2. 1
Pulsed anodic arc discharge for the synthesis of carbon nanomaterials
Carles Corbella1,*
, Sabine Portal1
, Denis B. Zolotukhin1,2
, Luis Martinez1
, Li Lin1
, Madhusudhan
N. Kundrapu3
, Michael Keidar1
1
Department of Mechanical & Aerospace Engineering, George Washington University, 800 22nd
Street,
Northwest, Washington, DC 20052, United States.
2
Department of Physics, Tomsk State University of Control Systems and Radioelectronics, 40 Lenina
Prospect, Tomsk, 634050, Russia.
3
Tech-X Corporation, 5621 Arapahoe Ave. Suite A, Boulder, CO 80303, United States.
*E-mail: ccorberoc@gwu.edu
Pulsed arc discharges can improve arc control and tailor the ablation process in the production of
1-D and 2-D nanostructures from carbon anodes. In this work, low-dimensional carbon
nanoparticles have been generated by means of anodic arc discharge in helium atmosphere
excited with a square-wave modulated signal (1-5 Hz, 10% duty cycle). The discharges were
performed between two graphite electrodes with maximal peak current of 250 A and maximal
voltage of 65 V. The erosion rates and conversion efficiency of the ablated anode are compared
to reference samples grown in DC steady arc mode. Ablation rates in pulsed arcs are typically of
the order of 1 mg/s. Combination of fast Langmuir probe diagnostics and optical emission
spectroscopy provided plasma parameters of the discharges at the arc column. Ranges of 1016
-
1017
m-3
for electron density and 0.5-2.0 eV for electron temperature are estimated. The obtained
samples were characterized with Raman spectroscopy and scanning electron microscopy. The
deposit on the cathode after pulsed arc consisted of carbon nanostructures such as graphene nano-
platelets and carbon nanotubes. Erosion dynamics of pulsed arc discharge has been described in
terms of a global model and compared to steady arc discharge. A correlation is identified among
discharge regimes, optical emission patterns and ablation modes. In conclusion, pulsed anodic arc
discharge is a very efficient source of carbon nanomaterials. The large control of the discharge
characteristics will permit to tailor accurately the production and the properties of carbon
nanotubes and graphene. This deposition method is promising for the fabrication of
semiconducting nanomaterials with tuneable electrical and optical properties.
Keywords: anodic arc discharge; carbon nanostructures; plasma diagnostics; pulsed power.
Page 1 of 36 AUTHOR SUBMITTED MANUSCRIPT - PSST-102916.R1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
AcceptedM
anuscript
3. 2
1. Introduction
Vacuum and atmospheric arc discharges find many applications in material processing
technology [1–3]. In particular, the production of carbon nanotubes by arc discharge constitutes a
classical example of plasma synthesis application in the field of nanotechnology [4]. The extreme
plasma conditions found in anodic arc discharges, namely high plasma densities and
temperatures, are favourable for the nucleation of nanoparticles with excellent structural and
functional properties [5-8]. For instance, samples with large crystalline domains such as graphene
networks have been fabricated with this method [9,10]. Also, depending on the target material,
the growth process can take place in a non-reactive atmosphere, so that the material precursor can
consist solely of a solid state anode with well-defined properties. These facts, typically combined
with high ablation rates of the anode and the possibility to work at atmospheric pressure [3],
make anodic arc discharge an attractive state-of-the-art technique for nanomaterial synthesis.
Current research in the synthesis of nanomaterials by plasma arcs is focused in addressing
technological issues, like arc instabilities, which hinder control and reproducibility of the
deposition process [11]. Oscillations of arc current and voltage, which were studied in
combination with fast camera imaging, showed chaotic behaviour resulting in poor selectivity in
carbon nanotube growth [12]. Another relevant problem is the high gas temperature associated to
the arc column, which locally overcomes 6000 K [13], thereby challenging the integrity of
electrode materials. Even refractory materials such as tungsten can be damaged due to prolonged
contact with arc plasmas. Concerning the process efficiency, it should be pointed out the eventual
production of undesired powder basically consisting of amorphous carbon during arc processes,
which reduces both the quality and the production rate of the fabricated nanomaterials [14]. This
Page 2 of 36AUTHOR SUBMITTED MANUSCRIPT - PSST-102916.R1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
AcceptedM
anuscript
4. 3
issue can affect the plasma density as well. Therefore, there is plenty of room for improvement in
control, efficiency, thermal load and selective growth in anodic arc discharge nanosynthesis. The
search for new cost-effective methods to produce nanomaterials by arc discharge is thus justified.
Pulsed discharges have been successfully implemented in different applications involving plasma
deposition methods [15,16]. For instance, plasma-enhanced chemical vapour deposition
(PECVD) of diamond-like carbon films has benefited from the adoption of pulsed-DC power
sources. Indeed, plasma parameters could be tuned by selecting the adequate peak voltage and
pulse frequency [17,18]. In another example, dense and high-quality coatings consisting of metals
or ceramics have been produced by high-power impulse magnetron sputtering (HiPIMS) [19,20].
This physical deposition method uses very intense and energetic plasma pulses during short time
intervals in order to sputter target material showing high ionization rates. A further advantage
characteristic of low- and middle frequency pulsed plasmas is their ease of up-scaling due to the
optimal control of the deposition processes [21,22]. These examples have demonstrated the
versatility of pulsed plasma techniques towards the design of deposition recipes of thin films and
coatings with tailored surface properties for multiple applications. Such recipes are being
systematically transferred to industrial processes nowadays [23].
This paper explores the characteristics of anodic arc discharges in pulsed mode (1-5 Hz) aimed to
the synthesis of 1-D (nanotubes) and 2-D (graphene) carbon nanomaterials. Standard
nanosynthesis processes require the use of hollow anodes filled with catalyst in order to promote
nanoparticle nucleation on hot substrates. Here, the main objective is the study of pulsed arc
discharges and, therefore, the experiments were performed with solid graphite anode. Stable and
self-sustained periodic arcs using carbon electrodes are feasible thanks to the special physical
Page 3 of 36 AUTHOR SUBMITTED MANUSCRIPT - PSST-102916.R1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
AcceptedM
anuscript
5. 4
properties of graphite: it shows high electric conductivity, which is required to generate arcs, but
on the other hand, its poor thermal conductivity restricts the plasma-heated region of the anode to
a small volume at the tip. Hence, the direct contact with the plasma arc keeps the anode so hot
that thermionic emission is supported even during the inactive period of discharge. This recursive
condition of hot anode is the key aspect to force continuity of the discharge during the whole arc
process. Here, we have characterized the pulsed carbon arcs by optical emission analysis and fast
Langmuir probe diagnostics. Evidence of the production of carbon nanostructures within pulsed
arc volume is provided by electron microscopy and Raman spectroscopy measurements on the
cathode surface. Finally, a global model describing the basic processes of carbon evaporation by
DC steady arcs and pulsed arcs is proposed. This study aims to set a milestone in the research of
carbon nanomaterials production by means of pulsed arc plasmas.
2. Experimental methods
2.1 Plasma chamber setup
The anodic arc discharges were performed in a plasma chamber thoroughly described elsewhere
[9]. Fig. 1 shows a schematic representation of the plasma arc chamber and the main electric and
gas flow connections. Briefly, it consisted of a cylindrical vacuum vessel with 270 mm in length
and 145 mm in diameter, filled with helium gas (99.999%) and pumped down by means of a
mechanical pump. If not otherwise stated, the working pressure was fixed to 300 Torr with a
background pressure lower than 0.1 Torr. The discharge was held between two floating
electrodes installed vertically and made of Poco EDM-3 graphite, namely an anode (3 mm
diameter) and a cathode (10 mm diameter) with a separation gap of between 2 mm and 5 mm.
