturation concentration [19,20].In contrast, in highly concentrated copper nitrate solutions, a part of the nitrate ions is distorted due to association or coordination with the copper ions.This results in additional Raman bands [20], which allow for the detection of a copper nitrate surface film despite the presence of sodium nitrate electrolyte.
Here we present the results of in-situ Raman analysis of electrochemical etching of copper in sodium nitrate solution.In addition, we investigate copper nitrate solutions of different concentrations regarding conductivity and pH to correlate these properties with their possible influence on the ECM process.

Experimental
For the in-situ Raman measurements, a process chamber was designed that allows the Raman beam to be incident perpendicular to the copper surface and the Raman light to be detected in the backscattered configuration.A scheme and an image are shown in Fig. 1.A two-electrode configuration was used.A copper wire with a diameter of 1 mm was used as the anode, the electrode distance was 1.3 mm, and the flow channel has a rectangular cross-section of 1.3 × 10 mm 2 .A flow rate of 0.4 m s -1 was applied by a peristaltic pump.Sodium nitrate electrolyte (200 g L -1 , 2.35 M) at pH 7 was used as prepared for the insitu experiments, and sodium nitrate and "copper nitrate trihydrate" at different concentrations were used for the ex-situ measurements (Raman, conductivity, and pH).All experiments were conducted at room temperature.Raman scattered light is excited with 532 nm laser light and detected with a 10× lens with NA = 0.25.The light is dispersed by a 600 g mm -1 grating and detected with a Silicon CCD centered at 1000 cm -1 (very similar Raman setup to [21]) .Integration times up to 1 s were used.All measurements were conducted at least two times.The experimental details of the structural analysis are given in the supporting information.

Raman measurements
This chapter consists of two parts.In the first part, ex-situ measurements of copper nitrate and sodium nitrate were conducted as reference measurements for the second part, which contains the in-situ measurements during ECM.An exemplary full Raman spectrum of copper and sodium nitrate is given in Fig. S1 in the supporting information.For all experiments, the most intense peak at about 1000 cm -1 was analyzed.The peak was fitted by two Lorentz peaks representing the two peaks measured in copper nitrate (according to [19]).

Ex-situ measurements
Two types of ex-situ Raman measurements were conducted.The first experiment investigated solutions of sodium nitrate (2.35 M) and copper nitrate (5 M) by Raman spectroscopy.The results are shown in Fig. 2. The spectrum of the sodium nitrate solution exhibits only one symmetrical peak at approximately 1053 cm -1 (υ 1 (A 1 `) = symmetric NOstretch of NO 3 -) and is consistent with previous literature findings [20,22].In contrast, the spectrum of the copper nitrate solution exhibits an asymmetric peak, indicating a second peak/shoulder at lower frequency.This suggests two different types of nitrate ions to be presentcausing a symmetric and a distorted NO-stretch that is weakened and redshifted by coordination to a Cu 2+ -ion.The intensity of this peak is expected to increase with increasing copper nitrate concentration [23].3. The dark green curve, obtained from solid copper nitrate, displays two distinct peaks.According to the literature, the peak at 1053 cm -1 originates (again) from a symmetric nitrate ion [20], while the peak at 1023 cm -1 is caused by a less symmetric nitrate moiety.The intensity of the band at 1023 cm -1 decreases with progressing "dilution" of the solid copper nitrate.As stated above, this observation is in agreement with the reports in the literature, which indicate a decreasing degree of asymmetry with decreasing copper nitrate concentration [23].
To investigate the nitrate ions in solid copper nitrate further, we conducted a complementary structural analysis of copper nitrate-2.5hydrateat 100 K and 300 K.This analysis confirmed the results of Fig. 1.Schematic drawing of the process chamber and image of the process chamber inside the Raman spectrometer.
L. Jakob et al.
Morosin [24] and Garaj [25,26], which showed that only 2.5 water molecules are present in the structure, in contrast to the common assignment as a trihydrate.Our results verify the structure obtained by Morosin at room temperature [24], and the main findings are summarized in Table 1.The structures at 100 K and 300 K exhibit no significant differences.The copper ion is coordinated in a square-planar fashion by two nitrate ions and two water molecules in trans configuration, each via an oxygen atom.Another outer-sphere water molecule is located on a special position and is thus present only "half" within the structure.
Both the nitrate ions and the water molecules exhibit different coordination environments and are thus structurally different.One of the nitrate ions exhibits only one contact to a water molecule from another trans-[Cu(NO 3 ) 2 (H 2 O) 2 ] unit, while the other nitrate ion is involved in a more complicated hydrogen-bonding network.This situation results in the presence of a more symmetric nitrate ion (involving N1, O-N bond lengths between 1.2365(8) and 1.2775(8) Å, as shown in Table 2) and a less symmetric one (involving N2, O-N bond lengths between 1.2171(8) and 1.3005(8) Å) within the structure of Cu(NO 3 ) 2 • 2.5 H 2 O both at 100 K and 300 K.
Compared to the calculated N-O bond lengths of water-coordinated nitrate (1.256 [27] / 1.26 [28] Å), the deviations in the second nitrate ion are doubled compared to the first nitrate ion.The same trend holds true for the O-N-O bond angles, which deviate twice as much from the totally symmetric nitrate ion (120 • ) for the second nitrate ion compared to the first.
The occurrence of two peaks in the Raman spectrum of Cu(NO 3 ) 2 • 2.5 H 2 O can be explained by these differences.As melting/dilution proceeds, the number of asymmetric copper-bound nitrate ions decreases, resulting in a reduction of the intensity of the second peak at 1023 cm − 1 .From these preliminary results, it can be concluded, that an asymmetric peak or even a second peak around 1050/1023 cm -1 is a clear hint for the presence of highly concentrated copper nitrate.

