Time-resolved optical emission spectroscopy of

Home

Search

Collections

Journals

About

Contact us

My IOPscience

Time-resolved optical emission spectroscopy of pulsed DC magnetron sputtering plasmas

This content has been downloaded from IOPscience. Please scroll down to see the full text. 2005 J. Phys. D: Appl. Phys. 38 1769 (http://iopscience.iop.org/0022-3727/38/11/018) View the table of contents for this issue, or go to the journal homepage for more

Download details: IP Address: 114.214.163.235 This content was downloaded on 15/10/2013 at 08:52

Please note that terms and conditions apply.

INSTITUTE OF PHYSICS PUBLISHING

JOURNAL OF PHYSICS D: APPLIED PHYSICS

J. Phys. D: Appl. Phys. 38 (2005) 1769–1780

doi:10.1088/0022-3727/38/11/018

Time-resolved optical emission spectroscopy of pulsed DC magnetron sputtering plasmas J Lopez1 , W Zhu1 , A Freilich1 , A Belkind1 and K Becker1,2 1

Department of Physics and Engineering Physics, Stevens Institute of Technology, Hoboken, NJ, 07030, USA 2 Center for Environmental Systems, Stevens Institute of Technology, Hoboken, NJ, 07030, USA

Received 11 November 2004, in final form 15 February 2005 Published 20 May 2005 Online at stacks.iop.org/JPhysD/38/1769 Abstract Reactive magnetron sputtering of dielectrics using pulsed DC power in the frequency range between 5–350 kHz provides a deposition process without arcing. We studied the optical emission spectra of aluminium, argon and oxygen during the magnetron sputter deposition of Al in Ar and the reactive sputter deposition of Al2 O3 in an Ar/O2 gas mixture using a fast intensified CCD camera. Time-resolved as well as time-averaged optical emission spectroscopic studies were carried out. The time-resolved studies focused on the temporal behaviour of the various emissions during the decay of the plasma after the power is turned off. Decay times ranging from 1 to 4 µs were observed. A detailed analysis of the various processes that contribute to the emission of a particular emission line and its decay was carried out and an attempt was made to relate the various decay times to the dynamics of, respectively, the decay of the fast electrons, the Ar metastables, the Al atoms (metallic mode) and the O atoms (oxide mode).

1. Introduction In a direct current (DC) magnetron sputter deposition process, the cathode potential is always negative. During deposition of dielectrics all inside surfaces of the deposition chamber are eventually coated with non-conductive layers, on which electrical charges will accumulate. These charges create arcs, resulting in a highly non-uniform removal of material from the anode and in the formation of particulates. The presence of particles and their ultimate incorporation into the deposited films usually severely limits the use of such thin films in many practical applications [1–7]. In addition, arcing may damage the power supply. Pulsed modulation of the DC magnetron plasma solves most of the arcing problems as discussed in the literature [8, 9]. Pulsed DC unbalanced magnetron sputtering is a welldeveloped deposition technique for coatings and thin films and is widely used in industry [10] to deposit thin films such as alumina, Al2 O3 . Alumina is a promising coating material for wear and corrosion protection [11, 12] and in microelectronics [13]. Thin films deposited using pulsed 0022-3727/05/111769+12$30.00

© 2005 IOP Publishing Ltd

DC reactive sputtering have a smooth surface structure as observed in cross-section SEM studies due to the absence of particles created by micro-arcs [9, 14]. Using closed-loop control systems, alumina can be deposited with deposition rates comparable to the deposition rate of metallic aluminium. A typical pulsed power train used for pulsed DC reactive sputtering with a single magnetron is shown in figure 1. A high negative voltage pulse of a few hundred volts is applied to the target (anode) during the ‘on-time’, τon , for periods of microseconds to tens of microseconds. A short ‘off-time’, 1 of the duration of τon , separates τoff , which is typically 15 – 10 subsequent negative pulses. The applied voltage during the off-time is either zero or slightly positive (up to approximately 20–50 V). Therefore, the off-time is often labelled ‘reverse time’, τrev . During the on-time, sputtering takes place and all dielectric surfaces are gradually charging up. To avoid arcing, the optimum pulsing frequency and reverse time are routinely determined empirically and are optimized for a particular application and magnetron device. Two requirements must be met: (i) the on-time must be sufficiently short to prevent arc formation due to the charge accumulation in a single

Printed in the UK

1769

J Lopez et al

rev

on

cycle

Figure 1. Schematic diagram of the pulsed DC voltage applied to a magnetron.

