Tunable electrical properties of Ti-B-N thin films sputter-deposited by the reactive gas pulsing process

: Titanium-boron-nitrogen (Ti-B-N) thin films were deposited by RF reactive magnetron sputtering using a titanium diboride (TiB 2 ) target in an argon + nitrogen mixture. The argon mass flow rate was kept constant whereas that of nitrogen was pulsed during the deposition. A constant pulsing period P = 10 s was used and the introduction time of the nitrogen gas (duty cycle dc ) was systematically varied from dc = 0 to 100% of the pulsing period. This reactive gas pulsing process allowed depositing Ti-B-N thin films with various boron and nitrogen concentrations. Such adjustable concentrations in the films also led to changes of their electronic transport properties. Boron and nitrogen contents exhibited a reverse evolution as a function of the nitrogen duty cycle which was correlated with the transition from a metallic-to-semiconducting-like behavior. A percolation model was applied to the electrical conductivity as a function of the nitrogen pulsing parameters assuming some correlations with the evolution of the Ti-B-N thin films nanostructure.


Introduction
The development of nanostructured thin films by reactive sputtering for a wide range of applications has nowadays become one of the challenging tasks for creating innovative functional materials.If binary compounds combining two metallic elements, or a single metal with a light element such as carbon, boron, nitrogen or oxygen, have been extensively investigated for these last decades, the addition of a third element to form ternary materials still remains relevant today since the association of three elements may lead to the formation of one, two or even multi-phase systems and so nanocomposite materials [1][2][3].As a result, one of the most interesting features of these multiphasic materials is that their properties can be tailored by playing with the size, volume fraction, distribution, composition and so on of the appropriate phases.Among the large family of ternary compounds, many investigations have been focused on combinations of two metals with a light and reactive element [4][5][6][7][8].It is quite common to associate two metals with nitrogen or carbon for hard coatings [9, 10], or with oxygen for optical applications [11,12].Other studies report on metallic oxynitrides MOxNy, oxycarbides MOxCy or carbonitrides MNxCy from one metal with the two corresponding elements (i.e., oxygen + nitrogen, oxygen + carbon, or nitrogen + carbon, respectively) [13][14][15][16][17].For these last thin film materials, one of the challenging tasks is the high reactivity of light elements towards the sputtered metal, which may restrain some reachable compositions and, thus limiting the range of final properties.
Among ternary compounds, the ternary metal boron nitrides, namely M-B-N, have been compiled by Rogl [18] for several metals.Such experimental data show that the phase diagram feature strongly depends on the metal affinity towards boron and nitrogen and these ternary systems can be classified with respect to the chemical ability of the metal element to form or not binary borides, binary nitrides and/or ternary boron nitride compounds.Furthermore, consistent with this compilation, most relevant works devoted to metal boron nitride thin films systematically claim about the key role played by the nitrogen incorporation during the processing stage.This is particularly true when metal boron nitride films are prepared by reactive sputtering where the control of the process (and so the resulting film properties) strongly depends on the nitrogen partial pressure [19][20][21][22].Titanium boron nitrides Ti-B-N are no exception and phase mixture consisting of materials with different properties can be produced [23][24][25].Since hard phases such as TiN, TiB2, and cubic BN may coexist with soft phases such as hexagonal BN or amorphous BN, the very large majority of studies about Ti-B-N films have been focused on their mechanical and tribological properties as hard and wear-resistant coatings [26][27][28][29].It is also worth noting that enhanced hardness and good wear resistance of Ti-B-N thin films sputter-deposited have been accordingly connected to the produced nanocomposite structure; the latter being closely related to the film composition and thus dependent on the nitrogen partial pressure implemented during the deposition process.
Although there are many investigations on understanding mechanical performances of Ti-B-N films, very little work has been performed on their electrical behaviors as a function of B and N contents.It is particularly interesting to know how electronic transport properties change as a function of B and N concentrations since pure TiB2 and TiN materials both exhibit metallic-like behaviors.Combining these two materials often showed optimized mechanical and tribological properties for some given amounts of TiB2 and TiN in an amorphous matrix of BN.Since BN is an insulating material and assuming a nanocomposite structure of Ti-B-N films, the motivation of this article is to understand how the electrical conductivity of Ti-B-N films can be tuned vs. light elements concentration, and to determine some correlations with the most relevant sputtering process parameters.
In this work, we prepared Ti-B-N thin films by reactive sputtering using the reactive gas pulsing process (RGPP) [30].Nitrogen mass flow rate is periodically supplied during the deposition stage with an increasing time of injection.The chemical composition is first determined so as to prove that B and N contents are adjustable by means of nitrogen gas pulsing.Electrical properties (resistivity, charge carrier mobility and concentration) are systematically measured and illustrate significant variations of Ti-B-N films conductivity for some given nitrogen pulsing parameters, and so as the nitrogen concentration in the films increases.Assuming a nanocomposite structure of Ti-B-N films, a percolation model is suggested to explain their electronic transport properties.