In-situ measurements
The process chamber described in the experimental section was used for the in-situ measurements.A preliminary experiment was conducted to analyze the current-voltage behavior in this setup and identify suitable processing voltages.The resulting curve is plotted in Fig. 4. The graph illustrates the typical behavior of a linear increase at low voltages, a limiting current plateau over several volts, and a steep increase in current density at high voltages [29,30].It is important to note that this steep increase is not due to an increased proportion of oxygen evolution at the anode, but to strongly accelerated metal dissolution [5].For the in-situ measurements, voltages of 2 V (below the limiting current density), 12 V (at the limiting current density plateau), and 18 V & 22 V (in the region of the steep rising current) were selected.According to previous theories, surface films should be present at very high currents / voltages (18 V and 22 V in our curve) [5], but there is no prediction for low and intermediate potentials (2 V and 12 V in our curves).The authors suggest that the mass transport limitation in the plateau region is likely due to the formation of the copper nitrate surface film, which would result in an asymmetric peak at the process voltage of 12 V.For the same reason, it is not expected that a surface film will form at 2 V.
As the analyzed surface moves downwards, due to the proceeding   etching of the copper, it was decided to perform the measurements according to Fig. 5. Prior to applying the external voltage, the Raman microscope`s field of focus (with a FWHM of approximately 12 µm in depth) was set approximately 50 µm below the copper surface.The absolute height level of the focus remains constant throughout the entire experiment.However, as the copper is etched, the copper surface moves through the field of focus of the objective lens.This causes the part of the sample from which scattered Raman signal is detected to change from copper (Fig. 5a) to a mixture of copper and copper nitrate surface film (Fig. 5b).As etching continues an increasing amount of the electrolyte is measured in addition to the surface film (Fig. 5c).Finally, in Fig. 5d, only the electrolyte is measured after the surface film has passed the focused area.Due to the low thickness of the surface film, which is postulated to be only several µm [11,31], the measurable spectrum is expected to always be a mixture of the surface film and electrolyte.If no surface film is present, only the signals stemming from the electrolyte should be measurable and should not contain any indication of the 1023 cm -1 -shoulder in the Raman signal.In principle it is possible that a copper oxide layer is formed anodically on the copper surface.However, a very low signal intensity is expected due to the low layer thickness.Therefore, no attempt was made to analyze the Raman spectra regarding copper oxide.Fig. 6 shows the results of the in-situ experiments conducted at 22 V, 18 V, 12 V and 2 V.The signal at approximately 1050 cm -1 is fitted with two Lorentz-Peaks, corresponding to the two peaks obtained in the preliminary experiments (Fig. 2b) [19].The area of the two peaks is plotted against the process time.Initially, the overall intensity is very low for all processing voltages, as the laser beam is primarily focused on and into the copper.
At 22 V and 18 V, an asymmetry in the nitrate vibration appears almost immediately after the voltage was switched on, lasting for approximately 10 and 15 s, respectively.The longer duration at 18 V is likely due to the lower current density and resulting lower etching rate.The intensity of the main peak, which originates from a mixture of surface film and pure electrolyte, begins to increase around the midpoint of the "surface film signal".The development of the two signal intensities is in line with the generally expected course, as shown in Fig. 5.
However, the "surface film signal" persists longer than expected for a surface film thickness of 1 -2 µm, given that the etching front moves at a rate of approximately 10 -15 µm s -1 .The surface film thickness is significantly affected by various parameters, particularly the electrolyte flow rate, resulting in substantial differences depending on the system applied.The low electrolyte flow rate in our case promotes the formation of thicker surface films.A smearing out of the surface film, meaning a continuously decreasing concentration of copper nitrate towards the electrolyte, is expected, which complicates the determination of a precise thickness even more.Determining the precise depth of focus of the laser, which is larger than 12 µm, is difficult due to its strong dependence on the refractive index of the measured media.Additionally, an unknown amount of side reactions introduces further uncertainties.Therefore, our results do not allow for a quantitative prediction of the surface film thickness.However, the data suggests that the surface film in our setup is thicker than 2 µm and may even exceed the previously postulated range by far, especially when including the "smear out region".Accurate determination of the film thickness requires further investigations and precise measurements.
It is also difficult to make quantitative statements regarding the concentration of the surface film.The maximum ratios between the shoulder and the main peak areas are 0.13 and 0.09 for 18 V and 22 V, respectively.These values are likely a result of a mixture of surface film and electrolyte rather than the pure surface film, as the Raman focus is quite broad.This is expected to reduce the ratio, which is still greater than for the ex-situ measurement (1:10 or 0.1, see Fig. 2b).This gives a