on-time period and (ii) the duration of the reverse pulse must be sufficiently long, so that the dielectric surfaces can be fully discharged prior to reestablishing the plasma in order to avoid charge accumulation during sequential ‘on’ and ‘off’ cycles. The sum of the longest on-time and the shortest off-time also determines the lowest pulse repetition frequency (f = 1/τcycle , where τcycle is the sum of τon and τrev ) which is often referred to as the critical frequency, fc , at which pulsed DC reactive sputtering can be accomplished arc free. From a practical point of view, the deposition rate in magnetron sputtering is linearly proportional to the power deposited into the plasma and (almost) linearly proportional to the duty cycle, which is defined as the ratio τon /τcycle [9, 15] in the frequency range from about 2 to 250 kHz. The power deposited into the discharge decreases with increasing frequency [9, 15]. In recent years, the first time-resolved investigations of pulsed DC magnetrons operating in the metallic mode were carried out using Langmuir probes to obtain the time dependence of the electron density and the electron temperature during the off-time and on-time as well as the time dependence of the plasma potential and the floating potential [16–19]. Furthermore, time-resolved studies using mass spectrometry were carried out to shed light on the dynamics of the energetic ions in the pulsed DC magnetron plasma [20]. Recently, the first time-resolved optical emission spectroscopic investigations of the pulsed DC magnetron were reported [21–23]. In this paper, we present results of time-resolved optical emission studies of a pulsed DC unbalanced aluminium target magnetron. The main focus of this work is to analyse the temporal behaviour of argon, oxygen and aluminium emissions during the off-time in a pulsed DC magnetron plasma in Ar (metallic mode) and in an Ar/O2 mixture (reactive or oxide mode). The temporal behaviour of the intensities of argon (carrier gas), oxygen (reactive gas) and aluminium (target) emission lines were investigated in the plasma afterglow using time-resolved optical plasma emission spectroscopy employing a fast intensified CCD (ICCD). The observed decay times, which range from 1 to 4 µs, provide a convenient means of monitoring the decay of the fast electrons, Ar metastables, 1770

Al atoms (metallic mode) and O atoms (oxide mode) after the plasma power is turned off.

2. Experimental technique 2.1. Apparatus Our experiments were performed in a cubic reaction chamber (‘box coater’) with a planar rectangular magnetron, HRC-817 (BOC Coating Technology). The magnetron was unbalanced by replacing the centre magnets with soft iron and the side magnets with magnets that were 2.5 times stronger than the original magnets. The tangential component of the magnetic field below the racetrack of the unbalanced magnetron was about 40% higher than that of a balanced magnetron. The magnetic field in our entire system was mapped using a Bell 615 Gaussmeter. The component of the magnetic field parallel to the target surface (parallel component) at the centre of the racetrack was 700 G. The perpendicular component was 800 G at the edges of the target and 400 G at the centre of the target. At the substrate, both the parallel and perpendicular components of the field were in the range of 5–10 G. Sputtering was done using a 99.999% aluminium target of size 5 × 17 . The base pressure in the system was in the range of (4–8) × 10−4 Pa. During sputtering, the pressure was measured using a capacitance manometer and controlled using a throttle valve. Sputtering of Al was done in an argon background gas, while reactive sputtering of Al2 O3 was done in the oxide mode in an Ar/O2 mixture (mixing ratio ranging from 1 : 4 Ar-to-O2 to 1 : 1) in the pressure range of 0.4–1.3 Pa. The pulsed DC power was provided by a combination of an Advanced Energy® MDX-10 DC power supply and an Advanced Energy® Sparc-le V pulse generator. The DC power supply was operated in the constant current mode at currents of about 4 A. The Sparc-le V was operated in the constant power mode at repetition frequencies of 20 kHz with reverse pulse durations from 1 to 10 µs. Past experience has shown that these operating parameters provide very stable plasma conditions and avoid both macro- and microarcs. The current and voltage pulse forms were recorded

Optical emission spectra of plasmas

Voltage (V)

0 -300 -600 -900 2

Trigger (V)

Current (A)

0 -2 -4 -6

2

Delay

0

-40

-20

0

20

40

Time (µs) Figure 2. Typical waveforms of the (a) voltage and (b) current applied to the magnetron. A typical trigger pulse for the ICCD camera detector is shown in (c).

using a Tektronix® TDS 3034 oscilloscope connected to a Tektronix® P5205 high voltage differential probe and a Tektronix® TCP 202 current probe. The oscilloscope was set up to allow observation of the reverse peak and the following period of plasma reestablishment of both the current and the voltage (see figure 2). The light emission from the plasma was collected with a fibre optics cable and imaged onto the entrance slit of a 0.5 m SPEX Model 1870 spectrometer equipped with a 1200 groove/mm grating. The entrance slit of the spectrometer was set at 500 µm to obtain an acceptable signal-to-noise ratio with a spectral resolution sufficient to observe the Ar, Al and O emission lines. The dispersed plasma emission spectra were recorded by a Roper Scientific ICCD camera in the exit plane of the spectrometer. A Roper Scientific ST-133 controller was used to control the data acquisition. An electrical trigger for the time-resolved spectroscopic measurements was obtained from the DC pulsed power supply system with a Tektronix® P5200 high voltage differential probe and run into a BNC Model 555 pulse/delay generator which produced a 2 V TTL trigger pulse. A typical trigger waveform is shown in sequence with the voltage and current signals in figure 2. In order to avoid the deposition of Al or Al2 O3 on the tip of the fibre optics cable, the cable was enclosed in a tube and recessed. This arrangement restricted the viewing area of the optical detection system to a section of the densest part of the plasma from the racetrack region nearest to the fibre optics