Materials and Methods
Ti-B-N thin films were prepared by RF reactive magnetron sputtering.The sputtering machine was a 110 L vacuum chamber evacuated by a cryogenic pump backed with a dry primary pump achieving an ultimate pressure of 510 -8 mbar.A TiB2 rectangular target (3811276.35mm and purity 99.9%) was fixed at 100 mm from the centre of the substrate holder.The argon flow rate was kept constant at qAr = 29 sccm and a constant pumping speed SAr = 95 L s -1 was used, leading to an argon partial pressure of 0.6 Pa.Before any deposition, an etching period of 15 min and 250 V of bias was applied to the substrates.Then, the TiB2 target was RF sputtered using a constant electric power of 1 kW.Furthermore, a pre-sputtering time was applied for 5 min to remove the contamination layer on the target surface and stabilize the process, leading to a target self-bias potential of VSB = 210 V. Nitrogen flow rate qN2 was pulsed during Ti-B-N deposition by means of the RGPP [30].A rectangular signal was used to pulse nitrogen gas with a constant pulsing period P = 10 s.This period was selected since such operating conditions correspond to a freedom of alternating between the boride and nitride poisoned state and vice-versa.For all depositions the nitrogen injection time tON was varied, from 0 to 10 s, corresponding to duty cycles dc (dc = tON/P) from 0 until 100% of P. The maximum nitrogen rate was qN2Max = 5 sccm.This value corresponds to the nitrogen amount required to completely avalanche the reactive sputtering process in the nitride sputtering mode [31].During the tOFF time, the nitrogen mass flow rate was completely stopped (qN2min = 0 sccm).The total sputtering pressure and TiB2 target voltage were alternated in the range of 6.0-6.310-3 mbar and 210-220 V for tOFF and tON times, respectively.During deposition, the substrate holder (191142 mm) was biased with a negative voltage of 20 V, and no external heating was added.Ti-B-N films were deposited on glass and (100) Si substrates.Deposition rate was measured from the film thickness and the corresponding deposition time.The latter was adjusted in order to get a constant film thickness of 300 nm.
Electrical resistivity measurements, from room temperature (25 o C) until 200 o C, were conducted with a homemade system using the four-probe van der Pauw method, equipped with an annealing hot plate.Carriers' concentration and carriers' mobility were assessed from setup when applying a magnetic field B = 0.8 T in a device again operating with the van der Pauw method.The infrared analysis was performed with a Perkin Elmer Spectrum Two Lita FT-IR spectrometer for the range of wavenumbers 400 until 4000 cm - 1 .The film composition was obtained by X-ray photoelectron spectroscopy (XPS) with a PHI VersaProbe 1 system using a monochromatized and focalized Al Kα X-ray source (hυ = 1486.7 eV, spot diameter = 200 micrometers).Base pressure during analysis is better than 5.10 -7 Pa.CasaXPS software is used for data treatment [32].

