1 Introduction

The study of thin-film carbon has a long and rich tradition [1,2,3,4,5]. Scientific analyzes of the thin-film carbon genome have primarily focused on characterizing the material properties associated with the various forms of this material [6,7,8]. Industrially oriented research, however, has tended to center on the plethora of applications associated with thin-film carbon, including its use in the automotive and drilling/milling industries, where it has been used to increase product lifetime [9,10,11,12,13]. While scientific reports on the material properties of thin-film carbon abound, much less has been written about the use of thin-film carbon for industrial applications. This is undoubtedly due to concerns related to the development and ownership of the intellectual property stemming from such research. We further speculate that the complexities associated with industrial applications, wherein a confluence of many concomitant factors are at play, may also act to further dampen the scientific interest into such matters, scientific studies tending to be performed under more pristine conditions.

Unfortunately, this has created a schism in the thin-film carbon field. Samples prepared for scientific analyzes are deposited onto planar substrates [14,15,16,17]. In contrast, for many industrial applications, non-planar substrates are frequently encountered, the thin-films of carbon employed for such applications having to conform with the geometric form of the underlying substrate. While it is often just implicitly assumed that the properties of thin-film carbon are invariant to the geometry of the underlying substrate, some research suggests that this is not the case [18,19,20]. Unfortunately, given the paucity of research performed on this topic, much remains unknown concerning how the properties of thin-film carbon are shaped by the geometry of the underlying substrate.

We opine that this deficiency in the scientific literature arises from three possible factors. (1) First, there may be a lot of self-censoring of the scientific record. Scientists who produce poor results (thin-films that poorly adhere to an underlying substrate, for example) are naturally reluctant to share them with the scientific community. It may be even worse than that. It is possible that authors who obtain fragile thin-films are reluctant to report this fact in a research article, as they may view that such a disclosure might potentially cast doubt on the credibility of the rest of the results that they are reporting. Unfortunately, through this self-editing of the scientific record, critical components are not documented. (2) Another reason may be scientific prejudice. Pure scientists, who want to focus on the most scientifically “important” issues, may view adhesion issues as being not “important” enough to in order to be worth their effort. They may view such issues as being more in the domain of industrially oriented research. (3) Finally, industries who sponsor research on adhesion may be unwilling to share results in the open literature out of concerns for their ownership over the related intellectual property. As a result of these factors, an incomplete picture emerges, even from a thorough read of the relevant current scientific literature.

Hints at a dependence of the properties of thin-film carbon on the geometry of the underlying substrate are found in research aimed at identifying the role that the substrate orientation angle plays in shaping the properties of the resultant forms of this material.Footnote 1 For the case of rf magnetron sputtering, the dependence of the properties of thin-films of carbon on the substrate deposition angle have been examined, in detail, by Bobzin et al. [18]. As rf magnetron sputtering is a physical vapor deposition (PVD) process, one might expect variations in the nature of the resultant thin-films of carbon as the orientation angle is varied, this variation reflecting the change in the direction of the flux of the incident carbon atoms with respect to the position of the substrate. A change in thickness, adhesion, and hardness, were in fact observed in the aforementioned study [18]; Bobzin et al. [18] find that angled substrates are found to be thinner (when prepared without hydrogen), between 20 and 30% harder, and exhibit a higher Young’s modulus (by about 20–30%) when contrasted with their planar counterparts. For the case of plasma enhanced chemical vapor deposition (PECVD), however, given that the deposition process is more chemical in nature, i.e., determined by chemical reactions between the growth surface and the deposition gases, one might expect thin-film carbon properties that are independent of the substrate orientation. Nevertheless, Nelson et al. [19] demonstrated that thin-films of carbon deposited through PECVD on horizontally placed substrates exhibit very different properties, in terms of surface roughness, thickness, and chemical bonding, from those deposited onto vertically placed substrates. Of particular note, Nelson et al. [19] find that angled substrates exhibit less roughness (sometimes by an order of magnitude) when contrasted with their planar counterparts, and that this varies with the deposition conditions, variations in the Raman spectra also being found to be significant. Clearly, in spite of its chemical nature, there is an underlying element of physicality found in the PECVD process, and this is influencing the form of the resultant thin-films of carbon. Further evidence in support of the important role that the substrate geometry plays in shaping the properties of a resultant thin-film of carbon is provided in the experimental results of Dowey et al. [20], wherein it was shown that thin-films of carbon grown through PECVD into the interior surface of stainless steel tubes, of different diameters, exhibit forms that are critically dependent upon the tube diameter and the deposition pressure. In particular, it was found that thin-film carbon coatings that extend deeper into the tubes were achieved for decreased tube diameters. These results collectively suggest that there may be some dependence of the form of thin-film carbon on the geometry of the underlying substrate.