cable. The shape of the area imaged was that of a truncated cone, 4 cm in length with top and bottom diameters of 1.7 cm and 2.3 cm, respectively. We estimate the time that it takes for species to diffuse out of this plasma volume by calculating an ‘effective length’, Leff , which is given as the ratio of the plasma volume to the plasma area. For our geometry, the effective length is about 1.5 cm. If Leff is larger than the mean free path, λ, the diffusion time, τD , is given by the ratio (Leff )2 /D, where D is the diffusion coefficient. If Leff is smaller than the mean free path, the diffusion time, τD , is given by the ratio Leff /vth , where vth is the mean thermal velocity of the species. Our operating conditions are such that Leff is comparable with the mean free path. In the case of the Ar atoms, the calculated values for the diffusion time, τD , range from 50 to 175 µs under these conditions, if we use the ‘high-pressure’ approximation (Leff λ). If we use the ‘low-pressure’ approximation (Leff λ), we obtain values of around 25 µs. Therefore, we estimate the actual diffusion times of the Ar atom for our operating conditions to be of the order of 30–40 µs. This is significantly longer than our typical off-times. Moreover, the viewing area is such that species will also diffuse into the viewing area of the detection system. Therefore, we do not expect a significant net change in the number of Ar atoms in the viewing range of the optical detection system during the off-time. By contrast, the residence time of the sputtered Al atoms in the viewing area of the optical detection systems is estimated to be only about 3.5 µs (assuming a kinetic energy of 1 eV) and is thus significantly shorter than the off-time. As the production of Al atoms stops essentially instantaneously when the power is turned off at the beginning of the off-time, diffusion of Al atoms into the viewing area can be neglected. Since the O atoms are produced by dissociation of O2 , their residence time depends critically on the kinetic energy imparted to the atoms in the dissociation process. This kinetic energy can range from nearthermal to an access kinetic energy of a few electronvolts, with the bulk of O atoms having kinetic energies below about 1.2 eV [24]. This results in residence times between about 1 and 20 µs, with the bulk of the O atoms having a residence time of around 3 µs in the viewing area of the detection system. However, O atoms will also diffuse into the viewing area and the rates of outflow and influx are fairly equal, so that there is probably no significant net change in the density of O atoms in the viewing area during the off-time. 2.2. Plasma emission spectroscopy Optical emission spectroscopy (OES) is one of the most widely used methods for determination of the concentration of plasma species and the diagnostics of plasma processes due, in part, to its relative simplicity and the fact that it is a non-intrusive technique [25–27]. Neutral particles and ions in the plasma can be identified by measuring the wavelengths and intensities of their emitted spectral lines. OES detects the intensity of spontaneously emitted radiation from excited plasma constituents. The recorded plasma emission intensity depends on instrumental parameters such as the solid angle of light collection, on the transmission and sensitivity of the optical detection system, on the transition probability of the observed transition and on the number density of emitting 1771

J Lopez et al

Table 1. Summary of the Ar, O and Al emission lines and their energy levels [28]. Species

Wavelength (nm)

Energy levels Ek –Ei (eV)

Transition

Argon I Argon I Argon I Argon I Argon I Oxygen I Oxygen I Aluminium I Aluminium I

404.44 750.39 751.47 763.51 772.38 777.19 777.42 394.40 396.15

14.69–11.62 13.48–11.83 13.27–11.62 13.17–11.55 13.15–11.55 10.74–9.15 10.74–9.15 3.14–0.000 3.14–0.014

3s2 3p5 (2P1/2 ) 5p 3s2 –3p5 (2P3/2 ) 4s 3s2 3p5 (2P1/2 ) 4p–3s2 3p5 (2P1/2 ) 4s 3s2 3p5 (2P3/2 ) 4p–3s2 3p5 (2P3/2 ) 4s 3s2 3p5 (2P3/2 ) 4p–3s2 3p5 (2P3/2 ) 4s 3s2 3p5 (2P3/2 ) 4p–3s2 3p5 (2P3/2 ) 4s 2s2 2p3 (4S˚) 3p–2s2 2p3 (4S˚) 3s 2s2 2p3 (4 S˚) 3p–2s2 2p3 (4 S)] 3s 3s2 (1S) 4s–3s2 (1S) 3p 3s2 (1S) 4s–3s2 (1S) 3p

species. Detailed discussions of the basic principles of OES and the advantages and disadvantages of its application to plasma diagnostics can be found in the literature (see, e.g. [25–27] and references therein to other publications), to which we refer the reader for further details. Only a brief synopsis will be given here, with particular emphasis on our particular application. A summary of the emission lines chosen for the present study along with the relevant energy levels involved in the transitions is presented in table 1. As can be seen, the minimum energies for the excitation of all emitting levels in argon and oxygen from the ground state of the respective atom are well above 10 eV, whereas the minimum excitation energy for Al is much smaller at around 3.15 eV [28]. Thus, the emission of the argon and oxygen lines depends on the presence of fast electrons in the plasma with energies well above 10 eV, whereas the aluminium emissions can also be excited by low-energy electrons with energies below 10 eV. We also note that some Ar lines terminate in a lower state that is short-lived (radiative), whereas others terminate in long-lived (metastable) lower states. In our studies, we selected two specific emission regions, each about 40 nm wide. The first emission window from 375 to 415 nm contains the aluminium emission lines at 394.4 and 396.15 nm along with the neutral argon emission line at 404.44 nm. The second window from 745 to 785 nm contains the neutral argon lines at 750.39, 751.47, 763.51 and 772.38 nm along with the oxygen lines at 777.19 and 777.42 nm. The number density of emitting atoms (Ar, O, Al) in our magnetron plasma is determined primarily by electron-driven processes such as electron impact excitation of the atom in its electronic ground state and by electron impact excitation of an excited metastable level that lies energetically below the emitting level. We note that the emitting O levels can also be populated by dissociative electron impact excitation of O2 which requires a minimum electron energy of more than 16 eV. The rate of formation of excited species via electron-driven processes depends essentially on three parameters, the electron density at a given electron energy (ne ), the electron temperature (Te or more precisely, the electron energy distribution function of the plasma electrons) and the density of the respective target species, NA (ground-state atom or metastable atoms). The emitting levels can also be populated radiatively via cascading from more higher-lying excited states that decay to the emitting level. Lastly, emitting levels can be populated collisionally by Penning excitation involving metastables with sufficient energy. In the present case, this applies to Penning excitation of oxygen by Ar metastables. The efficiency of forming 1772