Results
The deposition rate R of Ti-B-N films was first measured as function of the nitrogen duty cycle dc (Fig. 1).It was obtained from thickness measurements of the films (average of 10 measurements giving rise to error bars of the deposition rate) by mechanical profilometry (step height method) and recording the time of deposition.The evolution of deposition rate vs. duty cycle can be divided in three regions (I, II and III in Fig. 1).For duty cycles lower than 30% of P (region I) a nearly constant rate can be assumed.Nitrogen injection time (less than 30% of P = 10 s) is too short for a full nitriding of the TiB2 target surface.The boride sputtering mode prevails and one can expect Ti-B-N films exhibiting characteristics similar to the TiB2 compound, i.e., a metalliclike behavior.An increase of duty cycle from 30% to 55% of the pulsing period P (region II) gives rise to enhanced deposition rates from 10.7 to 14.6 nm min -1 , respectively.Deposition rate exhibits a maximum for dc = 55% of P and a further increase of duty cycle (region III) produces a continuous and linear drop of the deposition rate down to 9.3 nm min -1 when the nitrogen gas is constantly supplied (i.e., for dc = 100% of P. For duty cycles lower than 55% of P, the nitrogen injection time (tON time) is too short to completely set the reactive sputtering process in the nitride mode.Alternations between nitride and boride modes occur during tON and tOFF times, respectively.Nitride mode prevails for duty cycles higher than 55% of P and the TiB2 target surface becomes more and more covered by a nitride compound, which reduces the sputtering yield.Similarly, there is a progressive incorporation of nitrogen atoms into the Ti-B-N films as the nitrogen injection tON time is increased corresponding to the formation of a poisoning nitride layer on the target surface and a decrease of the deposition rate.This optimized deposition rate vs. nitrogen injection corroborates with results previously reported by Chaleix and Machet [33] and later by Pierson et al. [34] where a maximum rate was also obtained for a given range of nitrogen flow rates.In like manner, the authors assigned this maximum rate to a nitrogen incorporation into the growing films and a progressive nitriding of the target leading to a decrease of the deposition rate (typical of low reactive systems).