Table 2
Bond lengths and angles of the nitrate ions in Cu(NO 3 ) 2 • 2.5 H 2 O measured at 300 K and 100 K (the latter values in brackets).strong hint, that the surface film concentration is exceeding 5 M. Overall, our direct measurements confirm important parts of the postulated supersaturated surface film hypothesis [11].
At the plateau at 12 V the curves exhibit qualitatively the same behavior as above the plateau at higher voltage.The etching proceeds slower at 2 -3 µm s -1 due to the lower current density, leading to a later and longer appearance of the shoulder.This result demonstrates that the copper nitrate surface film is already present at the mass transport limiting plateau.The maximum ratio (shoulder:main peak) is slightly smaller compared to higher voltages (0.07), which is a weak indication for a less concentrated surface film in the plateau region.
The results obtained at 2 V processing voltage confirm the absence of a highly concentrated copper nitrate film below the limiting current plateau.Throughout the entire experiment, only the symmetric nitrate stretch was detected, with no visible shoulder.Hence, the etch rate is low enough to prevent the accumulation of dissolved copper at the anode interface.

Ex-situ investigation of copper nitrate
The pH and conductivity for sodium nitrate and copper nitrate solutions of varying concentrations were measured and the results are displayed in Fig. 7.The pH of the sodium nitrate electrolyte solution is slightly alkaline and almost constant over the measured concentration range.In contrast, the copper nitrate solutions exhibit an increasingly acidic pH with increasing concentration, reaching an apparent pH of nearly 0 in a 5 M solution.The high positive charge density of the copper ion causes a strong polarization of the coordinated water molecules, Fig. 6. Results of the in-situ Raman measurements at four different processing voltages.The plotted intensities correspond to the areas of the two fitted Lorentz peaks according to the symmetric and the distorted nitrate ions, as shown in Fig. 2b.The background colors refer to Fig. 5.In the red area, the focus is mainly in the copper, in the green area, the surface film is focused and in the blue area only the electrolyte is in focus.The time axis varies between 22 V / 18 V and 12 V / 2 V due to strongly increased etching rates at higher potentials.resulting in a high chemical potential of the oxygen-hydrogen bonds in the water molecules.The presence of a highly concentrated copper nitrate surface film during ECM leads to a highly acidic pH at the metal surface.To our knowledge, the surface models proposed in the literature [11], have neglected this fact.Differing results are reported on oxygen evolution during high-rate copper dissolution in the literature [11,30].It is, therefore, not clear if oxide films are present.The acidic surface film could significantly impact the presence and stability of metal oxides underneath supersaturated surface films, and therefore the oxygen evolution.We propose that the potentially present anodic oxides are subject to a dynamic situation including anodic formation and chemical dissolution.It can be concluded that the pH of the bulk electrolyte is likely of minor importance in the presence of the supersaturated and acidic surface film.
The conductivity of sodium nitrate increases steadily up to the highest concentration measured.However, for copper nitrate, the conductivity increases up to about 2.1 M, after which it begins to decrease.This behavior has been reported for other metal nitrates by Schneider and Lohrengel [4], who measured metal nitrate solutions at even higher concentrations, resulting in low conductivities of almost 0 mS cm -1 .Further investigation is required to determine the impact of a highly resistive layer on the electrochemical machining system.

