A new method for production of titanium vapor and synthesis of titanium nitride coatings

It is proposed to synthesize on machine parts and cutting tools wear-resistant titanium nitride coatings with the help of the hollow-cathode glow discharge, a molybdenum crucible for titanium evaporation being used as the anode of the discharge and a process vacuum chamber being used as the hollow cathode. The research revealed that at the anode surface area less than a critical value S*=(2m/M)S, where S is the area of the chamber walls,m is the mass of electrons andM is the mass of ions, the anode fall of potential is positive and grows from ∼50V at argon pressure p=0.2Pa to ∼2 kV at p=0.02Pa. At the discharge current I=0.6A electrons accelerated by the anode fall of 0.9 kV transport into the crucible with the inner diameter of 12mm the power of ∼0.54 kW, which allows the titanium evaporation and the coating deposition rate of 5mm·h 1 on a substrate distanced from the crucible at 100mm. After the argon is replaced with the nitrogen, titanium nitride coating without titanium droplets is synthesized the deposition rate amounting to about the same value.


Introduction
Wear-resistant TiN coatings substantially increase the useful life of machine parts, tools, and other products, because their hardness of ∼25 GPa and more [1] exceeds by many times hardness of the bulk materials.To improve properties of the coatings, they are usually bombarded during the coating synthesis by energetic particles accelerated from the gas discharge plasma [2] or using beam sources [3].
One of the methods widely used for the production of titanium vapor needed for synthesis of the above hard coatings is vacuum arc evaporation [4,5].It is remarkable for a high deposition rate of synthesized coatings and an adequate plasma density near their surfaces.The performance of coated products can be improved using combined treatment, which includes preliminary strengthening of the product surface and deposition on the surface of the hard coating using the vacuum arc evaporation of a titanium target [6].Drawbacks of the arc method are nonuniformity of the discharge plasma in the process chamber, as well as the soft metal droplets in the hard coatings.To prevent the droplets in the coatings, the vacuum arc plasma can be filtered [7].
Another method widely used for the production of titanium vapor is magnetron sputtering [8], which excludes the metal droplets.However, the magnetron sputtering is characterized by a low efficiency of the target material use, a low ionization degree of sputtered atoms and a low plasma density near the product surface.In this case, the growing coatings should be bombarded by beams of fast neutral atoms [9][10][11].
Due to development of magnetrons with rotating cylindrical targets [12] efficiency of the target material use was improved.The density of magnetron discharge plasma near the product surface and the sputtering rate are substantially enhanced when a pulsed dc magnetron [13] is used.When high-power pulsed magnetrons [14] are used, the ionization degree of sputtered atoms grows up to 60-80%.With the use of unbalanced magnetrons [15][16][17], which ensure an adequate ion current density on the product surface, the plasma density near the negatively biased product grows by many times.A serious drawback of the magnetron method is a great reduction of the coating deposition rate, while using a reactive gas, for instance, nitrogen, to synthesize a titanium nitride coating.It is due to deposition on the target of chemical compounds, whose sputtering rate is much less compared to the pure metals.
The metal vapor can also be produced due to sputtering a target placed at the bottom of a hollow cathode [18][19][20][21][22] even without using the magnetic field to maintain the glow discharge at the low gas pressure.In the latter case the target sputtering rate and coating deposition rate both greatly decrease, when a reactive gas is added to argon.The reason is the same as in the above magnetron method.
The coating deposition rate is independent of the reactive gas only when the metal atoms are evaporated from the target surface, for instance by the cathode spots of the mentioned above vacuum arc.Another way to evaporate the target is heating with an intense electron beam [23].There are no metal droplets in synthesized coatings in this case.However, an appreciable amount of the target material on its surface is liquid.For this reason, the products to be coated should be placed over the target.Negative biasing of the product allows bombardment of the growing coating by ions accelerated from the plasma produced by the electron beam.However, this plasma is also non-uniform, and its density is not enough for an adequate ion bombardment, the same as in the above conventional magnetron sputtering systems.
It was found in [24][25][26] that in the low-pressure glow discharge a small anode placed inside a large hollow cathode can be heated to the melting point of the anode material.The heating can be used for the metal vapor production and the coating deposition.However, this assumption has yet to be proved.The present work is devoted to the investigation of the coating deposition using titanium evaporation in a molybdenum crucible used as an anode of the low-pressure hollow-cathode glow discharge, when the process vacuum chamber is used as a hollow cathode.