In this paper, we will directly examine if the geometry of the substrate plays a role in shaping the form of the resultant thin-film of carbon. For the purposes of this study, we focus on how the substrate geometry influences the adhesion quality, which has yet to be considered in this context. Depositions of thin-films of carbon onto curved, i.e., rings, and flat substrates will be considered. We employ a titanium alloy with aluminium and vanadium, Ti6Al4V, for the purposes of this investigation [21]; this alloy, which is commonly found in industrial applications, corresponds to a mixture of titanium, aluminum, and vanadium atoms, the aluminum and vanadium weight percentages corresponding to about 6 and 4%, respectively. In actual industrial applications, an interlayer, or adhesion layer, is commonly introduced for the case of thin-films of carbon deposited onto Ti6Al4V substrates [22,23,24,25]. An interlayer adds to the complexity of the thin-film carbon characterization, and thus, has not been employed for the purposes of this work in order to allow for an observation of the resultant thin-films of carbon in their most basic form. The quality of the adhesion of the resultant thin-films of carbon will then be assessed through the application of the Scotch tape adhesion test; the Scotch tape adhesion test has been widely applied in the assessment of the adhesion of thin-films onto an underlying substrate [26,27,28,29,30,31,32], adhesion within forms of thin-film carbon long being the subject of investigation [33,34,35,36,37]. The Scotch tape adhesion test is only applicable to thin-films of carbon without an adhesion layer, this test not being applicable to thin-films which bond strongly to the underlying substrate, such as those that result as a consequence of the presence of an adhesion layer, i.e., the Scotch tape adhesion test lacks the required sensitivity. Rf magnetron sputtering and PECVD, two of the most common deposition approaches used for thin-film carbon preparation, are employed for the preparation of the samples considered in this work. The deposition parameters used in this work are representative of those found elsewhere in the literature [38, 39]. Finally, the vibrational spectra, corresponding to each thin-film of carbon, are assessed using Raman spectroscopy.

It should be noted that this paper does not aim to “optimize” the deposition parameters for thin-film carbon in order to get the best adhesion possible. Rather, it aims to explore as to whether or not the adhesion of “normal” forms of thin-film carbon is influenced by the underlying substrate geometry. As a consequence, a detailed consideration of the deposition conditions is not pursued for the purposes of this analysis. Rather, we demonstrate that our films exhibit properties, or are prepared using deposition conditions, that are representative of “normal’ thin-films of carbon and assess as to whether or not the resultant adhesion is dependent upon the geometry of the underlying substrate.

This paper is organized in the following manner. In Sect. 2, the experimental work, i.e., the sample preparation and measurements performed, are discussed. Results are then presented in Sect. 3. Finally, the conclusions of this study are summarized in Sect. 4.

2 Experiment

Thin-films of carbon are deposited onto Ti6Al4V substrates without an adhesion layer. Both rf magnetron sputtering and PECVD deposition processes are employed for the purposes of this analysis. For each deposition approach considered, a deposition onto a curved ring, with a radius of about 10 mm, and a deposition onto a round flat substrate, with a diameter of about 25 mm, are performed. Prior to each deposition, the substrates are first mechanically polished with a polishing compound on a buffing wheel. Next, each substrate is dipped into an ultrasonic bath of distilled water with 5% all purpose cleaner (TSP) for about 20 min. Following that, each substrate is dipped into an acetone ultrasonic bath for about 2 min. To allow for the streak-free drying of the substrate surface, the substrates are moved in the air in a circular motion. Detailed surface characterizations are not performed prior to the depositions.