The electrodes were initially in contact, and the discharge was initiated after energizing and then
Page 4 of 36AUTHOR SUBMITTED MANUSCRIPT - PSST-102916.R1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
AcceptedM
anuscript
6. 5
separating them. Although nanomaterials have been systematically grown on a substrate holder at
the lateral of the chamber, here we considered the nanostructures deposited directly on the
cathode surface. A Miller Gold Star 300SS DC power supply, used to ignite an support the arc
discharge, was remotely controlled by a waveform generator, which provided square pulse
signals with frequencies of 1 Hz, 2 Hz and 5 Hz and a duty cycle of 10% (pulse widths of 100
ms, 50 ms and 20 ms, respectively). The amplitude of the pulsed signal was set to the nominal
maximal voltage, 5 V, which allowed for full power arcs within the pulse width duration. The
time interval during which the electrodes are energized is labelled as ON, whereas the non-
energized electrodes interval is labelled as OFF. The resulting waveform of the arc discharge
current was measured with a current clamp. The waveforms of arc voltage, arc current and pulse
signal were registered with an oscilloscope. A port view with a strongly absorbing optical filter
attached was available to observe the discharge luminosity distribution.
FIG. 1. Schematic representation of the experimental setup. The arc power is controlled by a waveform
generator, which supplies square pulses with frequencies 1-5 Hz to the arc power supply. The following
signals are measured (oscilloscope): CH1 = cathode voltage; CH2 = anode voltage (CH2-CH1 =
discharge voltage); CH3 = discharge current; CH4 = pulse signal waveform.
Pump
Helium
Pump
Helium
Waveform
Generator
Oscilloscope
CH1 CH2 CH3 CH4
Current
Clamp
Page 5 of 36 AUTHOR SUBMITTED MANUSCRIPT - PSST-102916.R1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
AcceptedM
anuscript
7. 6
2.2 Discharge parameters
Fig. 2a shows the pulse parameters determining the amplitude and duration of the pulse. The
example shows the measured pulse signal and arc current waveforms. The evolution of discharge
current as a function of pulse frequency is depicted in Fig. 2b, where the discharge current
waveforms at 1 Hz, 2 Hz and 5 Hz are represented. All curves overlap during the rise time of the
pulse, i.e. the ON cycle, and they start a decay regime towards zero current just after the pulse
signal enters the OFF cycle. The characteristic rise and decay times are similar for all
frequencies. The arc current with the lowest frequency, 1 Hz, generally reaches a plateau of
around 250 A in 50 ms. Such saturation is also achieved in the case of 2 Hz. However, the arc
current evolution is interrupted during the current rise at 5 Hz, in which case the saturation value
is not reached because the ON phase only lasts 20 ms. Interestingly, the arc current corresponding
to DC arcs is smaller than the peak values above: it is stabilized at around 150 A. This value
measured for DC arcs, which is lower than the peak currents measured in pulsed arcs,
demonstrates that the arc current stabilizes towards a lower value in steady state conditions. It has
been observed that such stabilization takes place within 1 s after plasma ignition.
(a)
0.00 0.05 0.10 0.15
0
50
100
150
200
250
Arccurrent(A)
Time (s)
1 Hz
2 Hz
5 Hz
DC
(b)
0,0 0,2 0,4 0,6 0,8
Pulsed signal
Arc current
Signalvoltages(a.u.)
Time (s)
Duty cycle = ∆t/τ
τ
∆t
Page 6 of 36AUTHOR SUBMITTED MANUSCRIPT - PSST-102916.R1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
AcceptedM
anuscript
8. 7
FIG. 2. (a) Example of pulse signal and current waveforms, together with definitions of the parameters
pulse period and duty cycle. ON phase=∆t, OFF phase=τ-∆t (b) Discharge current waveforms obtained
at pulse frequencies of 1 Hz, 2 Hz and 5 Hz at full peak power. The arc current corresponding to steady
DC arc at full power is included, showing stabilization at a lower value.
Fig. 3 shows the average power, P, employed in steady DC discharges at different set currents.
The DC power values grow somewhat linearly from 1 kW to 10 kW within the current range 50-
150 A. This curve is compared with the power exhibited by the pulse counterpart, which showed
a power per pulse of approximately 1 kW independently of the pulse frequency (1-5 Hz). In the
case of pulsed arcs, the average power per pulse or power in pulse is defined as follows:
ttItVP d)()(
1
0
∫=
τ
τ
(1)
where V(t) is the arc voltage waveform, I(t) is the arc current waveform, and τ is the pulse period
(inverse of the pulse frequency). The inset in Fig. 3 shows example waveforms for arc voltage
and arc current which were taken to calculate the average power per pulse. Please check the
Supporting Information for further examples.
0 20 40 60 80 100 120 140 160
0
2
4
6
8
10
12
-0.4 -0.2 0.0 0.2 0.4
0
20
40
60
Dischargecurrent(A)
Dischargevoltage(V)
Time (s)
0
100
200
300
Dischargepower(kW)
Discharge current (A)
Average power
in pulsed arc
Page 7 of 36 AUTHOR SUBMITTED MANUSCRIPT - PSST-102916.R1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
AcceptedM
anuscript
9. 8
FIG. 3. Evolution of discharge power (Eq. 1) with current for steady arc. The dashed line shows the
average power in pulsed arc operation, which is calculated from current and voltage waveforms (inset:
example at 1 Hz, 10% duty cycle).
2.3 Characterization techniques
The composition and parameters of the plasma arcs were studied by Langmuir probe and optical
emission spectroscopy (OES) diagnostics. Current-voltage characteristics in the arc column and
its vicinity (1 cm away from arc centre) were taken with a fast Langmuir probe. It consisted of a
tungsten wire (0.5 mm in diameter and 20 mm in length) mounted on an actuator, whose motion
between arc centre and lateral was controlled with a waveform generator. The remaining time of
the electrostatic probe in the arc column was lower than 50 ms in order to minimize damaging of
the material (porosity increase and melting) [24]. A set of voltages between -40 V and 40 V were
provided by a second waveform generator to bias the electrostatic probe. The scanning frequency
of the triangular signal was 1 kHz. On the other hand, OES diagnostics was performed with a
UV-vis-IR StellarNet spectrometer within the spectral range of 191 nm and 851.5 nm with a
0.5 nm resolution. OES was performed by registering the plasma-emitted light transmitted
through a 2" glass window. The bonding structure of the deposited samples was characterized by
means of a Raman spectrometer Horiba LabRAM HR operated at a wavelength of 532 nm. The
morphological characterization was performed with a scanning electron microscope (SEM)
Tescan XEIA FEG SEM at an accelerating voltage of 10 kV.
3. Experimental results
3.1 Erosion dynamics
Page 8 of 36AUTHOR SUBMITTED MANUSCRIPT - PSST-102916.R1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
AcceptedM
anuscript
10. 9
Table 1 summarizes the main discharge parameters corresponding to steady DC arcs and pulsed
arcs. The DC arcs were performed at 60 A and 150 A, which correspond respectively to typical
values used in carbon nanoparticle synthesis and the large current value asymptotically reached in
full power operation (see Fig. 2b). The pulsed arcs were always performed at maximal peak
power of around 250 A. Consistently with the evolution shown in Fig. 3, the DC arc current
provided total power values of 2 kW and 10 kW for 60 A and 155 A, respectively. On the other
hand, the average power in pulsed operation remained always within ~1 kW range. The
resistance of the plasma arc was calculated by dividing the peak voltage with the peak current. In
general, all discharges showed minimal resistances comprised between 0.1 Ω and 1 Ω. In
particular, the pulsed discharges exhibited resistances two to three times lower than DC
discharges. The plasma conductivity at low frequencies, σp, scales with the plasma density, n0, as
)/(0
2
mp mne υσ = , where e is the elementary charge, m is the electron mass, and νm is the
electron-neutral collision frequency. Therefore, pulsed arcs might have plasma densities around
two to three times higher than in DC arcs assuming a constant νm value.
TABLE 1. Comparison of erosion parameters in DC and pulsed anodic arc discharges of carbon at 300
Torr He. The pulsed processes were held with a duty cycle of 10%.