Composition and structure
The films composition as a function of the duty cycle is first determined from XPS analyses.XPS spectra show clear signals corresponding to Ti 2p, N 1s and B 1s peaks.(Fig. 2).The position of these peaks agrees with results recorded by other investigators [35,36].A curve fitting procedure has been systematically performed for all recorded signals and for various duty cycles.Ti 2p peak consists of three components corresponding to the contributions from TiO2 (458.2 and 464.1 eV due to Ti 2p 1/2 and Ti 2p 3/2 , respectively), TiNO/TiN (456.7 and 462.3 eV) and TiB2 (454.3 eV), as illustrated in Fig. 2a.For N 1s spectra (Fig. 2b), decomposition leads to three contributions: NTi (396.9 eV), NB (398.3 eV) and NO (402.8 eV).B 1s spectra also display three contributions assigned to TiB2 (187.8 eV), BN (190.0 eV) and BO (192.1 eV).
For the 2p levels of titanium (Fig. 2a), the decomposition is more complicated.The contributions of TiO2 are still visible at 458.2 and 464.1 eV due to Ti 2p 1/2 and Ti 2p 3/2 , respectively.The main difficult issue remains determining the contributions of TiN/TiNO on one hand and TiB2 on the other hand.For each of these compounds, the forms are complex and cannot be reduced to a simple component.Moreover, since pure standards of TiN and TiB2 compounds are not accurately produced, the best prepared samples were TiN sample with 54% nitrogen and 46% titanium with no boron detected and a sample as close as possible to TiB2 with 58% boron, 34% titanium and 8% nitrogen (atomic concentrations).The maximum of intensity is at 456.7 eV for TiNO/TiN and at 454.3 eV for TiB2.
For the lowest duty cycles and up to dc = 40% of P, curve fitting of Ti 2p and B 1s peaks gives rise to signals corresponding to Ti-B bonds with TiB2 peak at 454 eV and 187 eV as shown in Fig. 2a and 2c.As expected, a significant contribution of the film oxidation is measured from Ti-O and B-O bonds with TiO2 peaks at 458 and 464 eV (2p 1/2 and 2p 3/2 signals in Fig. 2a), and also with BO peak at 192 eV (Fig. 2c).As duty cycle increases (dc = 60% of P and higher), TiB2 contribution recorded from Ti 2p and B 1s peaks reduces and vanishes for a constant supply of nitrogen.Similarly, the BN peak from N 1s (398 eV) and B 1s (190 eV) becomes more and more intense as well as the TiN peak at 396 eV (Fig. 2b).The film oxidation still remains for any duty cycle but it mainly originates from Ti signals (high reactivity of this metal towards oxygen) with a significant reduction of B-O and N-O bonds influence as shown with NO and BO peaks at 400 eV (Fig 2b) and 192 eV (Fig. 2c).As expected, increasing the duty cycle from 0 to 100% of P favors the amount of Ti-N and B-N bonds, whereas that of Ti-B bonds decreases without completely preventing the film oxidation due to the high reactivity between titanium and oxygen.
[Ti], [B] and [N] atomic concentrations in the films as a function of duty cycle (neglecting the oxygen amount due to surface oxidation, mainly) were determined from Ti 2p, N 1s and B 1s signals (Fig. 3).These three element concentrations exhibit a continuous and smooth variation as the nitrogen injection time becomes longer.B and N concentrations show a reverse evolution, while the Ti one slightly reduces (note that concentrations have been determined assuming homogeneous compositions through the film thickness, which is not completely relevant and induces an assumed inaccuracy of 3 at.% for the chemical concentration).Without nitrogen pulsing (dc = 0% of P), the [B]/[Ti] atomic concentration ratio is close to 1.8.This is lower than the stoichiometric TiB2 compound due to oxygen occurrence (more than 10 at.%) and a few at.% of nitrogen.When nitrogen is constantly injected (dc = 100% of P), nitrogen-rich Ti-B-N films are prepared with [N] higher than 44 at.% and [B] around 35 at.%.It is also worth noting that for duty cycles higher than 60% of P, the three element concentrations tend to stabilize.The reverse evolution of nitrogen and boron contents vs. nitrogen supply agrees with results previously reported by Pierson et al. [37] and Han et al. [38], who also sputter-deposited Ti-B-N films from a TiB2 target in Ar + N2 atmosphere.The authors also noticed the same kind of saturation for Ti, B and N atomic concentrations as the nitrogen flow rate increased (stabilization of the concentrations for nitrogen injections depending on the operating conditions).
The composition of the films is plotted within the Ti-B-N ternary phase diagram (Fig. 4).TiB2, TiN and BN stoichiometric compounds are also indicated as well as quasibinary tie lines between these individual compounds (dashed lines).Nitrogen pulsing does not produce a chemical composition evolution along the quasi-binary TiN-TiB2 line as reported for Ti-B-N films prepared by PACVD [39].It rather moves away from such a line as the duty cycle increases with a BN-enrichment of the films as the nitrogen supply tends to be constant.The same trend has ever been reported by others [40,41] giving rise to a comparable saturation of nitrogen content in the Ti-B-N films, even by sputtering a TiB2 target in a pure nitrogen atmosphere.At first and as proposed by Mayrhofer et al. [42], we may suggest the formation of a composite structure composed of TiN and TiB2 nano-crystals randomly distributed in the film, and embedded into a disordered and amorphous a-BNx compound.Assuming this nanocrystalline Ti-B-N structure, the corresponding phase proportions (mole fractions) are 42% TiN, 10% TiB2 and 48% BN in the equilibrium phase diagram for Ti-B-N films prepared with dc = 100% of P. As expected, reducing the duty cycle leads to lower contents of TiN and BN phases (e.g., for dc = 40% of P, phase proportions in mole fractions are to 28% TiN, 53% TiB2 and 19% BN).It is also worth noticing that using a short injection time of nitrogen (dc = 20% of P) leads to a significant incorporation of N in Ti-B-N films due to the high reactivity of nitrogen towards titanium and boron elements.
Occurrence of chemical bonds can be brought to the fore from FTIR analyses, especially for light elements like B and N. Figure 5 shows FTIR spectra recorded for Ti-B-N films deposited on (100) silicon and for various duty cycles.No clear signals are recorded for duty cycles lower than 45% of P. Some peaks appear as the nitrogen injection time increases and they become more significant when duty cycle exceeds 55% of P. A broad asymmetric band appears close to 1390 cm -1 and clearly develops as dc reaches 90% of P, which is mainly assigned to the characteristic absorption of hexagonal BN (in-plane B-N bond stretching) in Ti-B-N single layer [43,44].
Figure 5. FTIR spectra measured for Ti-B-N films prepared by pulsing the nitrogen gas with an increasing duty cycle from 0 to 90% of the pulsing period P = 10 s.
Similarly, a broad and significant signal is measured around 738 cm -1 related to the B-N-B bending mode occurring in amorphous boron nitride [37] and a second one is associated with the transverse optical mode of the cubic BN phase at 1108 cm -1 [45,46].A narrow and intense peak can also be noticed at 610 cm -1 with a few weak peaks between 500 and 600 cm -1 .This group of signals correspond to Ti-N bonds (typical stretching mode of TiN around 600 cm -1 [47]).All these FTIR signals support the formation and increasing amount of B-N and Ti-N bonds for high duty cycles, keeping an amorphous film structure.These results agree with X-ray diffraction analyses where no diffraction peaks due to boron nitride or other phases were detected (not shown here).