excited species via Penning excitation depends on—among other things—the density of metastables, which in turn is also determined by the rate of the electron-induced formation of metastables from ground-state atoms. Time-resolved OES in the afterglow of a pulsed DC magnetron plasma allows us to elucidate the temporal behaviour of these parameters (ne , Te and NA ) as the plasma decays. Among the disadvantages of OES are the comparatively poor spatial resolution relative to, e.g. laser-induced fluorescence (LIF) spectroscopy, the difficulty in determining absolute emission intensities, the possible reabsorption and trapping of emission lines that terminate in the ground state of an atom or molecule which may affect the measurement of intensity ratios of lines and the fact that the density of groundstate species can only be inferred indirectly.

3. Results and discussion A time-resolved emission spectrum in the wavelength range from 375 to 415 nm at a frequency of 20 kHz and a reverse time τrev = 10 µs recorded in the metallic mode is shown in Figure 4. Figure 5 displays a time-resolved emission spectrum in the range from 755 to 785 nm in the oxide mode under the same pulsing conditions. Electrical probe measurements of the same plasma that were carried out in our group previously [19] showed that the electron energy distribution during the on-time in the near-substrate region is in a combination of a Maxwellian distribution with a mean energy of 2–4 eV (slow electrons or bulk electrons) and a high-energy component with electron energies of up to 40 eV (fast electrons or beam electrons). It is reasonable to assume that in the pulsed DC mode, the beam electrons disappear in about 1 µs or less once the power is turned off. Subsequently, the plasma density represented by the bulk electrons decays much more slowly. Experiments [19] have shown that the temperature of the bulk electrons decays initially roughly exponentially from 4 to 2.5 eV with a time constant of about 6 µs, followed by an even slower decay with a time constant in the range from 30 to 40 µs. The presence of the beam electrons and the bulk electrons in the electron energy distribution is also reflected in the measured time-resolved emission intensity of the various emission lines. Both timeresolved spectra shown in figures 3 and 4 show an increase in the emission intensity when the power is turned on until the emission intensity is constant during the remainder of the on-time. Once the power is turned off, the emission intensity decays. Different emission lines decay with different time constants. We note that the observed emission decay times are


Figure 3. Time-resolved emission spectrum of the aluminium lines (394.40 and 396.15 nm) along with the neutral argon line (404.44 nm) at a plasma power of 1 kW, a frequency of 20 kHz, a reverse time of 10 µs and a pressure of 0.5 Pa in pure argon.

Figure 4. Time-resolved emission spectrum of the neutral argon lines (750.39, 751.47, 763.51 and 772.38 nm) along with the oxygen lines (777.19 and 777.42 nm) at a plasma power of 1 kW, a frequency of 20 kHz, a reverse time of 10 µs and a pressure of 0.5 Pa in an Ar/O2 mixture with a ratio of 1 : 4.

much longer than the radiative lifetimes of the emitting levels. Thus, the observed decay is determined entirely by the plasma dynamics and can be related to the energy required to excite the emitting level and the mechanisms by which it is populated. Subsequent measurements were carried out by recording the time-resolved emission intensity of individual emission lines. Decay times were obtained by fitting an exponential function to the recorded intensity. In some cases, a doubleexponential was required to properly fit the data. This is illustrated in figure 5. Figure 5(a) shows the time-resolved emission of the Al line at 396.15 nm obtained in the metallic mode together with a single exponential fit yielding a decay

time of 3.88 ± 0.3 µs. Figure 5(b) shows the time-resolved emission of the O 777.19 nm line obtained in the oxide mode. Here a double-exponential fit was required to adequately fit the recorded data, yielding two decay times of 0.74 ± 0.3 µs and 3.97 ± 0.3 µs, respectively. In these measurements, the emission intensities were measured every 0.5 µs with an integration time of 2 µs, which allows us to determine decay times down to about 1 µs, which is roughly equal to the decay time of the fast beam electrons [19] in the afterglow. The accuracy with which the decay times can be determined ranges from ±0.1 µs for intense emission lines to about ±0.3 µs for the very weak emission lines. 1773

J Lopez et al

(a)

1.0

Al 396.15 nm Line

Intensity (a.u)

0.8

0.6

τ = 3.88 µs

0.4

0.2

0.0 0

5

10

Time (µs) (b)

1.0

τ = 0.74 µ s

O 777.19 nm Line

Intensity (a.u)

0.8

0.6

0.4

τ = 3.97 µ s

0.2

0.0 0

5

10

Time (µs) Figure 5. (a) Time-resolved intensity of the aluminium line at 396.15 nm obtained in the metallic mode at a plasma power of 1 kW, a frequency of 20 kHz, a reverse time of 10 µs and a pressure of 0.5 Pa in pure argon along with a single exponential fit (——) to the data corresponding to a decay time of 3.88 µs; (b) same as (a) for the O line at 777.19 nm (oxide mode) except for the fact that a double-exponential fit is required to adequately fit the data corresponding to an initial decay time of 0.74 µs and a later decay time of 3.97 µs (see text for details).