Samples 1 and 2 correspond to those deposited through the use of conventional rf magnetron sputtering onto curved and flat Ti6Al4V substrates, respectively. A turbo pump is employed for the purposes of these depositions. Owing to the spatial constraints associated with our rf magnetron sputtering system, each deposition is performed separately. Both the curved and flat substrates are positioned at a distance of about 60 mm from the graphite target. The curved substrate is mounted onto a rotating rod with a 2 rpm rotational speed, while the flat substrate is held in a static position; a rotation of the flat substrate may have improved the uniformity and the adhesion, but our magnetron sputterer did not have the required rotational axis in order to perform such a rotation. The base pressure is 130 µTorr for both depositions. Prior to each deposition, the substrate surfaces are further cleaned in an argon plasma for about 30 min. The deposition pressure is kept at about 7 mTorr for both cases for the entire deposition process, the argon gas flow rate being set to 12.5 sccm throughout. Each deposition is performed over a period of exactly 120 min. The temperature is kept below 70 °C during each deposition [38]; the temperature is measured using a Kurt J. Lesker temperature gauge that was positioned about 20 mm from the substrate where the thin-film growth occurred, the substrate temperatures, as evaluated by this temperature gauge, being found to range between 30 and 70 °C for all depositions, these substrate temperatures being representative of that found elsewhere in the field (see, for example, Schwan et al. [38]). These deposition parameters are summarized in Table 1.

Table 1 Deposition parameters for all four of the samples considered in this study
Full size table

Samples 3 and 4 correspond to those deposited through PECVD onto curved and flat Ti6Al4V substrates, respectively. A small and economical PECVD system is employed for this purpose, and no turbo pump is used; a rotary vane pump system is employed for the purposes of these depositions. The depositions of both of these samples occurred at the same time. Both the curved and flat substrates are hung onto a mounting tree about 25 mm off the chamber bottom and are evenly distributed within the vacuum chamber. The base pressure is 25 mTorr for the purposes of this deposition. Prior to the deposition, the substrate surfaces are further cleaned in a hydrogen and argon plasma mixture for about 20 min; this cleaning process is performed uniformly over the entire substrate surface. The deposition pressure is kept at about 400 mTorr during the entire deposition process with argon and hexane gas flows of about 12 sccm each. The deposition is performed over a period of exactly 30 min. These deposition parameters are summarized in Table 1; the substrate temperature was not measured during our PECVD depositions as no temperature gauge was available on the PECVD deposition system that is employed for the purposes of this study.

All four samples are tested for their adhesion. Our first assessment on the quality of the adhesion is based on the visual appearance and the response of each thin-film to manual handling; nitrile gloves are used in order to handle the samples, such gloves providing a good grip on the thin-film carbon surface while preventing dirt contamination. Next, the Scotch tape adhesion test is applied. As has been mentioned before, the Scotch tape adhesion test is the standard approach used for adhesion assessment in the field [9, 40, 41]. To apply the test, a Scotch tape is firmly applied onto the thin-film of carbon and, after 2 min, sharply removed. This Scotch tape adhesion test is performed at room temperature.

To ascertain if there are differences in the nature of the chemical bonds that are present for the different thin-films of carbon considered, the vibrational spectra are examined through the use of Raman spectroscopy. The Raman spectra are acquired through the process presented by Laumer et al. [24]. Each thin-film of carbon is excited with 5 mW of power using a He–Ne laser, the wavelength being set to 633 nm and the magnification being set to 100. The Raman spectra are acquired over the spectral range from 900 to 2100 cm−1. For visualization purposes, the Raman signals are smoothed out with a local regression algorithm with a span of 10%. Baseline correction and smoothing, following the approaches outlined by Laumer et al. [24] and Laumer and O’Leary [25], are performed. The smoothing algorithm employed uses a weighted linear least-squares model.