Experimental setup
To carry out the investigations a 500-mm diameter and 600-mm-long vacuum chamber was used (Fig. 1).The chamber inner volume amounts to V = 0.118 m 3 and its inner surface area amounts to S = 1.46 m 2 .On the upper flange of chamber 1 is mounted high-voltage feedthrough 2 enabling immersion in gas-discharge plasma 3 of flat square anode 4 hanged up using a 1-mm-diameter molybdenum wire under the upper flange.The anode is made of 2-mmthick titanium sheet and its surface area S a is step-by-step diminished in successive experiments from ∼0.1 to ∼0.001 m 2 .
On the bottom flange of the chamber is mounted another feedthrough 5 with vertical steel rod 6 and fastened to the rod 30-mm-long hollow anode 7 with the inner diameter of 12 mm.Both the hollow anode and the rod are shielded by cylinder 8 with inner diameter of 30 mm connected to the grounded chamber.The upper end of the cylinder is 15 mm higher than the upper end of the hollow anode.
On the chamber wall is mounted third feedthrough 9, which is used for immersion in plasma 3 of substrate holder 10 distant from hollow anode 7 at 100 mm.The gas pressure in the chamber is measured by MKS-vacuummeter (USA) with transducer BARATRON in the pressure range 0.1-5 Pa and by VIT-3 vacuummeter with the ion gauge PMI-2 in the pressure range 0.001À0.1 Pa.The gas enters the chamber through connecting pipe 11 and is evacuated from the chamber through ground-ed grid 12 by a turbo-molecular pump, which ensures the residual gas pressure of 0.001 Pa.Using a gas supply system the pressure inside the chamber can be regulated from 0.001 to 5 Pa.Negative pole of DC power supply 13 is connected to the grounded chamber and its positive pole can be connected using switch 14 to square anode 4 or hollow anode 7. When switch 15 is closed, DC power supply 16 allows application of the negative bias voltage to substrate holder 10.When switch 15 is open, electrostatic voltmeter 17 allows measurement of the floating holder potential, which is practically equal to the cathode fall of potential.
On the chamber door there is a 20-mm-diameter quartz window (not shown in Fig. 1) used for remote temperature measurements of anode 4 and substrate holder 10 with pyrometer IMPAC IP 140 produced by the LumaSense Technologies GmbH (Germany).