3.1. Time-resolved emission spectroscopic studies in the metallic mode Figure 6 shows the time-resolved emission of the five Ar lines (404.44, 750.39, 751.47, 763.51 and 772.38 nm). For clarity of presentation and to facilitate a comparison of the decay times of different emission lines, the emission intensities of the various lines have been normalized to unity at the point in time when the power is turned off. As can be seen the three lines at 404.44 nm 750.39 and 751.47 nm decay very rapidly once the power is turned off, with a time constant of about 0.95 µs. This is not unexpected as the main route to populating the emitting levels of these three lines is electron impact excitation of ground-state Ar atoms, which requires electrons with energies of more than 13 eV. Those energies are primarily found in the fast beam electrons, which disappear within a microsecond into the off-time. Thus, the time dependence of 1774

these emission lines can be used to monitor the decay of the beam electrons in the afterglow. By contrast, the Ar lines at 763.51 and 772.38 nm decay more slowly, with a time constant of about 3.3 µs, which is, however, much shorter than the time it takes the excited Ar atoms to diffuse out of the viewing area of our optical detection system (see discussion earlier). Since the lower level of both lines is metastable, the emitting level—in addition to being (re)populated by electron impact excitation from the Ar ground state—can also be populated by electron impact excitation of the metastable Ar atoms which requires a minimum electron energy of about 2 eV in the case of the Ar 763.51 nm line. We estimated the temporal behaviour of the Ar 763.51 nm emission due to excitation out of the metastable level using the published cross-section data of Lin and co-workers [29, 30]. The plasma electrons were represented by a Maxwellian energy distribution function corresponding to an initial electron temperature of 4 eV which was allowed to decay to 2.5 eV over a period of 6 µs as discussed earlier [19]. Another process that we had to take into account was the decrease in the Ar metastable density due to ionization out of the metastable level. This process has a threshold energy of slightly more than 3 eV and a cross-section that is about 25 times larger than the excitation cross-section out of the metastable level at 6 eV. A calculation using the same temporal behaviour of the electron temperature of the bulk electrons yielded a temporal behaviour of the density of Ar metastables in the off-time that decays with a time constant of about 5 µs. Combining both processes resulted in an emission decay rate for the Ar 763.51 nm line that was comparable (to within 30%) with the observed slow decay constant of 3.3 µs of the Ar 764 nm line. We therefore attribute the measured decay time of 3.3 µs of the Ar 763.51 nm line to the decay of the density of Ar metastables in the afterglow. The decay of the Ar 763.51 and 772.38 nm lines can thus be used to monitor the decay of the Ar metastable density in the plasma during the off-time. We note that a careful inspection of the decay of the three initially fast-decaying Ar lines (404.44, 750.39 and 751.47 nm) throughout the entire 10 µs off-time shows that these lines also decay with longer time constants of about 4–5 µs (which are difficult to determine accurately because of the low signal levels) after the rapid initial decay (see figure 6 and also figure 7(a)). We attribute this to the fact that the emitting levels of these three Ar lines can in principle also be excited from the Ar metastable states by the slow bulk electrons, but with crosssections that are much smaller than those for excitation of the emitting levels of the Ar 763.51 and 772.38 nm lines out of the metastable levels [29, 30]. Figure 7(a) shows the emission decay of the aluminium 396.15 nm line compared with that of the neutral argon neutral lines at 750.39 nm and 763.51 nm in the metallic mode at a frequency of 20 kHz and a reverse time of 10 µs. Figure 7(b) shows the temporal behaviour of the voltage applied to the anode and the discharge current. The decay of the two Ar emission lines has been discussed in the previous paragraph. The 396.15 nm aluminium emission line decays even more slowly than the Ar 763.51 nm line with a time constant of 3.9 µs. As the excitation of this line requires only a minimum energy of slightly more than 3 eV, one would expect a decay constant that follows the disappearance of the slow bulk


750.39 751.47 404.44 763.51 772.38

Intensity (a.u.)

1.0

nm nm nm nm nm

0.5

0.0 0

5

10

Time (µs) Figure 6. Time-resolved emission intensity of the Ar lines at 763.51 and 772.38 nm in comparison with those of the Ar lines at 404.44, 750.39 and 751.47 nm.

electrons. However, the observed decay time is significantly shorter and, in fact, is comparable with the diffusion time of the excited Al atoms out of the viewing region of the optical detection system. This indicates that the decay of the Al emission line is mainly governed by the decay in the density of aluminium atoms in the afterglow. A similar time dependence was observed for the other aluminium line at 394.40 nm under the same operating conditions and also for other pulsing conditions. Thus, the temporal behaviour of the Al emission line can serve as a monitor for the density of Al atoms in the afterglow. We also see in figure 7(a) that the temporal behaviour of the three line emissions is very different when the plasma is reestablished at the end of the off-time when the power is turned on again. The Ar 750.39 nm line rises rapidly and overshoots the steady-state level of the emission intensity, which is reached in about 10 µs. The Ar 763.51 nm line exhibits a much less pronounced overshoot and reaches its steady-state intensity level after about 5 µs. By contrast, the intensity of the Al line increases much more slowly. There is no overshoot in the emission intensity and the line does not reach its steadystate intensity level until more than 10 µs after the voltage pulse has been applied. This can be understood when one considers the temporal behaviour of the voltage applied to the anode and the discharge current when the plasma is reestablished (figure 7(b)). When the voltage pulse is applied, we see a brief (2 µs) overshoot in the voltage (up to twice the nominal voltage setting) followed by a roughly 5 µs period during which the voltage gradually declines to its nominal value, which is reached about 7–8 µs after the voltage pulse is initiated. During this time, the discharge current at first oscillates for about 2 µs around a value that is about 50% of its steady-state

