INTRODUCTION

Sources of wide ion beams are used to modify the surfaces of products, including the implantation of the necessary chemical elements into them, etching a surface, polishing it, and accompanying the synthesis of various functional coatings. Each source contains a gas-discharge chamber in which a plasma emitter is created, as well as an ion-optical system for accelerating ions and forming a beam. The main disadvantage of numerous ion sources is that a plasma emitter is produced using thermionic cathodes, which fail in an environment of chemically active gases, e.g., oxygen [1, 2]. To produce ions of chemically active gases, sources with the generation of a plasma emitter in a high-frequency discharge or in a cold-cathode glow discharge are used [3, 4].

When processing dielectric products with ion beams, a positive charge accumulates on the surface and in the volume of a material. In many cases, this limits the scope of application of accelerated ions. In this regard, sources of beams of fast neutral atoms obtained via neutralization of the charge of accelerated ions have become widespread. As an example, paper [5] describes sources of fast neutral atoms with the generation of plasma emitters in a high-frequency discharge. Their disadvantages are the limited energy of accelerated particles and, consequently, the limited sputtering rate of materials as well as the need to match high-frequency power supplies to a plasma load and the negative effect of high-frequency radiation on electronic devices and the health of personnel.

The energy of accelerated particles and the rate of sputtering of materials by them are noticeably higher in sources of fast neutral atoms with the generation of a plasma emitter by a glow discharge with the confinement of electrons in an electrostatic trap [6–8]. A trap is usually formed by a hollow cathode and an emission grid that is negative with respect to it. In [6], argon ions with an energy of up to 3 keV fly into the vacuum chamber from a homogeneous plasma that fills a hollow cathode with a diameter of 21 cm through the holes of the accelerating grid. As a result of collisions with a charge exchange at an argon pressure of 0.2 Pa, ions turn into fast neutral atoms at a distance of ~10 cm from the grid. At lower pressures, the discharge voltage begins to rise and the discharge is quenched; at a pressure of more than 0.5 Pa, breakdowns occur through the grid holes between the plasma emitter in the hollow cathode and secondary plasma in the chamber. As a result, the acceleration of ions between these two plasmas stops. Thus, the volume of the electrostatic trap (3 L), which provides the operability of the source [6] in the range from 0.2 to 0.5 Pa, is minimal for the sources under consideration. To expand the pressure range in which these sources operate steadily, it is necessary to increase the hollow-cathode volume. Such sources can be used to modify the properties of coatings on metal products [9] and dielectrics [10] during their deposition.

When using a source of fast argon atoms for etching or polishing the surfaces of small substrates, e.g., with a diameter of 2 cm, the utilization factor of a beam with a diameter of 20 cm is ~1%. Its increase requires the beam compression so that its diameter at some distance from the grid only slightly exceeds the substrate diameter. For this purpose, the flat emission grid of the source with a diameter of 20 cm is replaced by a concave grid with the same diameter and a 20-cm radius of curvature of the surface. In this case, the diameter of the beam of fast neutral atoms decreases with increasing the distance h from the grid from 20 cm to a minimum value of ~1 cm at h = 20 cm and then increases again. By installing a substrate with a diameter of 2 cm at a distance of 18 or 22 cm from the grid, it is possible to increase the substrate etching rate by two orders of magnitude compared to etching with a wide beam formed by a flat grid.

However, it is impossible to use such a beam for polishing the surface. The fact is that polishing is effective only at large angles of incidence (α = 75°–85°) of accelerated particles on the treated surface [11]. When the beam is compressed using a round concave mesh with a surface diameter and radius of curvature equal to 20 cm, the angular spread of atoms is 60°, and it is impossible to install the substrate to be processed in the beam so that fast atoms bombard its surface at an angle of incidence of α = 75°–85°. The aim of this study was to develop a source of fast argon atoms with beam compression that increases the rate of polishing a substrate by atoms with a large angle of incidence (α = 80°) on its surface.

EXPERIMENTAL INSTALLATION

Figure 1 shows a structural diagram of the installation for polishing substrates with a beam of fast argon atoms. It consists of a water-cooled gas-discharge chamber with a diameter of 50 cm and a length of 21 cm, in which a plasma emitter of ions is formed, and a water-cooled rectangular housing with a length of 55 cm, a height of 14 cm, and a width of 40 cm with a removable flange. A hollow cylinder with a diameter of 3 cm and a length of 30 cm with a quartz window at the end is installed at the center of the removable flange. The temperature of the substrates installed inside the housing is measured through this window using an IMPAC IP 140 infrared pyrometer (LumaSense Technologies GmbH, Germany). The cylinder protects the window against deposition of metal films. The housing is mounted on the left wall of the chamber, which is pumped out through a flat grounded grid to a residual gas pressure of 0.001 Pa by turbomolecular and rotary pumps. The grounded grid does not allow plasma to penetrate into the vacuum-pumping system.