Electrical properties
DC electrical resistivity at room temperature 300K of Ti-B-N thin films deposited on glass continuously rises as a function of the duty cycle and spanning more than 6 orders of magnitude as dc changes from 0 to 100% of P (Fig. 6).Since the nitrogen pulsing period is very short (P = 10 s) and due to deposition rates varying from 9.3 to 14.6 nm min -1 (Fig. 1), one can first assume that a random N and B distribution is produced through the film thickness rather than a multilayered structure, which could influence electrical conductivity, as previously reported for metal/metal oxide coatings [48].For dc lower than 40% of P, resistivity is in the range of typical metals, i.e., below 10 -5 Ω m.Afterwards, Ti-B-N films abruptly become more resistive for duty cycles between 40 and 80% of P, and tend to stabilize to a few Ω m as the nitrogen gas is constantly supplied.Carrier mobility µ300K and carrier concentration n300K at 300 K are also both influenced by the nitrogen injection for a similar range of duty cycles.The carrier concentration is over 10 28 m -3 for films prepared with duty cycles lower than 40% of P, which typically corresponds to metallic-like behaviors.It loses several orders of magnitude as duty cycle changes from 60 to 100% of P where n300K = 2.910 19 m -3 .Similarly, the carrier mobility exhibits a reverse evolution with µ300K lower than a few 10 -4 m 2 V -1 s -1 for the lowest duty cycles, whereas it reaches 2.010 -2 m 2 V -1 s -1 for dc = 100% of P. As a result, this drop of conductivity is mainly assigned to the strong decrease of the free carrier concentration, which prevails on an enhanced mobility.These significant variations of electronic transport properties of Ti-B-N films correlate with their nitrogen enrichment and thus, with an increasing amount of amorphous BN phase.These results support conclusions previously reported by Rogl [18], who claimed that we may expect a metallic-like behavior for boron-rich compounds, but with an increasing nitrogen content semiconducting to insulating properties develop.In addition, it is interesting of noting that this abrupt increase of resistivity obtained for duty cycles in-between 40-80% of P also corresponds to the transition from absorbent (metallic-like) to transparent (semiconducting-like) Ti-B-N films in the visible region (optical transmittance spectra of Ti-B-N films deposited on glass not shown here).
Resistivity vs. temperature also support this gradual metal-to-semiconducting transition as the duty cycle and thus the nitrogen content in Ti-B-N films rises (Fig. 7).A nearly temperature independent resistivity can be noticed for the lowest duty cycles where  is kept nearly constant from 30 to 200°C with for example, a negative temperature coefficient of resistance at 300 K (TCR300K) of -8.610 -5 K -1 for films prepared without nitrogen pulsing.This is consistent with Pierson et al. [49] results, who also reported negative TCR values and the loss of metallic character even for Ti-B-N films prepared with low nitrogen flow rates.Their results also showed a significant increase of TiBN electrical resistance for a given range of nitrogen flow rates corresponding to negative TCRs, which is in some ways similar to the sudden increase of resistivity measured for duty cycles higher than 40% of P (Fig. 6).In our TiBN films, TCR300K becomes even more negative for dc = 50% of P with TCR300K = -3.910 - K -1 and a further increase of the nitrogen injection time leads to a semiconducting-like behavior with an exponential decrease of the film resistivity as the temperature rises.An Arrhenius plot (electrical conductivity vs. reciprocal temperature) gives rise to an activation energy Ea = 12 meV for dc = 60% of P and reaches 124 meV for a constant supply of the nitrogen flow rate (dc = 100% of P).This increasing activation energy is connected to the nitrogen-enrichment of Ti-B-N films and so, to the formation of an amorphous and insulating BN phase.However, the highest value of activation energy is quite low compared to typical semiconducting materials, but it remains in the order of magnitude of some nitride or boride semiconductors where the temperature dependent resistivity strongly depends on their composition [50].Our results clearly show that electronic transport characteristics of Ti-B-N films progressively but rapidly change from metallic-to-semiconducting behaviors as a function of the duty cycle, which are directly associated to variations of composition and structure.At first and as frequently modeled for Ti-B-N mechanical properties, we can suppose that films exhibit a nanocomposite structure [39,42,51].The latter is made of nanometric TiN and TiB2 crystallites (conductive phase) embedded in a disordered BN matrix (insulating medium).
Based-on the sudden change of electrical properties of Ti-B-N films for a given range of duty cycles (Fig. 6) and assuming a nanocomposite structure, a percolation model has been used to describe the evolution of thin film conductivity at 300 K 300K as a function of 1 -ndc, where ndc is defined as the normalized duty cycle (Fig. 8).Sputtering conditions with 1 -ndc = 0 corresponds to the most N-rich Ti-B-N films (i.e., dc = 100% of P) and so, the most resistive films.Conductivity abruptly rises when 1-ndc reaches 0.4 and afterwards tends to saturate.This 300K vs. 1 -ndc evolution typically behaves like a percolation phenomenon.Two electrically different media (insulating amorphous BN matrix and conducting TiN and TiB2 phases) are mixed in Ti-B-N films.The conducting phase proportion increases while increasing 1 -ndc.As a result, the insulating-conducting transition corresponds to the percolation threshold, which can be determined assuming an effective medium theory (EMT) [52].The percolation theory allows describing the electrical conductivity of a composite medium before and after the percolation threshold c through the following equations [53]: before the percolation threshold (i.e.,  < c), where  = 300K is the DC electrical conductivity at 300 K (S m -1 ), i the DC electrical conductivity of the insulating medium (S m -1 ),  = 1 -ndc the conducting phase ratio (arb.units), and s the critical exponent in the insulating region.
after the percolation threshold (i.e.,  > c), where c is the DC electrical conductivity of the conducting phase (S m -1 ), and t the critical exponent in the conducting region.
For the full range of conducting phase ratios , a phenomenological relationship can be used as suggested by McLachlan et al. [54]: with  = .
The best fittings have been performed from these equations as shown in figure 8 (red and green lines for equations ( 1) and ( 2), respectively).A percolation threshold c = 0.45 is obtained, which corresponds to dc = 55% of P. From the chemical compositions determined from XPS analyses (Fig. 3), this duty cycle produced comparable B and N concentrations in Ti-B-N films.This result supports Mayrhofer et al. who claimed that this range of compositions matches with a film composed of two nanoscale phases (2-3 nm) encapsulated by a high-volume fraction (about 50%) of disordered phase [38,41].It should also be pointed out that from the experimental data at  = conductivity of the conducting phase is c = 2.4710 5 S m -1 .This value is lower than that of single TiN and TiB2 films (TiN = 1.8510 6 S m -1 [55] and TiB2 = 3.5210 6 S m -1 [56]) since the conducting phase certainly contains defects and is made of small grains (a few nm), but it remains in the order of magnitude of metallic materials.About the insulating medium, results of the fits give i = 5.2110 -1 S m -1 .This value is higher than conductivities usually reported for BN compounds ( is below 10 -6 S m -1 for pure c-BN at room temperature [57]).This difference is understandable since electronic transport properties of BN strongly depend on crystallinity, preparation method, nature and type of doping [57].In our Ti-B-N films, the insulating matrix certainly contains a significant amount of defects and Ti, which favor mobility and concentration of free carriers and thus enhance the overall conductivity.
Critical exponents s and t are also of interest, and universal values in-between sun = 0.7-1 and tun = 1.6-2 are commonly reported for three dimensional systems with an inverse Swiss cheese model [52].For our Ti-B-N films, fittings from experimental data lead to s = 6.1 and t = 0.41, which are significantly different compared to universal values.These discrepancies can be assigned to an inhomogeneous structure of the films (percolation theory is based-on perfect spherical particles for both insulating and conducting media).Sputtering deposition at room temperature often leads to columnar architectures [58] and RGPP may produce periodic multilayers [59].As a result, our Ti-B-N films can be considered as an anisotropic system affecting the percolation law since the assumed homogeneity and sphericity of media are not fully established.However, fitting with equation (3) and using parameters obtained from each region allows a reliable description of the electrical conductivity of Ti-B-N films for the full range of duty cycles (blue line in Fig. 8).This two-exponent phenomenological percolation equation can be valuably used for assessing the complex DC electrical conductivity of Ti-B-N films exhibiting reverse B and N concentrations.