value and than gradually increases over the next 5 µs to its full value. The initial voltage overshoot produces a significant amount of energetic beam electrons, which in turn, leads to a significant population of the emitting level of the Ar 750.39 nm line during that period. Since the emitting level of the Ar 763.51 nm line is not as efficiently excited from the Ar ground state as that of the 750.39 nm, the overshoot in the emission intensity of the 763.51 nm line is less pronounced. The gradual increase in the emission intensity of the Al 396.15 nm line, on the other hand, is due to the slowly increasing Al number density, which is largely determined by the discharge current and, thus, rises only gradually, mimicking the increase in the discharge current. It has been demonstrated [21] that the ratio of the Ar emission lines at 750.39 and 751.47 nm depends critically on the electron temperature. An intensity ratio of 2 corresponds to an electron temperature of about 4 eV, whereas a ratio of less than 1.5 indicates an electron temperature of less than 1 eV. Thus, this intensity ratio is a sensitive monitor for changes in the electron temperature. Figure 8 shows the time-resolved intensities of the Ar 750.39 and 751.47 nm lines and their ratio obtained in our magnetron during the off-time. When the power is turned off, the intensity ratio drops from a value of about 2 (corresponding to an electron temperature of 4 eV) to a value of about 0.3 (electron temperature below 1 eV) with a time constant of about 1 µs. This observation confirms the rapid disappearance of the fast beam electrons in the afterglow when the power is turned off. In the above context it is important to keep in mind that the concept of an ‘electron temperature’ is an oversimplification in a pulsed DC magnetron plasma and should be used with caution. As discussed earlier, the electron energy distribution of the plasma electrons during 1775

J Lopez et al

Intensity (arb. units)

(a)

2.0

1.5

1.0

0.5

Ar, 750.39 nm Ar, 763.51 nm Al, 396.15 nm 0.0

2

(b) Voltage Current

0

-200 -2 -400

Current (A)

Voltage (V)

0

-4 -600 -6 -800 -10

0

10

20

30

Time (µs) Figure 7. (a) Time-resolved emission spectrum of the neutral argon lines at 750.39 nm ( ) and 763.51 nm ( ) along with the aluminium line ), and current, Ic (). Both data sets were taken at 1 kW, 20 kHz, 10 µs reverse at 396.15 nm (♦); (b) temporal behaviour of the voltage, Vc ( time and a pressure of 0.5 Pa in a pure Ar environment.

the on-time is a combination of the slow bulk electrons, whose energy distribution is near-Maxwellian, and the fast beam electrons, whose energy distribution is far from a Maxwellian. Thus, a ‘high electron temperature’ during the on-time is simply an indication of the presence of these fast beam electrons in the energy distribution. It should not be interpreted as a temperature associated with a Maxwellian distribution. By contrast, the low electron temperature during the latter part of the off-time is an indication that the fast beam electrons have disappeared and the remaining electron energy distribution is dominated by slow bulk electrons with a near-Maxwellian energy distribution. 1776

3.2. Time-resolved emission spectroscopic studies in the oxide mode Figure 9 is similar to figure 7 and shows the emission intensities of the neutral argon lines at 750.39 and 763.51 nm and the neutral oxygen line at 777.19 nm (figure 9(a)) as well as the temporal variations of the cathode voltage and the current. Both Ar emission lines exhibit essentially the same behaviour as in the metallic mode (see figure 7(a)) with decay constants of 1.2 µs (750.39 nm) and 3.3 µs (763.51 nm). The 777.19 nm O line appears to decay in a fashion similar to the Ar 763.51 nm line. However, a more detailed analysis (see figure 5(b) and


(a) 14000 12000


10000 8000 6000 4000 2000

750.39 nm 751.47 nm

0 -5

0

5

10

15

20

25

30

Time (µs)

Intensity Ratio 750.39 nm/751.47 nm

(b) 2.0

1.5

750.39 nm/751.47 nm 1.0

0.5

0.0 -5

0

5

10

15

20

25

30

Time (µs) Figure 8. (a) Time-resolved emission intensity of the Ar 750.39 and 751.47 nm lines obtained at 1 kW, 20 kHz, 10 µs reverse time and a pressure of 0.5 Pa in a pure Ar environment; (b) intensity ratio of the Ar 750.39 and 751.47 nm lines during the off-time (see text for further details).

discussion earlier) shows that the decay of this line is more complex and requires a double-exponential fit yielding two decay times of 0.8 ± 0.3 µs and 4.0 ± 0.3 µs, respectively. The first decay constant is consistent with the approximately 1 µs decay constant observed for the fast decaying Ar emission lines and can thus be attributed to the decay of the fast beam electrons in the plasma afterglow. The emitting O atoms can in principle be produced in a single step dissociative excitation process from ground-state O2 molecules as well as via a twostep process, in which the O2 molecule is first dissociated

by electron impact into two ground-state O atoms and the O atoms are subsequently excited by a second electron collision process. The first process requires a minimum electron energy of more than 16 eV. The O2 neutral dissociation proceeds primarily through the A 3u+ and B 3u− states, which require minimum energies of about 6 eV and 8.4 eV, respectively [32]. The subsequent excitation of the O atoms from the ground state requires more than 10 eV. Thus, both pathways to the formation of excited O atoms require electrons of comparatively high energy, and it is not surprising that the temporal decay of the 1777