Fig. 1.
figure 1

Diagram of the installation for processing substrates with a beam of fast atoms.

Full size image

Inside the housing, there is an accelerating grid manufactured in the form of a set of 11 plane–parallel titanium plates 0.5 mm thick, 300 mm wide, and 150 mm long. The plates with 4.5-mm-thick inserts between them are fastened together with tie studs. The accelerating grid is attached to the walls of the housing using four ceramic insulators protected from deposition of metal films by hollow cylindrical shields. Between the chamber and housing, there is a rectangular hole 31 cm wide and 6 cm high, which is covered with an accelerating grid. The grid is installed so that the edges of its plates lie in the plane of the chamber wall.

A high-voltage feedthrough is mounted on the top of the housing. It allows one to connect the grid to an accelerating-voltage source and change the negative voltage at the grid from zero to 5 kV. A rotation device is installed under the bottom of the housing, which consists of a rod with a diameter of 10 mm with a threaded M6 socket for connection through a hole in the bottom of the housing to the substrate holder. The rod that rotates at a speed of 8 rpm is located slightly below the bottom of the housing, and the angle between the bottom and the rod axis is 80°. The substrate holder is made in the form of a rod with an M6 thread from the bottom and a bowl with a depth of 3 mm and an inner diameter of 30 mm at the top. After connecting the holder to the rod of the rotation device and installing a substrate with a 30-mm diameter and a 6-mm thickness in its bowl, the center of the substrate surface appears to be on the axis of the housing at a distance of 25 cm from the plane–parallel grid plates.

At the bottom of the housing, there is also a stand for a vertical target 11 cm high and 34 cm wide made of a 1-mm-thick titanium sheet, which, after removing the substrate holder from the housing, can be moved from the removable flange of the housing to the plane–parallel grid plates. To measure the energy flux density of fast atoms, instead of a target, it is possible to fix a stainless-steel substrate with a diameter of 30 mm and a thickness of 2 mm on the stand, move it along the axis of the housing, and measure its temperature with a pyrometer.

Argon is supplied from the gas-supply device to the housing. It passes into the vacuum chamber through the gaps between the plane–parallel plates. The gas pressure in the chamber is measured by a Baratron sensor and regulated from the control panel. The gas-discharge power source and the accelerating voltage source are controlled from the same control panel.

Switching the power supply on between the anode and the gas discharge chamber at a gas pressure of 0.5 Pa leads to the ignition of a glow discharge with a voltage of U ~ 400 V and filling the chamber with homogeneous plasma. The chamber surface is separated from plasma by a layer of positive spaсe charge. The width d of this layer depends on the voltage U and the density of the ion current j from plasma to the chamber. According to Childe–Langmuir’s law [12],

$$j = (4{\text{/}}9){{\varepsilon }_{{\text{o}}}}{{(2e{\text{/}}M)}^{{1/2}}}{{U}^{{3/2}}}{\text{/}}{{d}^{2}},$$
(1)

where εo is the electric constant, e is the electron charge, and M is the ion mass.

When measuring the ion mass M = AMo in atomic mass units A = 1.66 × 10– 27 kg,

$$j = 5.46 \times {{10}^{{-8}}}{{(1{\text{/}}{{M}_{{\text{o}}}})}^{{1/2}}}{{U}^{{3/2}}}{\text{/}}{{d}^{2}}.$$
(2)

The argon ion mass is Mo = 40, and the average layer width d is 5 mm at the ion current density j = 2.76 A/m2 and the ion current Ii = jS = 2.76 × 0.72 = 2 А, where S = 0.72 m2 is the area of the chamber’s inner surface. For an ion energy of at most 0.5 keV, the coefficient of secondary ion–electron emission is γ ≈ 0.05 [13], while the ion current is virtually equal to the discharge current I. Therefore, for a discharge current I not higher than 2 A, the width of the layer between plasma and the chamber exceeds the width of the gaps between the grid plates, which is 4.5 mm, and is equal to the layer width between plasma and the accelerating grid.