Conclusions
Ti-B-N thin films 300 nm thick are sputter-deposited by RF reactive sputtering.A TiB2 target is sputtered in a reactive atmosphere composed of Ar and N2 gases.The reactive gas pulsing process (RGPP) is implemented to inject the nitrogen gas with a constant pulsing period P = 10 s.A systematic variation of the duty cycle dc (tON injection is carried out in order to tune the chemical composition of the films.A nitrogenenrichment is obtained as the dc gradually increases from 0 to 100% of P. In the same way, the boron concentration is inversely reduced keeping a nearly constant Ti content.It is also shown that increasing the nitrogen injection time favors the B-N bonds occurrence in the films with a phase proportion close to 48% of BN (mole fraction) when the nitrogen is constantly supplied (dc = 100% of P).
For duty cycles in-between 40-80% of P, electronic transport properties measured at room temperature significantly change with a resistivity enhancement of 4 orders of magnitude, while carrier mobility and concentration correspondingly change for the same range of dc.A regular but rapid transition from a metallic to semiconducting-like behavior is clearly brought to the fore as a function of the duty cycle, which is supported from resistivity vs. temperature measurements.Assuming a nanocomposite structure in Ti-B-N films made of nanometric TiN and TiB2 crystallites (conductive phase) embedded in a disordered BN matrix (insulating medium), a percolation model well describes the tunable films conductivity as a function of the nitrogen injection.
These results demonstrate that the nanocomposite structure successfully proposed for mechanical properties and wear resistance of Ti-B-N films, can also be applied to well describe the electrical properties of such films as a function of B and N concentrations.They also show that RGPP technique is a suitable approach to adjust electronic transport properties by means of a simple and easy change a pulsing parameter.A further incorporation of nitrogen in the films would be relevant to extend even more the multiphase nanostructuring growth.
Author Contributions: Conceptualization, data curation, C. Sakkas; methodology, software, J.M. Cote; formal analysis, validation, J. Gavoille; formal analysis, validation, J.Y. Rauch; formal analysis, validation, P.H. Cornuault; data curation, investigation, A. Krystianiak; formal analysis, validation, O. Heintz; writing-review and editing, funding acquisition, N. Martin.All authors have read and agreed to the published version of the manuscript.All authors have read and agreed to the published version of the manuscript.
Funding: Please add: This research was partially funded by Fonds Européen de Développement Régional -FEDER, grant number CTE 6059 in the framework of the TOOLEXPERT Interreg V project.
Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.