Frequency
(Hz)
Peak arc
current
(A)
Peak arc
voltage
(V)
Average
power
(kW)
Ablation
rate
(mg/s)
Rate per
pulse
(mg/s)
Efficiency
(g/Kwh)
Min R
(Ohm)
DC 60 35 2.1 2.1 2.1 3.5 0.6
DC 150 65 10 22 22 8 0.4
1 250 50 1.0 1.0 10 4 0.2
2 250 50 1.1 1.0 10 3.3 0.2
5 180 50 0.9 0.8 8 3.4 0.2
Page 9 of 36 AUTHOR SUBMITTED MANUSCRIPT - PSST-102916.R1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
AcceptedM
anuscript
11. 10
Discharge voltage has not been a constant parameter during DC arc although it has varied very
little. As stated earlier, the electrodes were initially separated (approximately 1 mm) to ignite the
discharge. After that, none of the electrodes was moved and the inter-electrode gap increased
naturally by anode consumption. As a consequence, the arc voltage increased gradually from
32 V to 35 V during the DC arc discharge at 60 A. Also smoothly, arc voltage increased from
63 V to 65 V during the DC arc discharge held at 150 A. The discharge characteristics are not
expected to change to a large extent due these small variations of arc voltage.
The ablation rate values listed in Table 1 were calculated by dividing mass variation of the
carbon anode for each process with the total arcing time. The anode masses before and after
arcing were measured with a precision microbalance. The highest ablation rate, ~20 mg/s, was
obtained in DC arcs at ~150 A, whereas values of ~2 mg/s were measured for DC arcs at 60 A.
Finally, values of ~1 mg/s were characteristic for pulsed arcs. The rate per pulse provides a
benchmark to compare erosion rates of pulsed and DC arcs, namely this parameter quantifies the
removed mass within the pulse ON phase. The rate per pulse values observed in pulsed arcs are a
few times higher than the ones measured in 60 A DC discharges. However, the full power DC
discharge held at 150 A shows the highest absolute rate. A similar conclusion can be extracted
from the efficiency values, which measure the total ablated material per unit of time and of
power. This parameter is of industrial interest, since it quantifies the power cost of mass erosion
by arc discharge. In summary, DC arcs at 60 A and pulsed arcs show similar efficiencies, which
suggests that pulsed anodic arc is a priori a cost-effective process which is competitive with
conventional DC processes. It is worth noting the extraordinarily high efficiency reached in the
case of 155 A DC arcs. However, it should be pointed out that such high-current process is at cost
Page 10 of 36AUTHOR SUBMITTED MANUSCRIPT - PSST-102916.R1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
AcceptedM
anuscript
12. 11
of system overheating and risk of electrode damaging, while pulsing mode provides controllable
ablation in more stable and reproducible processes.
Fig. 4 shows the optical emission dynamics observed in this study. Besides, further images are
provided in the Supporting Information. The following emission patterns are characteristic in DC
and pulsed arcs:
• Steady DC arc: Initially, there is a luminous region centered at the inter-electrode gap, which
comprises part of anode and of cathode: INTERMEDIATE ARC. After a few seconds, such
profile changes into a discharge preferentially surrounding the anode tip. In other words, the
luminous arc region leaves the inter-electrode gap to become an ANODIC ARC. This is the
arc optical emission at steady state.
• Pulsed arc: The discharge shows a strong emission at the intermediate region between anode
and cathode during the pulse ON phase. This emission profile is qualitatively identical to that
observed in the initial phase of the DC arc – INTERMEDIATE ARC. Following the end of the
ON phase, the OFF phase is characterized by the vanishing of the initial strong emission and,
next, a tiny glowing of the HOT ANODE tip. The discharge never enters into the anodic arc
regime observed in the steady state of the DC arc, but it remains showing an intermediate arc.
Maybe, for this reason, the average electric resistance in the pulsed arc discharge is lower
than in the DC arc discharge: charge transfer is more effective in pulsed arcs due to plasma
attachment on both electrodes, whereas there is no plasma attachment to cathode in the anodic
arc, and therefore electric resistance is higher in this case.
Page 11 of 36 AUTHOR SUBMITTED MANUSCRIPT - PSST-102916.R1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
AcceptedM
anuscript
13. 12
FIG. 4. Examples of optical arc emission, together with the observed glowing phases, and the resulting
anode erosion profiles for (a) steady DC arc at 150 A for ≈5 s and (b) pulsed arc at 1 Hz (10% duty cycle)
and 250 A peak for ≈30 s. There is a correlation between operation mode and anode tip ending (indicated
with an arrow) after the process: pulsed arcs provide anode tips with flat ending, whereas steady DC arcs
show the tendency to sharpen the anode tip
It is worth to mention that the anode tip shapes appear correlated with the observed emission
patterns (Fig. 4). On one side, DC arcs tend to leave rounded anode tips, being sharpened and
sometimes broken in case of high current arcs (150 A). On the other one, pulsed arcs tend to
leave flat-end anodes. Both effects of the arc discharges, i.e. sharpening and flattening effects,
could be explained in terms of the light emission profiles characteristic of each discharge. DC
arcs show most of the time anodic arc pattern, which would account for the rounded anode tip
(a) (b)
ON
OFF
Intermediate
arc
Anodic
arc
Pulsed, 1 HzSteady DC
Cathode Cathode
Anode Anode
Intermediate
arc
Hot anode
Page 12 of 36AUTHOR SUBMITTED MANUSCRIPT - PSST-102916.R1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
AcceptedM
anuscript
14. 13
shapes because the active part of the discharge is uniformly distributed around the anode tip.
Indeed, the whole tip undergoes sublimation due to the high-temperature plasma. Concerning
pulsed arcs, the flat-end tips might owe to the preferential presence of the discharge at the
intermediate region between electrodes. Hence, the erosion is basically restricted at the very
surface of the anode tip, leaving thus a flat-end tip. The sub-surface layers remain colder than in
the anodic arc scenario. Therefore, carbon is preferentially evaporated from the outermost layer
of the anode tip. Further results are provided in the Supporting Information.
3.2 Plasma diagnostics
The optical plasma emission has been analyzed by OES. Fig. 5 compares typical emission spectra
of DC arc discharge and pulsed (1 Hz) arc discharge at 300 Torr. All spectra, whose profiles were
corrected from the optical transmittance of the window, show characteristic emission lines from
molecular C2 superposed to a continuous background. Such background is associated with
thermal emission from hot anode and hot gas discharge. The DC discharge shows a stable
spectrum. Such discharge was performed within the range 40-60 A, which are usual values of arc
current to synthesize nanoparticles. In contrast, the pulsed discharge did not show a stable
emission spectrum, but its emission consisted of two characteristic phases:
(1) Strong emission phase connected to the ON cycle and consisting of emission lines and
continuous background. The presence of discrete and continuous contributions suggests that this
phase shows relevant plasma activity (formation of C2 molecules) within a hot gas atmosphere.
(2) Weak emission phase correlated to the OFF cycle with dominating thermal background,
suggesting that plasma activity is marginal within a colder gas atmosphere.
Since all measurements were performed at identical integration time of the spectrometer (1 ms),
the corresponding intensities can be compared. The total intensity of the emission during the ON
Page 13 of 36 AUTHOR SUBMITTED MANUSCRIPT - PSST-102916.R1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
AcceptedM
anuscript
15. 14
cycle of the pulsed discharge roughly doubles the emission intensity of the DC discharge. This
result might owe to the higher plasma density expected in pulsed discharges compared to DC
discharges. Such contrast in emission intensities is also consistent with the lower effective
impedances and higher effective ablation rates measured in pulsed discharges (0.2 Ω and
10 mg/s) in relation to DC discharges (0.6 Ω and 2 mg/s).
FIG. 5. Optical spectra of the plasma arc emission in DC mode (black line) and pulsed mode (red and
blue lines) at 300 Torr. Comparison of brightness and extension of the plasma during on and off cycles (1
Hz, 250 A peak). Plasma emission during ON cycle (red line) is dominated by C2 transitions, whereas
emission is dominated by thermal background during OFF cycle (blue line).