J Lopez et al

(a)

2.0


1.5

1.0

Ar, 750.39 nm Ar, 763.51 nm O, 777.19 nm

0.5

0.0

2

(b) 0

Voltage(V) Current (A) 0

-2 -400

Current (A)

Voltage (V)

-200

-4 -600 -6 -800 -10

0

10

20

30

Time (µs) Figure 9. (a) Time-resolved emission spectrum of the argon lines at 750.39 nm ( ) and 763.51 nm ( ) along with the oxygen line (♦) at 777.19 nm; (b) temporal behaviour of the voltage, Vc ( ), and current, Ic (). Both data sets were taken at 1 kW, 20 kHz, 10 µs reverse time and a pressure of 0.5 Pa in an Ar/O2 ratio of 1 : 4.

O emission line initially follows that of the disappearance of the fast beam electrons. The second decay constant of 4 µs is close to the decay constant obtained for the Ar 763.51 and 772.38 nm emission lines, whose decay is governed by the decay of the Ar metastable density. The fact that the O 777.19 nm line decays with a similar time constant suggests the presence of another route to populating the emitting O level that is related to the Ar metastable density, e.g. Penning excitation of oxygen by Ar metastables, which is exothermic by about 1 eV. Penning excitation processes with energy differences of up to 3 eV have been identified as important and abundant processes in 1778

planetary and cometary atmospheres (see, e.g. [33] and [34] and references therein to earlier work). Other channels that can populate the emitting O level include electron impact excitation out of the metastable O(1D) and O(1S) states, which requires also comparatively high minimum excitation energies of, respectively, 8.77 eV and 6.55 eV. While the cross-section for exciting the O(1 D) is among the largest excitation crosssections for the O atom with a peak value of 0.28 × 10−20 m2 at 6 eV [35], neither measured nor calculated cross-sections are available for electron excitation processes out of either the O(1 D) or the O(1 S). Population of the emitting O level via cascading from higher-lying O states is not expected to be


important. The cross-sections for excitation of the higherlying oxygen triplet state are much larger than those for exciting singlet or quintet states [35]. However, the higherlying triplet states cannot decay radiatively to the upper level of the O 777.19 nm line via dipole-allowed transition. The lack of reliable cross-section data renders it difficult, if not impossible, to asses to what extent the secondary routes that can populate the upper level of the O 777.19 nm line impact the observed O decay constants. Lastly, we note that O atoms which are produced with significant access kinetic energy in the O2 dissociation (more than 1.2 eV) have a residence time in the viewing area of our optical detection system that is comparable or even shorter than the observed 4 µs decay time. However, the bulk of the O atoms have access kinetic energies below about 1.2 eV [24] and residence times ranging from about 6 to 20 µs. It is difficult to assess quantitatively to what extent, if at all, these residence times impact the measured decay time of the oxygen line.

process and explains the second decay constant, which is close to that time constant that we found for the decay of Ar metastables. However, other mechanisms that may contribute to the population of the emitting level of the O 777.19 nm line such as excitation of the metastable O state can also not be ruled out.

Acknowledgments This work was supported in part by the US National Science Foundation (NSF). Partial financial support of this work by the NSF through an AGEP grant to JL is also gratefully acknowledged. JL would also like to further acknowledge a MAC Doctoral Fellowship from the State of New Jersey. Lastly, the authors would like to thank both reviewers of the original manuscript for their many constructive comments and helpful suggestions.

References 4. Conclusions We present results of time-resolved optical emission studies of a pulsed DC unbalanced magnetron plasma with an aluminium target with a focus on the temporal behaviour of argon, oxygen and aluminium emissions during the off-time in a pulsed DC magnetron plasma in Ar (metallic mode) and in an Ar/O2 mixture (reactive or oxide mode). The temporal behaviour of the intensities of argon (carrier gas), oxygen (reactive gas) and aluminium (target) emission lines were investigated in the plasma afterglow using time-resolved optical plasma emission spectroscopy employing a fast ICCD. The observed decay times, which range from 1 to 4 µs, provide a convenient means of monitoring the decay of the fast electrons, Ar metastables, Al atoms (metallic mode) and O atoms (oxide mode) after the plasma power is turned off. In the metallic mode, we found two decay times for two groups of Ar emission lines. A fast decay for the Ar 404.44, 750.39 and 751.47 nm lines with a time constant of about 1 µs represents the decay of the fast beam electrons in the plasma after the power is turned off. A slower decay time of about 3.2 µs for the Ar 763.51 and 772.38 nm lines is indicative of the decay of the Ar metastable density in the plasma afterglow. The 396.15 nm aluminium emission line decays with a time constant of 3.9 µs, which we attribute to the decay in the density of aluminium atoms in the afterglow. Thus, the temporal behaviour of the Al emission line can serve as a monitor for the density of Al atoms in the afterglow. In the oxide mode, the decay of the Ar emission lines is essentially identical to that in the metallic mode. The 777.19 nm O line shows a two-step decay with decay times of 0.8 and 4.0 µs. The emitting O atoms are produced via two electron impact channels, dissociative excitation of O2 and neutral dissociation of O2 followed by excitation of the O atom. Both pathways require electrons of comparatively high energy and it is thus not surprising that the temporal decay of the O emission line initially follows that of the disappearance of the fast beam electrons. The fact that the O 777.19 nm line at later times decays with a time constant similar to that of the Ar 763.51 and 772.38 nm lines suggests that the Penning excitation of oxygen by Ar metastables may be an important