3 Results and discussion

Figure 1 presents visual images corresponding to a sample of thin-film carbon deposited onto a curved Ti6Al4V substrate through the use of rf magnetron sputtering, i.e., Sample 1. The untreated thin-film of carbon is depicted in Fig. 1a, and shows the state of the thin-film immediately following deposition. The sample is then handled and scrubbed with nitrile gloves to test if some of the thin-film peels off. No film detachment is observed, however, a residue, probably associated with the use of the nitrile groves, is apparent, as may be seen in Fig. 1b. The Scotch tape adhesion test is then performed. Based on our visual observations, half of the film is removed with this first Scotch tape, as is shown in Fig. 1c. A subsequent application of the Scotch tape adhesion test is noted to remove most of the remaining thin-film. The final resultant form of Sample 1, following the application of the second Scotch tape adhesion test, is depicted in Fig. 1d.

Fig. 1
figure 1

a A thin-film of carbon deposited onto a curved Ti6Al4V substrate without an adhesion layer, i.e., Sample 1. The deposition is performed using an rf magnetron sputtering system with a turbo pump. This image represents the thin-film of carbon right after the deposition without any handling. The online version is depicted in color. b A thin-film of carbon deposited onto a curved Ti6Al4V substrate without an adhesion layer, i.e., Sample 1. The deposition is performed using an rf magnetron sputtering system with a turbo pump. This image represents the thin-film of carbon after rough handling, i.e., dropping onto a table, and rubbing with nitrile gloves. The online version is depicted in color. c A thin-film of carbon deposited onto a curved Ti6Al4V substrate without an adhesion layer, i.e., Sample 1. The deposition is performed using an rf magnetron sputtering system with a turbo pump. This image represents the thin-film of carbon after the first Scotch tape adhesion test. The online version is depicted in color. d A thin-film of carbon deposited onto a curved Ti6Al4V substrate without an adhesion layer, i.e., Sample 1. The deposition is performed using an rf magnetron sputtering system with a turbo pump. This image represents the thin-film of carbon after the second Scotch tape adhesion test. The online version is depicted in color.

Full size image

A thin-film of carbon deposited onto a flat Ti6Al4V substrate, i.e., Sample 2, is prepared with the same deposition conditions as that employed for the preparation of Sample 1. The untreated thin-film of carbon is shown in Fig. 2a. Immediately following deposition, the film started blistering off and detached from the underlying substrate. Any touching or handling of the thin-film led to the removal of the thin-film of carbon from the underlying substrate. A Scotch tape adhesion test removed the film completely, as shown in Fig. 2b. Clearly, there is a difference in the quality of the adhesion between thin-films of carbon prepared onto curved and flat substrates for the case of rf magnetron sputtering, even though the depositions conditions are identical. This difference is believed to be related to the geometric form of the underlying substrate.

Fig. 2
figure 2

a A thin-film of carbon deposited onto a flat Ti6Al4V substrate without an adhesion layer, i.e., Sample 2. The deposition is performed using an rf magnetron sputtering system with a turbo pump. This image represents the thin-film of carbon right after the deposition without any handling. The online version is depicted in color. b A thin-film of carbon deposited onto a flat Ti6Al4V substrate without an adhesion layer, i.e., Sample 2. The deposition is performed using an rf magnetron sputtering system with a turbo pump. This image represents the thin-film of carbon after the Scotch tape adhesion test. The online version is depicted in color

Full size image

Figure 3a shows a thin-film of carbon deposited onto a curved Ti6Al4V substrate using our economical PECVD system without a turbo pump, i.e., Sample 3. We note that this deposition produces a thin-film in the middle of the substrate area, but that no film is observed towards the edges of the curved substrate. Nelson et al. [19] suggest that sharp edges, such as those near the edge of the ring in Sample 3, enhance the electric field strength there during the PECVD deposition. Enhanced electric field strength leads to greater ion energies, which have been suggested by some to lead to an etching effect. We suspect that this etching may account for the absence of film at the ring edges of Sample 3; Sun et al. [42] envisage the deposition process that occurs within the scope of a thin-film carbon PECVD deposition process as corresponding to a balance between deposition and etching, Kim and Lee  [17] presenting observations which support this supposition of Sun et al. [42]. The resultant thin-film of carbon has been shown to exhibit good adhesion, no film being removed with the Scotch tape adhesion test, as is depicted in Fig. 3b.