The spectral emission lines observed in the carbon arc discharges correspond to the Swan system
of the C2 vibrational transitions [25,26]. Fig. 6 displays an example emission spectrum together
with the assigned transitions of molecular C2 from upper state ν'' to lower state ν'. The intensities
300 400 500 600 700 800
0
1000
2000
3000
4000
Intensity(a.u.)
Wavelength (nm)
Cycle ON
Cycle OFF
40 A
Current ON
(100 ms)
Current OFF
(900 ms)
Arccurrent
Time (s)
Emission
dominated by
C2 transitions
Emission
dominated
by thermal
background
300 400 500 600 700 800
0
1000
2000
3000
4000
Intensity(a.u.)
Wavelength (nm)
Cycle ON
Cycle OFF
40 A
Current ON
(100 ms)
Current OFF
(900 ms)
Arccurrent
Time (s)
Emission
dominated by
C2 transitions
Emission
dominated
by thermal
background
DC
Page 14 of 36AUTHOR SUBMITTED MANUSCRIPT - PSST-102916.R1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
AcceptedM
anuscript
16. 15
of the spectral lines, Iν'ν'', are correlated to the energies of the vibrational transitions, Evib,
according to the Boltzmann relationship [26]:
vibB
vib
'''
4
'''
'''
ln
T
E
const
S
I
κσ υυυυ
υυ
−=
(2)
Here, σν'ν'' is the inverse of the wavelength; Sν'ν'' is the band strength, whose values corresponding
to the transitions are tabulated elsewhere [25]; Tvib is the vibrational temperature of the gas, and
κB is the Boltzmann constant. The values of Evib for ν''=0,1,2,3,4 and ν'=0,1,2,3,4 were extracted
from literature [27]. Linear fit analysis on the Boltzmann plot provides vibrational temperatures
comprised between 4000 K (0.3 eV) and 12000 K (1.0 eV). A more accurate estimation is not
possible because the analyzed volume of the plasma arc includes regions with important gradients
in plasma density and temperature [13,26]. Nevertheless, this analysis has provided an average of
≈0.7 eV for the gas temperature in the plasma arc, which will be compared to the electron
temperature measurements by Langmuir probe diagnostics reported below.
FIG. 6. Optical spectrum of the plasma arc emission, where the characteristic peaks of the C2 Swan
system are identified. On the right, Boltzmann representation of Swan band peaks. The slope on the linear
300 400 500 600 700 800
0
500
1000
1500
2000
Intensity(a.u.)
Wavelength (nm)
∆ν=-2∆ν=-1
∆ν=0
∆ν=2
∆ν=1 C2 transitions
d3Πg a3Πu
C2 Swan band
300 400 500 600 700 800
0
500
1000
1500
2000
Intensity(a.u.)
Wavelength (nm)
∆ν=-2∆ν=-1
∆ν=0
∆ν=2
∆ν=1
300 400 500 600 700 800
0
500
1000
1500
2000
Intensity(a.u.)
Wavelength (nm)
∆ν=-2∆ν=-1
∆ν=0
∆ν=2
∆ν=1 C2 transitions
d3Πg a3Πu
C2 Swan band
0 2 4 6 8 10
Ln[Iν''ν'
/(σν'ν''
4
Sν'ν''
)]
Evib
(x10
3
K)
vibB
vib
'''
4
'''
'''
ln
T
E
const
S
I
κσ υυυυ
υυ
−=
0 2 4 6 8 10
Ln[Iν''ν'
/(σν'ν''
4
Sν'ν''
)]
Evib
(x10
3
K)
vibB
vib
'''
4
'''
'''
ln
T
E
const
S
I
κσ υυυυ
υυ
−=
Page 15 of 36 AUTHOR SUBMITTED MANUSCRIPT - PSST-102916.R1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
AcceptedM
anuscript
17. 16
fit analysis provides an average vibrational temperature of ≈0.7 eV for the plasma arc volume in DC arc
discharges and in pulsed arc discharges.
Plasma parameters at the arc column (electrode gap) and its vicinity were obtained by means of
Langmuir probe diagnostics using a fast electrostatic probe. Fig. 7a shows the measuring
positions of the probe. Due to the small surface area of the probe, it was required to operate the
arc discharges at 10 Torr He atmospheres in order to measure currents with high signal-to-noise
ratio. The anodic arc discharges were held within the arc current interval 40-60 A to avoid
melting of the probe components. Please note that such plasma conditions are not compatible
with pulsed operation mode, which requires higher pressures (>100 Torr) and higher peak (>60
A) currents. Therefore, arcs were sustained only in DC mode. Fig. 7b shows an example of V-I
characteristics measured during the arc discharge. The recorded characteristics taken at the arc
column and 1 cm away exhibit similar shapes. A few examples of V-I plots can be found in the
Supporting Information. These characteristics show saturation at the positive voltage branch,
which is ascribed to electron saturation. On the other hand, the negative voltage branch does not
reach saturation as it would be expected or ion current region. Instead, a large total current
exceeding the value observed for electron saturation is measured. This excess in ion current is not
consistent with the theory of cold Langmuir probes in non-reactive atmospheres. The very large
currents at the negative branch owe to extra electron emission generated by the following
mechanisms, besides ion current: (1) Auger emission due to influx of helium metastables [28]
and (2) thermionic emission due to the high gas temperatures (emissive probe effect). The
separation of electron fluxes due to ion current, Auger emission and thermionic emission is out of
the scope of this article, and it will be studied in a future work. However, it should be pointed out
Page 16 of 36AUTHOR SUBMITTED MANUSCRIPT - PSST-102916.R1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
AcceptedM
anuscript
18. 17
that these effects are higher than those observed earlier probably due to the discharge proximity
[28].
FIG. 7. Plasma parameters of steady arc at 60 A and 10 Torr have been measured by means of a fast
Langmuir probe within a region < 1 cm from the arc column. (a) Image of the fast probe in motion (top
view of the reactor). (b) Representative V-I characteristics together with the main contributions to the
probe current. The plasma parameters were obtained from the electron saturation region because ion
saturation is masked by further contributions active at the negative branch of bias voltage.
Plasma parameters have been evaluated from the positive voltage branch of the V-I curves by
assuming a one-temperature Maxwellian electron distribution. Analysis of the V-I plots has
provided electron densities of the orders of 1016
-1017
m-3
and electron temperatures between 0.5
eV and 2.0 eV. These electron densities are much smaller compared with typical values of 1020
m-3
reported in the literature [13]. Nonetheless, such small values of plasma density might be
compensated by the high density of He metastables detected in the V-I curves. It is worth noting
that arc emission spectrum analysis above has provided average vibrational temperature of 0.7
eV. This temperature, associated to the vibrational energy of the C2 molecules, is on the lower
-40 -20 0 20 40
-0.3
-0.2
-0.1
0.0
0.1
Probecurrent(A)
Probe voltage (V)
(b)(a)
Electron
saturation
Ion saturation
+ Auger emission
+ Thermionic emission
Iarc = 60 A
p = 10 Torr
Protective
plate
Langmuir probe
in electrode gap
(50 ms)
Langmuir probe
away from
electrode gap
(950 ms)
Graphite
anode
Connection to
power supply
Page 17 of 36 AUTHOR SUBMITTED MANUSCRIPT - PSST-102916.R1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
AcceptedM
anuscript
19. 18
side and measured at 300 Torr but still within the electron temperature interval, suggesting thus
that the considered arc discharge is near local thermodynamic equilibrium (LTE) conditions.
3.3 Sample characterization
The material deposited onto the cathode by pulsed arc discharge has been characterized and
compared with the deposit formed by steady DC arc discharge. The analyzed region is the anode
tip, which is divided into two parts, namely a white zone and a dark zone (Fig. 8). As explained
below, the colour contrast between these surface zones provides information in regard to the
structure of the deposited material.
FIG. 8. Left side: Deposition of ablated carbon onto the cathode after pulsed arc discharge at 1 Hz. Right
side: The top view SEM image of the carbon cathode shows a variety of morphologies as consequence of
the deposited carbon nanostructures.