Experimental results
When power supply 13 is connected to square anode 4, switch 15 is open and the voltage between the anode and the chamber amounts to 1-1.5 kV, increase of the argon pressure up to p ∼ 0.5 Pa results in breakdown and ignition of the gas discharge.A stationary glow discharge is established with the discharge voltage U ∼ 400 V, and the chamber is filled with a quite homogeneous bright glow of plasma 3. The plasma is separated from the chamber walls, grid 12 and cylinder 8 with a cathode sheath 18 intensity of the glow in the sheath being less than in the plasma.The mean current density on the chamber walls at the maximal discharge current I = 2 A amounts to j = 1.37 A•m À2 , and at the minimal cathode fall of potential U c ≈ 400 V the mean width of the cathode sheath according to the Child-Langmuir law amounts to d = 7.1 mm.It exceeds diameters of the orifices in the grid 12 being equal to 5 mm, and for this reason the plasma does not penetrate into the pumping system in the whole range of parameters I = 0.2Ä2 А and U = 0.4Ä3 kV.
Prior to start measurements the chamber walls are bombarded during 30 minutes by 400-500-eV argon ions from the plasma filling the chamber at the argon pressure p = 0.1 Pa and the discharge current of 2 A. It is needed to reduce the percentage of impurities at low argon pressure.Figure 3 shows that with the pressure decrease from 0.4 to 0.01 Pa the anode fall of potential DU grows from ∼25 to ∼500 V at the discharge current I = 0.7 A.
Figure 4 shows the anode with the surface area S a = 0.002 m 2 at the same discharge current I = 0.7 A and different argon pressures.At the pressure of p = 0.4 Pa the positive anode fall of potential amounts to DU ∼ 25 V and is characterized with a bright ball of plasma near the anode surface (Fig. 4a).In contrast to the pale blue color of the homogeneous plasma filling the chamber, the anode plasma ball is yellow.Measurement with the pyrometer showed that the anode temperature in this case amounts to 250 °C.
With the pressure reduction the anode plasma ball grows in size and its yellow color turns to violet.Finally the anode plasma disappears, the anode fall of potential DU and the anode temperature T start to grow rapidly.At p = 0.015 Pa, the anode fall of potential amounts to DU = 250 V, measured with pyrometer temperature amounts to T = 1050 °C (Fig. 4b).At p = 0.008 Pa, the anode fall reaches DU = 600 V and the titanium anode starts melting (Fig. 4c).
When power supply 13 is connected to hollow molybdenum anode 7, the effective anode surface area S a < 0.0007 m 2 , which is 3 times less than in the case of the flat anode presented in Figure 4.For this reason at the  With the pressure decreasing the plasma ball grows in size, turns its yellow color to violet and disappears at p ∼ 0.03 Pa, the same as in the case of flat anode 4.
After the decay of the anode plasma, the anode fall of potential abruptly rises from DU ∼ 300 V to DU ∼ 1500 V and higher.If the molybdenum hollow anode is previously filled with small titanium cheeps, heating of the cheeps by accelerated electrons 20 results in the titanium melting and the production of the titanium vapor flow from the surface of liquid titanium 21 to substrate holder 10.The experimental curves presented in Figure 5 show that the anode fall of potential starts to rise at higher pressure with increasing the discharge current.
After the anode fall of potential DU reaches a certain value DU*, it suddenly falls down to DU ∼ 100 V, then grows to the previously reached value of DU* and falls down again.In this case oscillations of the discharge voltage U and anode fall of potential DU take place their frequency amounting to 0.2-0.4Hz.
The maximal experimentally achieved value of the anode fall of potential DU* is decreasing with increasing of the discharge current.Plasma 3 is separated from cylinder 8 by positive space charge sheath of ions 19 and from the anode 7 by negative space charge sheath of electrons 20 inside the cylinder.
Electrical power of the anode heating by electrons is equal to P = I Â DU* and with increasing of the discharge current I rises from 170 W at I = 0.1 A to 550 W at I = 0.8 A (Fig. 6).At the discharge current I > 0.4 A electrons accelerated in the negative space charge sheath transport into the crucible with the inner diameter of 12 mm about the same electrical power of P ∼ 0.55 kW.It means that the power density on the liquid titanium surface of 1.13 cm 2 amounts to ∼5Â10 6 W • m À2 , which is quite enough for titanium evaporation and synthesis of titanium nitride coatings on the substrate.
A HSS substrate with a mask on its polished surface was fastened under substrate holder 10.Before the coating deposition the substrate and the mask were cleaned and heated by argon ions.To carry out the cleaning, chamber 1 is filled with argon and switch 15 is closed.At the discharge current of 2 A and the argon pressure of 0.5 Pa the anode fall DU does not exceed 50 V and heating of the titanium inside the hollow anode is negligible.In this case application to the substrate holder of the negative bias voltage ∼900 V from power supply 16 (Fig. 1) results in the bombardment of the substrate by argon ions with energy ∼1000 eV, which effectively clean and heat it.
After the 10-min-long pre-treatment argon is replaced with nitrogen.At the gas pressure of 0.025 Pa and the discharge current of 0.6 A the anode fall of potential amounts to DU = 900 V, and on the substrate bombarded by 100-eV nitrogen ions during 1-hr is synthesized titanium nitride coating.To ensure the ion energy value of 100 eV, readings of voltmeter 17 at open switch 15 exceed by 100 V readings of this voltmeter at closed switch 15 and the bias voltage applied to the substrate from power supply 16.
After the coating deposition, the mask is removed from the substrate surface, and the height d of the step between covered by the mask and open surfaces of the substrate is measured using stylus profiler DectakXT produced by Bruker Nano, Inc. (USA).It amounts to 4.9 mm and the coating deposition rate is equal to ∼5 mm•h À1 .The coating color is gold, and using X-ray diffractometer EMPYREAN produced by PANALYTICAL (Netherlands) it was found that the coating contains only one phase of TiN.After the nitrogen is replaced with argon, titanium coating is deposited, and the deposition rate amounts to about the same value of 5 mm•h À1 .

