PISCES-RF: a liquid-cooled high-power steady-state helicon plasma device

Figure 1 shows the computer-aided design (CAD) model of the experimental device, pointing out some of the important aspects of the experiment. During actual plasma operation, the source assembly (helicon antenna area in figure 1) is covered with RF shielding to prevent RF noise escaping into the room, which can disrupt or damage sensitive diagnostics. Also, not shown is the matching box that is connected to the leads of the antenna (which are shown as copper rods sticking out at an angle of ∼45 degrees). The new RF source assembly with the DI water cooled RF transparent window (detailed description is given in the next section 2.2) was incorporated on to an existing linear magnetized plasma chamber, located in the Plasma Interaction Surface Component Experimental Station (PISCES) laboratory at the University of California at San Diego (UCSD). Since the primary goal of this new device would be to produce steady state plasmas at high densities to generate extremely high fluences for PMI studies, we shall henceforth call this machine PISCES-RF.

Zoom In Zoom Out Reset image size

The main cylindrical plasma chamber, along with the external magnets and pumping system belonged to the erstwhile controlled shear decorrelation experiment (CSDX) [17, 27–38]. Previously in CSDX, argon plasmas with high density and relatively low temperatures (n > 10¹⁹ m⁻³, T_electron ∼ 5 eV and T_ion < 1 eV [39]) were produced using an m = 1 helicon antenna, operating at 13.56 MHz at RF powers of ∼1.5–2 kW. The current plasma device consists of a 3 m long, 20 cm diameter stainless steel (SS) chamber immersed in an external magnetic field that can go up to 0.240 T. In this work, the magnetic field used is almost uniform for most of the length of the machine, except at the location of the helicon source, which has a small ripple due to the extra separation of the two magnetic field coils around the RF antenna (please see figure 2). This extra space allowed us to safely install the high-power RF antenna assembly and build a grounded RF shield around it (please see figure 3). This also ensured that the plasma following the magnetic field lines gets radially terminated at the center of the ceramic (as shown in figure 2), thus allowing us to test the worst possible condition for ceramic heating.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

A turbomolecular pump, placed at the downstream end of the chamber, backed by two other mechanical pumps allows the base pressure to be in the high 10⁻⁷ Torr. A manually controlled valve placed near the pumps allow us to change the neutral pressures in the device independently from the gas flow rate. Depending on the gas used, to achieve the helicon mode of operation (W), we typically need the neutral gas to flow at 20–100 sccm (standard cubic centimeters per minute) which gives us neutral pressures of 2–10 mTorr in the plasma chamber. In the current set up, we have added a specially designed end flange (please see figure 2 for the details) upstream of the RF source that allows upstream end gas injection. This gas injection configuration has shown to critically affect the neutral gas management in Proto-MPEX [40] and is also the choice of operation for MPEX. Hence, we incorporated this change on PISCES-RF.

To operate the device with up to 20 kW of RF power, the helicon window must be actively liquid cooled. In our design, we used two concentric cylindrical dielectric tubes with deionized (DI) water flowing between them. This double walled RF transparent window material would require dielectrics with a very low loss tangent. Hence, we chose the inner window to be alumina (Al₂O₃), having a thickness of ∼6.35 mm (0.25 inches), while the outer window is quartz with thickness of ∼3 mm (0.117 inches). The annular water channel in between the two ceramics is ∼3.35 mm (0.133 inches) wide. The inner diameter of the alumina cylinder is ∼12 cm (4.75 inches). The RF transparent ceramic window assembly has an outer diameter of ∼14.6 mm (5.75 inches). This assembly is attached to conflat flanges on the two ends so that the whole assembly is compatible to both the CSDX chamber at UCSD and the Proto-MPEX chamber at ORNL.

DI water is used as the coolant to reduce RF absorption by the coolant itself. We used a recirculating heat exchanger to cool the DI water and re-route it back to the source. Since DI water can be corrosive to copper or brass, we designed the whole recirculating chiller and the corresponding plumbing system using PVC piping and SS adapters. Note that the RF antenna itself is made from copper, and hence the cooling lines for the antenna itself are kept separate from the DI water coolant lines for the ceramic. Moreover, to reduce impurities in the plasma, a ceramic-to-metal joint is expected to be used in MPEX for the water-to-vacuum boundary. Hence, we have used a ceramic to metal sealing technology using titanium brazing, assembled for us by MPF Inc.