Nanoparticle morphology and bonding structure have been explored by SEM and Raman
spectroscopy, respectively. Different regions at the carbon anode tip have been distinguished in
the case of pulsed arcs (Fig. 9). SEM images taken from the cathode deposition conforming the
white region evidence nucleation of carbon nanotubes and typical landscapes of graphene nano-
platelets networks. Consistently, in these locations, Raman spectra show the presence of
Cathode
Anode
Page 18 of 36AUTHOR SUBMITTED MANUSCRIPT - PSST-102916.R1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
AcceptedM
anuscript
20. 19
prominent G' (or 2D) peak at ≈2700 cm-1
, which constitutes the signature of carbon
nanostructures. Other two characteristic peaks, D (≈1350 cm-1
) and G (≈1600 cm-1
), appear sharp
and well-separated, which is expected in sp2
-rich carbon samples [29]. The G, G' and G* (≈2450
cm-1
) bands are associated with phonon modes in defect-free, sp2
hexagonal carbon networks,
whereas the D band indicates presence of defects [30]. On the other hand, the dark region of the
tip still shows graphene nano-platelet morphology but without nanotubes spread on the surface.
The peaks G* and G' are detected but have become less important than in the white region.
Simultaneously, the D and G bands are dominant and appear overlapped, giving account of a
disordered, cross-linked network typical of amorphous carbon.
1000 1500 2000 2500 3000
0
50
100
150
G*(?)
G'
G
Pulsed-1Hz-10DC-5V-30s-volcano ring-50X_Raman.txt
Intensity(cts)
Raman shift (cm
-1
)
D
1000 1500 2000 2500 3000
0
50
100
150
G*(?)
G'
G
Pulsed-1Hz-10DC-5V-30s-volcano ring-50X_Raman.txt
Intensity(cts)
Raman shift (cm
-1
)
D
1000 1500 2000 2500 3000
0
50
100
150 Pulsed-1Hz-10DC-5V-30s-black (outer) region-50X_Raman.txt
Intensity(cts)
Raman shift (cm
-1
)
1000 1500 2000 2500 3000
0
50
100
150 Pulsed-1Hz-10DC-5V-30s-black (outer) region-50X_Raman.txt
Intensity(cts)
Raman shift (cm
-1
)
D
G
G*
G'
(a) (b)
(c) (d)
Page 19 of 36 AUTHOR SUBMITTED MANUSCRIPT - PSST-102916.R1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
AcceptedM
anuscript
21. 20
FIG. 9. SEM images showing the cathode deposit after pulsed arc (1 Hz) at the (a) central (white) region
and (b) peripheral (dark) region. The observed morphology in (a) corresponds to a landscape of carbon
nanotubes. The respective bonding structures, which were analyzed by means of Raman spectroscopy,
show that: (c) the white region surface has gathered the carbon nanostructures (nanotubes and graphene
nano-platelet network) synthesized in the arc discharge, and that (d) the dark region surface is rich in
amorphous carbon.
The morphological and structural properties of carbon samples produced in pulsed arc discharges
show features typical of carbon samples produced in DC arc discharges as well [31,32], namely:
(i) the white region surface is rich in carbon nanostructures (nanotubes and graphene nano-
platelets) nucleated in the pulsed discharge; and (ii) the dark region surface is rich in amorphous
carbon with some contribution of carbon nanostructures. From these data, one can conclude that
pulsed anodic arc discharge is a very efficient source of carbon nanomaterials. The large control
of the technological parameters of discharge will permit to tailor the production rate and the
properties of carbon nanotubes and graphene networks grown on dedicated substrates.
Deposition of carbon powder on the chamber walls is substantially reduced when arc discharge is
operated in pulsed mode. Fig. 10 shows the surface of a vertical blind flange exposed to different
arc plasma atmospheres. This flange was located approximately 10 cm away from the electrodes.
The concentration of deposited macroparticles is qualitatively assessed by observing the contrast
between the exposed surface and a region masked with Kapton tape. In the case of DC arc held at
60 A during 10 s, anode mass removal is around 20 mg (see Table 1), which resulted in the
powder formation shown in Fig. 10a. On the other hand, pulsed arc discharge at 2 Hz and 10%
duty cycle during 50 s results in around 50 mg of mass removal. Despite this larger amount of
Page 20 of 36AUTHOR SUBMITTED MANUSCRIPT - PSST-102916.R1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
AcceptedM
anuscript
22. 21
eroded material, in particular a factor 2-3 with respect to DC arc, the density of deposited powder
is much smaller compared with the DC case (Fig. 10b), demonstrating thereby that pulsed arcs
produce very little quantity of carbon powder during nanoparticle growth process. Moreover, the
relatively low plasma impedance measured in pulsed arcs may account for the reduced density of
powder (see Table 1).
FIG. 10. Photographs of the blind flange showing the surfaces masked and exposed to plasma arcs.
Sample (a) was exposed to DC arc discharge at 60 A for 10 s (≈20 mg eroded from anode), and sample
(b) was exposed to 2 Hz-pulsed discharge for 50 s (≈50 mg eroded from anode). The inner diameter of the
flange is 45 mm.
4. Modelling of arc processes
In order to approach the basic physical mechanisms governing pulsed arc discharges, the carbon
arc process is studied in the framework of a global model based on pressure balance equations.
The measurement of the total pressure variations along arc operation has permitted to identify
different arc phases. The measured temporal evolution defines a dynamics in pressure which has
(a) (b)
Page 21 of 36 AUTHOR SUBMITTED MANUSCRIPT - PSST-102916.R1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
AcceptedM
anuscript
23. 22
been fitted to rate equations and discussed in the context of the experimental results above. Please
note that total pressure considered here is the average pressure in the chamber.
4.1 Discharge phases in DC arcs
Fig. 11 shows the time evolution of total pressure during carbon DC arc discharges at 60 A of arc
current. An initial pressure of 300 Torr He was set by a fixed aperture of the pumping valve. A
total of 5 regimes of pressure evolution can be distinguished during the discharge: (i) Initial or
equilibrium pressure, (ii) strong pressure increase, (iii) pressure decay, (iv) strong pressure drop,
and (v) pressure increase or stabilization. Such behaviour in gas pressure is characteristic of all
atmospheric anodic arc discharges conducted with carbon electrodes.
FIG. 11. Temporal evolution of total pressure during anodic arc discharge of carbon held at 60 A within a
He atmosphere at 300 Torr. The different regimes of pressure permit to define the following arc phases:
(i) steady state pumping;(ii) breakdown; (iii) anodic arc; (iv) gas trapping, and (v) recovery.
0 20 40 60 80 100 120
220
240
260
280
300
320
340
360
380
400
300 Torr He
60 A arc current
Pressure(Torr)
Time (s)
ii
iii
iv
v
i DC arc stops
DC arc
starts
Page 22 of 36AUTHOR SUBMITTED MANUSCRIPT - PSST-102916.R1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
AcceptedM
anuscript
24. 23
Fig. 12 shows the pressure evolution in arc discharges at 60 A and 150 A, together with the fitted
curves from the model outlined here. This analysis provides an approach to the macroscopic
physical processes in anodic arc discharges of carbon.
i. Equilibrium phase: Initially, prior to starting the arc discharge, the system with volume V
(4500 cm3
) is in steady state, i.e. its pressure p is constant. Pumping rate is in balance with
the inlet flow rate of helium. The expression for changes in total pressure is [33]:
Sp
t
p
V −=
•
He
d
d
φ (3)
which is equal to zero in steady state. Here, S is the pumping rate and He
•
φ is the gas
throughput, which is equal to the gas flow rate times pressure. S is approximately 350 cm3
/s
in the case of He atmosphere at around 300 Torr.