Figure 1 .
Figure 1.Ti-B-N deposition rate R as a function of the duty cycle dc when a TiB2 target is RF sputtered and injecting the nitrogen gas with a pulsing period P = 10 s.Deposition rate gradually changes and exhibits a maximum for dc close to 55% of P.

Figure 2 .
Figure 2. XPS a) Ti 2p, b) N 1s and c) B 1s spectra of Ti-B-N thin films prepared with various duty cycles dc.

Figure 4 .
Figure 4.The composition of Ti-B-N films (blue symbols) sputter-deposited by pulsing the nitrogen gas from dc = 0 to 100% of the pulsing period P = 10 s within a ternary phase diagram.Positions of TiB2, TiN and BN stoichiometric compounds are indicated as well as quasi-binary tie lines between the individual compounds (dashed lines).

Figure 6 .
Figure 6.DC electrical resistivity 300K, carrier mobility µ300K and carrier concentration n300K measured at room temperature as a function of the duty cycle dc of Ti-B-N thin films.

Figure 7 .
Figure 7. DC electrical resistivity  vs. temperature T measured on Ti-B-N thin films deposited on glass for various duty cycles (dc = 0 to 100% of the constant pulsing period P = 10 s).

Figure 8 .
Figure 8. DC electrical conductivity at room temperature 300K vs.  = 1 -ndc for Ti-B-N thin films (ndc is the normalized duty cycle).Red and green lines are related to the conductivity calculated from the fitting before (equation (1)) and after (equation (2)) the percolation threshold.The blue line is the best fit obtained from equation (3).