The gas in the chamber is ionized mainly by electrons emitted by its walls. They are accelerated in the layer near the chamber wall, fly through plasma, and are reflected back into the plasma in the layer at the opposite wall. These electrons can fly through the plasma up to 100 times; the length of their path from the chamber wall to the anode is L = 4V/Sa, where V is the chamber volume and Sa is the anode surface area [14], and may exceed the chamber width by two orders of magnitude. For a direct current, e.g., I = 2 А, the discharge voltage U is independent of the gas pressure p until it decreases to the critical value po. For the pressure po, the path Λ, during the passage of which the chamber-emitted electrons spend their entire energy eU on gas ionization, increases to L. Each electron forms N = eU/W ions, where W is the gas ionization cost, which is W = 26 eV for argon [15]. The comparable contribution to the gas ionization is made by fast electrons that are produced in the layer near the chamber wall.

When an accelerating voltage Ua of up to 5 kV is applied to the grid, the width of the space-charge layer between the grid and plasma increases. Ions from the plasma that are accelerated in this layer fly into the gaps between the grid plates. Due to the inhomogeneity of the electric field at the edges of the plates, they are scattered at small angles. Therefore, ions touch the side surfaces of the plates and, as a result, turn into fast atoms of a beam with a rectangular cross section with a height of ~5 cm and a width of ~28 cm, which is injected into the housing.

The grid of plates with rectilinear edges that lie in the plane of the inner surface of the chamber (see Fig. 1) can be replaced with a grid with curved edges shown in Fig. 2. The sections of the plates of this grid that border the plasma emitter have the shape of a circular segment with a radius of 50 cm. Therefore, the trajectories of ions accelerated in the layer between the plasma emitter and the grid and flying into the gaps between the grid plates, as well as the trajectories of fast atoms, which fly out of the grid after neutralization of the ion charges, are directed to the center of this circle.

Fig. 2.
figure 2

Diagram of an installation with compression of a beam of fast atoms (top view).

Full size image

The width of the converging beam of accelerated particles, which is approximately equal to 30 cm near the plasma emitter, should decrease tenfold at a distance of 5 cm from the center of the circle, i.e., at a distance of 30 cm from the plane–parallel plates of the grid and at a distance of 10 cm from the removable flange. As a result, at a distance of 30 cm from the plane–parallel plates of the grid, the flow density of fast argon atoms in the beam increases tenfold compared to the beam without compression. Accordingly, the rate of substrate sputtering by them increases tenfold.

In view of the tenfold increase in the flow density of fast atoms, the power density transferred by them to the substrate at the emission current density  j = 2.76 А/m2 of ions with the energy eU = 400 eV, which fly into the grid with a transparency of η = 0.9, can be estimated as w = 10ηjU = 10 × 0.9 × 2.76 × 400 ≈ 104 W/m2. The surface power density of thermal radiation from the substrate R* is determined by its temperature T:

$$R{\text{*}} = \beta {{\sigma }_{{{\text{S}} - {\text{B}}}}}{{T}^{4}},$$
(3)

where σS–B = 5.7 × 10–8 W/(m2 K4) is the Stephan–Boltzmann constant, β is the substrate emissivity, and the temperature T is measured in kelvins. Considering that the substrate absorbs energy on one side and radiates in both directions, the balance of absorption and emission can be represented as:

$$w = 2R{\text{*}} = 2\beta {{\sigma }_{{{\text{S}} - {\text{B}}}}}{{T}^{4}}.$$
(4)

The emissivity β of the stainless-steel substrate was measured with an IMPAC IP 140 pyrometer. The substrate was heated to a temperature of 100°C, the pyrometer was pointed at it, and the emissivity regulator was rotated until the pyrometer reading coincided with the substrate temperature. The emissivity value was β = 0.8. With the power density w = 104 W/m2 transferred by fast atoms to the substrate, the substrate temperature that follows from Eq. (4) is T = (w/(2βσS–B))1/4 = (104/(2 × 0.8 × 5.7 × 10−8))1/4 = 575 K = 302°С.

To obtain the density distribution of the flow of fast atoms over the beam width, a titanium target in the form of a strip with a thickness of 2 mm, a length of 340 mm, and a width of 25 mm was installed on the stand inside the housing. Its polished surface that faced the accelerating grid was covered along the entire length with a mask in the form of 10-mm-wide titanium strip. After etching the target with a beam for 0.5–1 h, the substrate was cooled in a vacuum and removed from the housing; the mask was then removed from it and transverse profilograms of the substrate surface were obtained using a Dektak XT stylus profilometer manufactured by Bruker Nano, Inc. (United States). Measuring the height of the step between the sections of the profilogram corresponding to the open surface of the target, which was sputtered with fast atoms, and the surface covered with the mask gives the thickness of the layer that was removed by fast atoms. Dividing the thickness of the removed layer by the processing time gives the etching rate.