[1] Schiller S, Goedickie K, Reschke J, Kirchhoff V, Schneider S and Milde F 1993 Surf. Coat. Technol. 61 331 [2] Scholl R 1993 Proc. 36th Annu. Technol. Conf. Soc. Vac. Coaters (SVC) p 405 [3] Scholl R 1994 Proc. 37th Annu. Technol. Conf. SVC p 312 [4] Sproul W D, Graham M E, Wong M S, Lopez S, Li D and Scholl R A 1995 J. Vac. Sci. Technol. A 12 1188 [5] Sellers J 1998 Surf. Coat. Technol. 98 1245 [6] Safi I 2000 Surf. Coat. Technol. 127 203 [7] Carter D, Walde H, McDonough G and Roche G 2002 Proc. 45th Annu. Technol. Conf. SVC p 470 [8] Westwood W D 1990 Handbook of Plasma Processing Technology ed S M Rossnagel et al (Park Ridge, NJ: Noyes) p 233 [9] Belkind A, Freilich A and Scholl R 1998 Surf. Coat. Technol. 108–109 558 [10] Rossnagel S M 1990 Handbook of Plasma Processing Technology ed S M Rossnagel et al (Park Ridge, NJ: Noyes) p 160 [11] Schneider J M, Sproul W D, Voevodin A A and Matthews A 1997 J. Vac. Sci. Technol. A 15 1084 [12] Zhu S, Wang F, Lou H and Wu W 1995 Surf. Coat. Technol. 71 9 [13] Bathia C S, Guthmiller G and Spool A M 1989 J. Vac. Sci. Technol. A 7 1298 [14] Cremer R, Reichert K, Neuschütz D, Erkens G and Leyendecker T 2003 Surf. Coat. Technol. 163–164 157 [15] Belkind A, Freilich A and Scholl R 1999 Using pulsed direct power for reactive sputtering of Al2 O3 J. Vac. Sci. Technol. A 17 1934 [16] Bradley J W, Bäcker H, Kelly P J and Arnell R D 2001 Surf. Coat. Technol. 135 221 [17] Bradley J W, Bäcker H, Kelly P J and Arnell R D 2001 Surf. Coat. Technol. 142–144 137 [18] Bäcker H, Henderson P S, Bradley J W and Kelly P J 2003 Surf. Coat. Technol. 174–175 909 [19] Zhao Z, Belkind A, Carter D, McDonough G, Roche G and Scholl R 2001 Time-resolved investigation of pulsed DC magnetron plasma Abstracts, Int. Conf. on Metallurgical Coatings and Thin Films (ICMCTF) (San Diego, CA: USA) p 5 [20] Miˇsina M, Bradley J W, Bäcker H, Aranda-Gonzalvo Y, Karkari S K and Forder D 2002 Vacuum 68 171 [21] Cornett M, George M, Walde H, Casson L and Pini R 2002 Proc. 45th Annu. Technol. Conf. SVC p 335 [22] Pajdarová A D, Vlˇcek J, B˘elshy P and Musil J 2003 Proc. 24th Int. Conf. on Phenomena in Ionized Gases (ICPIG) (Greifswald, 2003) vol 1, p 139

1779

J Lopez et al

[23] Lopez J, Zhu W, Freilich A, Belkind A and Becker K 2004 Optical studies of the spatio-temporal characteristics of a pulsed DC magnetron plasma 57th Gaseous Electronics Conf. (GEC) (Bunratty, Ireland) Bull. APS 49 (5) 74 [24] Cosby P C 1993 J. Chem. Phys. 98 9560–9 [25] Donnelly V M 1989 Plasma Diagnostics vol 1 ed O Auciello and D L Flamm (Washington, DC: Academic) p 1 [26] Grill A 1993 Cold Plasma in Materials Fabrication (Piscataway, NJ: IEEE Press) p 138 [27] Griem H R 1997 Principals of Plasma Spectroscopy (Cambridge, UK: Cambridge University Press) [28] NIST Atomic Spectra Database Lines Database, http://physics.nist.gov/cgi-bin/AtData/lines form

1780

[29] Lin C C 2004 Control. Plasma Phys. 44 405–12 [30] Boffard J B, Piech G A, Gehrke M F, Anderson L W and Lin C C 1999 Phys. Rev. A 59 2749 [31] Deutsch H, Becker K, Matt S and Märk T D 1999 J. Phys. B: At. Mol. Opt. Phys. 32 4249 [32] Becker K, Schmidt M, Viggiano A A, Dressler R and Williams S 2004 Air plasma chemistry Non-Equilibrium Air Plasmas at Atmospheric Pressure ed K Becker et al (Bristol, UK: Institute of Physics Publishing) pp 124–82 [33] Johnson R E 1999 Energetic Charged-Particle Interactions with Atmospheres and Surfaces (Berlin: Springer) [34] Johnson R E 1989 Sputtering of a planetary regolith Icarus 78 206 [35] Zecca A, Karwasz G P and Brusa R S 1996 La Rivista Nuovo Cimento 19 1 and references therein