In figure 3, we show the main aspects of the new source assembly. The SS chambers on either side are shown in dark gray. The external magnets are shown in dark yellow. The ceramic RF transparent window is shown in light blue (with white inner ceramic), over which the water-cooled helicon antenna (shown in golden) is wrapped. DI water is introduced in between the two concentric ceramic layers through bores drilled into the SS flanges, at 4 symmetrically placed azimuthal locations. The upstream end flange has been specially designed with four angular ports for infrared (IR) camera access to get a full azimuthal view of the inner ceramic surface. The end flange is also fitted with ultra-high vacuum compatible valves for end gas injection. We also have a water-cooled end plate, which acts as a shutter (position controlled by the black knob at the right end in figure 3.) for the IR windows when the IR camera is not being used and also as an upstream dump plate for the plasma.

In an RF source, we need a matching network for efficient coupling of the RF power from the power supply to the antenna. In the helicon plasma source design investigated in this paper, a layer of DI water fully separates the helicon antenna and the plasma volume. We are interested in the fraction of RF power absorbed in this fully enclosing layer and its impact on high-density plasma production.

We used a vector network analyzer (VNA) to measure the RF losses due to the DI water coolant, by measuring the return loss of the RF matching circuit when attached to the antenna with and without the coolant in between the two ceramic layers. The return loss, which is given by

$$\begin{equation}\mathrm{R}\mathrm{e}\mathrm{t}\mathrm{u}\mathrm{r}\mathrm{n}\mathrm{L}\mathrm{o}\mathrm{s}\mathrm{s}\enspace \left(\mathrm{d}\mathrm{B}\right)=10\;.\;\mathrm{l}\mathrm{o}{\mathrm{g}}_{\mathrm{10}}\left({P}_{\mathrm{r}\mathrm{e}\mathrm{fl}\mathrm{e}\mathrm{c}\mathrm{t}\mathrm{e}\mathrm{d}}/{P}_{\mathrm{f}\mathrm{o}\mathrm{r}\mathrm{w}\mathrm{a}\mathrm{r}\mathrm{d}}\right),\end{equation} \tag{ 1 }$$

is measured as a function of the input frequency in the absence of plasma. At the resonant condition, the reflection is minimum, indicating maximum absorption of RF power by the loaded resonant circuit. The quality factor (Q_L) of a loaded resonant circuit is related to the width (Δf) of the return loss curve, which in turn is related to the total resistance of the resonant circuit (R_T), by the equation

$$\begin{equation}{Q}_{\mathrm{L}}={f}_{0}/{\Delta}f=\left({\omega }_{0}L\right)/\left(2{R}_{\mathrm{T}}\right).\end{equation} \tag{ 2 }$$

Here, R_T = R_V + R_D + R_P, where R_V is the antenna vacuum resistance, R_D is the resistance due to the dielectric window and R_P is the plasma resistance, f₀ is the resonant frequency, ω₀ is the angular frequency at resonance and L is the inductance of the RF antenna. The width of the return loss curve can be calculated from the values of the input frequencies at the 3 dB points and the corresponding resistance is given by

$$\begin{equation}{R}_{\mathrm{T}}=\pi L{\Delta}f.\end{equation} \tag{ 3 }$$

Thus, by measuring the 3 dB width of the return loss curve using the VNA in the absence of plasma (R_P = 0), with and without the coolant, we can calculate the difference in the RF loading (R_T), which gives an estimate of the RF losses due to the coolant.

To measure the RF losses in the presence of plasma, we used a Pearson transformer to directly measure the current through the antenna. From the measured forward power to the RF antenna, we can calculate the effective resistive loading. Again, we perform these experiments with and without the coolant to measure the difference in the resistive plasma loading (R_T). This method can also be used to measure the resistive vacuum loading (R_V).

To measure the temperature of the plasma facing inner surface of the ceramic (alumina), we used an FLIR T420 camera operating at 30 Hz, with a spatial resolution of 240 × 320 pixels for IR imaging. Due to mechanical constraints, the IR camera was placed at a shallow angle, as shown in figure 2 (the line to the center of the ceramic is ∼17° with respect to the horizontal plane). This ensured that the IR view encompasses the full axial range of the inner surface of the ceramic. Four such symmetrically placed angular ports allowed us to obtain a full azimuthal view of the inner ceramic surface. From the IR camera location, the cylindrical ceramic surface looked like a part of a truncated cone, which was then mathematically mapped to a z–θ plane. Each individual azimuthal position of the IR camera recorded a little more than ±90° azimuthal view of the inner surface of the cylindrical ceramic. With IR imaging data from four such locations, we mathematically reconstructed the full z–θ thermal maps of the cylindrical surface. We also found that the shallow angle of the viewport allowed the IR emission from the uncooled, hot chamber walls to reflect off the polished white ceramic surface back onto the IR camera. This gave a spurious contribution to the IR data, which had to be corrected for. We performed calibration by cooling the chamber walls using liquid nitrogen. The mathematical mapping algorithms and the methods used to take care of the spurious IR reflection, that could otherwise lead to erroneous interpretation, are being presented elsewhere [41].