Fig. 3
figure 3

a A thin-film of carbon deposited onto a curved Ti6Al4V substrate without an adhesion layer, i.e., Sample 3. The deposition is performed using a PECVD system without a turbo pump. This image represents the thin-film of carbon right after the deposition without any handling. The online version is depicted in color. b A thin-film of carbon deposited onto a curved Ti6Al4V substrate without an adhesion layer, i.e., Sample 3. The deposition is performed using a PECVD system without a turbo pump. This image represents the thin-film of carbon after the Scotch tape adhesion test. The online version is depicted in color

Full size image

The thin-film of carbon deposited onto a flat Ti6Al4V substrate through the use of PECVD, i.e., Sample 4, using the same deposition conditions as for its curved counterpart, i.e., Sample 3, exhibits a similar form, and is depicted in Fig. 4a. Parts of the film peel off due to reasons that were not investigated in this study and remain unexplained at the present moment. The remaining thin-film of carbon exhibits a similar adhesion quality as the thin-film of carbon deposited onto the curved substrate, i.e., Sample 3. No thin-film is removed with the Scotch tape adhesion test, as is visually seen in Fig. 4b.

Fig. 4
figure 4

a A thin-film of carbon deposited onto a flat Ti6Al4V substrate without an adhesion layer, i.e., Sample 4. The deposition is performed using a PECVD system without a turbo pump. This image represents the thin-film of carbon right after the deposition without any handling. The online version is depicted in color. b A thin-film of carbon deposited onto a curved Ti6Al4V substrate without an adhesion layer, i.e., Sample 4. The deposition is performed using a PECVD system without a turbo pump. This image represents the thin-film of carbon after the Scotch tape adhesion test. The online version is depicted in color

Full size image

The vibrational response of the different samples is now examined through the use of the Raman spectroscopy. Figure 5a, b depict the Raman spectra corresponding to three of the samples (Samples 2, 3, and 4) and that corresponding to another sample that was prepared using the same desposition conditions as those employed for the preparation of Sample 1 (also deposited onto a curved substrate).Footnote 2 Figure 5a corresponds to the smoothed Raman spectra while Fig. 5b corresponds to the baseline corrected smoothed Raman spectra. These spectra are acquired on different spots, randomly selected over the sample surface. A comparison of the Raman spectra corresponding to the curved and flat substrates prepared through the use of rf magnetron sputtering, i.e., Samples 1 and 2, respectively, is given in the lower half of Fig. 5a, b, while that corresponding to the samples prepared through PECVD, i.e., Samples 3 and 4, is given in the upper half of Fig. 5a, b. It is noted that the smoothed baseline corrected Raman spectra that we depict in Fig. 5b are of a similar form to those found elsewhere in the thin-film carbon literature, and we take this to suggest that our thin-films of carbon are representative of that found elsewhere in the thin-film carbon field, at least with respect to their Raman spectra [43, 44]. It is observed that the variation in the Raman signal within Sample 1 itself seems comparable with the variation from Sample 1 to Sample 2. This suggests that there is very little variation in the form of the chemical bonding within our rf magnetron sputtering deposited thin-films of carbon. The same can be observed for our samples deposited through PECVD, i.e., Samples 3 and 4. That is, variations in the Raman signals found within Sample 3 are found to be comparable with the variations observed between Samples 3 and 4. Once again, there appears to be very little difference in the nature of the chemical bonding between these thin-films of carbon. What is noticeable, however, is the difference in the form of the Raman spectra associated with the thin-films of carbon deposited using the different techniques. We suspect that differences in the deposition pressures employed, i.e., our rf magnetron sputtering deposition process used a lower deposition pressure than our PECVD deposition process, the presence of hydrogen in the PECVD process, and other as yet to be identified factors, may be responsible for this difference in the form of the Raman spectra; the underlying Ti6Al4V substrate is expected to play a relatively minor role in these Raman results as: (1) our thin-films are estimated to be thicker than the penetration depth of the Raman source signal used on our thin-films of carbon, and (2) Raman measurements performed on bare Ti6Al4V substrates yielded a very weak response.