Conclusion
This publication presents experimental evidence for the postulated supersaturated metal-nitrate surface film formed during high-rate metal dissolution in sodium nitrate electrolyte.Additionally, it contributes to a deeper understanding of the influence of this surface film on the electrochemical etching process.Raman spectroscopy enables to distinguish between sodium nitrate (free nitrate) and copper nitrate (free and copper-coordinated nitrate) due to the partial distortion of the symmetry of the nitrate ion in highly concentrated copper nitrate solutions.In-situ Raman spectroscopy confirmed for the first time the presence of a copper nitrate film at and above the limiting current plateau.The in-situ signal exhibits a clear asymmetry, as expected for highly concentrated copper nitrate solutions based on preliminary ex-situ measurements.While, a quantitative assessment of thickness or concentration of the surface film is not feasible, it is qualitatively evident that the thickness is well above the previously postulated 1 -2 µm, and the concentrations are higher than 5 M. In addition, ex-situ measurements of conductivity and especially pH can provide valuable information on the potential influence of the surface film on the etching mechanism.The highly concentrated copper nitrate solutions have very low pH, which may directly impact the stability of copper oxide at the copper surface.It is unlikely, that a stable oxide film will form under these conditions.These findings are likely applicable to other metals.
The deeper understanding of the mechanisms of electrochemical etching, especially the surface situation at the anode, facilitate further technological progress of ECM in the future.In particular, the transportlimiting nature of surface films can be utilized to adjust the processes towards the desired process goal.Such contributions can govern the current distribution in the process and, therefore, the local etching rate.Further investigations in this direction will be the subject of future research in our group.
Fig. 2b visualizes the fitting of the asymmetric Cu(NO 3 ) 2 peak by two Lorentz peaks, representing the symmetric (violet) and distorted (blue) nitrate ions.The fit slightly overestimates the intensity on the right edge of the peak, resulting in an underestimation of the shoulder.The area ratio of the two fits is 1:10 (shoulder:main peak).To confirm the findings, a crystal of Cu(NO 3 ) 2 • 2.5 H 2 O was analyzed via Raman spectroscopy for several minutes under ambient conditions.Due to the hygroscopic nature of copper nitrate, the crystal converts first into Cu(NO 3 ) 2 • 6 H 2 O and finally into a liquid copper nitrate solution/melt.The corresponding Raman spectra are shown in Fig.

Fig. 2 .
Fig. 2. (a) Ex-situ Raman measurements of sodium nitrate (2.35 M) and copper nitrate (5 M) solutions.(b) Measured signal of copper nitrate (5 M) and the total fitted curve (green) as addition of two separate Lorentz fits according to the symmetric (violet) and the distorted (blue) nitrate ions.

Fig. 3 .
Fig. 3. Time series of the "melting-process" of a Cu(NO 3 ) 2 • 2.5 H 2 O crystal transforming first to Cu(NO 3 ) 2 • 6 H 2 O and finally to liquid copper nitrate solution/copper nitrate melt at ambient conditions.Time progresses from green to red over roughly 5 min.

Fig. 4 .Fig. 5 .
Fig. 4. Current density -voltage behavior of the etching process in the specially designed process chamber.The voltages chosen for the Raman investigations are marked.

Fig. 7 .
Fig. 7. Conductivity and apparent pH of various sodium nitrate and copper nitrate solutions.

Table 1
Cell parameters of the crystal structure of Cu(NO 3 ) 2 • 2.5 H 2 O measured at 100 K and 300 K.