We also used two thermocouples to measure the inlet and the outlet temperatures of the coolant DI water, as a function of time. From the difference of these measurements, we could use calorimetry to calculate the net amount of heat that was being removed by the coolant. This also gave us the timescales of how fast the system reached steady state. This information turned out to be an important parameter in understanding and determining the steady state temperature from the IR camera analysis, that allowed us to get rid of the spurious contribution from the back reflection of the hot chamber during IR imaging.

Finally, we used two standard RF-compensated, voltage-swept Langmuir probes to measure the typical plasma parameters such as the plasma density, electron temperature and plasma potentials at two axial locations, 0.8 m and 1.5 m downstream of the RF antenna.

The foremost concern regarding this liquid cooled RF antenna design was the fact that there is a fully azimuthal blanket of liquid under the RF antenna. This section explores the effect this has on RF absorption and high-density plasma production.

As explained in section 3, we used the VNA to measure the return loss of the matching circuit, loaded by the RF antenna, both with and without the DI water coolant. In figure 4, one example of such a return loss measurement is shown. Red circles are data with the coolant flowing, while the black squares are without any coolant in between the two ceramics, and this same scheme is used for figures 4–8. From the return loss curves shown in figure 4, we calculated the widths using the upper and lower frequencies at the 3 dB cut off points and used the resonance frequency and equations (2) and (3) to calculate the quality factor (Q_L) and the vacuum resistance (R_V). In this case, when we do not have the DI water coolant in the RF source, Q_L = 125 and R_V = 0.164 Ω. When we introduce DI water coolant, we get Q_L = 112 and R_V = 0.184 Ω. Thus, the effect of introducing the DI water coolant was a reduction of the quality factor by 11% and increase in the vacuum loading resistance by ∼0.02 Ω, which is not very concerning. For many scans over multiple days, this factor ranged from 7% to 11%. We also notice that the resonance frequency shifts by ∼0.08 MHz. This is expected, as introduction of the DI water coolant slightly changes the impedance of the RF antenna assembly. But this small change in the resonance frequency could be easily taken care of by the tuning capacitors in the matching circuit.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

In figures 5 and 6, we show the results of RF loading measurements using the Pearson transformer in the cases with no plasma (RF power was turned on, but the neutral gas was not injected into the chamber) and with plasma operation, respectively. For each chosen RF power input, we measured the corresponding current through the leads of the RF antenna and calculated the resistive loading. We show the normalized resistive loading with and without the DI water coolant. We found that the effect of introducing the DI water coolant was to effectively increase the resistive loading, which is consistent with what we found using the VNA. For vacuum conditions, the increase in the loading varied by 5% to 9% of the original antenna loading (without the DI water coolant).

As in most typical RF driven helicon sources, during operation, we observe several mode-jumps from the capacitively coupled plasma (E mode: CCP) to inductively coupled plasma (H mode: ICP) and finally the helicon mode (W). We can observe the effect of these mode jumps on the resistive loading of the matching circuit, even though these differential measurements can introduce a lot of uncertainty (as we are taking a small difference between two large numbers). These RF power values of the mode jumps, as seen in figure 6, are consistent with the typical visual characteristics of mode jumps in similar helicon devices [16, 17]. More importantly, we can clearly see the effect of the DI water coolant on the resistive loading. In our regime of interest (high power helicon mode of operation), the effect of DI water coolant on the loading is to increase it by 9 ± 1%. Since this experiment required operating the source without the DI water coolant, we did not operate the machine with higher than 3 kW of RF power, for safety issues.