ii. Breakdown phase (strong pressure increase – intermediate arc): This phase takes place
during the time interval when intermediate arc prevails. Anode evaporation is assumed to be
the main source of pressure increase. As observed in Fig. 10, this phase of steep increase
lasts for a few seconds. Eq. 3 is completed with an additional source term, C
•
φ :
CHe
d
d ••
+−= φφ Sp
t
p
V (4)
This new throughput term C
•
φ gives account of pressure increase due to ablation of the
carbon anode, which generates the carbon plasma arc. The values corresponding to 60 A
and 150 A discharges are calculated from the measured ablation rates, which are ≈2 mg/s
and ≈20 mg/s, respectively (see Table 1). To this end, we equate C
•
φ to the flux of carbon
atoms ejected from the solid anode:
Page 23 of 36 AUTHOR SUBMITTED MANUSCRIPT - PSST-102916.R1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
AcceptedM
anuscript
25. 24
t
V
p
d
d
C =
•
φ (5)
From the ideal gas equation approximation, one can develop the last expression into:
R
M
T
C
C
κ
φ =
•
(6)
where R is the ablation rate, MC is the carbon atom mass and T is the local gas temperature
near the anode tip, which is approximated to ≈5000 K.
iii. Anodic arc phase (pressure decay – powder generation): This phase of gradual pressure
decrease, which takes place in the anodic arc regime, has been associated to particle
nucleation and is triggered when a threshold in pressure is reached. Consistently, abundant
formation of carbon powder is reported in Section 3.3. Pressure variation has been fitted
with the following expression:
trap0CHe )(
d
d •••
−−−+−= φφφ ppGSp
t
p
V (7)
Two sink terms have been added to Eq. 4. The first sink term, G(p-p0), is introduced to give
account of carbon powder generation: a possible mechanism is particle coagulation giving
rise to macroparticles and dusty gas [34]. This term is proportional to the density of carbon
atoms, and G is a fitted constant that quantifies the powder generation rate. The origin of
the second sink term, trap
•
φ , is more difficult to explain. It may appear as consequence of
trapping of gas species by the generated macroparticles. The trapping of gas atoms by
growing aggregates of nanoparticles could lead to a pressure drop due to gas depletion.
Such encapsulation effect on gas species has been considered in earlier studies of carbon
nanoparticle engineering [35,36]. Although the probability of this event is very low in the
case of nanoparticles as hosting species, gas trapping cannot be ruled out when the size of
Page 24 of 36AUTHOR SUBMITTED MANUSCRIPT - PSST-102916.R1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
AcceptedM
anuscript
26. 25
the formed particles overcomes the nanoscale. The introduction of the loss term trap
•
φ in Eq.
7 is legitimate since it will be observed in the next arc phase and must be fitted consistently.
iv. Gas trapping phase (strong pressure drop): The arc discharge is switched off. As a result,
the arc-associated terms of powder generation and carbon ablation vanish, and only the gas
trapping mechanism prevails for 1-2 seconds. This effect is fitted with the same trap
•
φ term
introduced above:
trapHe
d
d ••
−−= φφ Sp
t
p
V (8)
This effect may explain the rapid and short-termed pressure decrease during the plasma
afterglow. As explained above, the trapping of gas species can take place in the gas volume
by forming dust particles, which are then deposited onto the chamber walls or pumped out
of the plasma reactor. When the arc plasma is no longer active, the gas located in the former
arc region cools down from about 0.7 eV (see section 3.2), enhancing thereby aggregation
of particles locally for a very short time interval.
v. Recovery phase (pressure stabilization): In this last step, the pressure increases gradually to
the initial equilibrium conditions. All plasma-associated terms vanish and only the normal
pumping and gas injection terms remain in the rate equation (Eq. 3). However, this
expression is not zero since pumping and flow rate terms are initially unbalanced due to the
important depletion of gas species in phase iv. The increase rate of pressure is determined
by the gas residence time.
Page 25 of 36 AUTHOR SUBMITTED MANUSCRIPT - PSST-102916.R1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
AcceptedM
anuscript
27. 26
FIG. 12. Temporal evolution of total pressure during anodic arc discharges of carbon held at 300 Torr He
at arc currents of (a) 60 A and (b) 150 A. The different arc phases are indicated. The fitting lines have
been computed using the balance equations of the global model. The throughput C
•
φ has been derived
from the measured ablation rate of the anode (Eq. 6). The pumping rate S is independently measured. The
rest of variables, G and trap
•
φ , are fitting parameters. S is allowed to vary within its order of magnitude,
i.e. around 50%, to improve the fitting, whereas C
•
φ is taken as exact value. The respective accuracies of
both G and trap
•
φ are estimated to be approximately 1%.
4.2 Discharge phases in pulsed arcs
200
300
400
500
600
0 10 20 30 40 50
Time (s)
Pressure(Torr)
250
270
290
310
330
350
370
0 20 40 60 80 100 120
Time (s)
Pressure(Torr)
(a)
(b)
300 Torr He
60 A arc current
S = 350 cm
3
/s
G = 300 cm3
/s
trap
•
φ = 7 Pa m
3
/s
C
•
φ = 7 Pa m
3
/s
300 Torr He
150 A arc current
S = 350 cm
3
/s
G = 350 cm
3
/s
trap
•
φ = 68 Pa m
3
/s
C
•
φ = 70 Pa m3
/s
ii
ii
iii
i
i
iii
iv
iv
v
v
Page 26 of 36AUTHOR SUBMITTED MANUSCRIPT - PSST-102916.R1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
AcceptedM
anuscript
28. 27
Fig. 13 shows the pressure evolution of pulsed arc discharges held at 1 Hz and 5 Hz with 10%
duty cycle. The observed behaviour deviates substantially from the one observed in steady DC
arcs. The initial pressure increase is associated to gas desorption by hot electrodes: Joule heating
of the electrodes occurs as they are short-circuited to trigger the arc discharge. As soon as the
first discharge strikes between the electrodes, the system enters into the pulsed arc regime. In the
case of 1 Hz discharges, the total pressure fluctuates between 330 and 370 Torr as consequence
of the pulsed power regime. In the case of 5 Hz trials, the lower variability in pressure values
between start and stop of arc is probably due to the faster variations in supplied power to the
discharge. Finally, the gas pressure is stabilized towards a lower value as soon as the arc stops.
The sustained increase of the final pressure and its mismatch with the initially set pressure are
due to thermal drifts within the plasma chamber.
280
300
320
340
360
380
0 20 40 60 80 100 120 140
Time (s)
Pressure(Torr)
280
300
320
340
360
380
40 50 60 70 80 90 100 110 120 130 140
Time (s)
Pressure(Torr)
Electrodes
shortcircuited
Pulsed arc
starts
Pulsed arc
stops
Electrodes
shortcircuited
Pulsed arc
starts
Pulsed arc
stops
(a)
(b)
300 Torr He
1 Hz pulses
300 Torr He
5 Hz pulses
Page 27 of 36 AUTHOR SUBMITTED MANUSCRIPT - PSST-102916.R1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
AcceptedM
anuscript
29. 28
FIG. 13. Temporal evolution of total pressure during pulsed anodic arc discharges of carbon held at 300
Torr He at (a) 1 Hz and (b) 5 Hz. The different events during plasma operation are indicated.
In order to track the total pressure with better time resolution, this parameter has been recorded
with an oscilloscope. Fig. 14a shows the pressure waveform and the corresponding arc current
waveform within one pulse period (1 Hz). The total pressure oscillates periodically following the
frequency of the pulsed power. At the onset of arc current, pressure increases and it keeps
growing as long as the pulse is active. This evolution is connected with the progressive increase
of anode temperature and the resulting evaporation of anode material. Just after the pulse activity
ceases, i.e. in 0.1 s, pressure decreases gradually to its original value, which is always higher than
the original set pressure presumably due to thermal drifts, as stated earlier, and due to degassing
of hot anode, which is glowing during the OFF cycle. Although the arc current reaches saturation
at 1 Hz with 10% duty cycle, the pressure amplitude does not. Such result has implications with
the effective ablation rate in pulsed arc discharge, so this issue is treated in the following.