Discussion
It was found in [28] that decreasing losses of fast electrons from a hollow cathode and increasing the cathode sizes it is possible to reduce operating pressure of the hollow-cathode glow discharge by two orders of magnitude down to ∼0.01 Pa.This finding greatly enlarged technological capabilities of the glow discharge, and it was already used for the processing of machine parts and cutting tools by the broad beams of accelerated electrons, ions and fast neutral atoms, as well as for the plasma immersion ion implantation.When a vacuum process chamber is used as a big hollow cathode, the gas is ionized by electrons emitted by the chamber walls and fast electrons produced in the cathode sheath near the chamber surface [28].Length of the way L = 4V/S a , here V is the chamber volume and S a is the anode surface area, which those electrons repelled from the chamber walls hundreds of times pass before reaching the anode, exceeds the chamber width a hundreds of times.
Oscillating inside the chamber electrons visit all the parts of the chamber volume on their way to the anode.Hence, inside the chamber, which plays the role of an electrostatic trap for the electrons, probability of the gas ionization by those electrons is distributed uniformly.
At a high anode surface area S a the cathode fall of potential U c is equal to the discharge voltage U.At a constant discharge current, it is independent of the pressure in the range from p o to p * .The upper limit p * of this pressure range meets equation l N (p) = a, here a = 4V/S is the chamber width, S is the surface area of its inner walls and l N (p) is the mean pass of emitted by the chamber electrons with initial energy eU c between ionizing collisions at the gas pressure p.The lower limit p o of the above pressure range meets equation L = L(p) ≡ (eU c /W)l N (p), here W is the ionization cost, and L(p) is the mean way, which the electrons have to pass at the gas pressure p in order to spend all their initial energy eU c on the gas ionization.The lower limit p o decreases proportionally to S a .
In the above pressure range the mean pass l N of emitted by the chamber electrons between ionizing collisions exceeds the trap width a, and the trap length L exceeds the energy relaxation length L of those electrons.In these conditions electrons spend all their initial energy on the gas ionization, and contribution to the gas ionization of fast electrons produced in the cathode sheath is independent of the pressure and determined only by the ratio d/a of the cathode sheath width d to the chamber width a.When the discharge current grows up, the sheath width and contribution to the gas ionization of electrons produced in the sheath both decrease.
At p < p o the length L exceeds the mean way L of electrons emitted by the chamber to the anode, they spend only a part of their energy on the gas ionization and heat the anode with the rest of their energy.For keeping the discharge current with the pressure decreasing the contribution of fast electrons produced in the cathode sheath should be increased, and for this purpose the ratio d/a should be raised.In accordance with the Child-Langmuir law [29] the cathode sheath width d is proportional to U c 3/4 , and for this reason at L > L the discharge voltage U = U c abruptly rises, when the pressure is decreasing from p o to the pressure p ex of the discharge extinguishing (full curves 1-3 in Fig. 2).
When the anode surface S a is less than a critical value S a * = (2m/M) 1/2 S, here m and M are the electron mass and the ion mass and S is the surface area of the chamber walls, the positive anode fall of potential DU appears [30].In this case the discharge voltage amounts to U = U c + DU (full curves 4-6 in Fig. 2) and substantially exceeds the cathode fall of potential U c (dotted curves 4-6 in Fig. 2).At the gas pressure p > 0.1 Pa the anode fall of potential DU does not exceed 100 V and is concentrated between the discharge plasma filling the chamber and the plasma ball near the anode surface (Fig. 4a).After the decay of the plasma ball DU grows by an order of magnitude and is concentrated immediately between the discharge plasma and the anode.In the latter case all electrons produced in the discharge are accelerated by the anode fall of potential DU and spend all their energy on the anode heating (Fig. 4b) and melting (Fig. 4c).
At a constant discharge current the anode fall of potential DU and the electrical power heating the liquid titanium in the molybdenum anode-crucible increase with the gas pressure decreasing.It results in growth of the titanium vapor pressure near the anode and a decrease of DU.Reduction of the heating power leads to a decrease of the liquid titanium temperature and the titanium vapor pressure.It results in growth of DU, the titanium pressure rises again, and so on.To exclude these oscillations the growth of DU should be limited with a certain value of DU*.It turned out that a stable evaporation of titanium in the molybdenum anode-crucible can occur at the electrical power heating the anode not exceeding ∼0.55 kW (Fig. 6), which corresponds to the mean power density not exceeding ∼5 Â 10 6 W•m À2 .
In order to raise the flow of titanium atoms from the crucible to the products under processing it is needed to increase the surface of evaporating liquid metal.In our case the positive anode fall of potential DU also takes place at the anode surface area S a = 80 cm 2 , which is about 40 times higher than the effective surface ∼2 cm 2 of the anode-crucible with outer diameter of 1.6 cm.It seems that an increase in the anode-crucible diameter up to 4 cm could heighten the evaporating surface of the liquid titanium and the flow of the titanium atoms by 6 times at the same mean value ∼500 W•cm À2 of the heating power density and the total heating power of 3-4 kW.