EXPERIMENTAL RESULTS

Studying imprints of the beam, which was formed by the grid with rectilinear edges (see Fig. 1), on the target has shown that, for an argon pressure of p = 0.05 Pa and a current of I = 2 A in the circuit of the gas-discharge chamber, the beam cross section virtually does not change with an increase in the distance h between the grid and target. The beam imprint width increases only slightly from 28 cm for h = 2 to 29 cm for h = 40 cm. The print height also increases from 5 cm for h = 2 to 5.5 cm for h = 40 cm. At the same time, as h increases, the imprint boundaries become diffuse. When the accelerating voltage Ua = 5 kV is applied to the grid, imprints become sharper, and their width and height retain constant values of 28 and 5 cm with increasing h.

Measurements of the stainless-steel substrate temperature T indicate that the temperature weakly depends on the distance h to the accelerating grid, and, when an accelerating voltage of Ua = 5 kV is applied to the grid, it rises from 55 to 350°С. According to (4), this corresponds to an increase in the energy flux density of substrate-heating fast argon atoms from 103 to 1.35 × 104 W/m2.

After a grid of plates with curved edges is installed, the temperature T of the stainless-steel substrate at an argon pressure of p = 0.05 Pa and a current of I = 2 А in the chamber circuit rapidly increases with an increase in the distance h to the plane–parallel plates of the grid from T = 410°С at h = 2 cm to T = 830°С at h = 30 cm (Fig. 3). The dependence of the temperature T on the distance h is due to the compression of a beam of fast neutral argon atoms that fly out of the grid and an increase in the energy flux density of substrate-heating fast argon atoms.

Fig. 3.
figure 3

Dependence of the temperature T of the substrate heated by fast atoms on the distance h to the plane–parallel plates of the grid.

Full size image

Beam imprints on flat targets show that, as the distance h increases, the height of the imprint retains a constant value of H = 5 cm, and the imprint width W decreases from W = 20 cm for h = 2 cm to W = 6 cm for h = 25 cm (Fig. 4).

Fig. 4.
figure 4

Dependence of the width W of the imprint of a beam of fast atoms on a flat target on the distance h to the plane–parallel grid plates.

Full size image

Figure 5 shows a profilogram of the surface of a titanium target in the form of a strip 34 cm long and 25 mm wide, which was sputtered together with the mask by a beam formed by the grid with rectilinear edges of the plates. Two sections are visible on it that correspond to the surface that was covered with a mask (on the left) and the open target surface of the target that was sputtered with fast atoms for 0.5 h. The height of the step between these areas is equal to the thickness of the removed titanium layer (2.2 μm). This value remains constant in the section of the 28-cm-long titanium strip, thereby indicating a uniform distribution of the flow density of fast argon atoms across the beam width. Dividing the thickness of the removed titanium layer by the time of its sputtering by fast argon atoms, we obtain the etching rate v = 4.4 μm/h.

Fig. 5.
figure 5

Profilogram of the target surface with a step between the surface covered by a mask (left) and the open surface sputtered by fast atoms (right).

Full size image

Figure 6 shows the distribution of the etching rate of the titanium strip by a beam formed by a grid with curved edges of the plates. The distribution was obtained at an argon pressure of p = 0.05 Pa, a current in the chamber circuit of I = 2 A, and an accelerating voltage of Ua = 5 kV at a distance from the plane–parallel grid plates of h = 25 cm. The etching rate retains a constant value of \({v}\) = 23 μm/h over the entire beam width (6 cm).

Fig. 6.
figure 6

Dependence of the etching rate v of the titanium strip located at the distance h = 25 cm from the plane–parallel plates of the grid on the distance R to the beam axis.

Full size image

After sputtering the polished titanium strip by fast argon atoms at an angle of incidence close to zero, the roughness of its surface increased significantly. To study the polishing process at a large angle of incidence, α = 80°, a titanium substrate with a diameter of 30 mm and a thickness of 6 mm was first fixed on a stand located at a distance of h = 25 cm from the accelerating grid, and its surface was sputtered for 1 h at a zero angle of incidence of argon atoms at a current in the chamber circuit of I = 2 A, an accelerating voltage of Ua = 5 kV, and an argon pressure of p = 0.05 Pa. The stand with the substrate was then removed from the housing and the surface roughness of the substrate was measured using a Dektak XT stylus profiler, which was Ra = 0.42 μm.