Fig. 5
figure 5

a Several Raman spectra corresponding to randomly selected locations on the thin-films of carbon deposited onto curved and flat Ti6Al4V substrates without an adhesion layer, i.e., Samples 1, 2, 3, and 4. The depositions of Samples 1 and 2 are performed using an rf magnetron sputtering system with a turbo pump, while those corresponding to Samples 3 and 4 are performed using a PECVD system without a turbo pump. These spectra have been smoothed but not baseline corrected. In reference to the Sample 1 results, please note footnote 2. The online version is depicted in color. b Several Raman spectra corresponding to randomly selected locations on the thin-films of carbon deposited onto curved and flat Ti6Al4V substrates without an adhesion layer, i.e., Samples 1, 2, 3, and 4. The depositions of Samples 1 and 2 are performed using an rf magnetron sputtering system with a turbo pump, while those corresponding to Samples 3 and 4 are performed using a PECVD system without a turbo pump. These spectra have been smoothed and baseline corrected. Representative benchmark spectra, from Lajaunie et al. [43] and Zou et al. [44], are also depicted. In reference to the Sample 1 results, please note footnote 2. The online version is depicted in color

Full size image

It is interesting to note that while there are clearly differences observed in the adhesion between thin-films of carbon deposited onto curved and flat Ti6Al4V substrates for the case of rf magnetron sputtering, very little difference is observed between the Raman spectra. We interpret this to suggest that the observed differences in the adhesion are not related to the underlying chemical content of the thin-films but are rather purely related to the geometric form of the substrate. Given the differences in the adhesion quality corresponding to thin-films of carbon that are prepared through rf magnetron sputtering, it seems that this PVD intensive process is more susceptible to the influence of the underlying substrate geometry than its PECVD counterpart, where the deposition is more shaped by the chemical reactions with the growth surface. Further work will have to be performed in order to clarify how exactly this occurs.

Finally, we would like to note that the results reported on here are a sub-set of the collection of results that were obtained by us related to adhesion. Indeed, we have performed this study over a much broader array of samples than the ones reported on here. A variety of depositions have occurred, using both our magnetron sputtering and PECVD deposition systems, with conditions very similar and different from those reported on here. For the samples that were prepared under similar conditions, the results are, within the range of experimental error, found to be similar to those reported on within the current manuscript. We thus conclude that our adhesion results are reproducible.

4 Conclusions

By depositing thin-films of carbon onto both curved and flat Ti6Al4V substrates, and then performing a Scotch tape adhesion test on the resultant films, we assessed the role that the geometry associated with the underlying substrate plays in shaping the quality of the adhesion of the resultant thin-films of carbon. Both rf magnetron sputtering and PECVD deposition techniques were employed for the purposes of this study. For the specific case of rf magnetron sputtering, we found that thin-films of carbon deposited onto curved Ti6Al4V substrates adhere better to the underlying substrate than those deposited onto flat Ti6Al4V substrates. In contrast, for the case of PECVD, the thin-films of carbon deposited onto curved Ti6Al4V substrates exhibit the same adhesion quality as those deposited onto flat Ti6Al4V substrates. Through the use of Raman spectroscopy, for the specific case of rf magnetron sputtering, it was demonstrated that there was very little difference in the chemistry of the resultant thin-films of carbon corresponding to the different substrate geometries. This suggests that it is the geometry of the substrate itself that is playing a role in determining the nature of the adhesion for the case of rf magnetron sputtering. We suspect that the PVD nature of rf magnetron sputtering may be responsible for this observed difference, PECVD depositions being more shaped by the chemical reactions with the growth surface.Footnote 3

Now that a dependence of the adhesion of thin-film carbon onto the underlying substrate geometry has been demonstrated, further work must be performed in terms of quantitatively characterizing this dependence. In order to perform this work properly, we envisage a set of experiments in which thin-films of carbon are deposited onto rings of various radii. We would expect that the flat substrate results would replicate the properties of the curved, i.e., ringed, substrate results in the limit that the curvature radii become very large. Detailed surface characterizations, prior and following the depositions, would be advisable. Rotation of the flat and ringed substrates would be performed throughout these depositions. A careful monitoring of the deposition process must be performed throughout the entire deposition cycle, the deposition pressure, the substrate temperature, the species present within the deposition chamber, and other factors related to the deposition, being continuously examined. Such a study will have to be performed in the future.