FIG. 14. (a) Periodic waveform of total pressure and discharge current during pulsed anodic arc
discharge of carbon held at 300 Torr He and 1 Hz (10% duty). Pulse starts at t=-0.1 s. (b) Comparison of
-0,4 -0,2 0,0 0,2 0,4
250
300
350
400
450
500
Dischargecurrent(A)
Pressure(Torr)
Time (s)
0
50
100
150
200
250
0,1 0,3 0,5 0,7 0,9
300
350
400
450
500
Pressure(Torr)
Time (s)
1 Hz, 50% duty
1 Hz, 10% duty(a) (b)
∆∆∆∆p50%
∆∆∆∆p10%
Page 28 of 36AUTHOR SUBMITTED MANUSCRIPT - PSST-102916.R1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
AcceptedM
anuscript
30. 29
periodic waveforms of total pressure during 1 Hz-discharges with 10% and 50% duty cycles. Pulse starts
at t=0.0 s. The respective amplitudes of total pressure, ∆p10% and ∆p50%, are indicated.
Fig. 14b compares the same 1 Hz-pressure curve at 10% duty cycle with the evolution of an arc
process held at 1 Hz and 50% duty cycle. In this second process, the total pressure does reach a
plateau 0.3 s after the pulse starts, and it decreases again just after the pulse ceases, i.e. at 0.5 s.
Please note that the pressure amplitude in the 50% duty cycle case, ∆p50, is around 120 Torr,
while the pressure amplitude in the case of 10% duty cycle, ∆p10 process only reaches 60 Torr.
This difference in pressure amplitudes by a factor two might explain the difference between
ablation rate for the DC arc at 150 A, 20 mg/s, and the lower effective rate for the pulsed arc with
1 Hz frequency, 10 mg/s (see Table 1). Indeed, the DC process at high power shows double
efficiency in ablation compared to its pulsed counterpart probably because of the anode heating
dynamics. In DC arcs, the supply of constant arc current heats up the anode more efficiently.
Therefore, although we have no evidence of anode reaching a higher temperature point in DC
arcs, it is plausible to assume that the anode temperature in DC arcs have a longer stable regime
than pulsed arcs judging by discharge current distribution. Also, erosion rates suggest that
sublimation temperature at the anode is kept at least above the threshold value longer in DC arcs.
In summary, the duty cycle is the limiting factor to achieve full ablation rate conditions in pulsed
arc discharges. For this conclusion, we should assume that the ON phase of the pulse produces
stable temperature regimes similar to DC arcs.
4.3 Pulsed vs. DC anodic arc discharges
Fig. 15a shows qualitatively the kinetics of total pressure in both DC and pulsed processes. A
major difference observed between DC and pulsed discharges is the smaller range of pressures
Page 29 of 36 AUTHOR SUBMITTED MANUSCRIPT - PSST-102916.R1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
AcceptedM
anuscript
31. 30
visited by the latter: pulsed discharge is periodically interrupted within phase ii, i.e. breakdown
phase. This event is supported by the corresponding light emission profiles displayed on the same
Fig. 15a. The optical emission profile at the ON cycle in pulsed discharges shows the same
pattern as the one observed in starting phase ii in DC discharges. After the pulsed power is
interrupted, the system remains in a hot anode afterglow state, and the breakdown is resumed as
soon as the forthcoming ON cycle starts. In the explored range of frequencies, pulsed arc
discharges have never entered into phases iii - powder generation and gas trapping; iv - gas
trapping, and v - pressure recovery. Instead, the system always exhibits the characteristic features
of breakdown state, namely an increase in pressure and an optical emission associated to the
initial arc discharge striking. Consistently, the pulsed arc emission does never show the anodic
arc profile characteristic of phase iii. This result is coherent with the observed anode tip shapes
after arc discharges: DC anodic arcs provide rounded and irregular tips, while pulsed arcs provide
flat-ended tips (see Fig. 4). Fig. 15b illustrates this correlation between anode tip shape (ablation
mode) and arc operation mode. In summary, pressure dynamics together with plasma emission
reveal that pulsed discharge is confined in and does not evolve from the breakdown phase ii of
anodic arc discharge.
Pressure
Time
Breakdown
Anodic arc
Periodic
breakdown
DC discharge
Pulsed discharge
(a) (b)
Cathode
Anode
Anodic arc (DC)
Cathode
Anode
Pulsed arc
Page 30 of 36AUTHOR SUBMITTED MANUSCRIPT - PSST-102916.R1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
AcceptedM
anuscript
32. 31
FIG. 15. (a) Schematic plot of pressure kinetics for the cases of DC (dotted line) and pulsed (solid line)
arc discharges. The different regimes are accompanied with representative images of the corresponding
plasma arc showing the characteristic emission patterns. DC arcs go through all phases outlined in
Section 4.1: breakdown, anodic arc (powder formation) and gas trapping phases, whereas pulsed arcs
remain in breakdown phase during the whole process. (b) Ablation modes: schematic representation of
the anode tip shape evolution as consequence of its interaction with the plasma arc in DC and pulsed
conditions.
The permanence of pulsed arc discharge in the breakdown phase suggests some technological
advantages in the production of high-quality nanomaterials. First, it may address the issue of
oscillations in current and voltage observed in DC steady arcs due to erratic arc path around
anode tip: the intermediate position of the pulsed discharge between electrodes favors a
preferential ablation on the central part of anode top surface, in opposition to the less productive
lateral ablation mode of DC discharge reported in ref. [12]. This hypothesis of less oscillations in
pulsed discharge should be tested by means of electric measurements combined with fast optical
imaging. Second, pulsed arc discharge prevents undesired generation of macroparticles and
powder since the system does not operate in the dusty plasma conditions of phase iii. This has
been verified by the very small, almost negligible deposition of dust on the reactor walls (see Fig.
10). Simultaneously, the selective formation of nanomaterials is promoted, as proved in Section
3.3. Such nanosynthesis is selective because the system does not proceed beyond the breakdown
phase and, therefore, the coagulation in bigger clusters is hindered. In a different application
concerning controlled production of particles [37,38], pulsed arc signals could be tailored to grow
carbon nanostructures with a determined size distribution. Thus, pulsed anodic arc discharge
Page 31 of 36 AUTHOR SUBMITTED MANUSCRIPT - PSST-102916.R1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
AcceptedM
anuscript
33. 32
technique is promising for the synthesis of monodisperse ensembles of nanoparticles with
accurate control of the particle size.
5. Conclusions
Here, we have proved that anodic arc discharges with carbon electrodes can be performed in
pulsed mode without external triggering. The selected pulse frequencies lie within the Hz range
in order to achieve full development of the discharge current. Pulsed arc processes are more
stable and slower than DC steady arc processes, thus providing better control in deposition of
carbon nanostructures. The anode tip after pulsed arc operation is always flat-ended, while it
appears rounded or sharpened after DC operation. Moreover, the resulting shape of anode tip is
correlated with the plasma emission profile, which is centred at the inter-electrode gap and at the
anode tip in the case of pulsed arc and DC arc, respectively. Plasma parameters at the arc region,
which is near LTE conditions, are similar in both steady DC and pulsed arcs. OES shows a
substantially higher emission in pulsed plasmas compared to their DC counterpart. Fast Langmuir
probe diagnostics indicated a high density of metastable helium atoms combined with strong
electron emission by the hot probe. SEM and Raman spectroscopy have assessed the morphology
and quality of the carbon nanostructures deposited onto the cathode. Finally, the development of
a simple global model based on pressure kinetics has permitted to identify different arc phases
during discharge. In particular, it has been proved that DC arcs take place normally in anodic arc
regime, which causes important production of undesired carbon macroparticles (powder) due to a
dominant dusty plasma state. On the other side, pulsed arcs undergo periodically a breakdown
Page 32 of 36AUTHOR SUBMITTED MANUSCRIPT - PSST-102916.R1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
AcceptedM
anuscript
34. 33
phase which minimizes powder generation and, therefore, optimizes the synthesis of
nanomaterials from ablation of the anode material.
In summary, this study has proved that anodic arc discharge in pulsed mode is a robust method
for the synthesis of carbon nanomaterials at atmospheric pressure. Future work will be devoted to
explore the growth of such nanomaterials within the plasma environment as well as on a
dedicated substrate. This deposition method can be also translated to the production of further
materials for special electrical and optical applications. Finally, the study in real time of the
plasma properties of pulsed anodic arcs will provide insight of the detailed discharge processes
occurring in reduced time scales. Main objective is to understand the physics behind anodic arc
discharge in transient, non-stationary conditions. Spatially- and time-resolved anode erosion
should be also investigated to complete the spatially-averaged model described in this paper.