After installing the substrate holder at the bottom of the housing, a substrate with the measured surface roughness was inserted into its bowl and the surface of the substrate rotating at a speed of 8 rpm was sputtered for an hour at an angle of incidence of α = 80° of fast argon atoms at the same current in the chamber circuit I = 2 A, a voltage of Ua = 5 kV, and a pressure of p = 0.05 Pa. The substrate was cooled in a vacuum and removed from the bowl of the holder, and the roughness of its surface was measured, which was Ra = 0.08 μm. The substrate was then inserted back into the bowl of the holder and sputtered for another 2 h. As a result, the roughness decreased to Ra = 0.01 μm.

In order to study the possibility of polishing a surface with a higher initial roughness, substrates with a diameter of 30 mm and a thickness of 6 mm were manufactured using the additive technology method. A fine powder for additive manufacturing, which was obtained by gas atomization from a heat-resistant CoCrMo alloy, was selected as the starting material for their manufacture. The substrates were manufactured at an industrial EOS M 400 additive manufacturing plant using the selective laser-melting technology. Measurements using the Dektak XT stylus profilometer showed that the surface roughness of the manufactured substrate is Ra = 4.9 μm. After the surface of this substrate is sputtered for 2 h at an angle of incidence of α = 80° of fast argon atoms at the same rotational speed, the same current in the chamber circuit I = 2 A, and an accelerating voltage at the grid of Ua = 5 kV, the roughness decreased to Ra = 1.6 μm. As a result of subsequent processing of this substrate for another 2 h, the roughness decreased to Ra = 0.5 μm. After the third treatment of the substrate for another 2 h, the roughness decreased to Ra = 0.1 μm.

DISCUSSION

The obtained results confirmed the efficiency of polishing a metal surface with a beam of fast neutral argon atoms with an angle of incidence on the surface of 80°. The surface roughness of the rotating titanium substrate was reduced from Ra = 0.42 to 0.01 μm, which corresponds to the fourteenth surface finish class. To ensure this roughness during an acceptable processing time, a flow density of argon atoms that sputter the surface is needed, which exceeds the current density of argon ions from the plasma emitter by an order of magnitude. For this purpose, a beam of accelerated particles was compressed with a reduction by an order of magnitude of the beam cross-section width of ions extracted from the plasma emitter at a constant beam cross-section height of 5 cm. The plasma emitter is created by a glow discharge with electrostatic confinement of electrons in the gas-discharge chamber with a diameter of 50 cm. The ion current density on its wall remains constant at a distance of up to 15 cm from the chamber axis and only then begins to decrease to a minimum value at a distance of 25 cm from the axis in the angle between the flat and cylindrical surfaces of the chamber. This ensures the high uniformity of plasma near the hole in the wall with a width of 31 cm and a height of 6 cm, which is overlapped with an accelerating grid in the form of a set of plane–parallel plates.

Ions are accelerated from the plasma emitter by a voltage of negative polarity at the grid, fly into the gaps between its plates, and cause potential emission of electrons that neutralize the ion charge when approaching their surfaces. The transparency of the grid of 0.5-mm-thick plates spaced 4.5 mm apart is 90%. Therefore, the equivalent current of the beam of fast atoms flying out of the grid is up to 90% of the current in the grid circuit, and their energy is equal to the ion energy corresponding to the accelerating voltage between the plasma emitter and the grid.

If the sections of the grid plates facing the plasma emitter are made in the form of a segment of a circle, the ions and fast atoms formed inside the grid move to the center of this circle. This makes it possible to reduce the beam width by one order of magnitude and increase both the flow density of fast atoms and the rate of substrate sputtering by them. The sputtering rate also increases due to an increase in the sputtering coefficient with an increase in the energy of fast atoms to 5.5 keV. At the same time, the sputtering rate of the flat substrate surface is several times higher than the sputtering rate of its cylindrical surface since the sputtering coefficient increases with an increase in the angle of incidence of fast atoms on the sputtered surface.

CONCLUSIONS

1. The use of an accelerating grid in the form of a set of plane–parallel plates under a high negative voltage makes it possible to obtain beams of fast neutral atoms with the energy corresponding to the applied voltage and a small angular spread.

2. When sections of the grid plates that border the plasma emitter are made in the form of a circular segment, it is possible to reduce the beam width by one order of magnitude and increase the flow density of fast atoms and the rate of sputtering the substrate by them.

3. The beam compression significantly increases the efficiency of polishing the surface with a beam of fast neutral argon atoms with an angle of incidence on the surface of 80° and reduces the roughness of the titanium-substrate surface from Ra ∼ 0.4 to 0.01 μm, which corresponds to the fourteenth surface finish class.

4. With an increase in the initial surface roughness of the substrate (made of CoCrMo alloy) to Ra ∼ 5 μm, the time required for polishing it increases to an unacceptable value of ~10 h.