As a final test of whether the DI water coolant affects high density plasma production, we show the radial profiles of the plasma density and the electron temperature with and without the DI water coolant, as measured by the RF compensated, voltage swept Langmuir probe in figures 7 and 8 respectively, obtained ∼0.8 m downstream of the RF antenna, in argon plasma. For 2 kW of RF power, with ∼4 mTorr (∼0.53 Pa) of neutral argon gas flowing at 50 sccm and with a uniform magnetic field of 0.1 T, we achieve a peak electron density of ∼1.95 × 10¹⁹ m⁻³. The shots are very repeatable. The effect of the DI water coolant in the RF source on plasma production is negligible. The very slight differences can probably be attributed to the different matching conditions in the two cases, as also found in the VNA results. Introduction of the DI water coolant changes the effective RF antenna impedance and hence can slightly shift the resonance matching conditions, which can easily be taken care of by manual or automatic matching networks with variable capacitors. However, it is clear that the presence of the DI water coolant does not hinder high density plasma production.

In a companion paper [42], we show systematic scans of argon plasma density, electron temperature and ion temperature as measured by laser induced fluorescence (LIF), as well as numerically computed profiles of the argon neutral density and temperatures. We show that the RF source design is stable enough to allow us to use a novel compact portable LIF design [43], which needed several minutes of integration of the emitted light for good signal-to-noise ratios to measure the ion velocity distribution functions of argon. Having confirmed this crucial step, henceforth we always performed experiments in the PISCES-RF with the DI water coolant flowing through the ceramic cylinders. This allowed us to safely increase the RF power to the helicon antenna to produce steady state plasmas at higher powers and perform thermal studies of the source.

We used IR imaging to investigate the thermal issues of the plasma facing inner ceramic surface, for steady state operations of the helicon source at RF powers of up to 20 kW. Strategically placed IR cameras allowed measurements of the spatio-temporal variation of the temperatures of the inner plasma facing ceramic surface. The IR camera was used to obtain two important pieces of information. We could calculate the effective heat deposited by the plasma on the inner ceramic. When this heat from the plasma was effectively removed by the DI water coolant, the system came to a thermal equilibrium within some characteristic time scale. This gave the final steady state temperature distribution (in the axial and azimuthal directions) of the inner ceramic surface.

The temperature, at each pixel, is measured as a function of time: T(t). To calculate the effective heat deposited on the inner ceramic surface, we used the theory of heat conduction through a semi-infinite solid [44, 45]. The plasma turned on in less than a second and acted as a source of a constant heat flux on any point on the inner ceramic surface. Near the beginning of this constant heat flux, when the heat front has not yet had the time to reach the outer cooled surface of the ceramic, the approximation of a semi-infinite solid is valid. Since bodies of finite thickness can be treated as infinitely wide at the beginning of a constant heat pulse, the analytical solutions for this case could be applied (see section 2.3.3.1 of [44]). In that approximation, the temperature, at each point on the inner ceramic surface, is given by

$$\begin{equation}T\left(t\right)={T}_{0}+\left(\frac{2}{\sqrt{\pi }}\right)\cdot \left(\frac{{Q}_{0}}{b}\right).\sqrt{t},\end{equation} \tag{ 4 }$$

where $T\left(t\right)$ is the instantaneous temperature, T₀ is the initial temperature, ${\dot {Q}}_{0}$ is the constant heat flux and b is a constant which is determined by the thermal properties of the ceramic itself. Denoting ρ as the density, λ as the thermal conductivity and C_p as the specific heat of the material at constant pressure, we have b² = ρλC_p. Rearranging and squaring equation (4), we get

$$\begin{equation}{\left[T\left(t\right)-{T}_{0}\right]}^{2}=K\cdot t\end{equation} \tag{ 5 }$$

which is effectively a straight line with slope K. Thus, from the experimentally measured slope K and using the known material constants to calculate b, we could infer the steady state, constant heat flux on each point (experimentally, for each pixel on the IR camera screen that corresponds to a point on the ceramic) of the inner plasma facing ceramic surface from the expression:

$$\begin{equation}{Q}_{0}=\left(\frac{\sqrt{\pi }}{2}\right).\;b\cdot \sqrt{K}.\end{equation} \tag{ 6 }$$

Finally, taking such inferred values of the steady state, constant heat flux from each pixel of each of the IR camera positions, we could construct the z–θ maps of the heat fluxes. In figure 9, we show the steady state, constant heat flux for 10 kW of RF power in argon plasma. We also show the footprint of the m = 1 RF antenna when unraveled and plotted on the z–θ plane. It is clearly seen that the largest heat deposition is under the straps of the helicon antenna, connected to the live, non-grounded, side of the RF power supply. This steady state, constant heat flux distribution occurs even though the plasma directly contacted the middle of the ceramic, near z = 0 (as the plasma would follow the slight expansion in the magnetic field flux lines due to the extra separation of the magnets near the RF antenna, as shown in figure 2). When integrated over the total surface of the cylinder, the total steady state, constant heat deposited is ∼1.68 kW (16.8% of the input RF power).