Acknowledgements
This work was supported by the U.S. Department Of Energy, Office of Science, Fusion Energy
Sciences program Award Number DESC0015767. The authors thank the assistance of Maryland
NanoCenter (University of Maryland at College Park) for the access to the SEM facility.
References
[1] Handbook of Vacuum Arc Science and Technology, Ed. by Boxman R L, Sanders D M, and
Martin P J, Noyes Publications, Park Ridge, NJ, 1995
[2] Jüttner B. Cathode spots of electric arcs. 2001 J Phys D: Appl Phys 34 R103–R123
[3] Plasma Engineering, 2nd Ed., Keidar M and Beilis I I, Elsevier, London, 2018
[4] Iijima S. Helical microtubules of graphitic carbon. 1991 Nature 354(6348) 56–58
Page 33 of 36 AUTHOR SUBMITTED MANUSCRIPT - PSST-102916.R1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
AcceptedM
anuscript
35. 34
[5] Keidar M, Shashurin A, Volotskova O, Raitses Y and Beilis I I. Mechanism of carbon
nanostructure synthesis in arc plasma. 2010 Phys Plasmas 17 057101
[6] Shashurin A, Keidar M. Synthesis of 2D materials in arc plasmas. 2015 J Phys D: Appl Phys
48(31) 314007
[7] Yao-Wen Y, Raitses Y, Koel B E and Yao N. Stable synthesis of few-layered boron nitride
nanotubes by anodic arc discharge. 2017 Sci Rep 7 3075
[8] Stratton B C, Gerakis A, Kaganovich I, Keidar M, Khrabry A, Mitrani J, Raitses Y, Shneider M
N, Vekselman V and Yatom S. In situ diagnostics for nanomaterial synthesis in carbon arc
plasma. 2018 Plasma Sources Sci Technol 27 084001
[9] Fang X, Shashurin A and Keidar M. Role of substrate temperature at graphene synthesis in an arc
discharge. 2015 J Appl Phys 118 103304
[10] Fang X, Corbella C, Zolotukhin D B and Keidar M. Plasma-enabled healing of graphene nano-
platelets layer. Front Chem Sci Eng https://doi.org/10.1007/s11705-018-1787-7 (2019, Accepted)
[11] Gershman S and Raitses Y. Unstable behavior of anodic arc discharge for synthesis of
nanomaterials. 2016 J Phys D: Appl Phys 49 345201
[12] Yatom S, Selinsky R S, Koel B E and Raitses Y. “Synthesis-on” and “synthesis-off” modes of
carbon arc operation during synthesis of carbon nanotubes. 2017 Carbon 125 336-343
[13] Kundrapu M and Keidar M. Numerical simulation of carbon arc discharge for nanoparticle
synthesis. 2012 Phys Plasmas 19 073510
[14] He D, Zhao T, Liu Y, Zhu J, Yu G and Ge L. The effect of electric current on the synthesis of
single-walled carbon nanotubes by temperature controlled arc discharge. 2007 Diamond Relat
Mater 16 1722-1726
[15] Anders A. Fundamentals of pulsed plasmas for materials processing. 2004 Surf Coat Technol 183
301–311
[16] Bradley J W and Welzel T. Physics and phenomena in pulsed magnetrons: an overview. 2009 J
Phys D: Appl Phys 42 093001
Page 34 of 36AUTHOR SUBMITTED MANUSCRIPT - PSST-102916.R1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
AcceptedM
anuscript
36. 35
[17] Andújar J L, Vives M, Corbella C and Bertran E. Growth of hydrogenated amorphous carbon
films in pulsed d.c. methane discharges. 2003 Diamond Relat Mater 12 98–104
[18] Corbella C, Rubio-Roy M, Bertran E, Portal S, Pascual E, Polo M C, Andújar J L. Ion energy
distributions in bipolar pulsed-dc discharges of methane measured at the biased cathode. 2011
Plasma Sources Sci Technol 20 015006
[19] Gudmundsson J T, Brenning N, Lundin D and Helmersson U. High power impulse magnetron
sputtering discharge. 2012 J Vac Sci Technol A 30 30801
[20] Anders A. Tutorial: Reactive high power impulse magnetron sputtering (R-HiPIMS). 2017 J Appl
Phys 121 171101
[21] Bräuer G, Szyszka B, Vergöhl M and Bandorf R. Magnetron sputtering - Milestones of 30 years.
2010 Vacuum 84 1354–1359
[22] Corbella C. Upscaling plasma deposition: The influence of technological parameters. 2014 Surf
Coat Technol 242 237–245
[23] Bobzin K, Brögelmann T, Kruppe N C, Engels M, von Keudell A, Hecimovic A, Ludwig A,
Grochla D and Banko L. Fundamental Study of an Industrial Reactive HPPMS (Cr,Al)N Process.
2017 J Appl Phys 122 015302
[24] Fang X, Shashurin A, Teel G and Keidar M. Determining synthesis region of the single wall
carbon nanotubes in arc plasma volume. 2016 Carbon 107 273–280
[25] Danylewych L L and Nicholls R W. Intensity measurements on the C2 (d3
Πg ─a3
Πu) Swan band
system. 1974 Proc E Soc Lond A 339 197-212
[26] Li J, Kundrapu M, Shashurin A and Keidar M. Emission spectra analysis of arc plasma for
synthesis of carbon nanostructures in various magnetic conditions. 2012 J Appl Phys 112 024329
[27] Phillips J G. Perturbations in the Swan System of the C2 Molecule. 1968 J Molec Spectroscopy 88
233-242
[28] Shashurin A, Li J, Zhuang T, Keidar M and Beilis I I. Application of electrostatic Langmuir probe
to atmospheric arc plasmas producing nanostructures. 2011 Phys Plasmas 18 073505
Page 35 of 36 AUTHOR SUBMITTED MANUSCRIPT - PSST-102916.R1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
AcceptedM
anuscript
37. 36
[29] Ferrari A C and Robertson J. Interpretation of Raman spectra of disordered and amorphous
carbon. 2000 Phys Rev B 61(20) 14095–14107
[30] Ferrari A C, Basko D M. Raman spectroscopy as a versatile tool for studying the properties of
graphene. 2013 Nature Nanotechnology 8(4) 235–246
[31] Keidar M and Waas A M. On the conditions of carbon nanotube growth in the arc discharge. 2004
Nanotechnology 15 1571–1575
[32] Ng J and Raitses Y. Role of the cathode deposit in the carbon arc for the synthesis of
nanomaterials. 2014 Carbon 77 80-88
[33] Industrial Plasma Engineering, Vol. 2, Roth J R, Institute of Physics Publishing, Bristol, 2001
[34] Yatom S, Khrabry A, Mitrani J, Khodak A, Kaganovich I, Vekselman V, Stratton B and Raitses
Y. Synthesis of nanoparticles in carbon arc: measurements and modeling. 2018 MRS
Communications 8(3) 842-849
[35] Prato J C H M and Wud F. There Is a Hole in My Bucky. 1995 J Am Chem Soc 117 7003-7004
[36] Vougioukalakis G C, Roubelakis M M and Orfanopoulos M. Open-cage fullerenes: towards the
construction of nanosized molecular containers. 2010 Chem Soc Rev 39 817–844
[37] Gordiets B F, Inestrosa-Izurieta M J, Navarro A and Bertran E. Nanoparticles in SiH4-Ar plasma:
Modelling and comparison with experimental data. 2011 J Appl Phys 110 103302
[38] Pilch I, Söderström D, Brenning N and Helmersson U. Size-controlled growth of nanoparticles in
a highly ionized pulsed plasma. 2013 Appl Phys Lett 102 033108
Page 36 of 36AUTHOR SUBMITTED MANUSCRIPT - PSST-102916.R1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
AcceptedM
anuscript