Conclusions
-The titanium evaporation in the anode-crucible by electrons accelerated from the glow discharge plasma allows synthesis in the same plasma of the titanium nitride coatings not containing the titanium droplets, which enables an appreciable improvement of the machine parts and the cutting tools quality.-In the glow discharge with a process vacuum chamber used as a hollow cathode at the anode-crucible surface less than a critical value S* = (2m/M) 1/2 S, where S is the inner surface area of the chamber, m is the mass of electrons and M is the mass of ions, the positive anode fall of potential appears and grows up to 1.5-2 kV with the pressure decreasing, which allows to heat and melt titanium in the crucible by accelerated electrons.-The power density of electrons heating the liquid titanium is limited by the metal vapor pressure near the anode and does not exceed 0.5 kW•cm À2 .In order to increase the flow of titanium atoms to the products under processing it is needed to heighten the crucible diameter.
The work was carried out using the equipment of the Center of collective use of MSTU "STANKIN".

Figure 2
presents dependencies of the discharge voltage U between anode 4 and chamber 1 as well as the cathode fall of potential U c measured between substrate holder 10 and the chamber with voltmeter 17 at the discharge current of I = 0.5 A in the anode circuit and various surface areas of the anode S a .They show that at S a ≥ 0.016 m 2 the cathode fall of potential U c is practically equal to the discharge voltage U, however at S a = 0.002 Ä0.008 m 2 U substantially exceeds U c .When at S a = 0.064 m 2 the argon pressure is reduced from p = 1 Pa to p* = 0.5 Pa, the cathode fall U c slightly decreases from 440 V to a minimal value of 405 V and keeps this value till p reaches p o 1 = 0.37 Pa and starts rising with the further pressure reduction.At S a = 0.032 m 2 the cathode fall U c keeps the minimal value of 405 V till p reaches p o 2 = 0.19 Pa and then U c starts rising with the further pressure reduction.At S a = 0.016 m 2 the cathode fall U c starts rising with the further pressure reduction at p o 3 = 0.09 Pa, at S a = 0.008 m 2 U c starts rising at 0.045 Pa (dotted curve 4 0 in Fig. 2), at S a = 0.004 m 2 U c starts rising at 0.022 Pa (dotted curve 5 0 ) and at S a = 0.002 m 2 U c starts rising at 0.011 Pa (dotted curve 5 0 ).Comparison of the dotted curves 4 0 , 5 0 and 6 0 of the cathode fall of potential U c with full curves 4, 5 and 6 of the discharge voltage U reveals a positive anode fall of potential DU being equal to the potential difference U-U c between the anode and the floating substrate holder.

Fig. 3 .Fig. 4 .
Fig. 3. Dependence of the positive anode fall of potential DU on the gas pressure p at the discharge current 0.7 A and the anode surface area S a = 0.002 m 2 .

Fig. 5 .
Fig. 5. Dependence of the anode fall of potential DU on the nitrogen pressure p at the discharge current I = 0.1Ä0.8A and the hollow anode filled with titanium.

Fig. 6 .
Fig. 6.Dependence of the electrical power P heating the hollow anode on the discharge current I.