Zoom In Zoom Out Reset image size

In figure 10, we show the spatial pattern of the steady state temperatures on the plasma facing inner surface of the ceramic, for 10 kW argon plasma and a DI water coolant flow rate of ∼22 GPM (0.0014 m⁻³ s⁻¹). Before the plasma was turned on, the ceramic was kept at a uniform surface temperature of 293 K. The DI water coolant can remove most of the heat as the ceramic surface does not get very hot. The hottest portions of the ceramic are at the very ends. This is due to the constraints of the ceramic-metal brazing technique where ∼1.5 cm of the edges of the ceramic are not directly in contact with the DI water coolant. We find that the regions under the helicon antenna strap remain colder even though they have the maximum heat deposited under them, thus signifying the effectiveness of the DI water coolant in removing the heat. Moreover, as mentioned earlier, the separation between the two magnets (as shown in figure 2) near the helicon antenna allows the magnetic field lines under the ceramic to bulge out. Thus, the plasma hits the ceramic at z = 0 which we envision to be the worst-case scenario for the MPEX source. Even then, we see that the maximum heating is under the straps of the helicon antenna and not at the spatial location where the plasma is supposed to hit the inner surface of the ceramic.

Zoom In Zoom Out Reset image size

Figure 11 shows the final z–θ map of the steady state, constant heat flux for 10 kW of H plasma. The magnetic field profile is the same as used in argon, but the magnitude is much lower (0.02 T) in order to ensure proper RF matching. Again, it is clearly seen that the largest heat deposition is under the straps of the helicon antenna, connected to the live (non-grounded) side of the RF power supply. When integrated over the total surface area of the cylinder, the total heat deposited is ∼2.76 kW, (27.6% of the input RF power). Thus, we find that the net heat deposited from the plasma is higher in case of H plasma when compared to argon plasma.

Zoom In Zoom Out Reset image size

In figure 12, we see the spatial pattern of the steady state temperatures on the inner surface of the ceramic for 10 kW of steady state H plasma and a DI water coolant flow rate of ∼22 GPM (0.0014 m⁻³ s⁻¹). The ceramic was at a uniform surface temperature of 293 K. We find that even though there is considerable heat deposition right under the helicon antenna straps (see figure 11), the DI water coolant is effective in cooling that area, but the hottest portions of the ceramic are near the very ends, which are not in direct contact with the DI water coolant, due to mechanical constraints of the titanium brazing. Even then, for 10 kW of RF power at steady state conditions, the maximum increase in the ceramic temperature is ∼30 K, which would not produce catastrophic thermal stresses in the ceramic.

Zoom In Zoom Out Reset image size

To measure the net heat being removed by the DI water coolant, we used two thermocouples to measure the inlet and the outlet temperatures of the DI water. Knowing the flow rate of the DI water coolant, and measuring the temperature difference, we have the net heat (Q), given by

$$\begin{equation}Q=m\cdot C\cdot {\Delta}T,\end{equation} \tag{ 7 }$$

where $\dot {m}$ is the mass flow rate of DI water, C is the specific heat capacity of DI water and ΔT is the measured temperature difference. Since we had to take the difference of two large numbers to get a very small value, these measurements were prone to many errors. The thermocouples are extremely sensitive to RF noise, hence we had to ensure proper RF shielding of the cables with proper grounding to reduce RF pickup.

For relatively low RF powers, if we used the maximum flow that was accessible to us using the recirculating chiller [∼22 GPM (0.0014 m⁻³ s⁻¹)], the increase in the DI water temperature was of the same order of the noise. Hence, we added another valve to reduce the flow to the lowest possible value [∼2 GPM (0.000 126 m⁻³ s⁻¹)], so that the measured temperature differential was above the noise. Figure 13 gives an example of such measurements, for relatively low RF powers, in high density argon helicon plasma. We see that as we increase the RF power, the steady state saturated value of the temperature differential also increases, which is consistent with expectations. Moreover, we also get an estimate of the time that the system takes to come to thermal equilibrium. For 2 GPM (0.000 126 m⁻³ s⁻¹) of DI water flow, it takes ∼2 min to reach the steady state. Using the data in figure 13 and equation (7), we find that the net heat removed by the DI water coolant for 3 kW, 4 kW and 5 kW of input RF power, is 756 W (25.2%), 972 W (24.3%) and 1134 W (22.7%), respectively.

Zoom In Zoom Out Reset image size

In figure 14, we show another example of thermocouple data, but for higher powers in argon plasma. For 10 kW of input RF power the DI water coolant had to remove a much larger amount of heat, and hence we could obtain a higher signal to noise ratio even at higher flow rates of the DI water coolant. Here we also show the consistency in the net heat removed by changing the flow rates of the DI water coolant for the same RF power plasma. For 10 kW of argon plasma operation, calorimetry at 10 GPM (0.000 63 m⁻³ s⁻¹) yields a net heat removal of 2.48 kW (24.8%) and for a flow rate of 22 GPM (0.0014 m⁻³ s⁻¹), yields 2.64 kW (26.4%), which are pretty consistent. We also notice that as we increase the flow rate, the time it takes to reach steady state levels decrease, as expected. For ∼10 GPM (0.000 63 m⁻³ s⁻¹), it takes ∼1 min and for ∼22 GPM (0.0014 m⁻³ s⁻¹), which was the maximum flow rate for the recirculating chiller we are using, it takes ∼30 s to reach saturation. Moreover, even for steady state conditions, the increase in the temperature of the DI water coolant is nominal (<1 K) which confirms the effectiveness of the cooling method.

Zoom In Zoom Out Reset image size

Having confirmed high density argon plasma production (please see figure 7) while ensuring non-catastrophic thermal loading of the ceramic plasma facing component by efficient cooling, we increased the RF power to 20 kW to obtain steady state, high density H plasmas for achieving fluences relevant to PMI. For general safety purposes (since the PISCES-RF chamber was not water-cooled), we had the device on for ∼2 min when operating at 20 kW. We ensured from calorimetry that the system reached steady state by then and also allowed us to finish recording all the data.

In figures 15 and 16, we show the radial profiles of the plasma density for steady state H plasmas as a function of RF power, measured 0.8 m and 1.5 m downstream of the RF helicon source, respectively. The source was operated with ∼6 mTorr (∼0.8 Pa) of neutral H gas flowing at 100 sccm and with uniform magnetic field of 0.02 T. Helicon sources have windows of operation in the operating parameter space (flow, pressure, magnetic field and RF power), and this above combination allowed us to achieve the helicon mode (W). We see from the plots that the helicon mode is achieved for RF power >8 kW in our device. The plasma density at r = 0 is ∼10¹⁷ m⁻³ for up to 7 kW and jumps to ∼0.3 × 10¹⁹ m⁻³ at 8 kW and increases thereafter. A sudden discrete jump of more than an order of magnitude in the plasma density, as a function of RF power along with centrally peaked density profiles, is the standard signature of the helicon mode [12–17]. For 20 kW in H, we achieve peak central plasma densities of ∼1.35 × 10¹⁹ m⁻³ at the port 0.8 m downstream of the helicon source. From figure 16, we see that even 1.5 m downstream of the plasma source, the peak central densities are ∼1 × 10¹⁹ m⁻³, which makes this device PMI relevant. Moreover, even at 20 kW of steady state operation, the increase in the DI water temperature is only slightly more than 1 K, with the DI water coolant flow rate being 22 GPM (0.0014 m⁻³ s⁻¹).

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

In figures 17 and 18, we show the effect of increasing RF power on the central H plasma densities and electron temperatures measured at the two ports, 0.8 m and 1.5 m downstream of the source. After we achieve the helicon core mode for H plasma, in the range of the RF powers used in this study (up to 20 kW), the plasma density is linear in power and has not yet reached saturation. On the other hand, the electron temperature seems to have reached saturation (∼8 eV at 0.8 m and ∼3 eV about 1.5 m away from the RF source) once we have reached the high-density helicon core mode. From the measured ion saturation current, we calculate that the ion fluxes to the probes are ∼3 × 10²³ m⁻² s⁻¹ at 0.8 m and ∼2 × 10²³ m⁻² s⁻¹ at 1.5 m away from the RF source, respectively. Note that the data shown (in figures 15–18) are taken for one particular set of plasma operating conditions: with ∼6 mTorr (∼0.8 Pa) of neutral H gas flowing at 100 sccm and with uniform magnetic field of 0.02 T. Plasma source parameter optimization to achieve the highest plasma densities or highest electron temperatures have not yet been done.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size